Split (1)

train · ~9.11M rows (showing the first 93.5k)

identifier stringlengths 11 32	pdf_url stringlengths 17 4.62k ⌀	lang stringclasses 120 values	error stringclasses 1 value	title stringlengths 2 500 ⌀	source_name stringlengths 1 435 ⌀	publication_year float64 1.9k 2.02k	license stringclasses 3 values	word_count int64 0 1.64M	text stringlengths 1 9.75M
https://openalex.org/W4249511088	https://www.qeios.com/read/G4XRR8/pdf	English	null	ROPN1 Positive	Definitions	2,020	cc-by	65	Qeios · Definition, February 2, 2020 Open Peer Review on Qeios Open Peer Review on Qeios ROPN1 Positive National Cancer Institute National Cancer Institute Qeios ID: G4XRR8 · https://doi.org/10.32388/G4XRR8 Source National Cancer Institute. ROPN1 Positive. NCI Thesaurus. Code C142843. National Cancer Institute. ROPN1 Positive. NCI Thesaurus. Code C142843. An indication that ROPN1 expression has been detected in a sample. Qeios ID: G4XRR8 · https://doi.org/10.32388/G4XRR8 1/1
https://openalex.org/W4392921190	https://www.frontiersin.org/articles/10.3389/friot.2024.1296599/pdf?isPublishedV2=False	English	null	Mind the FemTech gap: regulation failings and exploitative systems	Frontiers in the internet of things	2,024	cc-by	10,902	COPYRIGHT © 2024 Mehrnezhad, Van Der Merwe and Catt. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. KEYWORDS female-oriented technologies, FemTech, security, privacy, digital health, data protection regulations 1 femtech.health TYPE Original Research PUBLISHED 18 March 2024 DOI 10.3389/friot.2024.1296599 TYPE Original Research PUBLISHED 18 March 2024 DOI 10.3389/friot.2024.1296599 OPEN ACCESS EDITED BY Peter Novitzky, University College London, United Kingdom Peter Novitzky, University College London, United Kingdom REVIEWED BY Sine Nørholm Just, Roskilde University, Denmark Dharminder Chaudhary, Amrita School of Engineering, India Chandana Unnithan, Lifeguard Digital Health, Inc., Canada CORRESPONDENCE Maryam Mehrnezhad, maryam.mehrnezhad@rhul.ac.uk RECEIVED 18 September 2023 ACCEPTED 08 February 2024 PUBLISHED 18 March 2024 REVIEWED BY Sine Nørholm Just, Roskilde University, Denmark Dharminder Chaudhary, Amrita School of Engineering, India Chandana Unnithan, Lifeguard Digital Health, Inc., Canada Maryam Mehrnezhad1, Thyla Van Der Merwe2 and Michael Catt3 1Royal Holloway, University of London, Egham, United Kingdom, 2ETH Zurich, Zurich, Switzerland, 3Newcastle University, Newcastle upon Tyne, United Kingdom The security, privacy, and safety issues around Female-oriented technologies (FemTech) and data can lead to differential harms. These complex risks and harms are enabled by many factors including inadequate regulations, the non- compliant practices of the industry, and the lack of research and guidelines for cyber-secure, privacy-preserving, and safe products. In this paper, we review the existing regulations related to FemTech in the United Kingdom, EU, and Switzerland and identify the gaps. We run experiments on a range of FemTech devices and apps and identify several exploitative practices. We advocate for the policymakers to explicitly acknowledge and accommodate the risks of these technologies in the relevant regulations. Mehrnezhad M, Van Der Merwe T and Catt M (2024), Mind the FemTech gap: regulation failings and exploitative systems. Front. Internet. Things 3:1296599. doi: 10.3389/friot.2024.1296599 COPYRIGHT © 2024 Mehrnezhad, Van Der Merwe and Catt. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. 1 Introduction Speciﬁcally, we aim to focus on laws and regulations as they pertain to Europe, the United Kingdom and Switzerland, so as to complement the work that is ongoing regarding laws and regulations in the US (Scatterday, 2022; Rosas 2019). Speciﬁcally, we aim to answer the following research questions: Rosas (2019), explores the gaps without demonstrating how such gaps can be exploited (e.g., McMillan, 2023), focuses on user studies (e.g., Mcdonald and Andalibi, (2023), or is limited to a subset of FemTech solutions such as fertility tracker apps (Mehrnezhad and Almeida 2021). Although a wide range of regulations may concern the data types collected by FemTech, the sector is yet to be properly regulated. Such regulations include the California Consumer Privacy Act (CCPA)2, Health Insurance Portability and Accountability Act (HIPPA)3, Federal Food, Drug, and Cosmetic Act (FD&C Act)4, Federal Trade Commission Act5, the General Data Protection Regulations (GDPR)6, the Swiss Federal Act on Data Protection7, United Kingdom Medicines & Healthcare products Regulatory Agency (MHRA)8, and the EU Medical Devices regulation9. Note that there is a range of standards related to FemTech, e.g., the ISO 13485 Medical devices10 and ISO 3533:2021 Sex toys (Design and safety requirements for products in direct contact with genitalia, the anus, or both)11. Here we only focus on the related regulations with standardisation beyond the scope of this paper. • RQ1: What gaps exist in the applicable laws and regulations when it comes to female-related data? • RQ2: How do FemTech systems (apps, websites, IoT devices) misuse these gaps in the regulations, either intentionally or unintentionally? • RQ3: How do these systems violate the applicable laws and regulations? We review the existing regulations related to FemTech in the United Kingdom, EU, and Switzerland (as shown in Table 1). We run experiments on a range of FemTech devices, apps, and websites (as shown in Figure 1 and Table 2) and identify several exploitative practices. Our results show that there is indeed a gap in the existing laws and the current FemTech devices, apps, and systems are collecting a wide range of sensitive data about the users and others such as partner(s), baby/child, family and friends. We advocate for policymakers to explicitly acknowledge and accommodate the risks of these technologies in the relevant regulations. We conduct our studies in the United Kingdom and Switzerland. These two countries are particularly interesting since they are not EU members. 1 Introduction However, they have signiﬁcant business operating in the EEA which makes them relevant to the EU regulations including the general data protection laws and 2 oag.ca.gov/privacy/ccpa 3 cdc.gov/phlp/publications/topic/hipaa.html 4 fda.gov/regulatory-information/laws-enforced-fda/federal-food-drug- and-cosmetic-act-fdc-act 5 ftc.gov/legal-library/browse/statutes/federal-trade-commission-act 6 ico.org.uk/for-organisations/guide-to-data-protection/guide-to-the- general-data-protection-regulation-gdpr/ 7 fedlex.admin.ch 8 gov.uk/government/organisations/medicines-and-healthcare-products- regulatory-agency 9 ema.europa.eu/en/human-regulatory/overview/medical-devices 10 iso.org/iso-13485-medical-devices.html 11 iso.org/standard/79631.html 2 oag.ca.gov/privacy/ccpa 3 cdc.gov/phlp/publications/topic/hipaa.html 4 fda.gov/regulatory-information/laws-enforced-fda/federal-food-drug- and-cosmetic-act-fdc-act 5 ftc.gov/legal-library/browse/statutes/federal-trade-commission-act 6 ico.org.uk/for-organisations/guide-to-data-protection/guide-to-the- general-data-protection-regulation-gdpr/ 7 fedlex.admin.ch 8 gov.uk/government/organisations/medicines-and-healthcare-products- regulatory-agency 9 ema.europa.eu/en/human-regulatory/overview/medical-devices 10 iso.org/iso-13485-medical-devices.html 11 iso.org/standard/79631.html 1 Introduction Generally known and referred to as female-oriented technologies (aka female technologies or “FemTech”), FemTech is a term applied to the collection of digital technologies focused on women’s health and wellbeing, as the majority of the industry talks about its users. We, however, acknowledge that these products are available for people across all gender identities. FemTech products come in all forms of types and applications, ranging from mobile period apps to fertility-tracking wearables to IVF services on the blockchain. FemTech Analytics, a strategic analytics agency focused on the FemTech sector suggests several sub-sectors1. These sub-sectors have different market shares and include Pregnancy and nursing (21%), Reproductive health & contraception (17%), Menstrual health (14%), General healthcare (14%), Pelvic and uterine healthcare (10%), Sexual health (9%), Women’s wellness (7%), Menopause care (6%), Longevity (2%), and Mental health (2%). Predicted to be a $75-billion industry by 2025, this sector is booming. Consequently, they also introduce new risks and harms associated with the collection of sensitive health, medical, and sex data that are not identiﬁed and addressed in the related regulations. There is some research addressing the security and privacy (SP) risks that can originate from the mismanagement, misuse, and misappropriation of intimate data on issues such as abortion and (in)fertility (e.g., Mehrnezhad and Almeida, 2021). However, limited work has gone into exploring the laws, regulations, policies and standards surrounding FemTech’s SP risks. The existing work is either mainly around US regulations, e.g., Scatterday (2022); Frontiers in The Internet of Things Frontiers in The Internet of Things 01 frontiersin.org Mehrnezhad et al. 10.3389/friot.2024.1296599 TABLE 1 List of regulations in EU, United Kingdom, and Switzerland related to FemTech systems and data. Category Law Enforcement year Country General General Data Protection 2018 EU, United Kingdom Regulation (GDPR) General Swiss Federal Act on 1993 Switzerland Data Protection (FADP) Health & Medical MHRA Medical 2002 United Kingdom Devices Regulations Health & Medical Regulation (EU) 2017/745 2021 EU for Medical Devices medical and health ones. Speciﬁcally, we aim to focus on laws and regulations as they pertain to Europe, the United Kingdom and Switzerland, so as to complement the work that is ongoing regarding laws and regulations in the US (Scatterday, 2022; Rosas 2019). Speciﬁcally, we aim to answer the following research questions: medical and health ones. frontiersin.org 2 Background and related work Women have been discriminated against in medical and health research in many ways. For instance, in 1977, the US Food and Drug Administration (FDA) excluded women of childbearing age from taking part in drug trials leading to women being underrepresented in drug trials ever since (Nayeri, 2021). The same trend has been followed by technology companies where their solutions are mainly tailored to the male body. In response, FemTech solutions have stepped in and the COVID-19 pandemic has contributed to the massive digitisation of healthcare, including FemTech, too. FemTech products include mobile apps, connected devices and online services covering menstruation, menopause, fertility, pregnancy, nursing, sexual wellness, and reproductive healthcare, to name a few categories. The SP of FemTech can be investigated by 10 iso.org/iso-13485-medical-devices.html 11 iso.org/standard/79631.html Frontiers in The Internet of Things 02 frontiersin.org frontiersin.org Mehrnezhad et al. 10.3389/friot.2024.1296599 FIGURE 1 Examples of Femtech products (IoT, Apps) and their categories. Images have been borrowed from the products’ websites and modiﬁed for presentation in this paper. These categories are based on FemTech Analytics, a strategic analytics agency focused on the FemTech sector (femtech.health). FIGURE 1 Examples of Femtech products (IoT, Apps) and their categories. Images have been borrowed from the products’ websites and modiﬁed for presentation in this paper. These categories are based on FemTech Analytics, a strategic analytics agency focused on the FemTech sector (femtech.health). TABLE 2 Examples of FemTech digital solutions, categories, company’s country, and price. These categories are based on FemTech Analytics, a strategic analytics agency focused on the FemTech sector (femtech.health). TABLE 2 Examples of FemTech digital solutions, categories, company’s country, and price. These categories are based on FemTech Analytics, a strategic analytics agency focused on the FemTech sector (femtech.health). Frontiers in The Internet of Things frontiersin.org 3.1 Critical review of regulations performed on the security and privacy practices of FemTech. Examples of such system studies include analysis of the data collection practices of the period tracking app ecosystem and their policies (Shipp and Blasco, 2020), measuring the tracking practices of FemTech IoT devices Mehrnezhad et al., 2022b; Almeida et al., 2022), fertility apps and their compliance with the GDPR (Mehrnezhad and Almeida 2021), as well as trafﬁc analysis and policy review (with a focus on HIPPA) of a subset of iOS apps (Erickson et al., 2022). Limited work has gone into the SP assessment of FemTech IoT devices (Valente et al., 2019). The SP community has yet to properly investigate the data collection of FemTech ecosystems, (lack of) implemented security and privacy- enhancing technologies (PETs), the existing vulnerabilities, and potential SP measures to mitigate them. performed on the security and privacy practices of FemTech. Examples of such system studies include analysis of the data collection practices of the period tracking app ecosystem and their policies (Shipp and Blasco, 2020), measuring the tracking practices of FemTech IoT devices Mehrnezhad et al., 2022b; Almeida et al., 2022), fertility apps and their compliance with the GDPR (Mehrnezhad and Almeida 2021), as well as trafﬁc analysis and policy review (with a focus on HIPPA) of a subset of iOS apps (Erickson et al., 2022). Limited work has gone into the SP assessment of FemTech IoT devices (Valente et al., 2019). The SP community has yet to properly investigate the data collection of FemTech ecosystems, (lack of) implemented security and privacy- enhancing technologies (PETs), the existing vulnerabilities, and potential SP measures to mitigate them. Various aspects of FemTech data and systems make it challenging to point to one single law for the protection of FemTech data. The data collected by such technologies can be related to regulations around general data protection, work discrimination, software, apps, IoT, medical and health, and human rights. We focus on the general data protection laws and those concerning medical and health data. More speciﬁcally, we review the General Data Protection Regulations (GDPR), the Swiss Federal Act on Data Protection (FADP), the United Kingdom Medicines & Healthcare Products Regulatory Agency (MHRA), and the Regulation (EU) 2017/745 for Medical Devices. For each law, we go through its public documents and manually search for mentions of Fem-Tech data via a few keywords. 2 Background and related work Category Example Country Price (1) Pregnancy & nursing Elvie smart pump United Kingdom/United States £270 (2) Repro health & contraception Daysy cycle computer Switzerland/Germany 420 CHF (3) Menstrual health & fertility Lady Comp fertility tracker Switzerland/Germany 600 CHF (4) General healthcare HidrateSpark 3 water bottle United States 60 CHF (5) Pelvic & uterine healthcare Periﬁt kegel exerciser France 140 CHF (6) Sexual health Frida by Vibio sex toy United Kingdom £85 (7) Women’s wellness Livia menstrual pain reliever Israel $200 (8) Menopause care Balance menopause support app United Kingdom In-app purchases (9) Longevity Daviky Pill Organiser China £23 (10) Mental health Ivy (Bellabeat) health tracker for women United States $249 et al., 2022; Almeida et al., 2023), and when systems should also focus on the intersectional qualities of individuals and communities–differential vulnerabilities (Mehrnezhad and Almeida 2021). Multiple FemTech threat actors have been identiﬁed in Mehrnezhad et al. (2022b). These interested parties include, but are not limited to: (ex-)partner and family, employers and colleagues, insurance ﬁrms, advertising companies, political and religious organisations, governments, and medical and research companies. looking into IoT hardware, product websites, mobile apps, cloud datasets, etc. Figure 2 shows the FemTech ecosystem. However, the security and privacy of the user and data in FemTech are more complex than in some other contexts due to the nature of the data and the range of users (Coopamootoo et al., 2022). In Mehrnezhad et al., 2022b), it is suggested that FemTech privacy should be looked at via different lenses. These include the cases where somebody (e.g., a company) has user personal data but the user does not–inverse privacy (Erickson et al., 2022), when peer pressure causes people to disclose information to avoid the negative inferences of staying silent–unravelling privacy (Peppet, 2011), when the privacy of others (e.g., child, partner, family, friend) also matters–collective privacy (Almeida Such threat actors may exploit FemTech systems in various ways by performing attacks at different points of the ecosystem, e.g., human dimensions, hardware vulnerabilities, dataset attacks, app and website exploits, etc. A number of system studies have been Frontiers in The Internet of Things 03 frontiersin.org Mehrnezhad et al. 10.3389/friot.2024.1296599 FIGURE 2 FemTech IoT Ecosystem; data can be entered into these systems via user putting it in the app, website, or device, or automatically collected by the smart device, app, and website. 3.1 Critical review of regulations For building these keywords set we use the categories in Figure 1 and expand on it. Our keywords include, but are not limited to: Fem-Tech, women, period, fertility, pregnancy, abortion, fetus, baby, health, sex, menopause, mental health, reproductive, contraception, nursing, longevity, wellness, pelvic, uterine, breast, milk, female, cycle, birth, hormone, ovulation, lactation, menopause, etc. We identify the (lack of) related sections of each law regarding FemTech. IoT systems interact with more intimate aspects of our lives, bodies, and environments than other technologies; meaning their risks may lead to critical safety issues. IoT systems–which are yet to be regulated, create new opportunities for data collection than just apps and have the potential to compromise user security and privacy more signiﬁcantly. We argue that the intersection of health and medical solutions, user general data, and the data produced and collected by IoT devices and apps are putting and will continue to put FemTech users at greater risks, as evident by the ongoing research after the overturning of Roe vs. Wade (Mcdonald and Andalibi, 2023). 2 Background and related work Data is transferred to other places e.g., the company’s database in the cloud or sold to third parties. FIGURE 2 FemTech IoT Ecosystem; data can be entered into these systems via user putting it in the app, website, or device, or automatically collected by the smart device, app, and website. Data is transferred to other places e.g., the company’s database in the cloud or sold to third parties. 3 Methodology In this section, we explain our approach to investigating the data collection and privacy practices of a set of FemTech systems. In Figure 1, we have identiﬁed off-the-shelf products for the The methods we use fall into two groups: reviewing the regulations and conducting system studies. Frontiers in The Internet of Things 04 frontiersin.org Mehrnezhad et al. 10.3389/friot.2024.1296599 10.3389/friot.2024.1296599 10.3389/friot.2024.1296599 different FemTech categories. The products on the market can belong to multiple categories. For instance, a pelvic ﬂoor trainer can also be an intimate massager. Some of these products (e.g., a pill organiser) would also be categorised as general health solutions. Our system study experiments are performed in the United Kingdom between September 2022 to April 2023. We purchased these devices in either the United Kingdom or Switzerland by searching for FemTech products in each category. Table 3 shows that six of these devices (no.: 1,2,4,5,6,10) are connected to an app, one does not offer an app and is a standalone device (no.: 2), two are not connected to their apps (no.: 7,9), and one is only an app (no.: 8). These devices and apps are manufactured in various countries including United Kingdom, United States, Switzerland, Germany, France, Israel, and China and their price varies based on the product (from free apps with in-app purchases to £500–600). We chose this combination for two reasons. First, we wanted to cover a range of products from different brands with various functionalities and features. Second, some of these categories do not offer off-the-shelf IoT devices and are limited to apps or non-IT products only. This consent must involve some form of unambiguous positive action (e.g., ticking a box and clicking a link) and be separated from other matters (e.g., terms and conditions and privacy policy). In order to avoid “nudge behaviour,” the privacy consent should allow the user to make a choice, therefore it should include options such as Accept (Yes, Agree, Allow, etc.) and Reject (No, Disagree, Block, etc.). 12 eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX% 3A02002L0058-20091219 3.2.2 Privacy notice The ePrivacy Directive12 (“ePD,” aka “cookie law”) provides supplementary rules to the GDPR. According to the ePD website, publishers must rely on user consent when collecting and processing personal data using non-mandatory (not strictly necessary for the services requested by the user) cookies or other technologies. This is in accordance with the guidance given by the European Data Protection Board and the ICO. To comply with the GDPR, and according to the ICO guidelines, the online service providers (e.g., product websites and Android apps) are required to inform the users about tracking technologies (e.g., cookies), their purpose and reasons, and obtain the person’s consent to use the tracking data. 3 Methodology If a privacy notice only includes Accept and requires the user to engage with the notice and accept the settings before they can access an online service’s content, they are presenting the user with a tracking “wall.” Such user consent is not considered valid if the use of this tracking wall nudges the user to agree to their personal data being used by the company or any third parties as a condition of accessing the service. Similar to the above, the consent should not highlight Accept over Reject and other options. The online services should enable the user to withdraw the previously given consent with the same ease that they gave it. The service providers should not rely on the other control mechanisms (e.g., browser settings or mobile settings) as users’ opt-out mechanism. Pre-enabling the non- essential tracking technologies without users taking positive action before it is set on their device does not represent valid consent and is a violation. In order to highlight the non-compliant practices of these devices and systems, we followed the same methods we used in Mehrnezhad (2020); Mehrnezhad and Almeida (2021) and tested the websites and apps of these products for their tracking practices. For websites, we opened each website on Chrome on a MacBook laptop in order to observe (i) if there is a cookie (privacy) notice, and (ii) what the user control options were. For apps, when we installed each app on an Android device, we opened it for the ﬁrst time as well as later (a few times), and again to test if there was a cookie (privacy) notice and the control options. In order to review the privacy policies, when there was a link available, we followed the same approach used in the review of the regulations by looking for FemTech-related keywords. 3.2.3 Tracking practices To study the tracking behaviour of the websites of these devices, we used Brave13 (a privacy-oriented browser) to identify how many trackers are activated when the website is loaded for the ﬁrst time, and before any engagement with the cookie notice. Brave uses a block- by-design mechanism that blocks and reports ads and website trackers while the webpage is getting parsed. For identifying the app trackers, we use the Exodus Privacy app (a privacy audit platform for Android apps) 14 to ﬁnd the number and types of trackers within each app. Exodus uses static analysis (the evaluation of the app code without executing it) to ﬁnd the tracker’s code signature in an app’s APK. 3.2.1 Data collection We installed all the Android apps associated with these products from the Google Play App Store. In the case of IoT devices, we set them up, i.e., charging them, turning them on and connecting them to the Android app. We then started using these devices and their companion apps as an end-user. We observed what type of data each of these devices collect either via the user’s manual input (e.g., name and age) or automatic data collection via the device’s sensors and other resources, e.g., access to phone contacts. These data types are presented in Table 3. For these experiments, we followed the same structure of recent papers (Mehrnezhad et al., 2022b; Almeida et al., 2022; Mehrnezhad et al., 2022a). Two of the authors repeated this process for each app independently (on two Google Pixel 6 phones) and logged their observations. If there was an inconsistency in the result, the experiment was repeated jointly for a third time. 13 Brave.com 14 reports.exodus-privacy.eu.org/en/ Frontiers in The Internet of Things 4.1 General data protection regulation Switzerland is not an EU member, and nor is it a member of the larger European Economic Area (EEA). Swiss companies don’t have to obey the GDPR. However, they have to obey the GDPR when they are operating in the EEA. The main data protection law of Switzerland is the Federal Act on Data Protection (FADP). FADP’s deﬁnitions include a category of sensitive personal data. Sensitive personal data is deﬁned in four groups: data on 1) religious, ideological, political or trade union-related views or activities, 2) health, the intimate sphere or the racial origin, 3) social security measures, and 4) administrative or criminal proceedings and sanctions. Accordingly, in addition to valid consent for personal data, consent must be given expressly in the case of processing sensitive personal data or personality proﬁles. Similar to the GDPR, the FADP gives sensitive data more protection. Due to Brexit, and since the EU GDPR is an EU regulation and no longer applies to the United Kingdom. If a company operates inside the United Kingdom, they need to comply with the Data Protection Act 2018 (DPA 2018). According to the ICO, the provisions of the EU GDPR have been incorporated directly into United Kingdom law as the United Kingdom GDPR. In practice, there is little change to the core data protection principles, rights and obligations. In the GDPR, personal data is deﬁned as: “information that relates to an identiﬁed or identiﬁable individual.” The GDPR recognises some types of personal data as more sensitive, referred to as “special category data,” and gives them extra protection15. This data includes information that reveals racial or ethnic origin, political opinions, religious or philosophical beliefs, trade union membership, as well as genetic data and biometric data, and data concerning health, sex life, and sexual orientation. The GDPR prohibits the processing of special category data. This requirement is on top of all the other subject rights for general personal data. Switzerland is implementing new legislation to better protect its citizens’ data: the new Federal Act on Data Protection (nFADP), will come into effect on 1st September 2023. This revision was intended in particular to bring it closer to European data protection legislation. One of the main changes is in the deﬁnition of sensitive data. These categories of personal data will continue to be considered sensitive under the Revised FADP. 15 ico.org.uk/for-organisations/guide-to-data-protection/guide-to-the- general-data-protection-regulation-gdpr/special-category-data/what- is-special-category-data/ 4.1 General data protection regulation For instance, the Revised FADP will add two new categories: genetic data and biometric data that uniquely identify an individual. When we search in the GDPR articles and guidelines, Fem-Tech data categories are not mentioned directly. There is an overlap between FemTech data and some of the special categories of data, e.g., health, sex life, sexual orientation, and potentially genetic, biometric data, and even racial or ethnic origin, political opinions, religious or philosophical beliefs. The GDPR deﬁnes the following data: Both GDPR and nFADP mandate a Data Protection Impact Assessment (DPIA) on special categories and sensitive data. DPIA is a process to help companies identify and minimise the data protection risks of a project16. In general, by going through the guidelines and the description of data protection laws, we did not ﬁnd any explicit mention of the FemTech keywords in the FADP. We also observed that the FADP is less expanded, developed, speciﬁed, and potentially enforced when it comes to sensitive data. Health data: “data concerning health means personal data related to the physical or mental health of a natural person, including the provision of healthcare services, which reveal information about his or her health status.” Genetic data: “means personal data relating to the inherited or acquired genetic characteristics of a natural person which give unique information about the physiology or the health of that natural person and which result, in particular, from an analysis of a biological sample from the natural person in question.” 4 Applicable laws and regulations In this section, we provide the results of our review of the laws and regulations. Frontiers in The Internet of Things 05 frontiersin.org 10.3389/friot.2024.1296599 Mehrnezhad et al. 16 ico.org.uk/for-organisations/guide-to-data-protection/guide-to-the- general-data-protection-regulation-gdpr/accountability-and- governance/data-protection-impact-assessments/ 17 gov.uk/government/publications/medical-devices-software- applications-apps 4.3 UK medical devices regulations 2002 The Medicines and Healthcare Products Regulatory Agency (MHRA) is an executive agency of the Department of Health and Social Care in the United Kingdom which is responsible for ensuring the safety of medicines and medical devices. Their website provides a range of guidance17 and regulations concerning health and medical services. MHRA has a guidance document on medical device stand- alone software including apps. It was published in 2014 and updated in 2022. It is clariﬁed that “a medical purpose is determined by what the manufacturer states in the device’s labelling, instructions for use and any promotional materials.” It is a helpful document to guide developers in identifying how to progress within the regulatory environment and to distinguish whether the app falls within the scope of being a “medical device.” If the device or app is a medical Biometric data: “means personal data resulting from speciﬁc technical processing relating to the physical, physiological or behavioural characteristics of a natural person, which allow or conﬁrm the unique identiﬁcation of that natural person, such as facial images or dactyloscopic data.” It does not deﬁne data concerning sex life, sexual orientation, racial or ethnic origin, political opinions, or religious or philosophical beliefs. A few more focused guidelines and documents have been developed around the special category data including the European Data Protection Board (EDPB)’s guidelines for genetic data and biometric data. However, to the best of our knowledge, there aren’t any speciﬁc data protection regulations set for “Fem- Tech data” when collected and processed beyond health and medical clinics. Frontiers in The Internet of Things frontiersin.org 06 10.3389/friot.2024.1296599 Mehrnezhad et al. through the surface of the body.” This regulation also applies to clinical investigations concerning such medical devices. device then it must comply with the Medical Devices Regulations200218. This regulation is more than 20 years old and does not provide any content on the SP aspects of modern medical devices, e.g., apps and connected devices. There is also no mention of FemTech-related data. g g As general requirements for Electronic programmable systems, this document brieﬂy says that for software devices or those that incorporate software, the development and risk management (i.e., information security, veriﬁcation and validation) should be according to the state-of-the-art practices. The general safety requirements take into account the intended purpose which is set by the manufacturer. 4.3 UK medical devices regulations 2002 The parts related to risks and risk management are for safety risks and there is no mention of SP risks. Article 110 of this regulation is on data protection stating: “1) Member States shall apply Directive 95/46/EC to the processing of personal data carried out in the Member States pursuant to this Regulation. 2) Regulation (EC) No 45/2001 shall apply to the processing of personal data carried out by the Commission pursuant to this Regulation.” Note that the GDPR supersedes the Directive 95/46/EC and repeals Regulation (EC) No 45/2001. More recently, the MHRA has been working on a new Software and AI as a Medical Device Change Programme19 where one of its 11 work packages (WP5) is “Cyber Secure Medical Devices.” This WP’s deliverables include: 1) Secondary Legislation [Cybersecurity requirements for medical devices and IVDs (in vitro diagnostic medical devices)], 2) Regulatory Guidance (Guidance elucidating cybersecurity requirements for medical device and IVDs), 3) Best Practice Guidance (Management of unsupported software devices), 4) Processes (Reporting of relevant cybersecurity vulnerabilities). 18 legislation.gov.uk/uksi/2002/618/contents/made 19 gov.uk/government/publications/software-and-ai-as-a-medical- device-change-programme/ 4.4 Regulation (EU) 2017/745 for medical devices Overall, we did not ﬁnd any direct mention of FemTech-related data and its protection in these regulations. Similar to the United Kingdom MHRA, the European Commission also has a guidance document on Qualiﬁcation and Classiﬁcation of Software in Regulation20 (EU) 2017/745–MDR and Regulation (EU) 2017/ 746–IVDR, released in October 2019. In comparison, we found the United Kingdom guidance more comprehensive in terms of helping developers decide about the intended use of their product as a medical device. The United Kingdom has been complying with EU medical and health regulations for years. However, due to Brexit, the United Kingdom does not necessarily comply with EU medical regulations anymore. For medical devices, Switzerland follows what is speciﬁed by the EU system of compliance assessment and certiﬁcation, based on bilateral agreements. Hence, we also review the EU Regulation for Medical Devices. In the EU, medical devices must undergo a conformity assessment to demonstrate they meet legal requirements to ensure their safety and performance as intended. They are regulated at the EU Member State level, but the European Medicines Agency (EMA) is involved in the regulatory process. The Regulations on Medical Devices [Regulation (EU) 2017/745] and on In Vitro Diagnostic Devices [Regulation (EU) 2017/746] changed the European legal framework for medical devices, coming into effect in 2021 and 2022, respectively. In this section, we focus on the former. 5 Analysis of FemTech systems In this section, we present the results of data collection and tracking practices as well as the privacy policy review. 20 ec.europa.eu/docsroom/documents/37581?locale=en 5.1 Data collection For example Daysy Cycle Computer Hidrate Spark3 Smart TABLE 3 Data collected by FemTech IoT devices and apps. Devices with X are not connected to their associated apps. Android App categories include: Health and Fitness, HF; Medical, M; Entertainment, E; and Tools, T. Device/ App (1) Elvie pump (2) Daysy cycle (3) Lady comp fertility (4) Hidrate bottle (5) Periﬁt kegel (6) Frida sex toy (7) Livia pain reliever (8) Balance menopause (9) Daviky pill organiser (10) Ivy health tracker Device ✔ ✔ ✔ ✔ ✔ ✔ X no X ✔ App ✔ ✔ no ✔ ✔ ✔ ✔ ✔ ✔ ✔ Category HF M — HF HF E E HF T HF Download # 100k+ 50k+ NA 100k+ 100k+ 100k+ 10k+ 100k+ 500+ 1M+ User data User ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ Contact ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ Lifestyle ✔ ✔ ✔ ✔ Period ✔ ✔ ✔ ✔ ✔ Pregnancy ✔ ✔ Nursing ✔ ✔ Reproductive ✔ ✔ ✔ Sexual ✔ ✔ ✔ ✔ ✔ Medical info ✔ ✔ ✔ Physical ✔ ✔ ✔ Emotional ✔ ✔ ✔ Data about others Partner ✔ ✔ ✔ Social media ✔ ✔ ✔ ✔ Child ✔ IoT/Mobile device’s resources Storage ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ Contacts ✔ ✔ ✔ ✔ Accounts ✔ ✔ Settings ✔ ✔ ✔ Cam/mic ✔ ✔ ✔ ✔ WiFi ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ Location ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ Bluetooth ✔ ✔ ✔ ✔ ✔ ✔ NFC ✔ Sensors ✔ ✔ ✔ ✔ ✔ ✔ ✔ (e.g., happy, anxious). These systems also ask for or automatically collect data about others including: Baby/child (e.g., nursing, sleep cycles, fetal movements), Social media proﬁles, forums, or plugins (e.g., Facebook, Spotify), Partner (e.g., details of partnered sex activities, name, age, photo). These technologies might even ask about the medical history of the user’s family. Finally, these systems also have access to the devices’ resources, e.g., camera, microphone, (e.g., happy, anxious). These systems also ask for or automatically collect data about others including: Baby/child (e.g., nursing, sleep cycles, fetal movements), Social media proﬁles, forums, or plugins (e.g., Facebook, Spotify), Partner (e.g., details of partnered sex activities, name, age, photo). These technologies might even ask about the medical history of the user’s family. 5.1 Data collection This regulation deﬁnes “medical device” as “any instrument, apparatus, appliance, software, implant, reagent, material or other article intended by the manufacturer to be used, alone or in combination, for human beings for one or more of the following speciﬁc medical purposes.” They include diagnosis, prevention, prediction, monitoring, prognosis, treatment, alleviation, and compensation of disease, injury or disability, investigation, replacement or modiﬁcation of the anatomy or of a physiological or pathological process or state, providing information by means of in vitro examination of specimens derived from the human body (e.g., organ, blood and tissue donations). They add that “devices for the control or support of conception” shall also be deemed to be medical devices. The following products shall also be deemed to be medical devices. It deﬁnes an “invasive device” as “any device which, in whole or in part, penetrates inside the body, either through a body oriﬁce or In Figure 1, we have identiﬁed off-the-shelf products available for purchase in Switzerland and/or the United Kingdom, with a range of functionalities. While we purchased a device per category, these devices may belong to different categories and be advertised across categories as well as FemTech and/or general health. We examined what types of data these devices collect, as presented in Table 3. We broadly categorise this data into three groups: user data, data about others, and device/phone data. Our examinations show that user data include, but are not limited to: Name (e.g., photo, age, gender), Contact (e.g., mobile, email, address), Lifestyle (e.g., weight, diet, sleep), Period (e.g., cycle length, ovulation days), Pregnancy (e.g., test results, due dates, IVF), Nursing (e.g., time, volume, pain) Reproductive organs (e.g., cervical mucus, biofeedback, muscle strength), Sexual activities (e.g., date, contraceptives, orgasm), Medical information (e.g., medication type, blood pressure, lab reports scan). Physical symptoms (e.g., headache, constipation), Emotional symptoms Frontiers in The Internet of Things 07 frontiersin.org 10.3389/friot.2024.1296599 Mehrnezhad et al. (e.g., happy, anxious). These systems also ask for or automatically collect data about others including: Baby/child (e.g., nursing, sleep cycles, fetal movements), Social media proﬁles, forums, or plugins (e.g., Facebook, Spotify), Partner (e.g., details of partnered sex activities name age photo) These technologies might even ask device ﬁles/and storage, phone’s contacts and calls, communicational sensors (WiFi, Bluetooth, NFC), motion and environmental sensors from the phone or the device (e.g., temperature, pressure, Co2). Frontiers in The Internet of Things 5.2 Privacy consent, privacy policy, and tracking practices devices which collected minimal data. For instance, Lady Comp Fertility Tracker collects some form of user data (e.g., age), cycle information, sex, and has a thermometer to measure user basal temperature. This device does not offer an app and has a memory for a year. The manual suggests that this data can be backed up by connecting the device to a PC via a cable. However, via testing, we could not use such a feature. Note that Table 3 only represents the data collected by the device and app itself and does not show the data that may be collected via the websites, e.g., for purchasing, creating a proﬁle account for networking, and subscribing. For instance, the Livia Menstrual Pain Reliever device does not collect any data about the user, though its associated app (which is not connected to the devices) does. Also, its website requires user and contact information for purchasing and subscription and offers a review platform via a third-party service too. As demonstrated in Table 4, all the apps and websites that we studied appear to violate the GDPR in terms of asking for valid consent. They either nudge the user into accepting a highlighted accept, limit the access behind a privacy notice wall, bundle the privacy notice with other matters (e.g., terms and conditions), or don’t provide any privacy consent. The only exception is the Balance Menopause App which presented valid consent. However, its website did not. In addition, our review of the privacy policies of these apps indicates that 4 apps included a reference to or a description of FemTech-related data. For instance, the DaysyDay app highlights that “Within this framework of the contractual relationship between you and us, health data such as your body temperature, menstruation, etc. may also be processed. For such processing, we need your explicit consent. By submitting this data, you are granting us that consent.” Yet, they also say: “Our online services are not subject to HIPA.” These statements are problematic since explicit consent is in conﬂict with obtaining consent via submitting such data. Similarly, Priﬁt’s privacy policy explains “Sensitive personal data” which is in accordance with the GDPR special data category and lists the data items that the app collects. However, it does not clarify how such data is given extra protection. 5.1 Data collection Finally, these systems also have access to the devices’ resources, e.g., camera, microphone, device ﬁles/and storage, phone’s contacts and calls, communicational sensors (WiFi, Bluetooth, NFC), motion and environmental sensors from the phone or the device (e.g., temperature, pressure, Co2). For example, Daysy Cycle Computer, Hidrate Spark3 Smart Bottle, and Periﬁt Kegel Exerciser collected data in all categories (user, partner, and device) quite intensively. There were also some 08 Frontiers in The Internet of Things frontiersin.org 10.3389/friot.2024.1296599 Mehrnezhad et al. TABLE 4 Privacy notice of apps and websites and GDPR violations. The bold options in the Website cookie notice column is the highlighted option in the notice. No. Product FemTech data reference in privacy policy Android app privacy notice [place] Violation Website cookie notice & options Violation 1 Elvie Smart Pump No I agree to Elvie’s terms of use & privacy policy [Sign-up page (wall)] ✔ Accept All, Customise ✔ 2 Daysy Cycle Comp Yes (health, body temp, menstruation) I’ve seen the imprint & accept privacy policy [Welcome page (wall)] ✔ Accept ✔ 3 Lady Copm Fertility Tracker NA No App NA Accept ✔ 4 HidrateSpark Smart Bottle No I agree to terms of service & privacy policy [Sign-up page (wall)] ✔ Preferences, Accept ✔ 5 Periﬁt Kegel Trainer Yes (health, sex, menopause, health, gender, height, weight) . . ., you expressly agree to collection of your health data, . . . [Sign-up page (wall)] ✔ Allow all cookies, Cookie settings ✔ 6 Frida (Vibio) Sex Toy No I have read & understood the Terms & Conditions and Privacy agreement [Sign-up page (wall)] ✔ No Notice ✔ 7 Livia Menstrual Pain Reliever NA No privacy content ✔ No Notice ✔ 8 Balance Menopause Support App Yes (health, symptoms, medication, menopause) (i) View our privacy policy [(Welcome page)] (ii) I accept that you may use the data I share for the above purposes [(Second page)] No Save and close, Accept all cookies ✔ 9 Daviky Pill Organiser No No privacy content ✔ No Notice ✔ 10 Ivy (Bellabeat) Health Tracker Yes (health, exercise, steps, heart rate, pregnancy, weight, sleep) By continuing you agree to Bellabeat’s Terms & Conditions & Privacy Policy [Sign-up page (wall)] ✔ No Notice ✔ Frontiers in The Internet of Things 21 datamatters.sidley.com/2023/04/20/femtech-has-been-warned-uks- ico-indicates-closer-scrutinization-of-femtech-apps/ 6.1 FemTech risks for such data. However, again, it is not clear how such data is treated with care. Bellabeat’s policy has a similar content on deﬁning sensitive personal data to Priﬁt. It then says: “If the information we collect is health data or another special category of personal data subject to the European Union’s General Data Protection Regulation, we ask for your explicit consent to process that kind of data. We acquire this consent separately when you take actions resulting in our receiving the data, for instance when you use the menstrual calendar feature.” However, when trying to use the app by signing up via email, another Privacy Consent wall was shown which required the user to agree to tick two boxes: one general privacy policy and terms of use and one stating: “I agree to the processing of my personal health data for providing me Period Diary app functions, See more in Privacy Policy.” As we discussed earlier, multiple threat actors have been shown to be interested in such data (Mehrnezhad et al. 2022b) including(ex-)partner and family, employers and colleagues, insurance ﬁrms, advertising companies, political and religious organisations, governments, and medical and research companies. For instance, some of these products can have shared usage, e.g., via a remote partner mode. Access to such intimate data (e.g., reproductive health) without informed and continuous user consent may enable tech abuse such as external pressure to become pregnant (WHO, 2020; Mehrnezhad and Almeida, 2021) and/or cyberstalking (Stevens et al. 2021; Chan 2021). The existing regulations are yet to cover several aspects of online safety for collectively created and shared data. Table 5 shows the apps and websites of all these products and the trackers. Our Exodus analysis revealed that these apps have between 1 and 9 trackers. In addition, the majority of these websites are tracking the users before the user engages with the cookie notice. One particular website (hidratespark.com/) increased the number of these trackers to more than 70 (and counting) while we kept the website open and without any interaction with it. It also attempted to use motion sensors on a mobile device if accessed from one. In contrast, the Daysy Cycle Computer and Lady Comp Tracker both included only one tracker (Google Tag Manager). FemTech solutions have already found their way to organisational usage (Erickson et al. 6.1 FemTech risks 2022) There are concerns around how workplace monitoring threatens women’s equity (Brown 2021; Veliz 2022; Brown 2020) e.g., the case of pregnancy redundancies and impact on promotions (Maternity Action, 2019) or discrimination due to infertility (van der Berch 2010). Given that FemTech solutions (e.g., fertility apps) are already sharing these data with third parties including employers (Harwell 2019; Scatterday 2022) without user consent (Mehrnezhad and Almeida 2021), these technologies could be used to further gender inequality at work. Similarly, work-related regulations, policies and guidelines could be blindsided when it comes to SP of such data. Take the “BS 30416, Menstruation, menstrual health and menopause in the workplace–Guide” as an example22 with no content on the SP issues of the data related to these practices. 22 bsigroup.com/en-GB/our-services/events/events/2023/menstrual- and-menopausal-health-in-the-workplace/ 5.2 Privacy consent, privacy policy, and tracking practices Balance app’s policy has a dedicated section for “health data” by deﬁning it and explaining their approach regarding explicit consent As can be seen, not only do these systems collect data about the user (and others), but the majority of them have access to mobile and device resources too. Some of these permissions are marked as dangerous according to Google’s protection levels. If not justiﬁed well, the risks of access to storage, contacts, camera, microphone, and location are more visible. However, speciﬁc permissions such as access to system Settings and other Accounts on the device also impose SP risks. Similarly, there is a body of research [e.g., Delgado-Santos et al. (2022)] on how sensors can be used to break user privacy. This can become more critical in FemTech systems since they are associated with user health. Frontiers in The Internet of Things 09 frontiersin.org frontiersin.org Mehrnezhad et al. 10.3389/friot.2024.1296599 TABLE 5 Tracking practices of apps and websites. TABLE 5 Tracking practices of apps and websites. No. Product Exodus trackers & permissions Brave trackers 1 Elvie Smart Pump 2, 13 6 2 Daysy Cycle Computer 1, 35 1 3 Lady Comp Fertility Tracker NA 1 4 HidrateSpark Smart Bottle 7, 25 70+ 5 Periﬁt Kegel Trainer 8, 36 31 6 Frida (Vibio) Sex Toy 2, 39 3 7 Livia Menstrual Guide App (Associated with Pain Reliever) 7, 35 9 8 Balance Menopause App 2, 27 2 9 Daviky Pill Organiser 0, 6 2 10 Ivy (Bellabeat) Health Tracker 9, 23 10 6 Discussion Recently, there are some efforts to enforce the law in the FemTech space (e.g., the ICO’s recent project on auditing FemTech apps21). Here we discuss the risks of FemTech, our ﬁndings, and that a more proactive approach to policy-making and enforcement is needed in this sector. Health insurance discrimination on the basis of health status is already a pressing issue (Crossley 2005; Rosenbaum 2009) and FemTech data can be used for such applications (Scatterday 2022). When an insurance company has access to the health data Frontiers in The Internet of Things frontiersin.org 10 10.3389/friot.2024.1296599 10.3389/friot.2024.1296599 10.3389/friot.2024.1296599 Mehrnezhad et al. of users, parents, siblings and other relatives, it is directly putting the privacy of others at risk too. If a possibility of a pre-existing medical condition (e.g., infertility or breast cancer) is identiﬁed, such insurance ﬁrms will not support the person or do it at much higher rates. A situation that might continue even after the death of the users; impacting their offspring. In such situations, customers have almost zero agency in objecting to the decision since the existing practices (including the regulations) do not give such support either in principles and/or the enforcement. While GDPR gives extra protection to special category data, there are 10 exceptions: explicit consent, employment, social security and social protection (if authorised by law), vital interests, not-for-proﬁt bodies, made public by the data subject, legal claims or judicial acts, reasons of substantial public interest (with a basis in law), health or social care (with a basis in law), public health (with a basis in law), and archiving, research and statistics (with a basis in law). Special category data cannot be used for solely automated decision-making (including proﬁling) that has legal or similarly signiﬁcant effects unless there is explicit consent or substantial public interest conditions are met. The exceptions of data protection regulations (e.g., GDPR) are indeed debatable. While the analysis of these exceptions in the wild is beyond the scope of this paper, we believe that this is an area that will unfold signiﬁcantly in the future. For instance, consider the ﬁrst exception: explicit consent. Given the sensitive nature of FemTech data and its differential and complex risks and harms (Mehrnezhad and Almeida 2021), how do we guarantee that the user is fully aware of the consequences of such consent and makes an informed and later continues decision? 6.3 General data regulations vs. medical devices regulations From app-only data collection to sensor-enabled FemTech devices, with extra processing via advanced algorithms, e.g., AI, FemTech data reveal people better than they know themselves. Reportedly, these apps share sensitive data (e.g., sex activity) with third parties (e.g., Facebook) the moment the user opens the app, even without a Facebook account Int (2019). Apart from the academic research on FemTech tracking practices such as (Mehrnezhad and Almeida 2021; Shipp and Blasco 2020), news reports have also paid attention to this matter (Page 2022), including the cases of selling FemTech data by data marketplaces (Cox 2022), the interest of political and religious organisations in such data (Glenza 2019) and the potential for new opportunities for spreading health-related misinformation (Pennycook et al. 2021). Additionally, FemTech data can be particularly of interest to governments. The recent debates around the overturn of the abortion law in the US Supreme Court (Page 2022) has shown very well how FemTech (e.g., apps) can enable such a systematic tracking and controlling of women’s bodies (Alvarez 2019; Shoichet 2020). When reviewing the current general data protection regulations and the medical ones, we ﬁnd several gaps and disconnections between the two sets of regulations. We would expect a higher level of safeguarding in these products where personal health data is recorded, even if the app does not fall within current medical device deﬁnitions and regulations. This is supposed to be covered by the special categories of data in the general data protection laws. However, in practice and based on our experiments, it is not enforced properly. For instance, we did not ﬁnd any appropriate consent in apps and websites tested and whether or not any extra protection is given to sensitive FemTech data. As we discuss in Almeida et al. (2022), the fact that these products collect data about others (partner, baby/child, family, etc.) adds to these complexities. When registering an app in the app store, the developers select the most appropriate app category. However, due to the ambiguity in the deﬁnition of these categories, the doors are open to potential misuse and gaming by the registrant. At the time of this writing, there are 38 categories on the Google Play App Store including “Medical,” and “Health and Fitness” categories. 6 Discussion More research is needed to ﬁll in these research gaps. During the pandemic selling and sharing medical and health data, e.g., with medical and research companies became a common practice. This trend has contributed to the ongoing conversations about the sensitive nature of such data and legal complications (Powles and Hodson 2017; Solon 2018). FemTech data deals with a complex mix of health, medical, biometric, and genetic data, alongside sexual activities/ orientation, reproductive decisions, and even religious and political views (Mehrnezhad and Almeida 2021) and can be of particular interest to cyber-criminals e.g., for blackmailing. Data breaches in digital health and FemTech are even more serious because of the sensitive nature of the data (Veliz 2022; Rosas 2019), particularly when taking socio-cultural differences into account; more marginalised groups will have more to lose from such disclosures. 6.3 General data regulations vs. medical devices regulations Yet, as reported in Table 3, only one of these apps (#2) is listed as medical, 5 listed as health and ﬁtness, and the rest include “Entertainment” or “Tool.” Miscategorising an app which contains medical records (such as user’s medical conditions and medicines, or family history) as Health & Fitness or other groups would enable the developers to avoid the potential consequences, for example, of remaining in the app market without drawing signiﬁcant attention to it. As long as such apps and services make only general wellness claims -like tools, entertainment, health and ﬁtness, they do not need to be vetted by health regulators or as seriously as one expects by the mobile app store. Our review of the related regulations, in combination with our system studies, highlight that the existing risks can put the users at differential risks. Frontiers in The Internet of Things frontiersin.org 6.4 Non-compliant practices We identiﬁed a range of inappropriate SP practices in a subset of FemTech systems. We showed that they do not present valid consent, they do not give extra protection to sensitive data, and track users without consent. These are some of the non-compliant practices within the current regulations. In Mehrnezhad et al. (2022b), we discuss that not only is such intimate data collected by FemTech systems, but also this data is processed and sold to third parties23. In Mehrnezhad and Almeida (2021), Mehrnezhad and Almeida (2023), we have discussed at length that complex harms and risks such as the re-identiﬁcation of individuals based on health data (Goldacre et al. 2022) can differentially impact the users. In addition, most of these products do not need a wide range of information about the users to deliver their services. Yet they continue to collect such sensitive data. Some of these practices could be due to factors such as copying and pasting an app code by developers without considering privacy-by-design principles. For instance, the app associated with Livia period pain reliever (which is approved by the FDA), is simply a guide on the use of the device. While interacting with it, we did not notice any data collection or permission requests. Yet, when we checked the app’s permissions, we noticed that the camera, music and audio, notiﬁcations, and photos and videos are listed. If turned on, this app is able to collect such data. This is clearly a bad practice from the developer side. We are now seeing more efforts in the policymaking space to recognise these issues. For instance, the EU is aiming to foster common European data spaces. Data spaces are data ecosystems, often domain-speciﬁc, in which data sharing should be possible between actors. One of the data spaces is the European Health Data Space24. This proposal is still under review and it is unclear when and how it will be implemented and enforced, let alone what kind of organisations fall under these deﬁnitions. We believe that the medical and health space is in need of domain-speciﬁc and sectorial regulations with attention to the needs of marginalised user groups such as women and those with physical and mental ability limitations. 24 health.ec.europa.eu/ehealth-digital-health-and-careeuropean-health- data-space_en 23 ftm.eu/articles/your-intimate-data-is-being-sold 7 Conclusion The SP issues around FemTech can lead to differential harm where complex risks are enabled by many factors including gaps in the regulations, non-compliant practices, the lack of enforcement, and limited research and guidelines for secure, privacy-preserving, and safe products. We reviewed the regulations related to FemTech in the United Kingdom, EU, and Switzerland and identiﬁed the gaps. We ran experiments on a range of FemTech devices, apps, and websites and identiﬁed several exploitative practices. We discussed our results and suggested that policymakers explicitly acknowledge and accommodate the risks of these technologies in the relevant regulations. 6.2 Gaps in the related regulations Our critical review of FemTech-related regulations shows that they are inadequate in addressing the risks associated with these technologies. The EU and United Kingdom medical devices regulations don’t have any references to FemTech data and user protection. The GDPR and Swiss FADP have references to sensitive and special category data which overlap with FemTech data. Yet, there are several areas for expansion and improvement. The United Kingdom MHRA is developing a new Software and AI as a Medical Device Change Programme, where apart Frontiers in The Internet of Things 11 frontiersin.org 10.3389/friot.2024.1296599 Mehrnezhad et al. from a dedicated work package to cybersecurity (WP5), it also has one on Classiﬁcation (WP2). The problem statement says: “Currently, the Medical Device Regulations 2002 (as amended) do not classify software proportionate to the risk it might pose to patients and public safety.” We believe that such efforts are required immediately to protect citizens against these risks. certiﬁcation, compliance testing and policing/penalties. Accordingly, providers and developers need to be aware of the regulations, guidance and best practices and have appropriate tools to develop and evaluate products. Currently, there are no entities well-equipped to provide such services. We acknowledge that the legal framework of the medical and health sector is a combination of laws, standards, certiﬁcations, and beyond. For instance, ISO 13485 is speciﬁcally for products that fall within the criteria for a “medical device.” Implementation of ISO 13485 tends to draw with its alignment to data standards, as such products are subject to clinical trial validation, governed by ethics committees, who would likely question marginal data practices and so has a wider inﬂuence on the company and its marketing behaviour. Companies often deliberately frame their products as “non-medical” and, e.g., as “wellbeing” to avoid being subject to the medical device regulation. Hence, the period and cycle tracking apps are on the market free from regulation as it can be argued that the information is not used for clinical decision-making and guidance for treatment. Whereas ovulation tests (aka class I in medical devices regulations)/pregnancy tests (class II) are used and subject to regulation, even if ovulation tests are then associated with an app just for the purpose of cycle tracking. 6.4 Non-compliant practices That, together with better enforcement of the existing regulations discussed in this paper can lead to more effective practices to protect the citizens’ security, privacy, and safety, while enabling them to improve the quality of their lives including their health via using these technologies without any risk or fear. Non-compliance or poor adherence to laws and standards may arise for many reasons. There may be unintentional oversight or a deliberate attempt for commercial or other purposes. The developers themselves may be unaware of best practices and regulations in the area. Different solutions (websites, apps, IoT devices) developed in different territories may be subject to different regulations, yet regulators may not have the powers or resources to certify compliance or investigate potential non-compliance where no certiﬁcation process exists. This might be the time to focus on more sectorial and domain-speciﬁc data protection regulations as we discuss next. References Almeida, T., Mehrnezhad, M., and Cook, S. (2023). “The importance of collective privacy in digital sexual and reproductive health,” in The 17th annual UK Fertility Conference. arXiv preprint arXiv:2311.15432. How menstruation apps are sharing your data (2019). How menstruation apps are sharing your data. Priv. Int. Available at: https://privacyinternational.org/long- read/3196/no-bodys-business-mine-how-menstruations-apps-are-sharing-your- data. Almeida, T., Shipp, L., Mehrnezhad, M., and Toreini, E. (2022). Bodies like yours: enquiring data privacy in femtech. ACM Nord. doi:10.1145/3547522. 3547674 Maternity Action (2019). Discrimination during maternity leave and on return to work, Available at: https://maternityaction.org.uk/advice/discrimination-during- maternity-leave-and-on-return-to-work/ Maternity Action (2019). Discrimination during maternity leave and on return to work, Available at: https://maternityaction.org.uk/advice/discrimination-during- maternity-leave-and-on-return-to-work/ Alvarez, P. (2019). House judiciary committee asks former orr director to clarify testimony on pregnant minors. CNN. Available at: https://edition.cnn.com/2019/03/22/ politics/scott-lloyd-pregnant-minors/index.html Mcdonald, N., and Andalibi, N. (2023). “i did watch ‘the handmaid’s tale”’: threat modeling privacy post-roe in the United States. ACM Trans. Computer-Human Interact. 30, 1–34. doi:10.1145/3589960 Brown, E. (2020). Supercharged sexism: the triple threat of workplace monitoring for women. SSRN 3680861 McMillan, C. (2023). Rethinking the regulation of digital contraception under the medical devices regime. Med. Law Int. 23, 3–25. doi:10.1177/ 09685332231154581 Brown, E. (2021). The femtech paradox: how workplace monitoring threatens women’s equity. Jurimetrics. Mehrnezhad, M. (2020). “A cross-platform evaluation of privacy notices and tracking practices,” in IEEE EuroS&P Workshop (EuroUSEC). Chan, S. (2021). Hidden but deadly: stalkerware usage in intimate partner stalking. Introd. Cyber Forensic Psychol. Underst. Mind Cyber Deviant Perpetrators, 45–66. Mehrnezhad, M., and Almeida, T. (2021). “Caring for intimate data in fertility technologies,” in Acm CHI. Coopamootoo, K., Mehrnezhad, M., and Toreini, E. (2022). “i feel invaded, annoyed, anxious and i may protect myself”: individuals’ feelings about online tracking and their protective behaviour across gender and country. USENIX. Mehrnezhad, M., and Almeida, T. (2023). “my sex-related data is more sensitive than my ﬁnancial data and i want the same level of security and privacy”: user risk perceptions and protective actions in female-oriented technologies. EuroUSEC. Cox, J. (2022). Data marketplace selling info about who uses period tracking apps. Available at: https://www.vice.com/en/article/v7d9zd/data-marketplace-selling-clue- period-tracking-data. Mehrnezhad, M., Coopamootoo, K., and Toreini, E. (2022a). How can and would people protect from online tracking? Proc. Priv. Enhancing Technol. 1, 105–125. doi:10. 2478/popets-2022-0006 Crossley, M. (2005). Discrimination against the unhealthy in health insurance. U. Kan. L. Rev. 54, 73. Mehrnezhad, M., Shipp, L., Almeida, T., and Toreini, E. (2022b). “Vision: too little too late? Conﬂict of interest MM: Conceptualization, Funding acquisition, Investigation, Methodology, Project administration, Resources, Writing–original draft, Writing–review and editing. TV: Conceptualization, Funding acquisition, Investigation, Methodology, Project administration, Resources, Writing–original draft, Writing–review and editing. MC: Conceptualization, Funding acquisition, Investigation, Methodology, Project administration, Resources, Writing–original draft, Writing–review and editing. The authors declare that the research was conducted in the absence of any commercial or ﬁnancial relationships that could be construed as a potential conﬂict of interest. Data availability statement supported by the ETH Future Computing Laboratory, partially conducted when the ﬁrst and second authors were working at ETH Zurich in 2022. This work has been supported by the UK EPSRC PETRAS National Centre of Excellence for IoT Systems Cybersecurity (EP/S035362/1), and the UKRI funded project, AGENCY: Assuring Citizen Agency in a World with Complex Online Harms (EP/W032481/2). The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author. 6.5 Domain-speciﬁc regulations As discussed in this paper, two sets of regulations apply to FemTech solutions: general data protection regulations and medical and health regulations. However, as shown, alone or combined they fail to protect the user from malicious practices. In addition, a key complement to regulations is systems of Frontiers in The Internet of Things 12 frontiersin.org 10.3389/friot.2024.1296599 Mehrnezhad et al. Publisher’s note All claims expressed in this article are solely those of the authors and do not necessarily represent those of their afﬁliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. Funding The author(s) declare ﬁnancial support was received for the research, authorship, and/or publication of this article. This work was Mehrnezhad et al. References do the risks of femtech already outweigh the beneﬁts?,” in EuroUSEC 2022. Delgado-Santos, P., Stragapede, G., Tolosana, R., Guest, R., Deravi, F., and Vera- Rodriguez, R. (2022). A survey of privacy vulnerabilities of mobile device sensors. ACM Comput. Surv. (CSUR) 54, 1–30. doi:10.1145/3510579 Nayeri, F. (2021). Is ‘femtech’ the next big thing in health care?. Available at: https://www. nytimes.com/2021/04/07/health/femtech-women-health-care.html, [Accessed April 2022] Erickson, J., Yuzon, J. Y., and Bonaci, T. (2022). What you don’t expect when you’re expecting: privacy analysis of femtech. IEEE Transactions on Technology and Society. Page, C. (2022). As roe v. wade reversal looms, should you delete your period- tracking app? Available at: https://techcrunch.com/2022/05/05/roe-wade-privacy- period-tracking Glenza, J. (2019). Revealed: women’s fertility app is funded by anti-abortion campaigners. The Guardian Theguardian.com/world/2019/may/30/revealed-womens- fertility-app-is-funded-by -anti-abortion-campaigners Pennycook, G., Epstein, Z., Mosleh, M., Arechar, A. A., Eckles, D., and Rand, D. G. (2021). Shifting attention to accuracy can reduce misinformation online. Nature 592, 590–595. doi:10.1038/s41586-021-03344-2 Goldacre, B., Morley, J., and Hamilton, N. (2022). Better, broader, safer: using health data for research and analysis. A Review Commissioned by the Secretary of State for Health and Social Care. Peppet, S. R. (2011). Unraveling privacy: the personal prospectus and the threat of a full-disclosure future. Nw. UL Rev. Powles, J., and Hodson, H. (2017). Google deepmind and healthcare in an age of algorithms. Health and technology 7, 351-367. doi:10.1007/s12553-017-0179-1 Harwell, D. (2019). Is your pregnancy app sharing your intimate data with your boss? The Washington Post. Available at: https://www.washingtonpost.com/ technology/2019/04/10/tracking-your-pregnancy-an-app-may-be-more-public- than-you-think/. Harwell, D. (2019). Is your pregnancy app sharing your intimate data with your boss? The Washington Post. Available at: https://www.washingtonpost.com/ technology/2019/04/10/tracking-your-pregnancy-an-app-may-be-more-public- than-you-think/. Rosas, C. (2019). The future is femtech: privacy and data security issues surrounding femtech applications. Hastings Bus. Law J. 15. Frontiers in The Internet of Things 13 frontiersin.org Mehrnezhad et al. 10.3389/friot.2024.1296599 10.3389/friot.2024.1296599 Rosenbaum, S. (2009). Insurance discrimination on the basis of health status: an overview of discrimination practices, federal law, and federal reform options: executive summary. J. Law, Med. Ethics 37, 101–120. doi:10.1111/j.1748-720x.2009.00423.x Rosenbaum, S. (2009). Insurance discrimination on the basis of health status: an overview of discrimination practices, federal law, and federal reform options: executive summary. J. Law, Med. Ethics 37, 101–120. doi:10.1111/j.1748-720x.2009.00423.x Stevens, F., Nurse, J. R., and Arief, B. (2021). Cyber stalking, cyber harassment, and adult mental health: a systematic review. Cyberpsychology, Behav. Soc. Netw. 24, 367–376. doi:10.1089/cyber.2020.0253 Scatterday, A. (2022). References This is no ovary-action: femtech apps need stronger regulations to protect data and advance public health goals. N. C. J. Law Technol. 23. Scatterday, A. (2022). This is no ovary-action: femtech apps need stronger regulations to protect data and advance public health goals. N. C. J. Law Technol. 23. Valente, J., Wynn, M. A., and Cardenas, A. A. (2019). Stealing, spying, and abusing: consequences of attacks on internet of things devices. IEEE Secur. Priv. 17, 10–21. doi:10.1109/msec.2019.2924167 Shipp, L., and Blasco, J. (2020). How private is your period? a systematic analysis of menstrual app privacy policies. Proc. Priv. Enhancing Technol. 2020, 491–510. doi:10. 2478/popets-2020-0083 Shipp, L., and Blasco, J. (2020). How private is your period? a systematic analysis of menstrual app privacy policies. Proc. Priv. Enhancing Technol. 2020, 491–510. doi:10. 2478/popets-2020-0083 van der Berch, K. (2010). Courts’ struggle with infertility: the impact of hall v. nalco on infertility-related employment discrimination, 81. University of Colorado Law Review. Shoichet, C. (2020). In a horrifying history of forced sterilizations, some fear the us is beginning a new chapter. Available at: https://edition.cnn.com/2020/09/16/us/ice- hysterectomy-forced-sterilization-history/index.html. Shoichet, C. (2020). In a horrifying history of forced sterilizations, some fear the us is beginning a new chapter. Available at: https://edition.cnn.com/2020/09/16/us/ice- hysterectomy-forced-sterilization-history/index.html. Veliz, C. (2022). Privacy is power: why and how you should take back control of your data. Int. Data Priv. Law. World Health Organization (2020). Sexual and reproductive health: infertility. World Health Organization. Available at: https://www.who.int/reproductivehealth/topics/ infertility/keyissues/en/. Solon, O. (2018). Data is a ﬁngerprint: why you aren’t as anonymous as you think online. Available at: https://www.theguardian.com/world/2018/jul/13/anonymous- browsing-data-medica l-records-identity-privacy 14 Frontiers in The Internet of Things frontiersin.org 14
https://openalex.org/W4318715275	http://journal.wima.ac.id/index.php/JAKO/article/download/4033/3217	English	null	THE IMPACT OF OWNERSHIP STRUCTURE ON FIRM VALUE IN THE BANKING SECTOR COMPANIES	Jurnal Akuntansi Kontemporer	2,023	cc-by-sa	7,188	JURNAL AKUNTANSI KONTEMPORER (JAKO) – VOL 15 NO 1 – JANUARI 2023 – Halaman 25 - 36 JURNAL AKUNTANSI KONTEMPORER (JAKO) – VOL 15 NO 1 – JANUARI 2023 – Halaman 25 - 36 A B S T R A K Tujuan Penelitian. Penelitian ini bertujuan untuk mengetahui pengaruh kepemilikan manajerial, kepemilikan asing, dan kepemilikan keluarga terhadap nilai perusahaan. Metode Penelitian. Penelitian ini menggunakan metode kuantitatif. Populasi dalam penelitian ini adalah perusahaan perbankan yang terdaftar di Bursa Efek Indonesia dari tahun 2016 sampai dengan tahun 2019. Sampel ditentukan dengan teknik purposive sampling dan diperoleh 88 sampel penelitian berdasarkan kriteria tertentu. Metode analisis yang digunakan adalah analisis statistik deskriptif, analisis regresi data panel, uji asumsi klasik, dan analisis regresi linier berganda. Hasil dan Temuan Penelitian. Hasil penelitian menunjukkan bahwa kepemilikan Tujuan Penelitian. Penelitian ini bertujuan untuk mengetahui pengaruh kepemilikan manajerial, kepemilikan asing, dan kepemilikan keluarga terhadap nilai perusahaan. Metode Penelitian. Penelitian ini menggunakan metode kuantitatif. Populasi dalam penelitian ini adalah perusahaan perbankan yang terdaftar di Bursa Efek Indonesia dari tahun 2016 sampai dengan tahun 2019. Sampel ditentukan dengan teknik purposive sampling dan diperoleh 88 sampel penelitian berdasarkan kriteria tertentu. Metode analisis yang digunakan adalah analisis statistik deskriptif, analisis regresi data panel, uji asumsi klasik, dan analisis regresi linier berganda. j g g Hasil dan Temuan Penelitian. Hasil penelitian menunjukkan bahwa kepemilikan manajerial dan kepemilikan asing berpengaruh negatif tidak signifikan terhadap nilai perusahaan dan kepemilikan keluarga berpengaruh positif signifikan terhadap nilai perusahaan. THE IMPACT OF OWNERSHIP STRUCTURE ON FIRM VALUE IN THE BANKING SECTOR COMPANIES Siti Paras Ubuddiyah Kenny Ardillah * Institut Teknologi dan Bisnis Kalbis, Jln. Pulomas Selatan Kav. 22, Jakarta, Indonesia *l20617@lecturer.kalbis.ac.id A R T I C L E I N F O Article history: Received August 29, 2022 Revised December 14, 2022 Accepted January 5, 2023 A B S T R A C T Research Purpose. This research aims to determine the effect of managerial ownership, foreign ownership, and family ownership on firm value. Research Method. This research uses quantitative methods. The population in this study is banking companies listed on the Indonesia Stock Exchange from 2016 to 2019. The sample was determined by purposive sampling technique and obtained 88 research samples based on specific criteria. The analytical method used is descriptive statistical analysis, panel data regression analysis, classical assumption test, and multiple linear regression analysis. ownership, foreign ownership, and family ownership on firm value. Research Method. This research uses quantitative methods. The population in this study is banking companies listed on the Indonesia Stock Exchange from 2016 to 2019. The sample was determined by purposive sampling technique and obtained 88 research samples based on specific criteria. The analytical method used is descriptive statistical analysis, panel data regression analysis, classical assumption test, and multiple linear regression analysis. Key words: Firm Value; Managerial Ownership; Foreign Ownership; Family Ownership DOI: https://doi.org/10.33508/jako.v15i1.4033 g y Research Result and Findings. The results showed that managerial ownership and foreign ownership had an insignificant negative effect on firm value and family ownership had a significantly positive effect on firm value. INTRODUCTION As owners of the company, family members will try to maintain the company's viability so that it can be passed on to the next generation (Astuti & Muna, 2017). Family ownership causes unclear roles and responsibilities among family members, which can cause problems in the contractual relationship between the principal and the company agent (Arum & Darsono, 2020). Family ownership in a company can increase operations and reduces managerial takeovers, so it will increase the firm value. The research results conducted by Hersugondo (2018) prove that managerial ownership has no significant effect on firm value. Meanwhile, Ballo (2020) stated that managerial ownership negatively influences the firm value. It is different from the results of the study by Dewayanto & Riduwan (2020), Soewarno & Ramadhan (2020), Mappadang (2021), and Bernardin & Karina (2021), which suggest that managerial ownership has a significant positive effect on firm value. Buttang (2020) stated that foreign ownership has a negative and significant effect on firm value. On the other hand, research conducted by Hersugondo (2018), Kao et al. (2018), Fitri et al. (2019), Polii & Herawaty (2020), Natali & Herawaty (2020), and Soewarno & Ramadhan (2020) suggests that foreign ownership has a significant positive effect on firm value. Arum & Darsono's research (2020) explains that family ownership does not affect firm value. In comparison, Astuti & Muna (2017), Kao et al. (2018), Malelak et al. (2020), and Najahiyah et al. (2022) stated that family ownership positively affects firm value. y ) The firm value by increasing company’s share price can become an indication of the company’s success rate in increasing investor investment confidence (Ardillah, 2018). Many indicators influence the increase in firm value, one of which is the company's ownership structure. The ownership structure is the various ownership levels held by internal and external investors (Widanastiti & Rahayu, 2020). The ownership structure is considered to influence organizational activities, which will thus affect the organization's presentation to maximize firm value. An improved ownership structure can make administrative capacity work better and make boards more cautious about obtaining and overseeing credit (liabilities) because an increased size of liabilities will lead to financial difficulties (Sudana, 2011). The justification for choosing bank firms in this study is that the banking business sector is one of the industries that can be relied on to be developed in the future. INTRODUCTION Indonesia is a country with strong capabilities in national and international competition. The geographical location with natural energy and a growing population of young people has laid a solid foundation for Indonesia's macroeconomic development. In 2019, Indonesia experienced normal economic development of around 5%. The Central Statistics Agency has released information on Indonesia's economic development in 2020 with the result of 5.32% which was 0.3% higher than 5.02% in 2019. Indonesia's economic development is caused by several aspects that are in line with the country's development goals in the face of globalization. Various sectors support Indonesia's economic development, and each industry has its portion. profit, prosper shareholders, or achieve long-term performance (Ardillah & Chandra, 2021; Sinaga, 2021). Usually, funders will trust the company if it has a high firm value (Savitri & Ramantha, 2019). The following is the average firm value data for all banking companies listed on the Indonesia Stock Exchange, as shown in Figure 1. Figure 1. Firm Value of Banking Companies 0,00 2,00 4,00 2016 2017 2018 2019 PBV (Times) PBV of Banking Companies (2016-2019) Figure 1. Firm Value of Banking Companies The firm value can be considered by funders related to their goal contribute to maximize the 25 THE IMPACT OF OWNERSHIP STRUCTURE ON FIRM VALUE IN THE BANKING SECTOR COMPANIES Oleh: UBUDDIYAH, dan ARDILLAH Based on figure 1, shows that the Price to Book Value as a firm's value indicator in 2016 was 2.82, and in 2017 was 2.88. There was an increase of 0.06 times greater than the book value. From 2018 to 2019, there was a very significant decrease in firm value. The decline in the firm value of banking sector companies was triggered by various elements, one of which was the share price of the bank. An assessment of the strength of a bank can be a sign for financial statement users in making investment choices (Anggarsini & Suprasto, 2018). Good performance of a company can attract premiums and trust all parties to reach the normative goal to maximize the value of the company (Ardillah et al., 2022). Investors will react to the health of the top-notch bank by expanding the cost of shares which will affect the development of corporate rewards (Maheswari & Suryanawa, 2016). Most Indonesian companies are family- owned. INTRODUCTION The daily activities of the Indonesian people cannot grow if they are separated from banking services where banking companies make a significant contribution to state revenue by looking at the company's financial situation (Savitri & Ramantha, 2019). The conflicting consequences of previous studies make researchers need to rethink the result of the test hypothesis. This research was replicated by Buttang (2020) with additional independent variables to form ownership structures such as managerial ownership and family ownership. This research aims to prove the effect of The companies are trying their best to achieve their goals by providing opportunities for management to own shares by equalize the interests of managers and shareholders and the firm value (Susanti & Mildawati, 2018). Managers will be motivated to improve the company's performance which will also impact to increase the in stock prices (Hersugondo, 2018). Financial statement users can pressure managers and ask them to provide supervision. Foreign investors can provide new capital and recruit ready-to-work heads and help companies closest to them to register in the global market, thereby reducing the cost of raising capital (Soewarno & Ramadhan, 2020). 26 JURNAL AKUNTANSI KONTEMPORER (JAKO) – VOL 15 NO 1 – JANUARI 2023 – Halaman 25 - 36 value. If the company is to be sold, it will be reflected in the price of the company's outstanding shares. managerial ownership, foreign ownership, and family ownership on the firm value of banking sector companies. Managerial Ownership Managerial ownership is an ownership structure estimated based on the level of the value claimed by the commissioners, directors, and management (Adi et al., 2015). Managerial ownership is administrative investors (directors and commissioners) who function in organizational dynamics. Managerial ownership consists of company shares claimed by an internal meeting or board (Pramesti & Budiasih, 2017). Foreign Ownership Based on article 1, paragraph 3 of Law Number 25 of 2007 a speculation exercise of foreign ownership has been concerning as efforts to maintain the business in the Republic of Indonesia. Foreign investors use sole proprietorship or joint activities with domestic investments. Foreign share ownership is the number of company shares claimed by outsiders (Hermiyetti & Erlinda, 2016). Most shared ownership is foreign ownership because investors do not have the motivational power and capacity to manage management, so implementing investor supervision over control is weak. Ownership Structure The ownership structure is defined as various forms of ownership or the proportion of equity held by internal and external shareholders. The mechanism of ownership structure is to reduce conflict between management and shareholders (Natali & Herawaty, 2020). The ownership structure has a qualitative effect that refers to the nature of ownership (Ardillah, 2021). The ownership structure of the company is represented by directors (agents) appointed by shareholders (principals) in carrying out its activities. The ownership structure is considered to have the ability to influence the company's operations to maximize the company’s value. The ownership structure is one step to reducing agency costs because shareholders and management both have their proportions in the company and will minimize agency conflicts due to a clear division (Arwani et al., 2020) Literature Review Agency Theory The principals expect agents to act following the principal's interests. By hiring an agent, the principal versus the agent can maximize the value of utility wealth by developing high investment capital. In some cases, however, both the principal and the agent need to increase their respective utility so that it cannot be trusted that the agent will consistently act on a legitimate concern for the rules (Godfrey et al., 2010). The agency theory is related to selfishness above the interests of others. In this situation, the manager as an agent is more concerned with enrichment than realizing the principal's interests (Kurniansyah et al., 2018). p p ( y ) If identified with this research, agency theory can be used as a rationale to support ownership structure for strategic policies, so that company expansion is valuable and growing. Managerial ownership can be a way to minimize agency conflicts within the company by aligning the interests of management with the company's shareholders (Effendi, 2016). It is also more likely that corporations with foreign investors will adhere to good governance and corporate practices, have more diverse holdings, and have improved monitoring capacities to reduce the asymmetries through enhanced information levels (Jackson & Strange, 2008). The family control reduces agency problems between managers and shareholders but creates conflict between controlling families and minority shareholders if the protection of minority shares is weak in terms of family ownership, the majority shareholder can expropriate the company’s wealth from the minority shareholder (Richardson & Leung, 2011). Firm Value The firm value is beneficial because the higher the company's value, the higher the value of investors. The firm value can be measured by the stock price, where its development can be seen from the stock price on the stock exchange (Indrarini, 2019). If the stock price increases, the firm value will also increase (Buttang, 2020). Based on the definition presented, the researcher assumes that the company's value is the selling Family Ownership Family Ownership Family ownership is a company where most of the common shares are owned by actively 27 THE IMPACT OF OWNERSHIP STRUCTURE ON FIRM VALUE IN THE BANKING SECTOR COMPANIES Oleh: UBUDDIYAH, dan ARDILLAH Ha1: Managerial ownership has a positive effect on firm value. Ha1: Managerial ownership has a positive effect on firm value. Ha1: Managerial ownership has a positive effect on firm value. participating family members. The owners of the family business are the major shareholders, and their interests will take precedence over the interests of the minor shareholders (Yustisi & Putri, 2020). The largest controlling shareholder of the company is an individual with certain controlling rights. In that case, the listed company is classified as a family-controlled company establishing family-based ownership with the surname and whether there is a marital relationship. Relationship Between Foreign Ownership and Firm Value Foreign ownership refers to the proportion of the company's common stock claimed by outsiders such as the law, the government, and some companies or non-foreign persons such as legal entities and the government (Buttang, 2020). Foreign ownership would positively affect the company's value because the higher the share of foreign ownership in foreign meetings, the greater the possibility of an increase in the firm value (Soewarno & Ramadhan, 2020). If this happens, it can form a better administrative component, affect the lack of interest, and empathically influence the company's operations to increase its bidding costs or increase its price overall. Hersugondo (2018), Kao et al. (2018), Fitri et al. (2019), Polii & Herawaty (2020), Natali & Herawaty (2020), and Soewarno & Ramadhan (2020) suggest that foreign ownership has a significant positive effect on firm value. To facilitate understanding of this research, the author provides a research framework presented in Figure 2 to prove the effect of ownership structure based on managerial ownership, foreign ownership, and family ownership on firm value. H1 H2 H3 Figure 2. Research Conceptual Model Managerial Ownership (X1) Firm Value (Y) Foreign Ownership (X2) Family Ownership (X3) Managerial Ownership (X1) H2 Firm Value (Y) Ha2: Foreign ownership has a positive effect on firm value. Family Ownership (X3) Relationship Between Family Ownership and Firm Value Relationship Between Family Ownership and Firm Value Figure 2. Research Conceptual Model Family firms were derived from companies has significant ownership with family members (Astuti & Muna, 2017). Family Ownership Family ownership will increase firm value because with the family member governing the company’s ownership can further develop the supervision of the executives and make the right choices for the resilience of the company. It is also because it has more critical inspiration related to the firm value associated with family abundance. The family ownership structure may place family members in top management positions in the company and other strategic parts (Natali & Herawaty, 2020). Astuti & Muna (2017), Kao et al. (2018), Malelak et al. (2020), and Najahiyah et al. (2022) stated that family ownership positively affects firm value. Ha3: Family ownership has a positive effect on Hypothesis Development Relationship Between Managerial Ownership and Firm Value Hypothesis Development Relationship Between Managerial Ownership and Firm Value The existence of managerial interests will lead someone to conclude that the firm value is increasing due to an increase in managerial ownership. Many managers' holdings will allow managers to monitor company activities effectively (Suryaningsih et al., 2018). In the company, the greater the level of managerial ownership, the more management will maximize its performance for the benefit of investors and themselves. This managerial ownership helps unite interests so that both can benefit directly from the decision and share some of the consequences (Susanti & Mildawati, 2018). Bernardin & Karina (2021), Dewayanto & Riduwan (2020), Mappadang (2021), and Soewarno & Ramadhan (2020) suggest that managerial ownership has a significant positive effect on firm value. Ha3: Family ownership has a positive effect on firm value. Operational Definition and Variables Measurement Firm value is a value that refers to the market value of a company that is sold in a state of operation (Rahmatia, 2015). This study measured the firm value using Price to Book Value. This measurement refers to research conducted by Hersugondo (2018) and Soewarno & Ramadhan (2020), which uses it to measure value. The higher the Price to Book Value, the higher the company's stock price conversely (Brigham et al., 2010). Price to Book Value can be formulated as follows. Foreign Ownership = Foreign Share Ownership Number of Outstanding Shares 𝑥 100% …..(3) Price to Book Value = Price Per Share Book Value Per Share ………..(1) Family ownership is all individuals or companies with family shareholders. The researcher calculated family ownership using dummy variables with a score of 0 if the company didn’t have family shares ownership and a score of 1 if the company had family shares ownership. This study measured the independent variable of family ownership using the nominal scale (Arum & Darsono, 2020). This measurement refers to research conducted by Astuti & Muna (2017). Managerial ownership is the ratio of shares owned by company management (directors, commissioners, and managers) to shares held by outside parties (public, government, or other companies) (Hermawan, 2018). In this study, managerial ownership is measured using a ratio scale. This measurement refers to research conducted by Bernardin & Karina (2021), Dewayanto & Riduwan (2020), and Putra & Putra (2021). Managerial ownership can be systematically formulated as follows (Hermawan, 2018). Managerial Ownership = Managerial Share Ownership Number of Shares Outstanding 𝑥 100% ..(2) Population and Sample The population in this study is public banking companies listed on the Indonesia Stock Exchange (IDX). The following are the considerations used in determining the 28 JURNAL AKUNTANSI KONTEMPORER (JAKO) – VOL 15 NO 1 – JANUARI 2023 – Halaman 25 - 36 shares ownership during research period (4) banking companies that had complete data needed in research during the research period. Each of the above criteria or considerations is based on a reason related to the research topic. exploratory test to determine the number of research samples, including (1) listed banking companies in the Indonesia Stock Exchange for the 2016-2019 period, (2) banking companies that had managerial shares ownership during the research period, (3) banking companies that had foreign Table 1. Results of Sample Selection F i hi i f h di Number Criteria Description Amount 1. Banking companies listed on the Indonesia Stock Exchange in the 2016-2019 period 44 2. Banking companies that don’t have managerial shares ownership during the research period (15) 3. Banking companies that don’t have foreign shares ownership during the research period (3) 4. Banking companies that do not have complete data needed for research (4) Number of Research Samples 22 Total Research Data (22 x 4 periods) 88 Table 1. Results of Sample Selection Table 1. Results of Sample Selection Foreign ownership is part of the outstanding shares of foreign investors in companies owned by individuals, legal entities, and governments. Their role is identified as foreign (Buttang, 2020). In this study, the independent variable of foreign ownership was measured using a ratio scale. This measurement refers to research conducted by Polii & Herawaty (2020), Natali & Herawaty (2020), and Budiman et al. (2021). Foreign ownership can be formulated systematically as follows (Farooque et al., 2007 in Buttang, 2020). Results Family ownership (FAMILY) is the third independent variable measured using the family ownership ratio. Family ownership has a minimum value of 0.000, which describes no family ownership in a company, while the Bank Oke Indonesia Tbk owns a maximum value of 0.672 in the 2016 and 2017 periods. The average value (mean) for family ownership is 0.031, indicating ownership of the company's shares by the family is only about 3.1% of the company's total shareholding. It means that the level of family share ownership is still low. The standard deviation value for family ownership is 0.110 which is higher than the average value of 0.031, which means the range of variation in family ownership data is high. Descriptive Statistical Analysis Table 2 shows the descriptive statistics result of the 88 research samples for each variable. Table 2. Descriptive Statistical Analysis Results FV MO Foreign Family Mean 4.598 0.008 0.227 0.031 Max. 39.487 0.126 0.982 0.672 Min. 1.107 0.000 0.000 0.000 Std.Dev. 6.185 0.026 0.275 0.110 The firm value has an average value of 4.598, meaning the Price to Book Value is very averagely high for the banking industry sector. The firm value has a maximum value of 39.487 from Bank Syariah Indonesia Tbk in 2016. The firm value has a minimum value of 1.107 from Bank Maybank Indonesia Tbk in 2019. The standard deviation of firm value is 6.185 which is higher than the average value of 4.598, which means the range of variation in firm value data is high. Data Analysis Method The analytical method used in this research is descriptive statistical analysis, panel data regression analysis, estimation model selection, data quality test, and hypothesis testing. The descriptive statistical analysis explains various characteristics of the data in one sample. The data quality test in this study uses the classical assumption test, which includes the normality 29 THE IMPACT OF OWNERSHIP STRUCTURE ON FIRM VALUE IN THE BANKING SECTOR COMPANIES Oleh: UBUDDIYAH, dan ARDILLAH while the maximum value of 0.982 (98.2 %) was owned by Bank Mandiri (Persero) Tbk in the 2017 period. The average value (mean) for foreign ownership is 0.227, meaning that foreign ownership of the company's shares is only around 22.7 percent of the company's total shares. It means that the level of foreign share ownership is still high. The standard deviation value for foreign ownership is 0.275 which is higher than the average value of 0.227, which means the range of variation in foreign ownership data is high. test, multicollinearity test, and autocorrelation test. Panel data regression and estimation model selection had to be analyzed in this study to reform panel data linear regression analysis including correlation coefficient, coefficient of determination tests, and partial tests. The panel data regression equation can be formulated as follows. PBV = a + b1 MANON + b2FORON + b3FAMON ……..(4) Classic Assumption Test The probability value in the normality test is 0.295933, where the probability value is greater than 0.05, so 0.295933 > 0.05, then the data is normally distributed. The correlation value of managerial ownership, foreign ownership, and family ownership is less than 0.80. Thus, the data in this study can be identified that there is no multicollinearity problem between the correlation between managerial ownership, foreign ownership, and family ownership. The DW value of the formed regression equation is 1.826150. In contrast, the Durbin-Watson table value with n=88 and k=3, then the values of dL=1.5836 and dU=1.7243, so the value of 4-dU = 4 –1.7243 = 2.2757, which lies between 1.7243 (dU) < 1.826150 (DW) <2.2757 (4-dU) then it is concluded that the DW value of the regression model that is formed has no autocorrelation. g Managerial Ownership (MO) is measured using the managerial ownership ratio. The managerial ownership has a minimum value of 0.000, which describes no share ownership by management in a company. The managerial ownership has a maximum value of 0.126 from Bank Capital Indonesia Tbk in the research period. The average value (mean) for managerial ownership is 0.008, meaning that share ownership of the company by the management is only about 0.8 % of the total share ownership of the company. It means that the level of managerial share ownership is still low. The standard deviation value of managerial ownership is 0.026 which is higher than the average value of 0.008, which means the range of variation in managerial ownership data is high. Estimation Model Selection Chow Test Foreign Ownership (FOREIGN) is measured using the foreign ownership ratio. Foreign ownership has a minimum value of 0.000, which describes no foreign ownership in a company, The following are the test results based on the Chow test shown in table 3. 30 JURNAL AKUNTANSI KONTEMPORER (JAKO) – VOL 15 NO 1 – JANUARI 2023 – Halaman 25 - 36 Hausman Test The following are the test results based on the Hausman test shown in table 4. Table 4. Hausman Test Results Test Summary Chi-Sq. Statistic Prob. Cross-section random 4.059737 0.2551 Partial Significance Test The following are the results of partial hypothesis testing using panel data regression analysis using the common effects model in table 7. The random cross-section probability value is 0.2551 > 0.05, so H0 is accepted. It means that the Hausman Test shows that the Random Effect Model (REM) was selected. Table 7. Partial Test (Hypothesis Test) Results Variables Coefficient Prob. C 0.015813 0.4750 MO -0.005996 0.8024 FOREIGN 0.003857 0.2990 FAMILY 0.631542 0.0000 Lagrange Multiplier Test The following are the test results based on the Lagrange multiplier test shown in table 5. Table 5. Lagrange Multiplier Test Results Cross- section Test Hypothes is Time Both Breusch- Pagan 2.8724 (0.0901) 2.6951 (0.1007) 5.5676 (0.0183) The panel data regression equation can be formulated as follows. 𝑃𝐵𝑉= 0,015 −0,005𝑀𝐴𝑁𝑂𝑁+ 0,003𝐹𝑂𝑅𝑂𝑁+ 0,631𝐹𝐴𝑀𝑂𝑁 …………………. (5) The panel data regression equation can be formulated as follows. The Breusch-Pagan probability value is 0.0901 > 0.05, so it can be concluded that H0 is accepted. It shows the Common Effect Model (CEM) chosen from the Lagrange Multiplier (LM) test. With a degree of freedom of 0.05, a probability value of 0.8024 is obtained. Compared with 0.8024 > 0.05, the probability is greater than the significance level. Ha1 is rejected, meaning that managerial Ownership (X1) partially does not affect firm value (Y). Foreign Ownership (X2) has a significant level of 0.2990. Compared with 0.2990 > 0.05, the probability value is greater than the absolute level. It means that Ha2 is rejected, meaning that foreign Ownership (X2) does not partially affect the firm value (Y). Family ownership (X3) has a significant level of 0.0000. Compared with 0.0000 < 0.05, the probability value is smaller than the absolute level. It means that Ha3 is accepted, meaning that family ownership (X3) partially affects firm value (Y). Table 6. Test Results of Multiple Correlation Coefficient & Coefficient of Determination R-squared 0.270606 Adjusted R-squared 0.244556 Table 6. Test Results of Multiple Correlation Coefficient & Coefficient of Determination R-squared 0.270606 Adjusted R-squared 0.244556 The chi-square probability value is 0.9326 > 0.05, so H0 is accepted because the results of the Prob Cross-section F> from alpha. It means the model used in this study is the Common Effect Model. The R-square value is 0.270606 = 0.520198, indicating that managerial ownership, foreign ownership, and family ownership on firm value have a moderate relationship. The value of the coefficient of determination test based on R- squared is 0.270606, meaning that managerial ownership, foreign ownership, and family ownership contribute 27.06% to firm value. Multiple Correlation Coefficient Test Table 3. Chow Test Results Effects Test Statistic Prob. Cross-section F 0.555554 0.9326 Cross-section Chi-quare 14.951084 0.8254 The following are the results of hypothesis testing using panel data regression analysis using the common effects model in table 6. Discussion The Effect of Managerial Ownership on Firm Value Based on the probability results, it can be obtained that the probability of managerial ownership (X1) 0.8024 is greater than the significance level of 0.05, so Ha3 is rejected. That means managerial Ownership (X1) partially does not affect firm value (Y). The study results have no effect due to the low proportion of managerial ownership. The results of this study support the result of research from Ballo (2020) and Azizah et al. (2021). The managerial ownership did not affect firm value because few management parties owned company shares. The low number of managerial ownership causes the management to be more concerned with their interests than the company's interests according to agency theory. The amount of share ownership that has not been significant causes managers to be more concerned with their goals as a manager than as a shareholder (Ballo, 2020; Azizah et al., 2021). This study is not in line with the research conducted by Dewayanto & Riduwan (2020), Soewarno & Ramadhan (2020), Mappadang (2021), and Bernardin & Karina (2021). ( ) According to agency theory, the more significant the portion of foreign ownership, the more it is necessary to establish a better supervisory mechanism that impacts reducing conflicts of interest. The company can make a positive impact on company performance by increasing the company's share price directly. The weakness of partners from local investors makes foreign investors strong. It tends to encourage changes in the direction of share prices following foreign share ownership, resulting in a decrease in the firm value. When the company's stock price is high, foreign investors will tend to invest in companies with high stock prices, and the company's high value makes foreign investors confident about the company's future. The greater the foreign ownership, the more shareholders, cannot freely manage the company. It may result in a conflict of interest between the foreign shareholder and the controlling or majority shareholder. The management control attention of foreign investors will be relatively low. Foreign investors are not too exposed to the risks that will be accepted later. The percentage of company shares they own is small enough to affect management decision-making. So, indirectly, foreign shareholders do not monitor the company's performance and share price, which will reflect the company's value. Model Conclusion Based on the result of panel data regression testing from the three-panel data models, it is recommended to use the Common Effect Model (CEM). 31 THE IMPACT OF OWNERSHIP STRUCTURE ON FIRM VALUE IN THE BANKING SECTOR COMPANIES Oleh: UBUDDIYAH, dan ARDILLAH 0.05, so Ha2 is rejected. That means foreign Ownership (X2) partially does not affect the firm value (Y). The results of this study do not support the result of Buttang's (2020) and Soewarno & Ramadhan (2020) research. This result is inconsistent with the theory from Soewarno & Ramadhan (2020) which suggested that the more significant the portion of foreign ownership in foreign parties, the greater the possibility of an increase in firm value because foreign ownership can form a monitoring mechanism, which has an impact on reducing conflicts of interest and has a positive impact on firm performance so that it can increase the company's stock price. On the other hand, the results of this study are in line with research conducted by Hersugondo (2018), Kao et al. (2018), Fitri et al. (2019), Polii & Herawaty (2020), Natali & Herawaty (2020), and Soewarno & Ramadhan (2020). Discussion The Effect of Managerial Ownership on Firm Value Discussion ( ) The results of this study confirm that the proportion of managerial ownership didn’t make the management to give the effort to maximize its performance for the benefit of shareholders to increase firm value. These results cannot prove that giving the company’s shares to management will be able to increase the management’s effort to maximize the firm value. The ownership structure is believed to influence the company's running, which affects the company's performance in achieving company goals. It is due to the control they have. With this motivation, managers will try to maximize the firm value. The possibility of this happening is because when managerial ownership is too low can’t cause the management to be able to manage the company optimally. Because the company does not have a significant enough role in deciding on the company, it does not affect the level of firm value. The management, which directly impacts every decision they take because they are the company's owners, cannot show the similarity of interests between management and shareholders. REFERENCES Adi, P. S., Nasir, A., & Rusli. (2015). Pengaruh Profitabilitas, Risiko Keuangan, Nilai Perusahaan, Kepemilikan Manajerial, dan Dividend Payout Ratio Terhadap Praktik Perataan Laba (Studi Empiris pada Perusahaan Real Estate dan Properti yang Terdaftar di Bursa Efek Indonesia Tahun 2011-2013). JOM Fekon, 2(1), 1–15. ) ( ) Anggarsini, N. W., & Suprasto, H. B. (2018). Pengaruh Tingkat Kesehatan Bank dan Corporate Social Responsibility pada Nilai Perusahaan Perbankan Fakultas Ekonomi dan Bisnis Universitas Udayana (Unud), E- Jurnal Akuntansi Universitas Udayana, 25(2), 1308–1338. https://doi.org/10.24843/EJA.2018.v25.i02. p19. Ardillah, K. (2018). Effect of Intellectual Capital, Corporate Sustainability Disclosure, and Corporate Governance To The Value of The Company: Empirical Study on Registered SOE in Indonesia Stock Exchange. Jurnal Ilmu Manajemen dan Ekonomika, 10(2), 60-67. The Effect of Foreign Ownership on Firm Value The Effect of Foreign Ownership on Firm Value The Effect of Family Ownership on Company Value ff f g p Based on the probability results, it can be obtained that the probability of foreign ownership (X2) 0.2990 is greater than the significance level of ff f y p p y Based on the probability results, it can be obtained that the probability of family ownership (X3) 0.0000 is less than the significance level of 0.05, 32 JURNAL AKUNTANSI KONTEMPORER (JAKO) – VOL 15 NO 1 – JANUARI 2023 – Halaman 25 - 36 so Ha3 is accepted. Based on the positive coefficient result, it can be concluded that family ownership (X3) affect positively firm value (Y). The results of this study support the research of Astuti & Muna (2017), Kao et al. (2018), Malelak et al. (2020), and Najahiyah et al. (2022). The results of this study are not in line with research conducted by Arum & Darsono (2020) and in line with agency theory, which indicates that the higher the share ownership by the family in a company, the higher the company contributes to its firm value. These results indicate that family ownership will increase the firm value because a family that dominates the company's ownership will increase supervision over management and make the right decisions for the company's survival (Natali & Herawaty, 2020). The families that had greater motivation to make decisions about the value of companies were affiliated with the wealth of families. Thus, when the shares owned by the family increase, it will increase the firm value. Companies with family-owned shares that participate in controlling the company tend to act in the interest of increasing the value of the company. Because the wealth owned by the family is in the company, the dominant ownership held by the family can exercise control and have a great influence on company decisions (Astuti & Muna, 2017). ownership structures such as ownership concentration and public ownership. Future researchers can conduct a larger number of samples by extending the duration of the sample research to five years period to strengthen the results of previous studies. CONCLUSION The researchers got the following conclusions. Managerial Ownership does not affect firm value. It indicates that when the shares owned by the company's management decrease, the firm value will also decrease. The results of this study are in line with research from Ballo (2020), which shows that a decline in managerial ownership has a negative effect on firm value. Foreign ownership does not affect firm value. It indicates that when the shares owned by foreigners, both individuals and entities, decrease, the company's value will also decrease. Family ownership has a positive effect on firm value. It indicates that when the shares owned by the family increase, it will increase the firm value. Ardillah, K. (2021). How Do Social Contribution Value and Ownership Structure Influence Corporate Sustainable Growth in State- Owned Companies in Indonesia?. Journal of Accounting and Strategic Finance, 4(2), 172-190. https://doi.org/10.33005/jasf.v4i2.158. Ardillah, K., & Chandra, Z. (2021). Corporate Environmental Disclosure, Environmental Performance, and Corporate Governance Structures on Firm Value. Ultima Accounting, 13(2), 334-351. https://doi.org/10.31937/akuntansi.v13i2.2 407. Ardillah, K., Breliastiti, R., Setiawan, T., & Machdar, N. M. (2022). The Role of Ownership Structure in Moderating The Relationship Between Tax Avoidance, Corporate Social Responsibility Disclosure, and Firm Value. Accounting Analysis Journal, 11(1), 21-30. https://doi.org/10.15294/aaj.v11i1.58613. These independent variables as ownership structures indicators used in this research only use three variables such as managerial ownership, family ownership, and foreign ownership. This research only used 88 research samples from four years period from 2016-2019. Future researchers should conduct similar research by including additional independent variables to indicate 33 THE IMPACT OF OWNERSHIP STRUCTURE ON FIRM VALUE IN THE BANKING SECTOR COMPANIES Oleh: UBUDDIYAH, dan ARDILLAH Arum, D. N. S., & Darsono. (2020). Pengaruh Tata Kelola Perusahaan, Kepemilikan Keluarga, Kepemilikan Institusional, dan Kualitas Pelaporan Terhadap Nilai Perusahaan. Diponegoro Journal of Accounting, 9(4), 1–8. Dewayanto, M. A. R., & Riduwan, A. (2020). Pengaruh Struktur Kepemilikan pada Nilai Perusahaan : Kebijakan Dividen Sebagai Variabel Mediasi. Jurnal Ilmu dan Riset Akuntansi, 9(5), 1–20. Effendi, M. A. (2016). The Power of Corporate Governance: Teori dan Implementasi. Jakarta: Salemba Empat. Arwani, A., Ramadhan, M. N., & Restiara, V. (2020). Kepemilikan Manajerial dalam Agency Theory. Didapat dari http://repository.iainpekalongan.ac.id/id/e print/269. Fitri, E. R., Savitri, E., & L. A. A. (2019). Influence of Foreign Ownership, Ownership Concentrated, and Environmental Disclosure to Firm Value. Indonesian Journal of Economics, Social, and Humanities, 1(2), 91-96. https://doi.org/10.31258/ijesh.1.2.91-96. Astuti, A. D., & Muna, A. (2017). CONCLUSION L. (2015). Pengaruh Kepemilikan Manajerial terhadap Nilai Perusahaan: Tanggung Jawab Sosial sebagai Variabel Moderating. Jurnal Ilmu dan Riset Akuntansi, 4(3), 1-15. Maheswari, I. G. A. G., & Suryanawa, I. K. (2016). Pengaruh Tingkat Kesehatan Bank dan Ukuran Bank Terhadap Nilai Perusahaan. E- Jurnal Akuntansi Universitas Udayana, 16(2), 1319–1346. Malelak, M. I., Soehono, C., & Eunike, C. (2020). Corporate Governance, Family Ownership, and Firm Value: Indonesia Evidence. SHS Web Conference, 76, 1-7. https://doi.org/10.1051/shsconf/202076010 27. Richardson, G., & Leung. S. (2011). Family Ownership Control and Earnings Management: Evidence from Hongkong Firms. Corporate Ownership and Control, 8(4), 96-111. https://doi.org/10.22495/cocv8i4p6. Savitri, K. K., & Ramantha, I. W. (2019). Pengaruh Komponen Risk Based Bank Rating pada Nilai Perusahaan Perbankan. E-Jurnal Akuntansi, 29(2), 883-898. https://doi.org/10.24843/EJA.2019.v29.i02. p29. Mappadang, A. (2021). Managerial Ownership, Leverage, Profitability, Corporate Value: An Interactive Effect In Indonesia Stock Exchange. Widyakala Journal, 8(2), 54-60. https://doi.org/10.36262/widyakala.v8i2.44 3. Sinaga, E. M. (2021). Pengaruh CSR terhadap Nilai Perusahaan dengan Profitabilitas sebagai Variabel Moderating (Perusahaan Aneka Industri Sektor Tekstil dan Garmen Terdaftar di BEI 2016-2018). Jurnal Manajemen Akuntansi, 1(2), 210-216. https://doi.org/10.36987/jumsi.v1i2.2149. Najahiyah, N., Aisjah, S., & Dzajuli, A. (2022). The Effect of Family Ownership on Company Values Mediated by Financial performance and Corporate Governance. International Journal of Research in Business and Social Science, 11(6), 148–156. https://doi.org/10.20525/ijrbs.v11i6.1923. Soewarno, N., & Ramadhan, A. H. A. (2020). The Effect of Ownership Structure and Intellectual Capital on Firm Value with Firm Performance As An Intervening Variable. International Journal of Innovation, Creativity, and Change, 10(12), 215–236. https://www.ijicc.net/images/vol10iss12/1 01217_Soewarno_2020_E_R.pdf Natali, G. R. K., & Herawaty, V. (2020). Faktor yang Mempengaruhi Nilai Perusahaan dengan Kepemilikan Asing Sebagai Moderasi: Studi Kasus pada Perusahaan Manufaktur yang Terdaftar Di IDX Periode 2016-2018. Prosiding. Seminar Nasional Pakar Ke 3 Tahun 2020: Jakarta. Polii, Y., & Herawaty, V. (2020). Pengaruh Profitabilitas, Pembagian Dividen, Kebijakan CSR, dan Kepemilikan Asing terhdap Nilai Perusahaan dengan Stabilitas Keuangan Sebagai Variabel Moderasi (Study Empiris Perusahaan Property, Real-Estate, dan Construction yang Terdaftar Di BEI Tahun 2018. Jurnal Penelitian dan Karya Ilmiah Lembaga Penelitian Universitas Trisakti, 5(2), 79-98. https://doi.org/10.25105/pdk.v5i2.7363. p Sudana, I. M. (2011). Manajemen Keuangan Perusahaan Teori dan Praktik. Jakarta: Erlangga. gg Suryaningsih, I., Andini, R., & Oemar, A. (2018). Pengaruh Kepemilikan Manajemen, Ukuran Dewan Komisaris dan Kebijakan Dividen Terhadap Nilai Perusahaan Dengan Corporate Social Responsibility Sebagai Variabel Moderating (Studi Kasus pada Perusahaan Industri Perdagangan Barang dan Konsumsi di BEI Periode Tahun 2012- 2016). Jurnal of Accounting, 4(4), 1–20. Pramesti, I. A. J., & Budiasih, I. CONCLUSION Pengaruh Kepemilikan Keluarga & Efektivitas Dewan Komisaris terhadap Nilai Perusahaan yang Tergabung dalam LQ45 tahun 2012-2013. Jurnal Riset Keuangan dan Akuntansi, 3(1), 17– 28. https://doi.org/10.25134/jrka.v3i1.460. Godfrey, J., Hodgson, A., Tarca, A., Hamilton, J., & Holmes, S. (2010). Accounting Theory. New York: John Wiley & Sons Australia, Ltd. Azizah, A. N., Setyaningrum, A., & Nurhayati, E. (2021). Pengaruh Kepemilikan Manajerial Terhadap Nilai Perusahaan dengan Enterprise Risk Management sebagai Variabel Moderating. Humanis, 1(2), 141–155. Hermawan, A. D. (2018). Pengaruh Kebijakan Dividen, Growth Opportunity, dan Struktur Modal terhadap Nilai Perusahaan dengan Kepemilikan Manajerial sebagai Variabel Intervening. Sosio E-Kons, 10(2), 131-142. http://dx.doi.org/10.30998/sosioekons.v10i 2.2690. Ballo, F. J. (2020). Pengaruh Profitabilitas terhadap Nilai Perusahaan dengan Pengungkapan Kepemilikan Manajerial sebagai Variabel Moderating (Studi Empiris pada Perusahaan Pertambangan yang terdaftar di Bursa Efek Indonesia Tahun 2012-2016). Jurnal Keuangan dan Perbankan, 8(1), 28– 40. Hermiyetti & Erlinda, K. (2016). Analisis Pengaruh Kepemilikan Manajerial, Kepemilikan Institusional, Kepemilikan Asing, dan Komite Audit terhadap Transfer Pricing. Media Riset Akuntansi, 6(1), 1–19. Bernardin, D. E. Y., & Karina, E. (2021). Nilai Perusahaan: Perencanaan Pajak dan Kepemilikan Manajerial serta Transparansi Perusahaan Sebagai Pemoderasi. Jurnal Financia, 2(1), 1–12. https://ejurnal.ars.ac.id/index.php/financi a/article/view/410. Hersugondo, H. (2018). Struktur Kepemilikan dan Nilai Perusahaan pada Perusahaan Jas Keuangan yang Terdaftar di BEI Tahun 2013- 2016. Prosiding. Seminar Nasional Multi Disiplin Ilmu dan Call of Papers Unisbank ke-4: Semarang. Indrarini, S. (2019). Nilai Perusahaan Melalui Kualitas Laba (Good Governance dan Kebijakan Perusahaan). Surabaya: Scopindo. Brigham, Eugene, F., & Houston, J, F. (2010). Dasar-Dasar Manajemen Keuangan (Edisi ke- 11). Jakarta: Salemba Empat. ) p Budiman, S. H., Randa, F., & Tongli, B. (2021). Pengaruh Struktur Kepemilikan Asing dan Kualitas Audit Terhadap Nilai Perusahaan Dengan Manajemen Laba Sebagai Mediasi. Jurnal Riset Akuntansi Aksioma, 20(1), 46–70. https://doi.org/10.29303/aksioma.v20i1.12 5. Jackson, G., & Strange, R. (2008). Corporate Governance and International Business: Strategy, Performance and Institutional Change. New York: Springer. Kao, M. F., Hodgkinson, L., & Jaafar, A. (2018). Ownership Structure, Board of Directors, and Firm Performance: Evidence From Taiwan. Corporate Governance, 19(1), 189-216. https://doi.org/10.1108/CG-04-2018-0144. Buttang, M. E. (2020). Pengaruh Struktur Kepemilikan Asing dan Kualitas Audit terhadap Nilai Perusahaan yang Dimediasi oleh Myopic Behaviour. Atma Jaya Accounting Research (AJAR), 03(02), 188–218. https://doi.org/10.35129/ajar.v3i02.129. Kurniansyah, Deddy, K. S., & Rizqi, F. A. (2018). Teori Agency dalam Pemikiran Organisasi: Pendekatan Positivist dan Principle-Agent. Jurnal Riset Akuntansi dan Bisnis Airlangga, 3(2), 435–446. 34 JURNAL AKUNTANSI KONTEMPORER (JAKO) – VOL 15 NO 1 – JANUARI 2023 – Halaman 25 - 36 Rahmatia, T. CONCLUSION G. A. N. (2017). Pengaruh Asimetri Informasi, Leverage, Kepemilikan Manajerial, dan Kepemilikan Institusional pada Manajemen Laba. E-Jurnal Akuntansi Universitas Udayana, 21(1), 200–226. Susanti, R., & Mildawati, T. (2018). Pengaruh Kepemilikan Manajemen, Kepemilikan Institusional, dan Corporate Social Responsibility Terhadap Nilai Perusahaan. Jurnal Ilmu dan Riset Akuntansi, 3(1), 1–17. Putra, F. & Putra, A. A. (2021). Pengaruh Struktur Modal dan Profitabilitas Terhadap Nilai Perusahaan. Jurnal Politeknik Caltex Riau, 14(1), 1–10. https://doi.org/10.35143/jakb.v14i1.4469. 35 THE IMPACT OF OWNERSHIP STRUCTURE ON FIRM VALUE IN THE BANKING SECTOR COMPANIES Oleh: UBUDDIYAH, dan ARDILLAH Widanastiti, S. A., & Rahayu, P. (2020). Pengaruh Struktur Modal, Struktur Aset, dan Struktur Kepemilikan Terhadap Risiko Keuangan pada Perusahaan Property dan Real Estate Yang Terdaftar di BEI Periode 2015-2017. Jurnal Cendekia Akuntansi, 1(1), 9–17. https://doi.org/10.32503/akuntansi.v1i1.10 61. Yustisi, Y. P., & Putri, D. P. (2020). Manajemen Laba yang Dipengaruhi oleh Managerial Overconfidence dan Kepemilikan Keluarga. Ekombis Review: Jurnal Ilmiah Ekonomi dan Bisnis, 9 (1), 121-134. https://doi.org/10.37676/ekombis.v9i1.127 1. 36
https://openalex.org/W4389558705	https://www.nepjol.info/index.php/NJR/article/download/59965/44849	English	null	Bronchial Artery Embolization: IS NBCA/LIPIODOL Better Than PVA?	Deleted Journal	2,023	cc-by	3,778	Cite this paper: Cite this paper: p p Thapa M, Gupta A, Yadav A, Gupta A. Bronchial Artery Embolization: IS NBCA/LIPIODOL Better Than PVA?. Nepalese Journal of Radiology 2023;13(2):9-14. http://doi.org/10.3126/njr.v13i2.59965 Methods: This retrospective study was conducted at a tertiary care center from January 2012 to December 2020. The final analysis was performed on 123 patients [NBCA(n= 37) and PVA(n= 86)]. Technical and clinical success rates, complications and recurrence rates were compared between the two groups. Introduction: Bronchial artery embolization (BAE) is a minimally invasive interventional procedure, which is now considered the first-line management strategy and an alternative to surgery for massive and recurrent haemoptysis. The advances in embolic agents have led to a significant improvement in the success rates of the procedure, however, there has been no significant change in the recurrence rate of haemoptysis. Results: A total of 248 arteries were embolized. In the PVA group, clinical success was achieved in 84 out of 86 cases (97.6%) and with NBCA in 36 out of 37 patients (97.3%) (p >0.05). Of the 120 patients in whom BAE was clinically successful, recurrence was observed in 43 patients within the 12-month follow-up period. The study showed a statistically significant association between the embolizing agent used for BAE and the recurrence of hemoptysis (χ2 = 4.80, df = 1, p = 0.028). The use of PVA particles for BAE was found to have 2.62 times higher odds (95% CI 1.10 - 6.81) of recurrence of hemoptysis as compared to the use of NBCA glue. Accepted: November 1, 2023 Accepted: November 1, 2023 ORIGINAL ARTICLE http://doi.org/10.3126/njr.v13i2.59965 1Department of Intervention Radiology, Nepal Army Institute of Health Sciences, Kathmandu, Nepal 2Department of Intervention Radiology, Sri Ganga Ram Hospital, New Delhi, India 1Department of Intervention Radiology, Nepal Army Institute of Health Sciences, Kathmandu, Nepal 2Department of Intervention Radiology, Sri Ganga Ram Hospital, New Delhi, India Published: November 24, 2023 Received: October 5, 2023 Received: October 5, 2023 Licensed under CC BY 4.0 International License which permits use, distribution and reproduction in any medium, provided the original work is properly cited INTRODUCTION Hemoptysis was graded as mild (less than 100 mL of blood expectorated per day), moderate (100- 300 mL/day) and massive (more than 300 mL/day or any amount of expectorated blood that led to hemodynamic disturbance).10 Bronchial artery embolization (BAE) is a minimally invasive interventional procedure, which is now considered the first-line management strategy and an alternative to surgery for massive and recurrent haemoptysis.1,2 Since its initial descriptions, it has evolved in multiple aspects including its technique, indications as well and efficacy.1 The exclusion criteria were as follows: patients with hemoptysis due to iatrogenic causes, prior history of bronchial artery embolization, use of neither of the two embolizing agents or both PVA and NBCA in the same patient and non-availability of follow- up data. According to our institutional protocol, CT angiography of the thorax was performed in all patients presenting with haemoptysis, before bronchial artery embolization. The advances in embolic agents have led to a significant improvement in the success rates of the procedure, however, there has been no significant change in the recurrence rate of haemoptysis.3 In addition, no consensus has been reached on the best embolic agent for BAE. Polyvinyl-alcohol (PVA) and absorbable gelatin-sponge particles are the most common agents as they are easy to handle, cheap and readily available. Recently, n-butyl- 2- cyanoacrylate (NBCA) has also been shown to control bleeding from various organs. It offers certain advantages including complete and rapid occlusion of the vessels and hence, a more reliable embolization. Despite these advantages, its use has been avoided in haemoptysis control due to the concern that it may cause complications like reflux and non-target embolization(7).1,3,4,5,6,7 Bronchial artery embolization was performed using Philips Allura Xper FD20. The right common femoral artery was accessed by retrograde puncture using the Seldinger technique. Bronchial artery embolization of the included patients was carried out with one of the two embolizing agents- NBCA or PVA. In the NBCA group, approximately 0.2 ml of NBCA (Endocryl, Samarth Life Sciences)was withdrawn from a 0.5 ml ampoule and mixed with lipiodol (Lipiodol Ultra Fluid, Guerbet) in the ratio of 15-30% in a 2-ml syringe and 0.3-1 ml of this mixture was injected depending on the calibre of the abnormal vessel and visualization of the glue-cast. Immediately before glue injection, the microcatheter was flushed with 5% dextrose (D5W). Conclusions: BAE with NBCA provided higher hemoptysis-free survival rates compared with PVA particles without increasing complication rates. NJR VOL 13 No. 2 ISSUE 22 November, 2023 9 Keywords: Bronchial Arteries; Embolism; Hemoptysis; Survival Rate Correspondence to: Dr. Manish Thapa Department of Intervention Radiology Nepal Army Institute of Health Sciences Kathmandu, Nepal Email: emaildrmanish@gmail.com Licensed under CC BY 4.0 International License which permits use, distribution and reproduction in any medium, provided the original work is properly cited Keywords: Bronchial Arteries; Embolism; Hemoptysis; Survival Rate Licensed under CC BY 4.0 International License which permits use, distribution and reproduction in any medium, provided the original work is properly cited 9 Thapa M et al. Bronchial Artery Embolization INTRODUCTION In the PVA group, the PVA particles 300-500 microns or 500- 700 microns (Contour PVA Embolization Particles; Boston Scientific), from the vial were emptied into a sterile container and 10-15 ml of 100% non-ionic contrast was added till desired consistency was achieved. This mixture was filled in a 2-ml syringe and was injected into the abnormal vessel under fluoroscopy till complete stasis was seen. Although a few recent studies have shown that NBCA can be used successfully for BAE without major complications, most of these studies were limited by a short follow-up period and no comparison group was chosen to compare the results of the NBCA group.3,7,8,9 Therefore, our study aimed to retrospectively compare the effectiveness and safety of NBCA versus PVA particles for control of haemoptysis and to determine if the recurrence of hemoptysis is related to the embolizing agent used. Technical success was defined as the successful embolization of both bronchial as well as non- bronchial systemic collaterals. Clinical success was defined as the complete absence of haemoptysis within 24 hours of the procedure. Post embolization, the patients were followed up in the OPD for 1 year. If haemoptysis recurred, subsequent management was planned (repeat embolization, emergency bronchoscopy or surgery) as determined by the RESULTS In both groups, technical success was achieved in all the patients. In the PVA group, clinical success was achieved in 84 out of 86 cases (97.6%), whereas, it was achieved with NBCA in 36 out of 37 patients (97.3%) (p >0.05) A total of 273 patients underwent bronchial artery embolization in the study center during the study period. Among these patients, 150 patients were excluded based on the above-mentioned exclusion criteria. Thus, the analysis was performed on 123 patients(follow-up data available), out of which NBCA was used in 37 patients and PVA was used in 86 patients. Thus, haemoptysis could not be controlled in a total of 3 patients (2 in the PVA group and 1 in the NBCA group) (p>0.05). Among these, 2 patients underwent surgery for the underlying cause and 1 patient died due to uncontrolled haemoptysis. No other major complications were observed. Minor complications were observed in 34out of 123 patients (27.6%), which included chest pain in 15 (12.2%) patients[10 in PVA group (11.6%) and 5 in NBCA group(13.5%], nausea and vomiting in 14(11.4%) patients [10 in PVA group (11.6%) and 4 in NBCA group (10.8%)] and fever in 10(8.1%) patients [7 in PVA group (8.1%) and 3 in NBCA group (8.1%)]. The complication rate did not show statistical significance when comparing the 2 groups(p>0.05). All of these complications were managed conservatively. In patients who developed fever, blood cultures were negative and fever subsided with antipyretics and a course of empirical antibiotics. There was no procedure- related mortality. The mean (± SD) age of the study participants was 45.34 years (±15.34). Most of them (93, 75.61%) were males. Active tuberculosis was the most common cause of hemoptysis (55, 45.08%). Most of the study patients (70, 57.85%) had massive hemoptysis on presentation. On average, the study patients had a hemoptysis duration of 12.6 (SD ± 30.8) days at the time of presentation. The right lung was more frequently involved (67, 54.92%) and most of the patients had only one lobe involvement (92, 75.41%). A total of 248 arteries were embolized: 157 bronchial arteries (81 right, 76 left) and 91 non- bronchial systemic arteries. There was no significant difference in the number of bronchial arteries and non-bronchial systemic collaterals embolized in the 2 groups. METHODS This study was a retrospective study conducted at a multispecialty tertiary care centre. Institutional review board approval was waived off as it was a retrospective study. The available medical records of all the patients (age>18 years) who had undergone bronchial artery embolization for hemoptysis from January 2012 to December 2020 were reviewed. NJR VOL 13 No. 2 ISSUE 22 November, 2023 NJR VOL 13 No. 2 ISSUE 22 November, 2023 10 10 Thapa M et al. Bronchial Artery Embolization The right bronchial arteries alone were embolized in 56 cases (45.9%), the left bronchial arteries alone in 31 cases (25.4%) and both in 36 cases (28.7%). These characteristics did not differ significantly among the two groups (Table 1). multi-disciplinary team. However, these cases were not included in the study The right bronchial arteries alone were embolized in 56 cases (45.9%), the left bronchial arteries alone in 31 cases (25.4%) and both in 36 cases (28.7%). These characteristics did not differ significantly among the two groups (Table 1). Data collection and analysis were done by standard technique. NJR VOL 13 No. 2 ISSUE 22 November, 2023 RESULTS The use of PVA particles for BAE was found to have 2.62 times higher odds (95% CI 1.10 - 6.81) of recurrence of hemoptysis as compared to the use of NBCA glue (Table 2) Aspergilloma 4 3 7 5.74% Post TB sequelae 1 3 4 3.28% Others 1 3 4 3.28% Hemoptysis severity Massive 19 51 70 57.85% 0.485 Moderate 14 31 45 37.19% Mild 3 3 6 4.96% Hemoptysis duration (Mean ± SD) 8.91 ± 20.40 14.26 ± 34.34 12.6 ± 30.80 0.600 The extent of lung involvement One lobe 24 68 92 75.41% 0.222 Two lobes 12 18 30 24.59% The site of the lung involved Right lung 22 45 67 54.92% 0.774 Left lung 9 26 35 28.69% Both lungs 5 14 19 15.57% Embolized artery Right bronchial artery 18 38 56 45.90% 0.326 Bilateral bronchial artery 7 28 36 28.69% Left bronchial artery 11 20 31 25.41% NBCA n-Butyl Cyano- Acrylate glue PVA Polyvinyl Alcohol Particles Of the 120 patients in whom BAE was clinically successful, recurrence was observed in 43 patients within the 12-month follow-up period. In the NBCA group (n=36), 7 patients (19.4%) developed recurrence of hemoptysis, whereas in the PVA group (n=84), 36 patients (42.8%) developed hemoptysis recurrence. To test for an association of the particle embolized with the recurrence of hemoptysis, the Chi-Squared test of independence was used. It was found that there was a statistically significant association between the embolizing agent used for BAE and the recurrence of hemoptysis (χ2 = 4.80, df = 1, p = 0.028). The use of PVA particles for BAE was found to have 2.62 times higher odds (95% CI 1.10 - 6.81) of recurrence of hemoptysis as compared to the use of NBCA glue (Table 2). test of independence was used. It was found that there was a statistically significant association between the embolizing agent used for BAE and the recurrence of hemoptysis (χ2 = 4.80, df = 1, p = 0.028). The use of PVA particles for BAE was found to have 2.62 times higher odds (95% CI 1.10 - 6.81) of recurrence of hemoptysis as compared to the use of NBCA glue (Table 2). test of independence was used. RESULTS NJR VOL 13 N 2 ISSUE 22 N b 2023 Table 1: Characteristics of study patients Characteristics NBCA n = 37 PVA n = 86 Total p-value N % Age (Mean ± SD) 45.89 ± 14.13 45.1 ± 15.9 45.34 ± 15.34 0.786 Sex Male 28 65 93 75.61% 1 Female 9 21 30 24.39% Hemoptysis cause Active TB 17 38 55 45.08% 0.561 Pneumonia 8 16 24 19.67% Unknown etiology 3 13 16 13.11% Bronchiectasis 2 10 12 9.84% Table 1: Characteristics of study patients 11 Thapa M et al. Bronchial Artery Embolization Aspergilloma 4 3 7 5.74% Post TB sequelae 1 3 4 3.28% Others 1 3 4 3.28% Hemoptysis severity Massive 19 51 70 57.85% 0.485 Moderate 14 31 45 37.19% Mild 3 3 6 4.96% Hemoptysis duration (Mean ± SD) 8.91 ± 20.40 14.26 ± 34.34 12.6 ± 30.80 0.600 The extent of lung involvement One lobe 24 68 92 75.41% 0.222 Two lobes 12 18 30 24.59% The site of the lung involved Right lung 22 45 67 54.92% 0.774 Left lung 9 26 35 28.69% Both lungs 5 14 19 15.57% Embolized artery Right bronchial artery 18 38 56 45.90% 0.326 Bilateral bronchial artery 7 28 36 28.69% Left bronchial artery 11 20 31 25.41% NBCA n-Butyl Cyano- Acrylate glue **PVA Polyvinyl Alcohol Particles Of the 120 patients in whom BAE was clinically successful, recurrence was observed in 43 patients within the 12-month follow-up period. In the NBCA group (n=36), 7 patients (19.4%) developed recurrence of hemoptysis, whereas in the PVA group (n=84), 36 patients (42.8%) developed hemoptysis recurrence. To test for an association of the particle embolized test of independence was used. It was found that there was a statistically significant association between the embolizing agent used for BAE and the recurrence of hemoptysis (χ2 = 4.80, df = 1, p = 0.028). NJR VOL 13 No. 2 ISSUE 22 November, 2023 DISCUSSION benefit of NBCA is reduced in these cases due to the recruitment of new feeding arteries. Our study could not support this finding as the number of cases of the above-mentioned conditions was quite less in our study group (aspergilloma-5.7% and tuberculosis sequelae-3.2%) and hence meaningful statistical analysis could not be performed in these groups and more studies are required. In this retrospective study, we analysed 2 embolizing agents used in bronchial artery embolization for haemoptysis (NBCA vs PVA) and evaluated the effectiveness and safety of both agents. We achieved a technical success rate of 100% in both groups. The clinical success rate was 97.6% in the PVA group and 97.3% in the NBCA group (p >0.05). These results are in accordance with the previously published studies which show that both agents are effective embolizing agents for BAE.3,8,12,13,14 In our study, the major and minor complication rates did not show statistical significance when comparing the 2 groups (p>0.05). Neurological complications (inadvertent embolization of the spinal artery) are one of the most dreaded complications of NBCA embolization. This is because the risk of non- target embolization is high with NBCA, which can occur due to uncontrolled reflux of NBCA or detachment of the polymerized NBCA that is adhered to the tip of the microcatheter during its withdrawal. Fortunately, we did not experience any such complications. However, extreme caution is advised when using NBCA during the embolization procedures. Another concern regarding NBCA is the risk of causing tissue necrosis (bronchial necrosis). However, we did not encounter any symptoms related to pulmonary ischemia or any airway abnormality in the embolized cases. In addition, the few cases who underwent follow-up CT or bronchoscopy did not show any such sign. This result is in accordance with a few recent studies which show that tissue necrosis is rare with NBCA in BAE.7,9,12 In patients in whom BAE was clinically successful, recurrence was observed in 36 patients out of a total of 84 patients in the PVA group (42.8%) and in 7 patients out of a total of 36 patients in the NBCA (19.4%) group (p=0.028). Thus, our results show that NBCA is a better embolizing agent as compared to PVA with respect to the recurrence of haemoptysis. This result is also in accordance with a previously published study by Woo et al which showed increased effectiveness of NBCA as compared to PVA. RESULTS It was found that there was a statistically significant association between the embolizing agent used for BAE and the recurrence of hemoptysis (χ2 = 4.80, df = 1, p Of the 120 patients in whom BAE was clinically successful, recurrence was observed in 43 patients within the 12-month follow-up period. In the NBCA group (n=36), 7 patients (19.4%) developed recurrence of hemoptysis, whereas in the PVA group (n=84), 36 patients (42.8%) developed hemoptysis recurrence. = 0.028). The use of PVA particles for BAE was found to have 2.62 times higher odds (95% CI 1.10 - 6.81) of recurrence of hemoptysis as compared to the use of NBCA glue (Table 2). = 0.028). The use of PVA particles for BAE was found to have 2.62 times higher odds (95% CI 1.10 - 6.81) of recurrence of hemoptysis as compared to the use of NBCA glue (Table 2). = 0.028). The use of PVA particles for BAE was found to have 2.62 times higher odds (95% CI 1.10 - 6.81) of recurrence of hemoptysis as compared to the use of NBCA glue (Table 2). To test for an association of the particle embolized with the recurrence of hemoptysis, the Chi-Squared NJR VOL 13 No. 2 ISSUE 22 November, 2023 Table 2: Chi-Squared test of independence for the association of embolizing agent used with hemoptysis recurrence Embolized particle Hemoptysis recurrence Total Odds Ratio 95% Confidence Interval p-Value Yes No n-Butyl Cyano-Acrylate glue (NBCA) 7 29 36 2.61 1.10 - 6.81 0.028 Polyvinyl alcohol particle (PVA) 36 48 84 Total 43 77 120 NJR VOL 13 No. 2 ISSUE 22 November, 2023 Table 2: Chi-Squared test of independence for the association of embolizing agent used with hemoptysis recurrence Embolized particle Hemoptysis recurrence Total Odds Ratio 95% Confidence Interval p-Value Yes No n-Butyl Cyano-Acrylate glue (NBCA) 7 29 36 2.61 1.10 - 6.81 0.028 Polyvinyl alcohol particle (PVA) 36 48 84 Total 43 77 120 NJR VOL 13 No. 2 ISSUE 22 November, 2023 12 Thapa M et al. Bronchial Artery Embolization DISCUSSION A lower recurrence rate of hemoptysis after NBCA embolization can partly be explained by the fact that the level of embolization is different in both embolizing agents. NBCA reaches up to the distal vascular bed and occludes it, whereas, PVA particles aggregate and form plugs which causes premature embolization proximal to the level of the vascular bed.7 The most common etiology of hemoptysis in our study was active tuberculosis (45% of total cases). We performed a subgroup analysis (stratified to the cause of haemoptysis) and found that the lower recurrence rate in the NBCA group was also seen in the active tuberculosis group. Out of a total of 55 cases of active tuberculosis, 40 were treated with PVA and 15 (37.5%) had a recurrence, whereas 15 were treated with NBCA and 3 (20%) had a recurrence, P<0.05. In the other etiologies, the difference was not significant or could not be evaluated due to the small sample size. The study by Woo et al showed that even though the rate of recurrence of haemoptysis was low with NBCA use in their total study group (p<0.05) this was not the case in patients of aspergilloma and chronic tuberculosis.12 According to them, the Our study had a few limitations. Firstly, it was a retrospective study, which has its inherent limitations. Furthermore, even though statistically not significant, there were differences in the baseline characteristics of the 2 groups. In addition, the choice of the embolizing agent was random, based on the discretion of the interventional radiologist. Thus, in this study, the indication of NBCA was not defined, and this might have caused a bias in the interpretation. CONFLICT OF INTEREST None 8. Kish JW, Katz MD, Marx MV, Harrell DS, Hanks SE. N-butyl cyanoacrylate embolization for control of acute arterial hemorrhage. J Vasc Interv Radiol 2004;15(7):689-95. https://doi. org/10.1097/01.rvi.0000133505.84588.8c CONCLUSION In conclusion, this study shows that BAE using NBCA has a lower recurrence rate of haemoptysis as compared to PVA particles, without any increase NJR VOL 13 No. 2 ISSUE 22 November, 2023 13 Thapa M et al. Bronchial Artery Embolization 7. Baltacioğlu F, Cimşit NC, Bostanci K, Yüksel M, Kodalli N. Transarterial microcatheter glue embolization of the bronchial artery for life- threatening hemoptysis: technical and clinical results. Eur J Radiol 2010;73(2):380–4. https:// doi.org/10.1016/j.ejrad.2008.10.017 in the complication rate. However, randomized controlled trials are required to confirm these advantages of NBCA over other embolizing agents. None None SOURCES OF FUNDING None REFERENCES 1. Yoon W, Kim JK, Kim YH, Chung TW, Kang HK. Bronchial and nonbronchial systemic artery embolization for life-threatening hemoptysis: a comprehensive review. Radiographics 2002;22(6):1395-409. https:// doi.org/10.1148/rg.226015180 9. Ikoma A, Kawai N, Sato M et al. Pathologic evaluation of damage to bronchial artery, bronchial wall, and pulmonary parenchyma after bronchial artery embolization with N- butyl cyanoacrylate for massive hemoptysis. J Vasc Interv Radiol 2011;22(8):1212-5. https:// doi.org/10.1016/j.jvir.2011.02.001 2. Swanson KL, Johnson CM, Prakash UB, McKusick MA, Andrews JC, Stanson AW. Bronchial artery embolization: experience with 54 patients. Chest 2002;121(3):789-95. https:// doi.org/10.1378/chest.121.3.789 10. Panda A, Bhalla AS, Goyal A. Bronchial artery embolization in hemoptysis: a systematic review. Diagn Interv Radiol 2017;23(4):307. https://doi.org/10.5152/dir.2017.16454 3. Yoo DH, Yoon CJ, Kang SG, Burke CT, Lee JH, Lee CT. Bronchial and nonbronchial systemic artery embolization in patients with major hemoptysis: safety and efficacy of N- butyl cyanoacrylate. AJR Am J Roentgenol 2011;196(2):W199-204. https://doi. org/10.2214/ajr.10.4763 11. Sacks D, McClenny TE, Cardella JF, Lewis CA. Society of Interventional Radiology clinical practice guidelines. J Vasc Interv Radiol 2003 Sep;14(9 Pt 2):S199-202. https:// doi.org/10.1097/01.rvi.0000094584.83406.3e 12. Woo S, Yoon CJ, Chung JW et al. Bronchial artery embolization to control hemoptysis: comparison of N-butyl-2-cyanoacrylate and polyvinyl alcohol particles. Radiology 2013;269(2):594-602. https://doi.org/10.1148/ radiol.13130046 4. Hill H, Chick JF, Hage A, Srinivasa RN. N-butyl cyanoacrylate embolotherapy: techniques, complications, and management. Diagn Interv Radiol 2018;24(2):98-103. https://doi. org/10.5152/dir.2018.17432 13. Agmy GM, Wafy SM, Mohamed SAA, Gad YA, Mustafa H, Abd El-Aziz AES. Bronchial and Nonbronchial Systemic Artery Embolization in Management of Hemoptysis: Experience with 348 Patients. ISRN Vascular Medicine 2013;2013:e263259. https://doi. org/10.1155/2013/263259 5. Takeuchi Y, Morishita H, Sato Y et al. Guidelines for the use of NBCA in vascular embolization devised by the Committee of Practice Guidelines of the Japanese Society of Interventional Radiology (CGJSIR), 2012 edition. Jpn J Radiol 2014;32:500-17. https:// doi.org/10.1007/s11604-014-0328-7 14. Sopko DR, Smith TP. Bronchial Artery Embolization for Hemoptysis. Semin Intervent Radiol 2011;28(1):48-62. https://doi. org/10.1055/s-0031-1273940 6. Vaidya S, Tozer KR, Chen J. An overview of embolic agents. Semin Intervent Radiol 2008;25(3):204-15. https://doi. org/10.1055%2Fs-0028-1085930 NJR VOL 13 No. 2 ISSUE 22 November, 2023 14
https://openalex.org/W2108633424	https://europepmc.org/articles/pmc4060159?pdf=render	English	null	Redox Imbalance and Morphological Changes in Skin Fibroblasts in Typical Rett Syndrome	Oxidative medicine and cellular longevity	2,014	cc-by	8,212	Research Article Redox Imbalance and Morphological Changes in Skin Fibroblasts in Typical Rett Syndrome Cinzia Signorini,1 Silvia Leoncini,1,2 Claudio De Felice,3 Alessandra Pecorelli,1,2 Ilaria Meloni,4 Francesca Ariani,4 Francesca Mari,4 Sonia Amabile,4 Eugenio Paccagnini,5 Mariangela Gentile,5 Giuseppe Belmonte,6 Gloria Zollo,1,2 Giuseppe Valacchi,7 Thierry Durand,8 Jean-Marie Galano,8 Lucia Ciccoli,1 Alessandra Renieri,4,9 and Joussef Hayek2 1 Department of Molecular and Developmental Medicine, University of Siena, 53100 Siena, Italy 1 Department of Molecular and Developmental Medicine, University of Siena, 53100 Siena, Italy 2 Child Neuropsychiatry Unit, University Hospital Azienda Ospedaliera Universitaria Senese (AOUS), 53100 Siena, Italy 3 Neonatal Intensive Care Unit, University Hospital AOUS, Policlinico “S. M. alle Scotte,” 53100 Siena, Italy 4Medical Genetics, University of Siena, 53100 Siena, Italy 5 Department of Life Sciences, University of Siena, 53100 Siena, Italy 6Department of Medicine Surgery and Neuroscience, University of Siena, 53100 Siena, Italy 7 Department of Life Science and Biotechnologies, University of Ferrara, 44121 Ferrara, Ita 8 Institut des Biomol´ecules Max Mousseron (IBMM), UMR 5247-CNRS-UM I-UM II-ENSCM, BP 14491 34093, Montpellier Cedex 5, France p 9 Genetica Medica, Azienda Ospedaliera Universitaria Senese, 53100 Siena, Italy orrespondence should be addressed to Cinzia Signorini; cinzia.signorini@unisi.it and Claudio De Felice; genient Correspondence should be addressed to Cinzia Signorini; cinzia.signorini@unisi.it and Claudio De Felice; geniente@gmail.com Correspondence should be addressed to Cinzia Signorini; cinzia.signorini@unisi.it and Claudio De Felice; geniente@gmail.com Correspondence should be addressed to Cinzia Signorini Received 10 March 2014; Revised 9 May 2014; Accepted 12 May 2014; Published 29 May 2014 Received 10 March 2014; Revised 9 May 2014; Accepted 12 May 2014; Published 29 May 2014 Academic Editor: Ozcan Erel Academic Editor: Ozcan Erel Copyright © 2014 Cinzia Signorini et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Evidence of oxidative stress has been reported in the blood of patients with Rett syndrome (RTT), a neurodevelopmental disorder mainly caused by mutations in the gene encoding the Methyl-CpG-binding protein 2. Little is known regarding the redox status in RTT cellular systems and its relationship with the morphological phenotype. In RTT patients (n = 16) we investigated four different oxidative stress markers, F2-Isoprostanes (F2-IsoPs), F4-Neuroprostanes (F4-NeuroPs), nonprotein bound iron (NPBI), and (4-HNE PAs), and glutathione in one of the most accessible cells, that is, skin fibroblasts, and searched for possible changes in cellular/intracellular structure and qualitative modifications of synthesized collagen. Hindawi Publishing Corporation Oxidative Medicine and Cellular Longevity Volume 2014, Article ID 195935, 10 pages http://dx.doi.org/10.1155/2014/195935 Hindawi Publishing Corporation Oxidative Medicine and Cellular Longevity Volume 2014, Article ID 195935, 10 pages http://dx.doi.org/10.1155/2014/195935 Hindawi Publishing Corporation Oxidative Medicine and Cellular Longevity Volume 2014, Article ID 195935, 10 pages http://dx.doi.org/10.1155/2014/195935 Research Article Redox Imbalance and Morphological Changes in Skin Fibroblasts in Typical Rett Syndrome Significantly increased F4-NeuroPs (12-folds), F2-IsoPs (7.5-folds) NPBI (2.3-folds), 4-HNE PAs (1.48-folds), and GSSG (1.44-folds) were detected, with significantly decreased GSH (−43.6%) and GSH/GSSG ratio (−3.05 folds). A marked dilation of the rough endoplasmic reticulum cisternae, associated with several cytoplasmic multilamellar bodies, was detectable in RTT fibroblasts. Colocalization of collagen I and collagen III, as well as the percentage of type I collagen as derived by semiquantitative immunofluorescence staining analyses, appears to be significantly reduced in RTT cells. Our findings indicate the presence of a redox imbalance and previously unrecognized morphological skin fibroblast abnormalities in RTT patients. 1. Introduction determines the clinical severity of the disease is a typical feature of RTT [4], key clinical aspects include autistic traits, epileptic seizures, breathing abnormalities, gait ataxia, stereotypies, and loss of finalistic hands use [3]. Mutations in the gene encoding the Methyl-CpG-binding protein 2 (MECP2) account for approximately 90% of cases with typical RTT and are almost exclusively de novo [4]. Nine most Rett syndrome (RTT) predominantly affects females with an incidence of 1 in 10,000–15,000 female births [1, 2]. In its typical form, affected patients exhibit various neuropsychi- atric features after 6–18 months [3] of apparently normal neu- rodevelopment. Although phenotypical heterogeneity which Oxidative Medicine and Cellular Longevity 2 frequent mutations comprise more than 3/4 of all the reported pathogenic ones [5]; in addition, MECP2 gene muta- tions are usually categorized as missense or truncating, including nonsense, frameshift, and large deletions, accord- ing to the type of sequence change. isolated and cultured with standard protocols [28]. Cells were grown in DMEM (Biochrom), supplemented with 10% fetal bovine serum, 100 U/mL penicillin, 100 𝜇g/mL streptomycin, and 2 mM L-glutamine. Cells were incubated at 37∘C with 5% CO2 until 80–90% confluence and routinely passed by trypsin-EDTA (Irvine Scientific). Oxidative stress (OS) indicates a combination of events resulting in a damage to biological molecules due to an imbalance between cellular antioxidant defences and free radicals production. OS appears to be involved in a large number of human diseases, including cancer [6–8] neurode- generative diseases [9, 10] and inflammatory bowel disease [11]. Furthermore, cumulating evidence suggests the presence of a redox imbalance in autism spectrum disorders (ASDs), a condition with a high social impact due to the explosive increase in its prevalence over the last four decades [12–18]. Our group and other laboratories have reported enhanced OS markers levels in plasma and erythrocytes from patients with RTT, thus suggesting the presence of a systemic OS in the disease. However, to date, it is unclear not only why, but also when, and where this OS derangement may occur [19–26]. i Cells for analysis were seeded onto 100 mm tissue culture plates (for NPBI and IsoP evaluations) and 12-well plates (for 4-HNE PAs and GSH/GSSG ratio) containing the indi- cated medium formulations. Fibroblasts at low passage were employed for the analysis.l At confluence, the cells were scraped, transferred in a tube, and washed twice with ice-cold PBS pH 7.4. 1. Introduction Cells were then centrifuged at 600 g for 10 min and the pellet resuspended in a final volume of 2 mL PBS. 2.3. Nonprotein Bound Iron (NPBI). Nonprotein bound iron was determined as desferoxamine- (DFO-) chelatable free iron (DFO-iron complex, ferrioxamine). DFO 25 𝜇M was added to 1 mL of cell suspension to obtain a final concen- tration of 25 𝜇M DFO and the cells were ruptured by the addition of 1 mL water, freeze-thawing, and sonication. The samples were then ultrafiltered in centrifuge filter devices (VIVASPIN 4, Sartorius Stedim Biotech GmbH, Goettingen, Germany) with a 30 kDa molecular weight cutoff and the ultrafiltrate was stored at −20∘C until analysis. The DFO excess was removed by silica (Silicagel 25–40 𝜇m; Merck, Darmstadt, Germany) column chromatography (Varian Inc., CA, USA). The DFO-iron complex was determined by HPLC and the detection wavelength was 229 nm [29]. In particular to date, little is known regarding the oxidant- antioxidant status in RTT cellular systems and its possible relationship with the morphological phenotype. Here, we investigated the levels of different OS markers (F2-IsoPs, F4- NeuroPs, NPBI, and 4-HNE PAs) and glutathione in primary skin fibroblasts from patients with typical RTT, as well as possible changes in the cellular/intracellular structure and/or qualitative changes in the synthesized collagen. 2. Materials and Methods 2.1. Subjects. A total of 16 female patients with typical RTT (mean age: 13.06 ± 6.5), as well as 8 healthy female controls of comparable age (mean age: 13.2 ± 6.8), participated in the study. Skin biopsies from the control group were obtained from a skin biobank by selecting subjects without diagnosed skin or collagen diseases (responsibility: Joussef Hayek). 2.4. Total (Sum of Free Plus Esterified) F2-Isoprostanes and F4- Neuroprostanes Determinations. Butylated hydroxytoluene was added to 900 𝜇L of cell suspension as an antioxidant to obtain a final concentration of 90 𝜇M and the samples were frozen at −70∘C until analysis. All isoprostane and neuroprostane determinations were carried out by gas chro- matography/negative ion chemical ionization tandem mass spectrometry (GC/NICI-MS/MS) analysis after solid phase extraction and derivatization steps. F2-Isoprostanes and F4- neuroprostanes levels were normalized for the cell protein content. All patients were consecutively admitted to the Child Neuropsychiatry Unit of the University Hospital of Siena (Head: Joussef Hayek). For all patients the MECP2 mutation was demonstrated (mutation categories: missense mutation 𝑛= 8; early truncating mutation 𝑛= 6; late truncating mutation 𝑛= 2), and a clinical severity score was calculated using the previously reported scaling system [27] (mean severity score: 15.9 ± 6.47; range: 5.0–28.0). All the examined subjects were on a typical Mediterranean diet.t 2.5. Solid Phase Extraction and Derivatization Procedures. At the time of the determination, the samples were sonicated by ultrasound treatment for 30 seconds and then purified as pre- viously reported [29]. Briefly, aqueous KOH (1 mM, 450 𝜇L) was added to the cellular suspension. After incubation at 45∘C for 45 min, the pH was adjusted to 3 by adding HCl (1 mM, 450 𝜇L). Each sample was spiked with tetradeuterated prostaglandin F2𝛼(PGF2𝛼-d4) (500 pg in 50 𝜇L of ethanol), as an internal standard, and ethyl acetate (10 mL) was added to extract total lipids by vortex-mixing and centrifugation at 1,000 g for 5 min at room temperature. The total lipid extract was applied onto an NH2 cartridge and the isoprostanes in the final eluates were derivatized. The carboxylic group was derivatized as the pentafluorobenzil ester whereas the hydroxyl groups were converted to trimethylsilyl ethers [30]. 2.5. Solid Phase Extraction and Derivatization Procedures. At the time of the determination, the samples were sonicated by ultrasound treatment for 30 seconds and then purified as pre- viously reported [29]. 2. Materials and Methods After fixation and permeabilization for 10 min at room temperature with 0.1% Triton X-100, cells were blocked for 30 min at room temperature with PBS containing 5% BSA. Fibroblasts were then incubated with primary antibodies for collagen I and collagen III (Thermo Fisher Scientific Inc., Rockford, IL, USA) in PBS with 1% BSA at 4∘C overnight. After washing, cells were incubated with secondary antibodies Alexa Fluor 568 and Alexa Fluor 488 (Life Technologies Corporation, Monza, Italy) for 1 h at room temperature. Nuclei were stained with 1 𝜇g/mL DAPI (Sigma-Aldrich S.r.l., Milan, Italy) for 1 min. Coverslips were mounted onto glass slides using antifade mounting medium 1,4-diazabicyclooctane in glycerine (DABCO). Observations and photographs were made with a Leitz Aristoplan light microscope (Leica, Milan, Italy) equipped with fluorescence apparatus. Incubation in primary antibodies was omitted in negative controls.hl 2.8. Intracellular Redox Status. Usually, cellular glutathione (GSH) exists mainly in the reduced form whereas the oxi- dized disulfide form (GSSG) is present in small amounts. The GSH/GSSG ratio is often taken as an indicator of cellular redox status. Intracellular GSH and GSSG levels were determined by an enzymatic recycling procedure according to Tietze [32] and Baker et al. [33]. The SH group of the molecule reacts with 5,5󸀠-dithiobis(2-nitrobenzoic acid) (DTNB), producing a yellow-coloured 5-thio-2-nitrobenzoic acid (TNB), and the disulfide is reduced by NADPH in the presence of GSH reductase. GSSG was determined after derivatization step of GSH by reaction with 2-vinylpyridine. The rate of TNB formation was monitored at 420 nm. GSH and GSSG levels were normalized for protein content.lti At confluence, after medium removal, cultured fibroblasts were washed twice with PBS pH 7.4 and treated with 5% 5- sulfosalicylic acid (w/v) solution for 30 min at 4∘C. The acidic extracts were stored at −70∘C until the assay. The protein extracts were obtained by cellular lysis with NaOH 0.1 M. g The relative intensity of fluorescence was measured in regions of interest (ROI) by using the Software LEICA AF6000 (Leica Microsystems-Germany). 2.9. 4-HNE Protein Adducts. 4-HNE PAs are markers of protein oxidation due to aldehyde binding from lipid perox- idation sources [34], determined by Western blot technique. Cell proteins (30 𝜇g protein, as determined by using Bio-Rad protein assay; BioRad, Hercules, CA, USA) were resolved on 4–20% SDS-PAGE gels (Lonza Group Ltd., Switzerland) and transferred onto a hybond ECL nitrocellulose membrane (GE Healthcare Europe GmbH, Milan, Italy). 2. Materials and Methods Briefly, aqueous KOH (1 mM, 450 𝜇L) was added to the cellular suspension. After incubation at 45∘C for 45 min, the pH was adjusted to 3 by adding HCl (1 mM, 450 𝜇L). Each sample was spiked with tetradeuterated prostaglandin F2𝛼(PGF2𝛼-d4) (500 pg in 50 𝜇L of ethanol), as an internal standard, and ethyl acetate (10 mL) was added to extract total lipids by vortex-mixing and centrifugation at 1,000 g for 5 min at room temperature. The total lipid extract was applied onto an NH2 cartridge and the isoprostanes in the final eluates were derivatized. The carboxylic group was derivatized as the pentafluorobenzil ester whereas the hydroxyl groups were converted to trimethylsilyl ethers [30]. A 3 mm skin punch biopsy was performed after obtained written informed consent of either the parents or the legal tutors of the patients (responsibility: Joussef Hayek). Biopsies from control subject were performed in the Dermatology Unit of the University Hospital of Siena. The study was con- ducted with the approval of the Institutional Review Board of the University Hospital Azienda Ospedaliera Universitaria Senese. 2.2. Fibroblasts Isolation and Culture. Following informed consent signature, skin biopsies (about 3-4 mm3) were per- formed using the Punch Biopsy procedure. Fibroblasts were 3 Oxidative Medicine and Cellular Longevity 2.6. F2-Isoprostane GC/NICI-MS/MS. Measured ions were the product ions at m/z 299 and m/z 303 derived from the [M−181]−precursor ions (m/z 569 and m/z 573) pro- duced from 15-F2t-IsoPs and the tetradeuterated derivative of prostaglandin F2𝛼(PGF2𝛼-d4), respectively [29, 30]. for 2 hours at 4∘C, dehydrated in a graded series of alcohol, embedded in Araldite resins, and polymerized in oven for 48 hours at 60∘C. Sixty nm thin sections, obtained with a Reichert ultramicrotome, were routinely stained with uranyl acetate and lead citrate and observed with a TEM Fei Tecnai G2 spirit at 100 Kv. 2.7. F4-NeuroPs GC/NICI-MS/MS. Measured ions were the product ions at m/z 323 and m/z 303 derived from the [M−181]−precursor ions (m/z 593 and m/z 573) produced from oxidized DHA and the tetradeuterated derivative of PGF2𝛼, respectively [29, 31]. 2.11. Immunofluorescence Double Staining. The localization of two main type of collagen synthetized by skin fibroblasts was evaluated by an immunofluorescence assay. Primary human fibroblasts, grown on glass coverslips at a density of 2 × 104 cells/cm2, were fixed in 4% paraformaldehyde for 10 min at 4∘C. 2. Materials and Methods After blocking in 3% nonfat milk (BioRad, Hercules, CA, USA), the membranes were incubated overnight at 4∘C with goat polyclonal anti 4- HNE adduct antibody (cod. AB5605; Millipore Corporation, Billerica, MA, USA). Following washes in TBS Tween and incubation with specific secondary antibody (mouse anti- goat horseradish peroxidase-conjugated, Santa Cruz Biotech- nology Inc., CA, USA) for 1 h at room temperature, the membranes were incubated with ECL reagents (BioRad, Hercules, CA, USA) for 1 min. The bands were visualized by autoradiography. Quantification of the relevant bands was performed by digitally scanning the Amersham Hyperfilm ECL (GE Healthcare Europe GmbH, Milan, Italy) and mea- suring immunoblotting image densities with ImageJ software. 2.12. Statistical Analysis. Results were expressed as medians with interquartile ranges or means ± standard deviation. Differences between groups were evaluated by the non- parametric Mann-Whitney rank sum test, Wilcoxon rank test, or Kruskal-Wallis test analysis of variance (ANOVA), as appropriate. The MedCalc ver. 12.1.4 statistical software package (MedCalc. Software, Mariakerke, Belgium) was used for data analysis. Two-sided 𝑃values <0.05 were considered as significant. 3. Results 3.1. Cell Oxidant and Antioxidant Status. Biochemical signs of lipid and protein oxidative damage, together with the antioxidant cellular defense, were evaluated in skin fibroblasts from RTT and healthy control subjects. Increased lipid oxidative damage was evidenced as indicated by increased levels of total (i.e., sum of free and esterified form) F2-IsoPs (7.5-folds) and F4-NeuroPs (12-folds), both deriving from membrane polyunsaturated fatty acids, that is, arachidonic (AA) and docosahexaenoic (DHA) acid, respectively (Figures 1(a) and 1(b)). Biochemical signs of protein oxidative damage consequent to a lipid peroxidation event were indicated by increased (1.48-folds) 4-HNE PAs levels for which a significant increase was also detected (Figure 1(c)). Oxidative 2.10. Transmission Electron Microscopy (TEM). Cultured fibroblasts were fixed in 2.5% glutaraldehyde in 0.1 M cacody- late buffer pH 7.2 (CB) for 3 hours at 4∘C. After a rinse in CB, the material was postfixed in 1% osmium tetroxide in CB Oxidative Medicine and Cellular Longevity 4 4 g y 0 200 400 600 800 1000 (pg/mg protein) Healthy controls Typical RTT Total F2-IsoPs ∗ (a) 0 200 400 600 800 1000 (pg/mg protein) Total F4-NeuroPs ∗ Healthy controls Typical RTT (b) 0 5 10 15 20 25 4-HNE PAs (×106 a.u.) ∗∗ Healthy controls Typical RTT (c) 0 0.2 0.4 0.6 0.8 1 NPBI (nmol/mg protein) ∗ Healthy controls Typical RTT (d) Figure 1: Increased levels of total (i.e., sum of free and esterified form) F2-IsoPs, total F4-NeuroPs, 4-HNE PAs, and NPBI in RTT skin fibrob- last as compared to the control cells. ∗𝑃< 0.0001, ∗∗𝑃= 0.0013. Data are expressed as means ± standard deviation. Legend: F2-IsoPs, F2-isoprostanes; F4-NeuroPs, F4-neuroprostanes; 4-HNE PAs, 4-hydroxy-2-nonenal protein adducts; NPBI, nonprotein bound iron. 0 200 400 600 800 1000 (pg/mg protein) Healthy controls Typical RTT Total F2-IsoPs ∗ (a) 0 5 10 15 20 25 4-HNE PAs (×106 a.u.) ∗∗ Healthy controls Typical RTT (c) Figure 1: Increased levels of total (i.e., sum of free and esterified form) F2-IsoPs, total F4-NeuroPs, 4-HNE PAs, and NPBI in RTT skin fibrob- last as compared to the control cells. ∗𝑃< 0.0001, ∗∗𝑃= 0.0013. Data are expressed as means ± standard deviation. Legend: F2-IsoPs, F2-isoprostanes; F4-NeuroPs, F4-neuroprostanes; 4-HNE PAs, 4-hydroxy-2-nonenal protein adducts; NPBI, nonprotein bound iron. 3.2. Cell Morphology Study. To evaluate the effect, if present, of the oxidant/antioxidant imbalance evidenced in RTT 3. Results 0 0.4 0.8 1.2 1.6 Healthy controls Typical RTT Reduced GSH (nmol/mg protein) ∗ (a) 0 0.4 0.8 1.2 1.6 GSSG ∗∗ (nmol/mg protein) Healthy controls Typical RTT (b) Figure 2: Significant reduction in cellular GSH and significant increase of GSSG in RTT skin fibroblast as compared to control cells. ∗𝑃< 0.0001, ∗∗𝑃= 0.0033. Data are expressed as means ± standard deviation. Legend: GSH reduced glutathione; GSSG, oxidized glutathione. 0 0.4 0.8 1.2 1.6 GSSG ∗∗ (nmol/mg protein) Healthy controls Typical RTT (b) (b) (a) Figure 2: Significant reduction in cellular GSH and significant increase of GSSG in RTT skin fibroblast as compared to control cells. ∗𝑃< 0.0001, ∗∗𝑃= 0.0033. Data are expressed as means ± standard deviation. Legend: GSH reduced glutathione; GSSG, oxidized glutathione. damage was concomitant to an imbalance in the principal antioxidant cytoplasmic agent in so far as a significant reduction (−43.6%) in cellular GSH, a significant increase (1.44-folds) of GSSG (Figures 2(a) and 2(b)), and a significant reduction (−3.05 folds) of GSH/GSSG ratio were reported. Lipid peroxidation events in RTT skin fibroblasts were found to be related to the levels of NPBI, a prooxidant agent. NPBI was significantly increased (2.3-folds) (Figure 1(d)), with significant positive correlations observed between NPBI and total cellular F4-NeuroPs (Rho = 0.84, 𝑃= 0.001) and NPBI versus total cellular F2-IsoPs (Rho = 0.69, 𝑃= 0.019). No significant relationships between each of the investigated molecules (i.e., F2-IsoPs, F4-NeuroPs, 4-HNE Pas, GSH and GSSG) and the MECP2 mutation categories were observed (𝑃 value range: 0.461–0.981). 3.2. Cell Morphology Study. To evaluate the effect, if present, of the oxidant/antioxidant imbalance evidenced in RTT 5 Oxidative Medicine and Cellular Longevity M RER G V 1𝜇m (a) RER V MLB 1𝜇m (b) Figure 3: Transmission electron microscopy of control (a) and RTT (b) fibroblasts cultures. Skin fibroblasts, either from control subjects or RTT patients, show a flattened morphology with extensive tapering cytoplasmic processes. An euchromatic and oval-shaped nucleus was present in central position of the cells, with clumps of heterochromatin next to the nuclear envelope. The cytoplasm contains many vesicles with variable electron density, a prominent Golgi complex, and mitochondria. Rough endoplasmic reticulum (RER) cisternae in RTT fibroblasts appear more dilated than in control. Some large multilamellar bodies (MLB) are frequently detectable in the cytoplasm of the RTT fibroblast cells. (G) Golgi complex, (M) mitochondrion, and (V) vesicle. Bar = 1 𝜇m. 3. Results RER V MLB 1𝜇m (b) M RER G V 1𝜇m (a) (b) (a) Figure 3: Transmission electron microscopy of control (a) and RTT (b) fibroblasts cultures. Skin fibroblasts, either from control subjects or RTT patients, show a flattened morphology with extensive tapering cytoplasmic processes. An euchromatic and oval-shaped nucleus was present in central position of the cells, with clumps of heterochromatin next to the nuclear envelope. The cytoplasm contains many vesicles with variable electron density, a prominent Golgi complex, and mitochondria. Rough endoplasmic reticulum (RER) cisternae in RTT fibroblasts appear more dilated than in control. Some large multilamellar bodies (MLB) are frequently detectable in the cytoplasm of the RTT fibroblast cells. (G) Golgi complex, (M) mitochondrion, and (V) vesicle. Bar = 1 𝜇m. Healthy controls Typical RTT Col III Col III Col I Col I Figure 4: Double immunofluorescence staining shows the localization of type I collagen (central column, red color) and type III collagen (left column, green color). Images are merged in the right panel and the yellow color indicates overlap of the staining. The colocalization of types I and III collagen is reduced in RTT skin fibroblasts. Legend: Col I, type I collagen; Col III, type III collagen. Healthy controls Col III Col I Figure 4: Double immunofluorescence staining shows the localization of type I collagen (central column, red color) and type III collagen (left column, green color). Images are merged in the right panel and the yellow color indicates overlap of the staining. The colocalization of types I and III collagen is reduced in RTT skin fibroblasts. Legend: Col I, type I collagen; Col III, type III collagen. granules were also detectable (Figure 4). Likewise, in these same cells, fluorescence relative intensity for type I collagen was significantly reduced (𝑃= 0.00997) (Figure 5). fibroblasts, the physiological cellular condition was evaluated by observation of two key typical features: the morphology and the collagen distribution.i At transmission electron microscope significant ultra- structural differences between control and RTT fibroblasts were observed. In particular, a marked dilation of the rough endoplasmic reticulum cisternae was detectable in the skin RTT fibroblasts, along with evidence of cytoplasmic multil- amellar bodies (Figures 3(a) and 3(b)). Immunofluorescence double staining showed a reduced degree of colocalization of type III and type I collagen in RTT skin fibroblasts when compared to control cells. 3. Results Staining for type I collagen was found to be more evident in RTT cells, where perinuclear Oxidative Medicine and Cellular Longevity Oxidative Medicine and Cellular Longevity 6 0 100 200 300 400 500 600 700 Relative intensity N.S. Col III Col I ∗ Healthy controls Typical RTT Figure 5: Relative intensity of fluorescence for types I and type III collagen in RTT and control skin fibroblasts. Software LEICA AF6000 (Leica Microsystems-Germany). Data are expressed as median ± semiinterquartile range ∗𝑃= 0.0062; N.S.: no significant difference (𝑃= 0.4361). Legend: Col I, type I collagen; Col III, type III collagen. 0 100 200 300 400 500 600 700 Relative intensity N.S. Col III Col I ∗ this fatty acid represents a normal constituent of all cell membranes. In particular, human fibroblasts are able to synthetize DHA [51] and incorporate it in their membrane phospholipids [52]. Furthermore, variations of the DHA content in the membrane of fibroblasts have been implicated in neuropsychiatric disorders, with lowered levels evidenced in patients with schizophrenia and bipolar disorder [53]. Our results indicate an increased oxidation of DHA in RTT fibrob- lasts, thus underlying an increased susceptibility of these cells to oxidative damage. This process, in turn, could contribute to changes in the fatty acid composition of membranes with consequent cellular damage. g 4HNE, which is formed from arachidonic acid or other unsaturated fatty acids following free radical attack, can bind, by Michael addition, to proteins, particularly, to cysteine, his- tidine, or lysine residues. Thanks to its ability to form adducts to the proteins, 4HNE is not only considered as a reliable marker of OS, but also has a biological impact by changing protein function [34]. Our findings of increased 4HNE-PAs can be considered as a long-term consequence of enhanced lipid peroxidation, further contributing to cellular damage. Figure 5: Relative intensity of fluorescence for types I and type III collagen in RTT and control skin fibroblasts. Software LEICA AF6000 (Leica Microsystems-Germany). Data are expressed as median ± semiinterquartile range ∗𝑃= 0.0062; N.S.: no significant difference (𝑃= 0.4361). Legend: Col I, type I collagen; Col III, type III collagen. Glutathione, the main cellular antioxidant defence, pre- venting the damage to key cellular components induced by reactive oxygen species, exists in both reduced (GSH) and oxidized (GSSG) states. Oxidative Medicine and Cellular Longevity Fibroblasts have been extensively employed to investigate OS in different pathophysiological processes such as genetic neurodegenerative diseases [42], Parkinson’s disease [43], aging [44], and response to etiological agents [45, 46]. Moreover, lipid peroxidation events are known to be relevant to the fibroblast function, as relationship between mechanotransduction and lipid metabolites has been previ- ously investigated in keloids [47]. Our reported results of decreased GSH levels and GSH/GSSG ratio indicate a reduced antioxidant defence in RTT cells and suggest that oxidative events are likely to be chronic, thus determining a consumption of glutathione in its reduced form. An alternative explanation could be that the MECP2 mutation-harboring cells may have coexisting defects in the synthesis and/or recycling of glutathione, thus leading to an uncontrolled free radicals action. Of course, further study is needed in order to clarify this point.i Overall, our data indicate that a significant oxidation of DHA, the precursor acid for F4-NeuroPs, and AA, precursor of F2-IsoPs, likely triggered by NPBI, occurs in skin fibrob- lasts of patients with MECP2 mutation and clinical RTT and confirm the biological relevance of such key mediators of lipid peroxidation. Skin dermis is known to be consisting of 80% collagen type I, with its remaining fraction being made mostly of collagen type III [54]. In this context, the fibroblast shows a pivotal role for collagen production, [55]. The biochemical markers so far employed for measuring OS are not to be considered as equal in terms of the con- veyed information. In particular, F2-IsoPs, the gold standard molecules for the OS in vivo evaluation [35, 48], are the end-products of arachidonic acid (AA) oxidation, a polyun- saturated fatty acid abundant in both brain grey and white matter. On the other hand, F4-NeuroPs are the oxidative end-products of docosahexanoic acid (DHA), abundant in neuronal membranes [49]. NPBI is a prooxidant factor, asso- ciated with hypoxia, hemoglobin oxidation, and subsequent heme iron release [50]. Besides being a major component of the extracellular matrix in a variety of internal organs and skin in adults, type III collagen is critical for fibrillogenesis in the development of apparatus such as skin and cardiovascular system [56, 57]. Oxidative Medicine and Cellular Longevity In its reduced state, the thiol group of cysteine is able to donate a reducing equivalent (H+ + e−) to other unstable molecules, such as ROS, whereas in donating an electron, glutathione itself becomes reactive but is ready to react with another reactive glutathione to form GSSG. erythrocytes samples [26, 35]. Although enhanced plasmatic OS markers levels suggest a systemic OS status, to date no indications are present in order to infer what tissues/cellular systems other than blood could be potentially damaged by the OS imbalance in RTT. A preliminary suggestion that an oxidative process should occur at the cellular level was con- tained in our recent study where oxidative posttranslational modifications on SRB1 receptor in primary fibroblast cultures were observed [36]. Fibroblasts have been used in the RTT scientific research to evaluate gene mutation and epigenetic process [37], differentiation to induced pluripotent stem (iPS) [38], cellular response subsequent to MECP2 mutation [39], reduction of stathmin-like 2 [40], and the effects of NB54 and other rationally designed aminoglycoside derivatives as potential therapeutic agents for nonsense MECP2 mutations in RTT [41]. Fibroblasts have been extensively employed to investigate OS in different pathophysiological processes such as genetic neurodegenerative diseases [42], Parkinson’s disease [43], aging [44], and response to etiological agents [45, 46]. Moreover, lipid peroxidation events are known to be relevant to the fibroblast function, as relationship between mechanotransduction and lipid metabolites has been previ- ously investigated in keloids [47].h erythrocytes samples [26, 35]. Although enhanced plasmatic OS markers levels suggest a systemic OS status, to date no indications are present in order to infer what tissues/cellular systems other than blood could be potentially damaged by the OS imbalance in RTT. A preliminary suggestion that an oxidative process should occur at the cellular level was con- tained in our recent study where oxidative posttranslational modifications on SRB1 receptor in primary fibroblast cultures were observed [36]. Fibroblasts have been used in the RTT scientific research to evaluate gene mutation and epigenetic process [37], differentiation to induced pluripotent stem (iPS) [38], cellular response subsequent to MECP2 mutation [39], reduction of stathmin-like 2 [40], and the effects of NB54 and other rationally designed aminoglycoside derivatives as potential therapeutic agents for nonsense MECP2 mutations in RTT [41]. 4. Discussion Our findings demonstrate, for the first time, the presence of an extensive redox imbalance in primary skin fibroblasts cultures obtained from patients with RTT by adding new evidence to the concept of an OS imbalance as key pheno- typical features of MeCP2 deficiency in RTT. Earlier clues for an abnormal OS balance in RTT patients harbouring MECP2 gene mutations are to date limited to plasma and Conflict of Interests The authors declare that there is no conflict of interests regarding the publication of this paper. Abbreviations 4-HNE: 4-Hydroxy-2-nonenal 4-HNE PAs: 4-Hydroxy-2-nonenal protein adducts AA: Arachidonic acid DHA: Docosahexaenoic acid F2-IsoPs: F2-isoprostanes F4-NeuroPs: F4-neuroprostanes GSH: Reduced glutathione GSSG: Oxidized glutathione IsoPs: Isoprostanes MECP2: Methyl-CpG-binding protein 2—human gene Authors’ Contribution The conception of the paper was performed by Cinzia Sig- norini, Claudio De Felice, Silvia Leoncini, Lucia Ciccoli, Alessandra Renieri, and Joussef Hayek. The experimental design was performed by Cinzia Signorini and Claudio De Felice. MECP2 mutation determinations were performed by Ilaria Meloni, Francesca Ariani, Francesca Mari, Sonia Amabile, and Alessandra Renieri. Clinical assessement was performed by Joussef Hayek, Claudio De Felice, and Alessan- dra Renieri. Skin biopsies were performed by Joussef Hayek. Samples were prepared by Cinzia Signorini, Silvia Leoncini, Alessandra Pecorelli, Gloria Zollo, Giuseppe Belmonte, Mar- iangela Gentile, Eugenio Paccagnini, Sonia Amabile, and Ilaria Meloni. NPBI assays were performed by Silvia Leoncini. Isoprostanes and neuroprostanes assays were performed by Cinzia Signorini, Thierry Durand, and Jean-Marie Galano. 4-HNE-PAs assays were carried out by Alessandra Pecorelli, and Giuseppe Valacchi. Gluthatione assays were carried out by Silvia Leoncini and Gloria Zollo. Immunocytochem- istry was performed by Giuseppe Belmonte and Alessandra Pecorelli. TEM was done by Eugenio Paccagnini and Mari- angela Gentile. Data analysis was done by Claudio De Felice, Cinzia Signorini, and Silvia Leoncini. All the authors equally contributed in data interpretation, paper drafting, and paper approval. Given that OS is known to be a major determinant of aging [73, 74], it is conceivable that aging phenomena in RTT, including skin fibroblasts morphological changes, may be caused through OS mechanisms. Oxidative Medicine and Cellular Longevity Overall in such tissues, type III collagen is colocalized, within the same fibril, with the most abundant member of the family, type I collagen, and regulates the diameter of type I collagen fibrils, which have to meet the physiological requirements of different tissues at different developmental stages [54, 58–62].h Our data demonstrate that both AA and DHA undergo significant oxidation in RTT fibroblasts. Although DHA is known to be particularly abundant in neuronal membranes, The reported ultrastructural changes are to be considered as a new finding to be added to the mitochondrial changes 7 Oxidative Medicine and Cellular Longevity MeCP2: Methyl-CpG-binding protein 2—human protein NPBI: Nonprotein bound iron OS: Oxidative stress PUFAs: Polyunsaturated fatty acids ROS: Reactive Oxygen Species RTT: Rett syndrome TEM: Transmission electron microscopy. previously reported in RTT skin fibroblasts [63]. Since no abnormalities in wound repair and/or healing have been reported in RTT patients, these ultrastructural features are to be considered as subclinical characteristics of the disease. Marked dilation of the rough endoplasmic reticulum cister- nae and cytoplasmic multilamellar bodies were observed for the first time in this study. Dilated rough endoplasmic retic- ulum cisternae can be considered as a nonspecific adaptive response due to either increased secretory activity or as the result of the cellular response to noxae of different nature [64, 65]. On the other hand, the presence of multilamellar bodies could be interpreted as an expression of cellular damage. In particular, those structures suggest the occurrence of autophagy phenomena of unclear pathogenesis [66]. Disclosure Cofirsts: Cinzia Signorini, Silvia Leoncini, Claudio De Felice, and Alessandra Pecorelli. Colasts: Lucia Ciccoli, Alessandra Renieri, and Joussef Hayek. A condition of oxidative stress has been previously reported in aging skin [67]. Skin aging is known to be mainly due to fragmentation/loss of type I collagen fibrils, conferring strength, and resiliency in association with metalloproteinase activation, involved in type I collagen degradation [68]. Although, no evidence of wound healing alterations or derma laxity has not been reported in RTT, an accelerated ageing process is known to occur in the affected patients. Our results suggest the occurrence of a possible reduction in type I collagen, a feature of aging skin, while they show a clear redox imbalance in RTT skin fibroblasts. Therefore, it is conceivable that a premature skin aging may occur in RTT patients, a hypothesis which is in line with prior evidence of premature senescence phenomena in the disease [69–72]. 5. Conclusion Our study demonstrates for the first time the occurrence of lipid peroxidation in RTT fibroblasts, together with a reduced antioxidant cellular defense with a major impact on cell morphology. We speculate about a possible functional involvement of these changes in the skin fibroblasts of RTT patients which must be taken into account when evaluating this cellular model of the disease. Our findings suggest that OS is a generalized phenomenon in RTT, thus affecting cellular systems and tissues apparently unrelated to the central nervous system. References [1] B. Hagberg, J. Aicardi, K. Dias, and O. Ramos, “A progressive syndrome of autism, dementia, ataxia, and loss of purposeful hand use in girls: rett’s syndrome: report of 35 cases,” Annals of Neurology, vol. 14, no. 4, pp. 471–479, 1983. [18] S. J. James, P. Cutler, S. Melnyk et al., “Metabolic biomarkers of increased oxidative stress and impaired methylation capacity in children with autism,” The American Journal of Clinical Nutrition, vol. 80, no. 6, pp. 1611–1617, 2004. [2] T. Bienvenu, C. Philippe, N. De Roux et al., “The incidence of rett syndrome in France,” Pediatric Neurology, vol. 34, no. 5, pp. 372–375, 2006. [19] E. Grillo, C. Lo Rizzo, L. Bianciardi et al., “Revealing the complexity of a monogenic disease: Rett syndrome exome sequencing,” PLoS ONE, vol. 8, no. 2, Article ID e56599, 2013. [3] B. Hagberg, “Clinical manifestations and stages of Rett syn- drome,” Mental Retardation and Developmental Disabilities Research Reviews, vol. 8, no. 2, pp. 61–65, 2002. [20] E. Großer, U. Hirt, O. A. Janc et al., “Oxidative burden and mito- chondrial dysfunction in a mouse model of Rett syndrome,” Neurobiology of Disease, vol. 48, no. 1, pp. 102–114, 2012. [4] R. E. Amir, I. B. Van den Veyver, M. Wan, C. Q. Tran, U. Francke, and H. Y. Zoghbi, “Rett syndrome is caused by mutations in X-linked MECP2, encoding methyl—CpG—binding protein 2,” Nature Genetics, vol. 23, no. 2, pp. 185–188, 1999. [21] C. De Felice, C. Signorini, T. Durand et al., “Partial rescue of Rett syndrome by 𝜔-3 polyunsaturated fatty acids (PUFAs) oil,” Genes and Nutrition, vol. 7, no. 3, pp. 447–458, 2012. [5] J. Christodoulou, A. Grimm, T. Maher, and B. Bennetts, “Ret- tBASE: the IRSA MECP2 variation database-a new mutation database in evolution,” Human Mutation, vol. 21, no. 5, pp. 466– 472, 2003. [22] L. Ciccoli, C. De Felice, E. Paccagnini et al., “Morphological changes and oxidative damage in Rett Syndrome erythrocytes,” Biochimica et Biophysica Acta: General Subjects, vol. 1820, no. 4, pp. 511–520, 2012. [6] B. Halliwell, “Oxidative stress and cancer: have we moved forward?” The Biochemical Journal, vol. 401, no. 1, pp. 1–11, 2007. [23] S. Leoncini, C. De Felice, C. Signorini et al., “Oxidative stress in Rett syndrome: natural history, genotype, and variants,” Redox Report, vol. 16, no. 4, pp. 145–153, 2011. [7] B. Halliwell, “Free radicals and antioxidants—quo vadis?” Trends in Pharmacological Sciences, vol. 32, no. 3, pp. 125–130, 2011. References [24] C. De Felice, C. Signorini, T. Durand et al., “F2-dihomo- isoprostanes as potential early biomarkers of lipid oxidative damage in Rett syndrome,” Journal of Lipid Research, vol. 52, no. 12, pp. 2287–2297, 2011. [8] B. Halliwell, Free Radicals in Biology and Medicine, Oxford University Press, Oxford, UK, 4th edition, 2007. [25] C. De Felice, L. Ciccoli, S. Leoncini et al., “Systemic oxidative stress in classic Rett syndrome,” Free Radical Biology and Medicine, vol. 47, no. 4, pp. 440–448, 2009. [9] C. D. Aluise, R. A. S. Robinson, J. Cai, W. M. Pierce, W. R. Markesbery, and D. A. Butterfield, “Redox proteomics analysis of brains from subjects with amnestic mild cognitive impair- ment compared to brains from subjects with preclinical Alzheimer’s disease: insights into memory loss in MCI,” Journal of Alzheimer’s Disease, vol. 23, no. 2, pp. 257–269, 2011. [26] C. De Felice, C. Signorini, S. Leoncini et al., “The role of oxidative stress in Rett syndrome: an overview,” Annals of the New York Academy of Sciences, vol. 1259, no. 1, pp. 121–135, 2012. [10] A.-M. Enciu, M. Gherghiceanu, and B. O. Popescu, “Triggers and effectors of oxidative stress at blood-brain barrier level: relevance for brain ageing and neurodegeneration,” Oxidative Medicine and Cellular Longevity, vol. 2013, Article ID 297512, 12 pages, 2013. [27] A. Renieri, F. Mari, M. A. Mencarelli et al., “Diagnostic criteria for the Zappella variant of Rett syndrome (the preserved speech variant),” Brain and Development, vol. 31, no. 3, pp. 208–216, 2009. [28] N. Dracopoli, J. Haines, B. Korf et al., Current Protocols in Human Genetics, John Wiley & Sons, New York, NY, USA, 2000. [11] J. C. Brazil, N. A. Louis, and C. A. Parkos, “The role of polymorphonuclear leukocyte trafficking in the perpetuation of inflammation during inflammatory bowel disease,” Inflamma- tory Bowel Diseases, vol. 19, no. 7, pp. 1556–1565, 2013. [29] C. Signorini, L. Ciccoli, S. Leoncini et al., “Free iron, total F2-isoprostanes and total F4-neuroprostanes in a model of neonatal hypoxic-ischemic encephalopathy: neuroprotective effect of melatonin,” Journal of Pineal Research, vol. 46, no. 2, pp. 148–154, 2009. [12] A. Ghezzo, P. Visconti, P. M. Abruzzo et al., “Oxidative stress and erythrocyte membrane alterations in children with autism: correlation with clinical features,” PLoS ONE, vol. 8, no. 6, Article ID e66418, 2013. [30] C. Signorini, M. Comporti, and G. Giorgi, “Ion trap tandem mass spectrometric determination of F2-isoprostanes,” Journal of Mass Spectrometry, vol. 38, no. Oxidative Medicine and Cellular Longevity 8 [16] S. J. James, S. Rose, S. Melnyk et al., “Cellular and mitochondrial glutathione redox imbalance in lymphoblastoid cells derived from children with autism,” FASEB Journal, vol. 23, no. 8, pp. 2374–2383, 2009. professional singer Matteo Setti for his artistic contributions to the exploration of Rett syndrome and his many charity concerts dedicated to the Rett girls and their families. [17] J. Baio, “Prevalence of autism spectrum disorders—autism and developmental disabilities monitoring network, 14 sites, United States, 2008,” Morbidity and Mortality Weekly Report, vol. 61, no. 3, pp. 1–19, 2012. Acknowledgments This work was supported by the Grant no. RF-TOS-2008- 1225570—Bando Malattie Rare to AR. The present research project has been mainly funded by the Italian Health Ministry and Tuscan Region. This research is dedicated to all the Rett girls and their families. The authors thank Dr. Pierluigi Tosi, Dr. Silvia Briani, and Dr. Roberta Croci from the Administrative Direction of the Azienda Ospedaliera Senese for continued support to our studies and for prior purchasing of the gas spectrometry instrumentation. They also thank Oxidative Medicine and Cellular Longevity References Jassal et al., “Generation and character- isation of friedreich ataxia YG8R mouse fibroblast and neural stem cell models,” PLoS ONE, vol. 9, no. 2, Article ID e89488, 2014. [58] E. Vuorio and B. de Crombrugghe, “The family of collagen genes,” Annual Review of Biochemistry, vol. 59, pp. 837–872, 1990. [43] A. Ferretta, A. Gaballo, P. Tanzarella et al., “Effect of resveratrol on mitochondrial function: implications in parkin-associated familiar Parkinson’s disease,” Biochimica et Biophysica Acta, vol. 1842, no. 7, pp. 902–915, 2014. [59] K. Piez and A. Reddi, Eds., Extracellular Matrix Biochemistry, Elsevier, New York, NY, USA, 1984. [60] R. Mayne and R. Burgeson, Eds., Structure and Function of Collagen Types, Academic Press, New York, NY, USA, 1987. [44] D. M. Boesten, J. M. de Vos-Houben, L. Timmermans, G. J. den Hartog, A. Bast, and G. J. Hageman, “Accelerated aging during chronic oxidative stress: a role for PARP-1,” Oxidative Medicine and Cellular Longevity, vol. 2013, Article ID 680414, 10 pages, 2013. [61] P. Bornstein and H. Sage, “Structurally distinct collagen types,” Annual Review of Biochemistry, vol. 49, pp. 957–1003, 1980. [62] R. Fleischmajer, B. R. Olsen, R. Timpl, J. S. Perlish, and O. Lovelace, “Collagen fibril formation during embryogenesis,” Proceedings of the National Academy of Sciences of the United States of America, vol. 80, no. 11, pp. 3354–3358, 1983. [45] G.-Y. Yang, C.-L. Zhang, X.-C. Liu, G. Qian, and D.-Q. Deng, “Effects of cigarette smoke extracts on the growth and senes- cence of skin fibroblasts in vitro,” International Journal of Biological Sciences, vol. 9, no. 6, pp. 613–623, 2013. [63] E. Cardaioli, M. T. Dotti, G. Hayek, M. Zappella, and A. Federico, “Studies on mitochondrial pathogenesis of Rett syn- drome: ultrastructural data from skin and muscle biopsies and mutational analysis at mtDNA nucleotides 10463 and 2835,” Journal of Submicroscopic Cytology and Pathology, vol. 31, no. 2, pp. 301–304, 1999. [46] H. Yang, C. Liu, D. Yang, H. Zhang, and Z. Xi, “Comparative study of cytotoxicity, oxidative stress and genotoxicity induced by four typical nanomaterials: the role of particle size, shape and composition,” Journal of Applied Toxicology, vol. 29, no. 1, pp. 69–78, 2009. [47] C. Huang and R. Ogawa, “Roles of lipid metabolism in keloid development,” Lipids in Health and Disease, vol. 12, no. 1, article 60, 2013. [64] G. Stabellini, G. Berti, C. References 10, pp. 1067–1074, 2003. [13] L. Ciccoli, C. De Felice, E. Paccagnini et al., “Erythrocyte shape abnormalities, membrane oxidative damage, and 𝛽-actin alter- ations: an unrecognized triad in classical autism,” Mediators of Inflammation, vol. 2013, Article ID 432616, 11 pages, 2013. [31] C. Signorini, C. De Felice, S. Leoncini et al., “F4-neuroprostanes mediate neurological severity in Rett syndrome,” Clinica Chim- ica Acta, vol. 412, no. 15-16, pp. 1399–1406, 2011. [14] A. Pecorelli, S. Leoncini, C. De Felice et al., “Non-protein- bound iron and 4-hydroxynonenal protein adducts in classic autism,” Brain and Development, vol. 35, no. 2, pp. 146–154, 2013. [32] F. Tietze, “Enzymic method for quantitative determination of nanogram amounts of total and oxidized glutathione: appli- cations to mammalian blood and other tissues,” Analytical Biochemistry, vol. 27, no. 3, pp. 502–522, 1969. [15] A. Chauhan, T. Audhya, and V. Chauhan, “Brain region- specific glutathione redox imbalance in autism,” Neurochemical Research, vol. 37, no. 8, pp. 1681–1689, 2012. [33] M. A. Baker, G. J. Cerniglia, and A. Zaman, “Microtiter plate assay for the measurement of glutathione and glutathione Oxidative Medicine and Cellular Longevity 9 disulfide in large numbers of biological samples,” Analytical Biochemistry, vol. 190, no. 2, pp. 360–365, 1990. [49] L. J. Roberts II, J. P. Fessel, and S. S. Davies, “The biochemistry of the isoprostane, neuroprostane, and isofuran pathways of lipid peroxidation,” Brain Pathology, vol. 15, no. 2, pp. 143–148, 2005. [34] G. Valacchi, A. Pecorelli, C. Signorini et al., “4HNE protein adducts in autistic spectrum disorders: rett syndrome and autism,” in Comprehensive Guide to Autism, V. B. Patel, V. R. Preedy, and C. R. Martin, Eds., pp. 2667–2687, Springer, New York, NY, USA, 2014. [50] L. Ciccoli, S. Leoncini, C. Signorini, and M. Comporti, “Iron and erythrocytes: physiological and pathophysiological aspects,” in Oxidants in Biology, G. Valacchi and P. Davis, Eds., pp. 167–181, Springer, Heidelberg, Germany, 2008. [35] C. Signorini, C. De Felice, T. Durand et al., “Isoprostanes and 4-hydroxy-2-nonenal: markers or mediators of disease? focus on Rett syndrome as a model of autism spectrum disorder,” Oxidative Medicine and Cellular Longevity, vol. 2013, Article ID 343824, 10 pages, 2013. [51] S. A. Moore, E. Hurt, E. Yoder, H. Sprecher, and A. A. Spector, “Docosahexaenoic acid synthesis in human skin fibroblasts involves peroxisomal retroconversion of tetracosahexaenoic acid,” Journal of Lipid Research, vol. 36, no. 11, pp. 2433–2443, 1995. [52] E. R. Brown and P. V. References Subbaiah, “Differential effects of eicos- apentaenoic acid and docosahexaenoic acid on human skin fibroblasts,” Lipids, vol. 29, no. 12, pp. 825–829, 1994. [36] C. Sticozzi, G. Belmonte, A. Pecorelli et al., “Scavenger receptor B1 post-translational modifications in Rett syndrome,” FEBS Letters, vol. 587, no. 14, pp. 2199–2204, 2013. [53] S. P. Mahadik, S. Mukherjee, D. F. Horrobin, K. Jenkins, E. E. Correnti, and R. E. Scheffer, “Plasma membrane phospho- lipid fatty acid composition of cultured skin fibroblasts from schizophrenic patients: comparison with bipolar patients and normal subjects,” Psychiatry Research, vol. 63, no. 2-3, pp. 133– 142, 1996. [37] K. Miyake, C. Yang, Y. Minakuchi et al., “Comparison of genomic and epigenomic expression in monozygotic twins discordant for rett syndrome,” PLoS ONE, vol. 8, no. 6, Article ID e66729, 2013. [38] M. Amenduni, R. De Filippis, A. Y. L. Cheung et al., “IPS cells to model CDKL5-related disorders,” European Journal of Human Genetics, vol. 19, no. 12, pp. 1246–1255, 2011. [54] E. H. Epstein Jr. and N. H. Munderloh, “Isolation and char- acterization of CNBr peptides of human [𝛼1(III)]3 collagen and tissue distribution of [𝛼1(I)]2𝛼2 and [𝛼1(III)]3 collagens,” Journal of Biological Chemistry, vol. 250, no. 24, pp. 9304–9312, 1975. [39] C. Del´epine, J. Nectoux, N. Bahi-Buisson, J. Chelly, and T. Bienvenu, “MeCP2 deficiency is associated with impaired microtubule stability,” FEBS Letters, vol. 587, no. 2, pp. 245–253, 2013. [55] L.-J. Min, T.-X. Cui, Y. Yahata et al., “Regulation of collagen synthesis in mouse skin fibroblasts by distinct angiotensin II receptor subtypes,” Endocrinology, vol. 145, no. 1, pp. 253–260, 2004. [40] J. Nectoux, C. Florian, C. Delepine et al., “Altered microtubule dynamics in Mecp2-deficient astrocytes,” Journal of Neuro- science Research, vol. 90, no. 5, pp. 990–998, 2012. [41] M. Vecsler, B. Ben Zeev, I. Nudelman et al., “Ex vivo treatment with a novel synthetic aminoglycoside NB54 in primary fibrob- lasts from Rett syndrome patients suppresses MECP2 nonsense mutations,” PLoS ONE, vol. 6, no. 6, Article ID e20733, 2011. [56] B. R. Olsen, “The roles of collagen genes in skeletal development and morphogenesis,” Experientia, vol. 51, no. 3, pp. 194–195, 1995. [57] R. S. Wong, F. M. Follis, B. K. Shively, and J. A. Wernly, “Osteo- genesis imperfecta and cardiovascular diseases,” The Annals of Thoracic Surgery, vol. 60, no. 5, pp. 1439–1443, 1995. [42] C. Sandi, M. Sandi, H. References Calastrini et al., “Experimental model for studying the effects of 2-ethylhexyl-phthalate and dialysate on connective tissue,” The International Journal of Artificial Organs, vol. 23, no. 5, pp. 305–311, 2000. [48] E. S. Musiek, H. Yin, G. L. Milne, and J. D. Morrow, “Recent advances in the biochemistry and clinical relevance of the isoprostane pathway,” Lipids, vol. 40, no. 10, pp. 987–994, 2005. [65] G. Ellender and K. N. Ham, “Connective tissue responses to some heavy metals. III silver and dietary supplements of Oxidative Medicine and Cellular Longevity 10 ascorbic acid. Histology and ultrastructure,” British Journal of Experimental Pathology, vol. 70, no. 1, pp. 21–39, 1989. [66] V. M. Korkhov, “GFP-LC3 labels organised smooth endo- plasmic reticulum membranes independently of autophagy,” Journal of Cellular Biochemistry, vol. 107, no. 1, pp. 86–95, 2009. [67] G. J. Fisher, T. Quan, T. Purohit et al., “Collagen frag- mentation promotes oxidative stress and elevates matrix metalloproteinase-1 in fibroblasts in aged human skin,” Amer- ican Journal of Pathology, vol. 174, no. 1, pp. 101–114, 2009. [68] W. Xia, C. Hammerberg, Y. Li et al., “Expression of catalytically active matrix metalloproteinase-1 in dermal fibroblasts induces collagen fragmentation and functional alterations that resemble aged human skin,” Aging Cell, vol. 12, no. 4, pp. 661–671, 2013. [69] B. Hagberg, “Rett syndrome: long-term clinical follow-up expe- riences over four decades,” Journal of Child Neurology, vol. 20, no. 9, pp. 722–727, 2005. [70] N. Halbach, E. Smeets, C. Steinbusch, M. Maaskant, D. van Waardenburg, and L. Curfs, “Aging in Rett syndrome: a longitu- dinal study,” Clinical Genetics, vol. 84, no. 3, pp. 223–229, 2013. [71] C. S. Ward, E. M. Arvide, T.-W. Huang, J. Yoo, J. L. Noebels, and J. L. Neul, “MeCP2 is critical within HoxB1-derived tissues of mice for normal lifespan,” Journal of Neuroscience, vol. 31, no. 28, pp. 10359–10370, 2011. [72] T. Squillaro, N. Alessio, M. Cipollaro et al., “Reduced expression of MECP2 affects cell commitment and maintenance in neurons by triggering senescence: new perspective for Rett syndrome,” Molecular Biology of the Cell, vol. 23, no. 8, pp. 1435–1445, 2012. [73] S. Dato, P. Crocco, P. D. ’Aquila et al., “Exploring the role of genetic variability and lifestyle in oxidative stress response for healthy aging and longevity,” International Journal of Molecular Sciences, vol. 14, no. 8, pp. 16443–16472, 2013. [74] S. I. References Liochev, “Reactive oxygen species and the free radical theory of aging,” Free Radical Biology and Medicine, vol. 60, pp. 1–4, 2013.
https://openalex.org/W2974853085	https://hess.copernicus.org/articles/24/1677/2020/hess-24-1677-2020.pdf	English	null	On the assimilation of environmental tracer observations for model-based decision support	Hydrology and earth system sciences	2,020	cc-by	10,573	Correspondence: Matthew J. Knowling (mjknowling@gmail.com) Received: 21 August 2019 – Discussion started: 18 September 2019 Revised: 29 February 2020 – Accepted: 12 March 2020 – Published: 8 April 2020 Received: 21 August 2019 – Discussion started: 18 September 2019 Revised: 29 February 2020 – Accepted: 12 March 2020 – Published: 8 April 2020 Abstract. It has been advocated that history matching nu- merical models to a diverse range of observation data types, particularly including environmental tracer concentrations and their interpretations and derivatives (e.g., mean age), constitutes an effective and appropriate means to improve model forecast reliability. This study presents two regional- scale modeling case studies that directly and rigorously as- sess the value of discrete tritium concentration observations and tritium-derived mean residence time (MRT) estimates in two decision-support contexts; “value” is measured herein as both the improvement (or otherwise) in the reliability of forecasts through uncertainty variance reduction and bias minimization as a result of assimilating tritium or tritium- derived MRT observations. The ﬁrst case study (Heretaunga Plains, New Zealand) utilizes a suite of steady-state and tran- sient ﬂow models and an advection-only particle-tracking model to evaluate the worth of tritium-derived MRT esti- mates relative to hydraulic potential, spring discharge and river–aquifer exchange ﬂux observations. The worth of MRT observations is quantiﬁed in terms of the change in the un- certainty surrounding ecologically sensitive spring discharge forecasts via ﬁrst-order second-moment (FOSM) analyses. The second case study (Hauraki Plains, New Zealand) em- ploys paired simple–complex transient ﬂow and transport models to evaluate the potential for assimilation-induced bias in simulated surface-water nitrate discharge to an ecologi- cally sensitive estuary system; formal data assimilation of tritium observations is undertaken using an iterative ensem- ble smoother. The results of these case studies indicate that, for the decision-relevant forecasts considered, tritium obser- vations are of variable beneﬁt and may induce damaging bias in forecasts; these biases are a result of an imperfect model’s inability to properly and directly assimilate the rich informa- tion content of the tritium observations. The ﬁndings of this study challenge the advocacy of the increasing use of trac- ers, and of diverse data types more generally, whenever en- vironmental model data assimilation is undertaken with im- perfect models. This study also highlights the need for im- proved imperfect-model data assimilation strategies. Correspondence: Matthew J. Knowling (mjknowling@gmail.com) While these strategies will likely require increased model complex- ity (including advanced discretization, processes and param- eterization) to allow for appropriate assimilation of rich and diverse data types that operate across a range of spatial and temporal scales commensurate with a forecast of manage- ment interest, it is critical that increased model complexity does not preclude the application of formal data assimilation and uncertainty quantiﬁcation techniques due to model insta- bility and excessive run times. 1 Introduction Numerical models used to provide water resources manage- ment decision support are often subjected to data assimilation through history matching (or “calibration”). This is due to the large information deﬁcit accompanying the development of these models and the potential for the history-matching pro- cess to lead to an increased reliability of simulated outputs of management interest (herein referred to as “forecasts”) through variance reduction. Modeling for the purpose of de- cision support is the context in which the remainder of the paper is framed. Hydrol. Earth Syst. Sci., 24, 1677–1689, 2020 https://doi.org/10.5194/hess-24-1677-2020 © Author(s) 2020. This work is distributed under the Creative Commons Attribution 4.0 License. Hydrol. Earth Syst. Sci., 24, 1677–1689, 2020 https://doi.org/10.5194/hess-24-1677-2020 © Author(s) 2020. This work is distributed under the Creative Commons Attribution 4.0 License. On the assimilation of environmental tracer observations for model-based decision support M tth J K li 1 J T Whit 1 C th i R M 1 P l R k ki2 d K i H l 3 Matthew J. Knowling1, Jeremy T. White1, Catherine R. Moore1, Pawel Rakowski2, and Kevin Hayley3 1GNS Science, Lower Hutt, New Zealand 2Hawke’s Bay Regional Council, Napier, New Zealand 3Groundwater Solutions Ltd, Melbourne, Australia M. J. Knowling et al.: Model tracer-data assimilation for decision support In a review of approaches for model- ing environmental tracers in groundwater systems, Turnadge and Smerdon (2014) state that age data have been useful for constraining models; in particular, “simulation of envi- ronmental tracer transport that explicitly accounts for the accumulation and decay of tracer mass, has proven to be highly beneﬁcial in constraining numerical models”. Zell et al. (2018) showed the relative importance of water level, stream discharge and environmental tracers (including tri- tium, chloroﬂuorocarbons – CFCs and SF6) in the condi- tioning of groundwater travel time forecasts. They reported that, overall, tracer data were of considerable beneﬁt in terms of forecast uncertainty reduction. In a recent review paper, Schilling et al. (2019) state that assimilation of concentra- tion observations through surface-water–groundwater ﬂow model history matching “harbors huge potential”, based on the ﬁndings of previous studies, while the assimilation of tracer-derived residence time observations in these models also often help signiﬁcantly (where an appropriate approach is adopted; e.g., Sanford, 2011; Zuber et al., 2011). Herein, we focus speciﬁcally on the ramiﬁcations of as- similating the information contained within tritium con- centration observations and tritium-derived mean residence time (MRT) observations for model-based decision support concerning low ﬂow and nutrient transport at the regional scale in hydrological environments where young groundwa- ter components are decision relevant. Tritium is a popular tracer for the identiﬁcation of relatively young-age ground- waters (i.e., < 70 years old), for the following reasons: (i) unlike CFCs, tritium is not affected by microbial degrada- tion or contamination and (ii) unlike SF6, it is not affected by potential subsurface sources (e.g., Morgenstern and Daugh- ney, 2012; Cartwright and Morgenstern, 2012; Beyer et al., 2014). However, the extent to which the assimilation of diverse data types (including environmental tracers) is beneﬁcial has previously been investigated only from a somewhat theoreti- cal standpoint, i.e., neglecting the effects of model error. Di- rect evaluation of the likelihood term of Bayes’ theorem is predicated on a “perfect” simulator to appropriately condi- tion uncertain model parameters through data assimilation. In real-world modeling contexts, however, the presence of model error can invalidate even the most rigorous data as- similation techniques (e.g., Doherty and Welter, 2010; White et al., 2014; Oliver and Alfonzo, 2018). M. J. Knowling et al.: Model tracer-data assimilation for decision support 1678 It is widely advocated that the assimilation of multiple types of state observations (i.e., “diverse data”) is beneﬁ- cial in “constraining” models. In other words, as more data are used for history matching, and the more diverse those data are, the more reliable of the forecasts become. This is an intuitive stance arising from direct application of Bayes’ equation and from the recognized rich information content of diverse data types; this intuition is supported by many stud- ies (e.g., Sanford et al., 2004; Michael and Voss, 2009; Ginn et al., 2009; Li et al., 2009; Gusyev et al., 2013; Hansen et al., 2013). For example, Hunt et al. (2006) demonstrated the importance of unconventional observations including lake– aquifer exchange ﬂuxes, the depth of lake isotope plume and groundwater travel times in achieving “well-constrained pa- rameter values” (e.g., acceptable posterior variance) through history matching a regional-scale groundwater model. lation problem can result in severely biased results (e.g., Do- herty and Christensen, 2011; Knowling et al., 2019; White et al., 2020). The largely unknown ability of an imperfect regional-scale model to simultaneously assimilate diverse data types that operate over different spatial and temporal scales – and how these imperfections may affect model-based decision support in some contexts – serves as the motivation for the current study. To the best of the authors’ knowledge, this is the ﬁrst study to explore the beneﬁt or otherwise of the assimilation of tracer data into imperfect models in terms of both forecast bias and variance. A subtle, yet very important, distinction should be made at this point. There is no doubting that diverse data types, in par- ticular environmental tracers, have contributed signiﬁcantly to the understanding of catchment processes and properties (e.g., Kirchner et al., 2001; André et al., 2005; Stewart and Thomas, 2008; McDonnell et al., 2010; Morgenstern et al., 2010; Han et al., 2012; Leray et al., 2012; Siade et al., 2018). However, as discussed, this study focuses instead on the role of (imperfect) models in two selected decision-support con- texts and how the assimilation of environmental tracers in particular affects their utility in these contexts, i.e., by in- creasing (or otherwise) the reliability of forecasts. y g g g History matching to environmental tracer observations, in particular, is a widely regarded mechanism for improving the reliability of forecasts. Published by Copernicus Publications on behalf of the European Geosciences Union. Published by Copernicus Publications on behalf of the European Geosciences Union. M. J. Knowling et al.: Model tracer-data assimilation for decision support M. J. Knowling et al.: Model tracer-data assimilation for decision support Therefore, when an imperfect simulator is used in a data assimilation framework, extreme care must be taken to assure that the model imper- fections do not corrupt (through biased ﬁrst moments or un- derestimated second moments) the forecast posterior distri- butions. A number of recent works have shown that the fail- ure to appropriately frame the imperfect-model data assimi- The objective of this study is twofold. First, we investi- gate the theoretical worth of tritium-derived MRT observa- tions relative to other observation data types. This investiga- tion is performed using a case study (Heretaunga Plains, New Zealand) that adopts ﬁrst-order second-moment (FOSM) techniques; our analysis focuses on the relative worth of MRT observations in terms of changes in the uncertainty as- sociated with spring discharge forecasts at various locations that are of management interest due to their ecological signif- icance. This ﬁrst case study employs advective-only particle- tracking modeling approach to simulate MRT. Second, we explore the use of discrete tritium concentra- tion observations in data assimilation in the context of a con- trolled model simpliﬁcation experiment as a means to un- derstand what, if any, ill effects may be induced by using these information-rich data types in a simpliﬁed (i.e., imper- fect) model. This exploration is performed using a second www.hydrol-earth-syst-sci.net/24/1677/2020/ Hydrol. Earth Syst. Sci., 24, 1677–1689, 2020 1679 M. J. Knowling et al.: Model tracer-data assimilation for decision support tionship between ﬁeld observations of spring discharge and groundwater levels. case study that employs a recently presented paired simple– complex model analysis (White et al., 2020). The paired model analysis is used herein to allow for the identiﬁcation of possible (and otherwise undetectable) bias and uncertainty underestimation surrounding forecasts of nutrient load to an ecologically sensitive estuary system. This second case study simulates (tritium and nitrate) tracer concentrations directly – using a full advective–dispersive modeling approach that also accounts for ﬁrst-order reaction rates. – The coastal boundary condition is represented using the (head-dependent ﬂux) general-head boundary package. The general-head stage is speciﬁed using a density- corrected mean sea level (e.g., Morgan et al., 2012). – Groundwater abstraction rates, based on observed and estimated data, are represented using the (speciﬁed ﬂux) well package. For a more detailed description of the Heretaunga Plains models, the reader is referred to Rakowski and Knowling (2018). 2.2 Forecasts We focus on the following forecasts due to their ecological signiﬁcance and their potential to be impacted by groundwa- ter abstraction: – spring discharge rate during summer at two locations (one in the central Heretaunga Plains and one in the up- per reaches of the catchment) (Fig. 1) 2 First case study The ﬁrst case study serves to investigate the ability of tritium- derived MRT observations to constrain ecologically sensitive spring discharge forecasts (i.e., the “worth” of these observa- tions) using a model of the groundwater system of the Here- taunga Plains (New Zealand; Fig. 1). The model was con- structed primarily for the purposes of groundwater allocation management decision support. 2.1 The model The model comprises 302 rows and 501 columns (uniform 100m×100m horizontal grid discretization). Two layers are used for ﬂow simulations, whereas six layers are used to gen- erate more vertically detailed cell-by-cell ﬂow budgets for particle-tracking simulations. MODFLOW-2005 (Harbaugh, 2005) is used to simulate groundwater ﬂow under steady- state and transient conditions. Separate simulations are con- ducted for data assimilation and forecasting purposes span- ning different time periods (and temporal resolutions) of in- terest (e.g., separate transient ﬂow simulations are conducted using annual stress periods for the period 1980–2015 and us- ing monthly stress periods for the periods 1997–1999 and 2011–2015). MODPATH (Pollock, 2012) is used to simulate advection-only (i.e., neglecting diffusion, dispersion and re- tardation) reverse particle tracking, thereby providing a ba- sis for assimilating tritium-derived MRT estimates (Fig. 1). Speciﬁcally, the mean particle exit time corresponding to each observation location is compared with tritium-derived MRT estimates (e.g., Sanford, 2011; Gusyev et al., 2014). – spring discharge rate during winter at the central Here- taunga Plains location (Fig. 1). 2.3 Observations for assimilation Data assimilation is undertaken notionally via FOSM tech- niques using the following observations: – 6167 groundwater levels (comprising time-averaged water levels; absolute and deviation-from-mean annual, monthly and daily water levels; long-term differences in water level; and vertical head differences); – 92 surface-water–groundwater ﬂuxes (time-averaged and transient river gain and loss ﬂuxes and spring dis- charge ﬂuxes, obtained using a range of techniques in- cluding ﬂow gauging, electrical conductivity and tem- perature surveys, water isotopic analyses, etc.; Wilding, 2017); and Relevant aspects of the model are the following: – 52 groundwater MRT estimates derived from tritium concentrations using lumped-parameter models. Specif- ically, a combination of exponential piston-ﬂow models (EPMs) and binary-mixing models (BMMs) (that com- prise two EPMs) were used. BMMs were employed for wells where long time series data are available for mul- tiple tracers and where an adequate ﬁt to different tracer signals could not be obtained on the basis of a single EPM. Relative EPM mixing fractions were speciﬁed on the basis of aquifer conﬁnement conditions and well screen length (mixing fractions of 80 %–95 % were ap- plied for wells with a long screen in unconﬁned con- ditions, whereas mixing fractions of 50 %–60 % were – Land-surface recharge estimates, derived from a daily soil water balance modeling assessment (Rajanayaka and Fisk, 2018), are speciﬁed using the (speciﬁed ﬂux) recharge package. – The interaction between groundwater and surface wa- ter (including rivers, streams and springs) is simulated using the (head-dependent ﬂux) river package. Time- varying river stage values are speciﬁed for the three main rivers in the region based on observed values. River bed conductance values are varied seasonally to reﬂect in an approximate manner the non-linear rela- www.hydrol-earth-syst-sci.net/24/1677/2020/ Hydrol. Earth Syst. Sci., 24, 1677–1689, 2020 M. J. Knowling et al.: Model tracer-data assimilation for decision support 1680 1680 M. J. Knowling et al.: Model tracer-data assimilation for decision support Figure 1. Heretaunga Plains model schematic, including the river boundary conditions (blue lines), top-layer conﬁnement status (unconﬁned areas shaded blue and conﬁned areas shaded yellow) and the coastal general-head boundary (magenta line). The location of groundwater tritium-derived MRT observations are shown as red triangles. The location of forecasts – spring discharge rates during summer and winter – are shown as white markers. Figure 1. Heretaunga Plains model schematic, including the river boundary conditions (blue lines), top-layer conﬁnement status (unconﬁned areas shaded blue and conﬁned areas shaded yellow) and the coastal general-head boundary (magenta line). The location of groundwater tritium-derived MRT observations are shown as red triangles. The location of forecasts – spring discharge rates during summer and winter – are shown as white markers. The theoretical underpinnings of FOSM-based uncertainty quantiﬁcation and data-worth assessment and details related to its application herein are presented in Appendix A. The theoretical underpinnings of FOSM-based uncertainty quantiﬁcation and data-worth assessment and details related to its application herein are presented in Appendix A. applied for wells with shorter screens in conﬁned con- ditions). The reader is referred to Morgenstern et al. (2018) for more details. Aspects that are relevant to the application of FOSM herein include: A highly parameterized approach was adopted (e.g., Hunt et al., 2007; Knowling et al., 2019), involving a total of 822 uncertain parameters. Spatially distributed parameteri- zation of hydraulic conductivity (horizontal and horizontal– vertical anisotropy ratio), effective porosity, speciﬁc storage and speciﬁc yield is achieved using pilot points (e.g., Do- herty, 2003). Spatially distributed river bed and boundary conductance parameters are deﬁned on a reach and zone ba- sis, respectively. We refer the reader to the Supplement for more information. – The prior parameter covariance matrix 6θ was spec- iﬁed as a block-diagonal matrix whereby geostatisti- cal correlation between pilot-point-based spatially dis- tributed parameters is represented through the use of an exponential variogram with a range of approximately 10 000 m and a sill proportional to the expected prior variance (the range of the square root of the diagonal elements of 6θ; i.e., the standard deviation of prior pa- rameter uncertainty is given in the Supplement). www.hydrol-earth-syst-sci.net/24/1677/2020/ Non- spatially and non-temporally distributed parameters are assumed to be uncorrelated and therefore occupy diag- onal matrix elements only. M. J. Knowling et al.: Model tracer-data assimilation for decision suppor dom measurement errors and model simpliﬁcation er- rors) based on model residuals (e.g., Doherty, 2015). tions. This i mation conta 1681 tions. This is generally in contrast to the more unique infor- mation contained within spring discharge observations. 3 Second case study This is due to the fact that the forecast is located where the aquifer is unconﬁned and receives rainfall and river recharge: these recharge rates are informed by MRT observations and have a large inﬂuence on the forecast. Similar results from a relative perspective are apparent for the summer–spring discharge forecast in the upper por- tion of the Heretaunga Plains. That is, transient head ob- servations and spring discharge observations during summer are of the highest worth, followed by observations of time- averaged heads, MRT and winter–spring discharge (Fig. 2b) – for reasons described above. The greater worth of MRT ob- servations for this forecast compared to the summer–spring discharge forecast located down gradient indicates that this forecast is more sensitive to (uncertain) model parameters that are conditioned through assimilating MRT observations. This is due to the fact that the forecast is located where the aquifer is unconﬁned and receives rainfall and river recharge: these recharge rates are informed by MRT observations and have a large inﬂuence on the forecast. Similar results from a relative perspective are apparent for the summer–spring discharge forecast in the upper por- tion of the Heretaunga Plains. That is, transient head ob- servations and spring discharge observations during summer are of the highest worth, followed by observations of time- averaged heads, MRT and winter–spring discharge (Fig. 2b) 3.1 The model The linked hydrologic-nutrient transport model simulates groundwater and surface-water ﬂow using MODFLOW- NWT (Niswonger et al., 2011); advective and dispersive transport of nitrate and tritium in the groundwater and surface-water system is simulated using the MT3D-USGS model (Bedekar et al., 2016). Denitriﬁcation and radioactive tritium decay processes are simulated using ﬁrst-order reac- tion rates. The model is described in detail in White (2018), and the vertical-discretization simpliﬁcation analysis is de- scribed in detail in White et al. (2020). – for reasons described above. The greater worth of MRT ob- servations for this forecast compared to the summer–spring discharge forecast located down gradient indicates that this forecast is more sensitive to (uncertain) model parameters that are conditioned through assimilating MRT observations. This is due to the fact that the forecast is located where the aquifer is unconﬁned and receives rainfall and river recharge: these recharge rates are informed by MRT observations and have a large inﬂuence on the forecast. For the winter–spring discharge forecast, the worth of MRT observations is lower than that of other observations (Fig. 2c). This indicates a low relevance of the spatially and temporally integrated information contained in MRT observations with respect to a forecast concerning higher- frequency and higher-magnitude signals. This is also sup- ported by the relatively low worth of the time-averaged head observations due to the temporally integrated nature of these quantities. As expected, a signiﬁcantly greater worth of spring discharge observations during winter is evident for this forecast due to the unique and directly relevant infor- mation content associated with discharge observations that capture high-ﬂow transience signals. Herein, we focus on a single forecast: the cumulative load of nitrate discharging from the surface-water system to the Firth of Thames – an ecologically sensitive estuary system – over a 10-year projection scenario involving present-day (2018) ﬂow and transport model forcing conditions. This forecast aggregates ﬂow paths across the entire model do- main (i.e., it represents the only nitrate-ﬂux sink of the sys- tem). This forecast is referred to herein as the “Firth fore- cast”. 2.4 Uncertainty quantiﬁcation and data-worth exploration Here we employ FOSM techniques (e.g., Tarantola, 2005; Doherty, 2015) to investigate the theoretical worth of var- ious observation data types in terms of the their inﬂuence on the uncertainty variance surrounding forecasts following data assimilation. Application of FOSM in this context re- quires only consideration of the relative differences in esti- mated forecast variance as a result of conditioning on dif- ferent observation data types. The use of FOSM in relative contexts has been shown to be especially robust (e.g., Daus- man et al., 2010; Herckenrath et al., 2011; Knowling et al., 2019). – The Jacobian matrix J was populated using 1 % two- point derivative increments. – The diagonal elements of the epistemic noise covariance matrix 6ϵ (see Appendix A) was speciﬁed on the basis of observation “weights” adjusted in such a way that the measurement objective function equals the number of non-zero weighted observations, in order to approxi- mate epistemic noise (i.e., the combined impact of ran- www.hydrol-earth-syst-sci.net/24/1677/2020/ Hydrol. Earth Syst. Sci., 24, 1677–1689, 2020 www.hydrol-earth-syst-sci.net/24/1677/2020/ 3 Second case study For the summer–spring discharge forecast in the central Heretaunga Plains, MRT observations display a worth that is considerably less than that of spring discharge observa- tions during the summer months (i.e., when lower ﬂows per- sist) and transient head observations (Fig. 2a). This is not surprising given that the forecast and the summer–spring discharge observations are of the same type and represent the same temporal condition, and transient head observa- tions are plentiful (5704), spanning different time periods at annual, monthly and daily resolutions. The worth of MRT observations is greater than winter–spring discharge obser- vations, indicating a higher relevance of the spatially and temporally integrated information contained within MRT ob- servations for this low-ﬂow-related prediction compared to the higher-frequency and higher-magnitude signals captured within spring discharge observations during winter. The second case study serves to evaluate how assimilating discrete groundwater tritium concentration observations may affect the robustness of forecasts in the context of a con- trolled model simpliﬁcation experiment, where the simpliﬁ- cation is related to model vertical discretization (we refer the reader to White et al., 2020, for an exploration of the appro- priateness of reduced-discretization models in decision sup- port more generally). In contrast to the ﬁrst case study, which focused on the theoretical worth of derived tritium observa- tions in terms of changes in forecast variance, this case study proceeds with repeated data assimilation in a paired simple– complex model analysis both with and without assimilating tritium observations. Through these paired-model analyses, any potential biases or underestimation of variances arising from the assimilation of tritium observations with a sim- pliﬁed model can be exposed. A linked hydrologic-nutrient transport model of the Hauraki Plains (New Zealand) (Fig. 3) is used as a basis for the model simpliﬁcation experiment. p g g g Similar results from a relative perspective are apparent for the summer–spring discharge forecast in the upper por- tion of the Heretaunga Plains. That is, transient head ob- servations and spring discharge observations during summer are of the highest worth, followed by observations of time- averaged heads, MRT and winter–spring discharge (Fig. 2b) – for reasons described above. The greater worth of MRT ob- servations for this forecast compared to the summer–spring discharge forecast located down gradient indicates that this forecast is more sensitive to (uncertain) model parameters that are conditioned through assimilating MRT observations. 3.2 Data assimilation and uncertainty quantiﬁcation As described in White et al. (2020), data assimilation was undertaken via history matching three versions of the model, each with a different vertical-discretization scheme; his- tory matching was performed using the iterative ensemble smoother PESTPP-IES (White, 2018). Across the three forecasts, a signiﬁcantly larger worth is evident when MRT observations are added to the observa- tion dataset compared to when MRT observations are re- moved from the observation dataset (red versus blue; Fig. 2). This indicates that correlation occurs between the informa- tion contained within MRT observations and other observa- History matching was conducted using 100 stochastic pa- rameter realizations. An ensemble size of 100 was deemed www.hydrol-earth-syst-sci.net/24/1677/2020/ www.hydrol-earth-syst-sci.net/24/1677/2020/ Hydrol. Earth Syst. Sci., 24, 1677–1689, 2020 M. J. Knowling et al.: Model tracer-data assimilation for decision support 1682 1682 M. J. Knowling et al.: Model tracer-data assimilation for decision support Figure 2. Worth of different observation groups (%DW) with respect to forecasts: (a) spring discharge ﬂux during summer in the central Heretaunga Plains, (b) spring discharge ﬂux during summer in the upper portion of the Heretaunga Plains and (c) spring discharge ﬂux during winter in the central Heretaunga Plains (see Fig. 1 for locations). The %DW value is quantiﬁed as both the increase in forecast uncertainty variance following the removal of an observation group available for conditioning (red) and the decrease in forecast uncertainty variance following the addition of an observation group available for conditioning (blue) (see Appendix A). Note the different scales on the y axes. Figure 2. Worth of different observation groups (%DW) with respect to forecasts: (a) spring discharge ﬂux during summer in the central Heretaunga Plains, (b) spring discharge ﬂux during summer in the upper portion of the Heretaunga Plains and (c) spring discharge ﬂux during winter in the central Heretaunga Plains (see Fig. 1 for locations). The %DW value is quantiﬁed as both the increase in forecast uncertainty variance following the removal of an observation group available for conditioning (red) and the decrease in forecast uncertainty variance following the addition of an observation group available for conditioning (blue) (see Appendix A). Note the different scales on the y axes. Figure 2. Worth of different observation groups (%DW) with respect to forecasts: (a) spring discharge ﬂux during summer in the central Heretaunga Plains, (b) spring discharge ﬂux during summer in the upper portion of the Heretaunga Plains and (c) spring discharge ﬂux during winter in the central Heretaunga Plains (see Fig. 1 for locations). The %DW value is quantiﬁed as both the increase in forecast uncertainty variance following the removal of an observation group available for conditioning (red) and the decrease in forecast uncertainty variance following the addition of an observation group available for conditioning (blue) (see Appendix A). Note the different scales on the y axes. for more information on parameterization and construc- tion of prior parameter covariance matrices. www.hydrol-earth-syst-sci.net/24/1677/2020/ sufﬁcient to avoid underutilization of observation data (i.e., “underﬁtting”) based on an exploration of the solution-space dimensionality using a subspace analysis (Moore and Do- herty, 2005; see the Supplement and Knowling et al., 2019, for more details). Following history matching, the 10-year projection scenario was evaluated with the 100 history- matched realizations (effectively a 100-member sample of the posterior distribution). From the resulting 100 scenario evaluations, a posterior probability density function (PDF) of the First forecast was constructed. – Observation data for assimilation. The history- matching experiments included 20 tritium concentration observations from the groundwater system (Fig. 3; see also the Supplement for observation locations per model layer). Other observations such as long-term averaged groundwater levels and surface-water ﬂows and tran- sient surface-water and groundwater nitrate concentra- tions were also used for history matching (see the Sup- plement for observation locations). The reader is referred to White (2018) and White et al. (2020) for a full description of the Hauraki Plains model data assimilation process; a brief overview is nevertheless pro- vided as follows: As shown in White et al. (2020), the reduced-discretization (one-layer and two-layer) model posterior PDFs for the Firth forecast display signiﬁcant bias compared to the correspond- ing seven-layer model posterior PDF (Fig. 4a, d, g). In White et al. (2020), it was hypothesized that the tritium observations were giving rise to the apparent bias in the one-layer and two-layer posterior PDFs through the phenomenon of (in- appropriate) parameter compensation (e.g., Clark and Vrugt, 2006; White et al., 2014) arising from history-matching mod- els with simpliﬁed model vertical discretization. Herein, we test this hypothesis by conditioning all three uniquely dis- cretized models again, but without using the discrete tritium observations, and then by comparing the resulting posterior PDFs to the corresponding PDFs in White et al. (2020). Any apparent difference in the posterior PDFs for the Firth fore- cast is therefore directly attributable to the exclusion of the tritium observations during history matching. – Model parameterization. Spatially distributed param- eterization of (horizontal and vertical) hydraulic con- ductivity, effective porosity, recharge rate, ﬁrst-order denitriﬁcation rate, initial concentration and dispersiv- ity is achieved using a combination of cell-based and zone-based multipliers. The nitrate loading rate and ab- straction well rate are parameterized using cell-by-cell and well-based multipliers, respectively. Streamﬂow- routing (SFR) elements are parameterized on a stream- segment basis. www.hydrol-earth-syst-sci.net/24/1677/2020/ www.hydrol-earth-syst-sci.net/24/1677/2020/ This parameterization approach gives rise to a problem dimensionality of 141 268, 50 180 and 29 050 for the seven-layer, two-layer and one-layer model history-matching experiments, respectively. We refer the reader to White (2018) and White et al. (2020) www.hydrol-earth-syst-sci.net/24/1677/2020/ Hydrol. Earth Syst. Sci., 24, 1677–1689, 2020 M. J. Knowling et al.: Model tracer-data assimilation for decision support 1683 Figure 3. Hauraki Plains model extent (red dashed line), layer-one inactive area (shaded) and surface-water network (blue lines). The terminal surface-water reaches that discharge to the Firth of Thames are shown as blue triangles. The location of groundwater tritium measurements are shown as magenta dots. is largely in contrast to the case where data assimilation is undertaken without the tritium observations, which leads to much more subtle differences in posterior PDFs across the uniquely discretized models (Fig. 4b, e, h). q y ( g ) The bias apparent in the posterior difference PDFs for the reduced-layer models relative to the seven-layer model (Fig. 4c, f, i) is directly attributable to the use of tritium ob- servations in the data assimilation process. The difference between the Firth forecast PDFs resulting from data assim- ilation with and without tritium is most pronounced for the one-layer model (Fig. 4i). In this case, excluding tritium ob- servations from the history-matching results in a decrease in simulated nitrate discharge of 2×107 to 4×107 kg – approx- imately a 40 % decrease in simulated mean nitrate discharge. We attribute the apparent one-layer PDF bias to the loss of simulated vertical ﬂow and associated deeper groundwater ﬂow paths. Brieﬂy, this occurs due to the aggregation of nu- merical discretization effects – the ﬂow paths of a coarser- layer model will be a smoother and averaged representation of those derived from a ﬁner-layer model. While these deeper ﬂow paths are not important for simulating the nitrate trans- port cycle (given the relatively high denitriﬁcation rates in the Hauraki system), it is apparently important for assimilating the tritium concentration observations. The biases identiﬁed reﬂect the sensitivity of the Firth forecast to uncertain parameters that were conditioned by tri- tium concentration observations. This occurs due to the spa- tially integrated nature of the Firth nitrate-load forecast and because the tritium observations provide insight into spatially and temporally averaged recharge and lateral ﬂux rates in the upgradient portion of the domain, where most of the surface- water–groundwater exchange occurs. Figure 3. www.hydrol-earth-syst-sci.net/24/1677/2020/ Hauraki Plains model extent (red dashed line), layer-one inactive area (shaded) and surface-water network (blue lines). The terminal surface-water reaches that discharge to the Firth of Thames are shown as blue triangles. The location of groundwater tritium measurements are shown as magenta dots. 3.3 Results The process of history matching with and without available groundwater tritium concentration observations yields sub- stantial differences in the posterior PDFs of the Firth fore- cast (Fig. 4). In the case of the seven-layer “complex model” (Fig. 4a, b), excluding the tritium observations results in a posterior PDF with a larger second moment and a slightly larger ﬁrst moment compared to including tritium observa- tions for history matching; the difference between the Firth forecast posterior PDFs with and without assimilating tritium observations is between 0 and 2 × 107 kg of nitrate (Fig. 4c). The larger second moment of the posterior PDF when ex- cluding tritium observations represents an intuitive and ex- pected outcome: using fewer observations for parameter con- ditioning through history matching should (theoretically) re- sult in a larger posterior variance for the forecasts that depend on those parameters. www.hydrol-earth-syst-sci.net/24/1677/2020/ 4 Discussion and conclusions This study explores the ramiﬁcations of assimilating tri- tium concentration and tritium-derived interpretation ob- servations, speciﬁcally in the context of two examples of decision-support modeling. The beneﬁt or otherwise of tri- tium data in other contexts such as site system character- ization and understanding and conceptual-model develop- ment is therefore not the focus of the current study; this study is concerned with a model’s ability to “predict” (in two decision-support contexts) rather than “explain” (observed system behavior), as contrasted by Shmueli (2010). The ﬁrst case study presented herein serves to demon- strate that assimilating the rich information contained within tritium-derived MRT observations may be of variable worth in terms of improving the reliability of forecasts, especially where MRT observations are correlated with other available state observations (e.g., where hydraulic data are widespread, given the apparent spatially and temporally integrated infor- mation content of MRT observations, as supported by Ginn et al., 2009). Moreover, the worth of MRT observations is Herein, for the purposes of identifying bias, the seven- layer model is considered to represent the best-available es- timate of the Firth forecast. Using this construct, we see that there are signiﬁcant differences in posterior PDFs across the uniquely discretized models arising from data assimilation that included the tritium observations (Fig. 4a, d, g). This www.hydrol-earth-syst-sci.net/24/1677/2020/ M. J. Knowling et al.: Model tracer-data assimilation for decision support 1684 1684 M. J. Knowling et al.: Model tracer-data assimilation for decision support Figure 4. Comparison of posterior probability density functions (PDFs) for the Firth forecast. The left column (a, d, g) allows for the identiﬁcation of bias as a result of both model simpliﬁcation and tritium (Tr) assimilation. By comparing to the middle column (b, e, h), model simpliﬁcation-induced bias can be separated from that induced by assimilating tritium observations. The isolation of tritium assimilation- induced bias evident with different simpliﬁed models is shown in the right column (c, f, i). Including tritium observations in the conditioning of the one-layer and two-layer models (g, d) yields signiﬁcant bias compared to the seven-layer PDF (a). However, if tritium observations are excluded from conditioning, the one-layer and two-layer PDFs (h, e) have considerably less bias compared to the corresponding seven-layer PDF (b). The differences in the PDFs (c, f, i) show that the tritium observations have the greatest biasing effect on the Firth forecast for the one-layer model. Figure 4. Comparison of posterior probability density functions (PDFs) for the Firth forecast. The left column (a, d, g) allows for the identiﬁcation of bias as a result of both model simpliﬁcation and tritium (Tr) assimilation. By comparing to the middle column (b, e, h), model simpliﬁcation-induced bias can be separated from that induced by assimilating tritium observations. The isolation of tritium assimilation- induced bias evident with different simpliﬁed models is shown in the right column (c, f, i). Including tritium observations in the conditioning of the one-layer and two-layer models (g, d) yields signiﬁcant bias compared to the seven-layer PDF (a). However, if tritium observations are excluded from conditioning, the one-layer and two-layer PDFs (h, e) have considerably less bias compared to the corresponding seven-layer PDF (b). The differences in the PDFs (c, f, i) show that the tritium observations have the greatest biasing effect on the Firth forecast for the one-layer model. shown to vary between forecasts in such a way that reﬂects the underlying physics represented by the model (e.g., the MRT observations are of greatest worth for forecasts that are located where the aquifer is receiving recharge); these physics dictate the “information ﬂow” rather than the spa- tial proximity of the MRT observations and the forecast. www.hydrol-earth-syst-sci.net/24/1677/2020/ Hydrol. Earth Syst. Sci., 24, 1677–1689, 2020 M. J. Knowling et al.: Model tracer-data assimilation for decision support M. J. Knowling et al.: Model tracer-data assimilation for decision support It should be noted, however, that even when the forecast is well “aligned” with observation data (i.e., the forecast is de- pendent on parameters residing in the solution space), some degree of parameter compensation will inevitably occur; all models are gross simpliﬁcations, and therefore model param- eters do not perfectly represent real-world properties (e.g., Clark and Vrugt, 2006; White et al., 2014). However, if the data used for assimilation are commensurate with the fore- casts, then the ill effects of model error may be expected to be negligible (e.g., Doherty and Christensen, 2011; Watson et al., 2013). pp g p Furthermore, we expect the above-mentioned issues asso- ciated with imperfect-model data assimilation to be relevant and largely transferrable to the assimilation of other environ- mental tracers, other information-rich observations and di- verse data types more generally. This is because we consider the primary barrier to the appropriate assimilation of tritium observation data encountered in the second case study to be fundamental challenges associated with extracting appropri- ate information from spatially discrete concentration obser- vations when using upscaled or simpliﬁed representations of hydraulic properties within a regional-scale model that sim- ulates tracer concentrations using the advection–dispersion equation (e.g., Zheng and Gorelick, 2003; Riva et al., 2008). To the extent that simulated outputs corresponding to ob- served tracer concentrations are sensitive to model details or parameters that are “missing” in a simpliﬁed model (e.g., White et al., 2014), parameter compensation will occur (e.g., Clark and Vrugt, 2006). To the extent that the forecast of management interest is dependent on these biased parameter estimates, the forecast will also become biased, potentially leading to resource mismanagement. The ubiquitous nature of model error and the challenges in appropriately account- ing for differences in, e.g., representative spatial scales be- tween ﬁeld observations and model-derived quantities, sug- gests that the ill effects identiﬁed in this study such as bias induced by history matching are not unique to the speciﬁcs of our study (e.g., consideration of tritium as a tracer). The similar ﬁndings and recommendations of Brynjarsdóttir and O’Hagan (2014) and He et al. (2018) in the disciplines of statistics and petroleum reservoir engineering, respectively, The above ﬁndings and recommendations suggest that there is a signiﬁcant need to develop improved strategies to assimilate diverse observation types including tracer con- centration and tracer interpretation observations in numeri- cal models for decision support. M. J. Knowling et al.: Model tracer-data assimilation for decision support 1685 also support the potential for the transferability in our ﬁnd- ings and recommendations to data assimilation in other envi- ronmental modeling contexts. reader to Knowling et al. (2019) and White et al. (2020) for a broader exploration of the consequences of model sim- pliﬁcation (in the form of parameterization reduction and vertical-discretization coarsening, respectively) in terms of the decision-relevant forecast bias–variance trade-off and its implications for management decision making more gener- ally. reader to Knowling et al. (2019) and White et al. (2020) for a broader exploration of the consequences of model sim- pliﬁcation (in the form of parameterization reduction and vertical-discretization coarsening, respectively) in terms of the decision-relevant forecast bias–variance trade-off and its implications for management decision making more gener- ally. If diverse and information-rich data such as tritium and MRT observations are available and data assimilation through history matching is deemed necessary and/or appro- priate, then a targeted modeling approach is needed that iden- tiﬁes which of these data are relevant to the forecast. This is critical to avoiding the ill effects of model error in the con- text of decision-support modeling (e.g., White et al., 2014; Knowling et al., 2019), as well as avoiding unnecessary com- plexity (through processes and parameters) needed to simu- late the equivalent values of the diverse data for assimilation purposes, which may greatly increase the computational cost of the modeling analysis. Collectively, these results suggest that the assimilation of tritium and tritium-derived observations through history matching with an imperfect model should be strategic and approached with caution. It is recommended that these information-rich observations should not indiscriminately be incorporated in a data assimilation framework, given that this study has shown that such an approach (i) may be of variable apparent beneﬁt, depending on the forecast being made, and (ii) when using imperfect models, which may produce far worse forecast outcomes than those that would have been ar- rived at without assimilating these observations at all. This recommendation is similar to those by Brynjarsdóttir and O’Hagan (2014) and He et al. (2018). We consider this rec- ommendation to be in stark contrast to what we believe is a common view among practitioners that “calibrating to more data improves the model and its predictions”; we therefore consider this recommendation to be of signiﬁcant implication to decision-support environmental modeling practitioners. M. J. Knowling et al.: Model tracer-data assimilation for decision support The forecast-speciﬁc nature of observation worth has also been reported previously (e.g., Dausman et al., 2010; Fienen et al., 2010; White et al., 2016). The worth of MRT observations relative to various hydraulic potential and discharge observa- tions across the different forecasts are, in general terms, sim- ilar to those reported by Hunt et al. (2006), Masbruch et al. (2014), Oehlmann et al. (2015), and Zell et al. (2018) (espe- cially when considering the discussion point in the following paragraph). presented herein would generally be larger for forecasts that are dependent on both uncertain hydraulic and transport pa- rameters (e.g., particle travel times). This is notwithstand- ing that the uncertainty variance for such forecasts may be larger given the additional source of uncertainty associated with porosity. These ﬁndings are nevertheless highly relevant in that MRT observations are widely used and often regarded to be of beneﬁt in constraining uncertain model parameters more generally (Schilling et al., 2019). The second case study serves to demonstrate that as- similating tritium concentration observations with simpliﬁed (i.e., imperfect) numerical models may induce signiﬁcant bias in forecasts; this is bias that is undetectable without a simple–complex model pair (e.g., Doherty and Christensen, 2011; White et al., 2014; Knowling et al., 2019). The forecast bias revealed in the second case study occurs as a result of the vertical-discretization simpliﬁed model’s inability to appro- priately assimilate the rich information content of the tritium observations. Generally, the observed pattern of simpliﬁca- tion and the resulting forecast bias implies that as the simpli- ﬁcation of the model increases, the dangers of assimilating rich and diverse data types also grows. This result is highly relevant to decision-support modeling practitioners, since all numerical models are gross simpliﬁcations of real environ- mental systems that they attempt to simulate. We refer the While the particle-tracking model used in the ﬁrst case study provides a mechanism for MRT observations to in- form uncertain model parameters, including aquifer porosity (which is otherwise uninformed by other historical ﬁeld ob- servations), it is important to note that the forecasts are insen- sitive to porosity. That is, the information contained within MRT observations is spread between parameters that both do and do not play a role in constraining forecasts – effec- tively “diluting” the information available for conditioning. It is therefore expected that the worth of MRT observations www.hydrol-earth-syst-sci.net/24/1677/2020/ Hydrol. Earth Syst. Sci., 24, 1677–1689, 2020 M. J. www.hydrol-earth-syst-sci.net/24/1677/2020/ M. J. Knowling et al.: Model tracer-data assimilation for decision support Knowling et al.: Model tracer-data assimilation for decision support Appendix A: First-order second-moment (FOSM) methodology σ 2 s = yT 6θy (A2) σ 2 s = yT 6θy (A2) and σ 2 s = yT 6θy (A2) This section provides a description of the FOSM approach used in the ﬁrst case study to quantify uncertainty variance and assess data worth. and σ 2 s = yT 6θy. (A3) (A3) The covariance matrix of uncertain model parameters 6θ can be approximated using the Schur complement as follows (Golub and Van Loan, 1996; Tarantola, 2005): The worth of data, expressed as a percentage, is given by The worth of data, expressed as a percentage, is given by %DW = σ 2 ±obs −σ 2 base min σ 2 base,σ 2 ±obs × 100, (A4) 6θ = 6θ −6θJT h J6θJT + 6ϵ i−1 J6θ, (A1) (A4) (A1) where 6θ is the prior parameter covariance matrix, which is speciﬁed based on expert knowledge pertaining to site sys- tem characteristics; 6ϵ is the epistemic observation noise co- variance matrix (often assumed to have non-zero diagonal el- ements only), which includes the effects of model structural errors and measurement errors; and J is the Jacobian ma- trix of partial ﬁrst derivatives (i.e., sensitivities) of simulated model outputs with respect to parameters. The Schur comple- ment can be considered a linearized form of Bayes’ equation to estimate the second moment of the parameter and fore- cast posterior distribution (e.g., Goldstein and Wooff, 2007; Christensen and Doherty, 2008; Dausman et al., 2010). where σ 2 ±obs is the increase or decrease in forecast uncer- tainty variance as a result of the removal or addition of one or more observations or observation groups used for parame- ter conditioning, respectively, and σ 2 base is either the forecast uncertainty calculated on the basis of all observation data or zero observation data, depending on whether data worth is being quantiﬁed by adding or removing observations. where σ 2 ±obs is the increase or decrease in forecast uncer- tainty variance as a result of the removal or addition of one or more observations or observation groups used for parame- ter conditioning, respectively, and σ 2 base is either the forecast uncertainty calculated on the basis of all observation data or zero observation data, depending on whether data worth is being quantiﬁed by adding or removing observations. Herein, we quantify %DW as a result of both the removal and addition of observation groups. M. J. Knowling et al.: Model tracer-data assimilation for decision support Such strategies will likely require increased model complexity (including advanced discretization, process representation and parameterization) such that information-rich and diverse data types that operate across a range of spatial and temporal scales commensurate with a given forecast can be properly assimilated. However, an important and challenging compromise will be encountered: the need for enough model complexity to appropriately assimilate rich and diverse observations while simultaneously ensuring that this level of complexity does not preclude the application of formal data assimilation and uncertainty quantiﬁcation techniques due to the associated numerical instability and excessive run times. The naviga- tion of this trade-off is central to effective and efﬁcient decision-support modeling practice. In the meantime, tracer- data model assimilation should involve processing or trans- forming of concentrations into quantities that may be more useful and may guard against ill effects of history matching imperfect models (e.g., by integrating observations in space and time) (e.g., Rasa et al., 2013; Knowling et al., 2019; White et al., 2020). Hydrol. Earth Syst. Sci., 24, 1677–1689, 2020 www.hydrol-earth-syst-sci.net/24/1677/2020/ M. J. Knowling et al.: Model tracer-data assimilation for decision support (FOSM) 1686 M. J. Knowling et al.: Model tracer-data assimilation for decision support M. J. Knowling et al.: Model tracer-data assimilation for decision support FOSM) M. J. Knowling et al.: Model tracer-data assimilation for decision support Christensen, S. and Doherty, J.: Predictive error dependencies when using pilot points and singular value decomposition in groundwater model calibration, Adv. Water Resour., 31, 674– 700, https://doi.org/10.1016/j.advwatres.2008.01.003, 2008. Data availability. The data ﬁles, scripts and software used herein are available upon request from GNS Science. Clark, M. P. and Vrugt, J. A.: Unraveling uncertainties in hydrologic model calibration: Addressing the problem of compensatory parameters, Geophys. Res. Lett., 33, L06406, https://doi.org/10.1029/2005GL025604, 2006. Supplement. The supplement related to this article is available on- line at: https://doi.org/10.5194/hess-24-1677-2020-supplement. Dausman, A., Doherty, J., Langevin, C., and Sukop, M.: Quantify- ing data worth toward reducing predictive uncertainty, Ground Water, 48, 729–740, 2010. Author contributions. MJK, JTW and CRM contributed to the con- cept. MJK and JTW undertook the modeling analyses. MJK pre- pared the paper with input from JTW and CRM. PR and KH con- tributed to the underlying Heretaunga Plains models. Doherty, J. and Christensen, S.: Use of paired simple and complex models to reduce predictive bias and quan- tify uncertainty, Water Resour. Res., 47, W12534, https://doi.org/10.1029/2011WR010763, 2011. Competing interests. The authors declare that they have no conﬂict of interest. Doherty, J. and Welter, D.: A Short Exploration of Structural Noise, Water Resourc. Res., 46, W05525, https://doi.org/10.1029/2009WR008377, 2010. Doherty, J. E.: Ground water model calibration using pilot points and regularization, Ground Water, 41, 170–177, 2003. Acknowledgements. The authors wish to thank Ty Ferre, Chris Tur- nadge and the anonymous reviewer for their helpful comments. Doherty, J. E.: PEST and its utility support software, Theory, Wa- termark Numerical Publishing, Brisbane, Australia, 2015. Fienen, M. N., Doherty, J. E., Hunt, R. J., and Reeves, H. W.: Us- ing prediction uncertainty analysis to design hydrologic monitor- ing networks: Example applications from the Great Lakes water availability pilot project, U.S. Geological Survey Scientiﬁc In- vestigation Report 2010-5159, Middleton, WI, USA, p. 44, 2010. Financial support. This research has been supported by the Min- istry of Business Innovation and Employment, as part of both the Te Whakaheke o te Wai and Smart Models for Aquifer Management research programs (grant nos. C05X1803 and CONT-41982-ETR- GNS, respectively), with co-funding from Hawke’s Bay Regional Council and Waikato Regional Council. Ginn, T. R., Haeri, H., Massoudieh, A., and Foglia, L.: Notes on Groundwater Age in Forward and Inverse Modeling, Transport Porous Med., 79, 117–134, https://doi.org/10.1007/s11242-009- 9406-1, 2009. Review statement. This paper was edited by Fabrizio Fenicia and reviewed by Ty P. A. Appendix A: First-order second-moment (FOSM) methodology We primarily focus on %DW values based on the removal of an observation group from an otherwise full observation dataset available for as- similation, given that these values reﬂect the unique (i.e., uncorrelated) information content of observations. However, the difference between %DW values arising from these dif- ferent data-worth quantiﬁcation approaches is used herein to comment on the level of information uniqueness or redun- dancy within observation groups. Equation (A1) assumes a linear relation between model parameters and simulated outputs (i.e., the sensitivities en- capsulated within the J matrix are independent of the param- eter values θ). It also assumes that parameter and epistemic uncertainty distributions are Gaussian (i.e., normal). While the posterior parameter and forecast uncertainty variances yielded by FOSM may only be approximate (de- pending on the validity of the linear assumption), the com- putational efﬁciency with which a large number of different number of conditioning “experiments” can be performed is unparalleled; these experiments facilitate the rapid evaluation of the worth of different types of observations to reduce fore- cast variance. In addition, a number of studies have shown support for its usage especially in a relative second-moment sense (e.g., Dausman et al., 2010; Herckenrath et al., 2011; Knowling et al., 2019). It is important to note that each FOSM-based data-worth assessment is conducted with respect to a single forecast (notwithstanding that we evaluate the worth of different ob- servation data with respect to a number of different fore- casts). We consider this to be a side beneﬁt of this approach, especially given the need for decision-support modeling to be undertaken in a forecast-targeted manner, as discussed re- cently by White (2017). The prior and posterior uncertainty variance surrounding a forecast σ 2 s can be expressed by mapping uncertainty from parameter to forecast “space”. This is achieved by computing the sensitivity of the forecast to model parameters, compris- ing the vector y (i.e., a row of J). That is Hydrol. Earth Syst. Sci., 24, 1677–1689, 2020 www.hydrol-earth-syst-sci.net/24/1677/2020/ 1687 M. J. Knowling et al.: Model tracer-data assimilation for decision support Ferre and one anonymous referee. Goldstein, M. and Wooff, D.: Bayes linear statistics, theory and methods, John Wiley & Sons, Chichester, UK, 2007. Golub, G. and Van Loan, C.: Matrix Computations, Johns Hop- kins Studies in the Mathematical Sciences, Johns Hopkins Uni- versity Press, available at: http://books.google.com/books?id= mlOa7wPX6OYC (last access: 20 July 2019), 1996. www.hydrol-earth-syst-sci.net/24/1677/2020/ M. J. Knowling et al.: Model tracer-data assimilation for decision support Geological Sur- vey Techniques and Methods 6-A37, Reston, VA, USA, p. 44, 2011. Knowling, M. J., White, J. T., and Moore, C. R.: Role of model parameterization in risk-based decision support: An empirical exploration, Adv. Water Resour., 128, 59–73, https://doi.org/10.1016/j.advwatres.2019.04.010, 2019. Oehlmann, S., Geyer, T., Licha, T., and Sauter, M.: Reducing the ambiguity of karst aquifer models by pattern matching of ﬂow and transport on catchment scale, Hydrol. Earth Syst. Sci., 19, 893–912, https://doi.org/10.5194/hess-19-893-2015, 2015. Leray, S., de Dreuzy, J.-R., Bour, O., Labasque, T., and Aquilina, L.: Contribution of age data to the characteri- zation of complex aquifers, J. Hydrol., 464–465, 54–68, https://doi.org/10.1016/j.jhydrol.2012.06.052, 2012. Oliver, D. S. and Alfonzo, M.: Calibration of imperfect mod- els to biased observations, Comput. Geosci., 22, 145–161, https://doi.org/10.1007/s10596-017-9678-4, 2018. Li, H., Brunner, P., Kinzelbach, W., Li, W., and Dong, X.: Calibration of a groundwater model using pattern informa- tion from remote sensing data, J. Hydrol., 377, 120–130, https://doi.org/10.1016/j.jhydrol.2009.08.012, 2009. Pollock, D. W.: User guide for MODPATH version 6 – A particle- tracking model for MODFLOW: U.S. Geological Survey Tech- niques and Methods, version 6. edn., U.S. Dept. of the Interior, U.S. Geological Survey, Reston, Va, USA, 2012. Masbruch, M., Gardner, P., and Brooks, L.: Hydrology and numer- ical simulation of groundwater movement and heat transport in Snake Valley and surrounding areas, Juab, Millard, and Beaver Counties, Utah, and White Pine and Lincoln Counties, Nevada, U. S. Geological Survey Scientiﬁc Investigations Report 2014- 5103, Reston, VA, USA, https://doi.org/10.3133/sir20145103, 2014. Rajanayaka, C. and Fisk, L.: Irrigation water demand & land surface recharge assessment for Heretaunga Plains, Hawke’s Bay Regional Council, C16053. Aqualinc Research Limited, Christchurch, New Zealand, 2018. Rakowski, P. and Knowling, M.: Heretaunga aquifer system groundwater model: Development report, Hawke’s Bay Regional Council Report No. RM18-14 – 4997, Napier, New Zealand, 2018. McDonnell, J. J., McGuire, K., Aggarwal, P., Beven, K. J., Biondi, D., Destouni, G., Dunn, S., James, A., Kirchner, J., Kraft, P., Lyon, S., Maloszewski, P., Newman, B., Pﬁster, L., Rinaldo, A., Rodhe, A., Sayama, T., Seibert, J., Solomon, K., Soulsby, C., Stewart, M., Tetzlaff, D., Tobin, C., Troch, P., Weiler, M., Western, A., Wörman, A., and Wrede, S.: How old is streamwa- ter? Open questions in catchment transit time conceptualiza- tion, modelling and analysis, Hydrol. Process., 24, 1745–1754, https://doi.org/10.1002/hyp.7796„ 2010. Rasa, E., Foglia, L., Mackay, D. M., and Scow, K. M. J. Knowling et al.: Model tracer-data assimilation for decision support 1688 Moore, C. and Doherty, J. E.: Role of the calibration process in reducing model predictive error, Water Resour. Res., 41, 1–14, 2005. Harbaugh, A. W.: MODFLOW-2005, the U.S. Geological Survey modular ground-water model – the Ground-Water Flow Process, U. S. Geological Survery, Reston, VA, USA, vol. 6, 2005. Morgan, L. K., Werner, A. D., and Simmons, C. T.: On the in- terpretation of coastal aquifer water level trends and water bal- ances: A precautionary note, J. Hydrol., 470–471, 280–288, https://doi.org/10.1016/j.jhydrol.2012.09.001, 2012. He, J., Reynolds, A. C., Tanaka, S., Wen, X.-H., and Kamath, J.: Calibrating Global Uncertainties to Local Data: Is the Learning Being Over-Generalized?, Society of Petroleum Engineers, SPE Annual Technical Conference and Exhibition, 24–26 September 2018, Dallas, Texas, USA, https://doi.org/10.2118/191480-MS, 2018. Morgenstern, U. and Daughney, C. J.: Groundwater age for identiﬁcation of baseline groundwater quality and impacts of land-use intensiﬁcation – The National Groundwater Monitor- ing Programme of New Zealand, J. Hydrol., 456–457, 79–93, https://doi.org/10.1016/j.jhydrol.2012.06.010, 2012. Herckenrath, D., Langevin, C. D., and Doherty, J. E.: Predic- tive uncertainty analysis of a saltwater intrusion model us- ing null-space Monte Carlo, Water Resour. Res., 47, W05504, https://doi.org/10.1029/2010WR009342, 2011. Morgenstern, U., Stewart, M. K., and Stenger, R.: Dat- ing of streamwater using tritium in a post nuclear bomb pulse world: continuous variation of mean transit time with streamﬂow, Hydrol. Earth Syst. Sci., 14, 2289–2301, https://doi.org/10.5194/hess-14-2289-2010, 2010. Hunt, R. J., Feinstein, D. T., Pint, C. D., and Anderson, M. P.: The importance of diverse data types to calibrate a watershed model of the Trout Lake Basin, Northern Wisconsin, USA, J. Hydrol., 321, 286–296, https://doi.org/10.1016/j.jhydrol.2005.08.005, 2006. Morgenstern, U., Begg, J., van der Raaij, R., Moreau, M., Martin- dale, H., Daughney, C., Franzblau, R., Stewart, M., Knowling, M., Toews, M., Trompetter, V., Kaiser, J., and Gordon, D.: Here- taunga Plains aquifers: groundwater dynamics, source and hy- drochemical processes as inferred from age, chemistry, and sta- ble isotope tracer data, GNS Science Report; Lower Hutt, New Zealand, https://doi.org/10.21420/G2Q92G, 2018. Hunt, R. J., Doherty, J., and Tonkin, M. J.: Are Models Too Sim- ple? Arguments for Increased Parameterization, Groundwater, 45, 254–262, https://doi.org/10.1111/j.1745-6584.2007.00316.x, 2007. Kirchner, J. W., Feng, X., and Neal, C.: Catchment-scale ad- vection and dispersion as a mechanism for fractal scaling in stream tracer concentrations, J. Hydrol., 254, 82–101, https://doi.org/10.1016/S0022-1694(01)00487-5, 2001. Niswonger, R., Panday, S., and Ibaraki, M.: MODFLOW-NWT, A Newton formulation for MODFLOW-2005, U.S. References Gusyev, M. A., Toews, M., Morgenstern, U., Stewart, M., White, P., Daughney, C., and Hadﬁeld, J.: Calibration of a tran- sient transport model to tritium data in streams and simula- tion of groundwater ages in the western Lake Taupo catch- ment, New Zealand, Hydrol. Earth Syst. Sci., 17, 1217–1227, https://doi.org/10.5194/hess-17-1217-2013, 2013. André, L., Franceschi, M., Pouchan, P., and Atteia, O.: Us- ing geochemical data and modelling to enhance the under- standing of groundwater ﬂow in a regional deep aquifer, Aquitaine Basin, south-west of France, J. Hydrol., 305, 40–62, https://doi.org/10.1016/j.jhydrol.2004.08.027, 2005. Gusyev, M. A., Abrams, D., Toews, M. W., Morgenstern, U., and Stewart, M. K.: A comparison of particle-tracking and solute transport methods for simulation of tritium concentrations and groundwater transit times in river water, Hydrol. Earth Syst. Sci., 18, 3109–3119, https://doi.org/10.5194/hess-18-3109-2014, 2014. Bedekar, V., Morway, E., Langevin, C., and Tonkin, M.: MT3D- USGS version 1: A U.S. Geological Survey release of MT3DMS updated with new and expanded transport capabilities for use with MODFLOW, U.S. Geological Survey Techniques and Methods 6-A53, Reston, VA, USA, p. 69, 2016. Beyer, M., Morgenstern, U., and Jackson, B.: Review of techniques for dating young groundwater (< 100 years) in New Zealand, J. Hydrol., 53, 93–111, 2014. Han, D. M., Song, X. F., Currell, M. J., and Tsujimura, M.: Us- ing chloroﬂuorocarbons (CFCs) and tritium to improve concep- tual model of groundwater ﬂow in the South Coast Aquifers of Laizhou Bay, China, Hydrol. Process., 26, 3614–3629, https://doi.org/10.1002/hyp.8450, 2012. Brynjarsdóttir, J. and O’Hagan, A.: Learning about physical param- eters: The importance of model discrepancy, Inverse Probl., 30, 114007, https://doi.org/10.1088/0266-5611/30/11/114007, 2014. Hansen, A. L., Refsgaard, J. C., Christensen, B. S. B., and Jensen, K. H.: Importance of including small-scale tile drain discharge in the calibration of a coupled groundwater-surface water catchment model, Water Resour. Res., 49, 585–603, https://doi.org/10.1029/2011WR011783, 2013. Cartwright, I. and Morgenstern, U.: Constraining ground- water recharge and the rate of geochemical processes using tritium and major ion geochemistry: Ovens catch- ment, southeast Australia, J. Hydrol., 475, 137–149, https://doi.org/10.1016/j.jhydrol.2012.09.037, 2012. Hydrol. Earth Syst. Sci., 24, 1677–1689, 2020 www.hydrol-earth-syst-sci.net/24/1677/2020/ M. J. Knowling et al.: Model tracer-data assimilation for decision support 1689 White, J. T.: A model-independent iterative ensemble smoother for efﬁcient history-matching and uncertainty quantiﬁcation in very high dimensions, Environ. Model. Softw., 109, 191–201, https://doi.org/10.1016/j.envsoft.2018.06.009, 2018. Sanford, W. E., Plummer, L. N., McAda, D. P., Bexﬁeld, L. M., and Anderholm, S. K.: Hydrochemical tracers in the middle Rio Grande Basin, USA: 2. Calibration of a groundwater-ﬂow model, Hydrogeol. J., 12, 389–407, https://doi.org/10.1007/s10040-004- 0326-4, 2004. White, J. T., Doherty, J. E., and Hughes, J. D.: Quantify- ing the predictive consequences of model error with lin- ear subspace analysis, Water Resour. Res., 50, 1152–1173, https://doi.org/10.1002/2013WR014767, 2014. Schilling, O. S., Cook, P. G., and Brunner, P.: Beyond Clas- sical Observations in Hydrogeology: The Advantages of Including Exchange Flux, Temperature, Tracer Concentra- tion, Residence Time, and Soil Moisture Observations in Groundwater Model Calibration, Rev. Geophys., 57, 146–182, https://doi.org/10.1029/2018RG000619, 2019. White, J. T., Fienen, M. N., and Doherty, J. E.: A python framework for environmental model uncer- tainty analysis, Environ. Model. Softw., 85, 217–228, https://doi.org/10.1016/j.envsoft.2016.08.017, 2016. Shmueli, G.: To Explain or to Predict?, Stat. Sci., 25, 289–310, https://doi.org/10.1214/10-STS330, 2010. White, J. T., Knowling, M. J., and Moore, C. R.: Consequences of model simpliﬁcation in risk-based decision making: An analy- sis of groundwater-model vertical discretization, Groundwater, https://doi.org/10.1111/gwat.12957, online ﬁrst, 2020. Siade, A., Prommer, H., Suckow, A., and Raiber, M.: Us- ing Numerical Groundwater Modelling to Constrain Flow Rates and Flow Paths in the Surat Basin through Environmental Tracer Data, CSIRO Report, Australia, https://doi.org/10.25919/5b8055bfe3ea5, 2018. Wilding, T. K.: Heretaunga Springs: Gains and losses of stream ﬂow to groundwater on the Heretaunga Plains, Hawke’s Bay Regional Council Report No. RM18-13 – 4996, Napier, New Zealand, 2017. Stewart, M. K. and Thomas, J. T.: A conceptual model of ﬂow to the Waikoropupu Springs, NW Nelson, New Zealand, based on hydrometric and tracer (18O, Cl, 3H and CFC) evidence, Hy- drol. Earth Syst. Sci., 12, 1–19, https://doi.org/10.5194/hess-12- 1-2008, 2008. Zell, W. O., Culver, T. B., and Sanford, W. E.: Prediction un- certainty and data worth assessment for groundwater transport times in an agricultural catchment, J. Hydrol., 561, 1019–1036, https://doi.org/10.1016/j.jhydrol.2018.02.006, 2018. Tarantola, A.: Inverse problem theory and methods for model parameter estimation, Society for Industrial and Applied Mathematics Philadelphia, PA, USA, 339 pp., https://doi.org/10.1137/1.9780898717921, 2005. Zheng, C. and Gorelick, S. M.: Analysis of Solute Trans- port in Flow Fields Inﬂuenced by Preferential Flow- paths at the Decimeter Scale, Groundwater, 41, 142–155, https://doi.org/10.1111/j.1745-6584.2003.tb02578.x, 2003. Turnadge, C. and Smerdon, B. M. J. Knowling et al.: Model tracer-data assimilation for decision support M.: Ef- fect of different transport observations on inverse modeling results: case study of a long-term groundwater tracer test monitored at high resolution, Hydrogeol. J., 21, 1539–1554, https://doi.org/10.1007/s10040-013-1026-8, 2013. Riva, M., Guadagnini, A., Fernandez-Garcia, D., Sanchez-Vila, X., and Ptak, T.: Relative importance of geostatistical and transport models in describing heavily tailed breakthrough curves at the Lauswiesen site, J. Contam. Hydrol., 101, 1–13, https://doi.org/10.1016/j.jconhyd.2008.07.004, 2008. Michael, H. A. and Voss, C. I.: Estimation of regional- scale groundwater ﬂow properties in the Bengal Basin of India and Bangladesh, Hydrogeol. J., 17, 1329–1346, https://doi.org/10.1007/s10040-009-0443-1, 2009. Sanford, W.: Calibration of models using groundwater age, Hydro- geol. J., 19, 13–16, 2011. www.hydrol-earth-syst-sci.net/24/1677/2020/ Hydrol. Earth Syst. Sci., 24, 1677–1689, 2020 M. J. Knowling et al.: Model tracer-data assimilation for decision support D.: A review of methods for mod- elling environmental tracers in groundwater: advantages of tracer concentration simulation, J. Hydrol., 519, 3674–3689, 2014. Zuber, A., Ró˙za´nski, K., Kania, J., and Purtschert, R.: On some methodological problems in the use of environmental tracers to estimate hydrogeologic parameters and to calibrate ﬂow and transport models, Hydrogeol. J., 19, 53–69, 2011. Watson, T. A., Doherty, J. E., and Christensen, S.: Parameter and predictive outcomes of model simpliﬁcation, Water Resour. Res., 49, 3952–3977, https://doi.org/10.1002/wrcr.20145, 2013. White, J. T.: Forecast First: An Argument for Ground- water Modeling in Reverse, Groundwater, 55, 660–664, https://doi.org/10.1111/gwat.12558, 2017. Hydrol. Earth Syst. Sci., 24, 1677–1689, 2020 www.hydrol-earth-syst-sci.net/24/1677/2020/ www.hydrol-earth-syst-sci.net/24/1677/2020/
https://openalex.org/W2581009913	https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0167625&type=printable	English	null	Quantification of Abdominal Fat in Obese and Healthy Adolescents Using 3 Tesla Magnetic Resonance Imaging and Free Software for Image Analysis	PloS one	2,017	cc-by	6,163	RESEARCH ARTICLE Background and Aims Computed tomography, which uses ionizing radiation and expensive software packages for analysis of scans, can be used to quantify abdominal fat. The objective of this study is to measure abdominal fat with 3T MRI using free software for image analysis and to correlate these findings with anthropometric and laboratory parameters in adolescents. Editor: Raffaella Buzzetti, Universita degli Studi di Roma La Sapienza, ITALY Received: April 22, 2016 Accepted: November 17, 2016 Published: January 27, 2017 Editor: Raffaella Buzzetti, Universita degli Studi di Roma La Sapienza, ITALY Received: April 22, 2016 Accepted: November 17, 2016 Published: January 27, 2017 OPEN ACCESS Citation: Eloi JC, Epifanio M, de Gonc¸alves MM, Pellicioli A, Vieira PFG, Dias HB, et al. (2017) Quantification of Abdominal Fat in Obese and Healthy Adolescents Using 3 Tesla Magnetic Resonance Imaging and Free Software for Image Analysis. PLoS ONE 12(1): e0167625. doi:10.1371/ journal.pone.0167625 Data Availability Statement: All relevant data are within the paper and its Supporting Information files. Funding: The current study was funded by Conselho Nacional de Desenvolvimento Cientı´fico e Tecnolo´gico (CNPq) grant, resulted of Notice Universal 14/2012 with grant equivalent to US D: 13,000. * matteob@terra.com.br Methods Copyright: © 2017 Eloi et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. This prospective observational study included 24 overweight/obese and 33 healthy ado- lescents (mean age 16.55 years). All participants underwent abdominal MRI exams. Vis- ceral and subcutaneous fat area and percentage were correlated with anthropometric parameters, lipid profile, glucose metabolism, and insulin resistance. Student’s t test and Mann-Whitney’s test was applied. Pearson’s chi-square test was used to compare propor- tions. To determine associations Pearson’s linear correlation or Spearman’s correlation were used. Data Availability Statement: All relevant data are within the paper and its Supporting Information files. Quantification of Abdominal Fat in Obese and Healthy Adolescents Using 3 Tesla Magnetic Resonance Imaging and Free Software for Image Analysis Juliana Cristina Eloi1, Matias Epifanio1, Marı´lia Maia de Gonc¸alves2, Augusto Pellicioli2, Patricia Froelich Giora Vieira2, Henrique Bregolin Dias2, Neide Bruscato3, Ricardo Bernardi Soder4, João Carlos Batista Santana5, Marialena Mouzaki6, Matteo Baldisserotto7* a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 1 Pediatric Gastroenterologist, Pediatric Gastroenterology Service, Hospital São Lucas da Pontifı´cia Universidade Cato´lica do Rio Grande do Sul (PUCRS), Porto Alegre, Rio Grande do Sul, Brazil, 2 PUCRS, Porto Alegre, Rio Grande do Sul, Brazil, 3 Verano´polis, Rio Grande do Sul, Brazil, 4 Brain Institute (InsCer), PUCRS, Porto Alegre, Rio Grande do Sul, Brazil, 5 School of Medicine, Universidade Federal do Rio Grande do Sul (UFRGS), Porto Alegre, Rio Grande do Sul, Brazil, 6 Department of Pediatrics, Division of Gastroenterology, Hepatology and Nutrition, The Hospital for Sick Children, Toronto, Canada, 7 Imaging Center Coordinator, Brain Institute (InsCer), PUCRS, Porto Alegre, Rio Grande do Sul, Brazil Conclusions 3T MRI can provide reliable and good quality images for quantification of visceral and subcu- taneous fat by using a free software package. The results demonstrate that WC is a good predictor of visceral fat in obese adolescents and visceral fat area is associated with total cholesterol/HDL ratio, insulin and HOMA-IR. 3T MRI to Measure Abdominal Fat in Adolescents with visceral fat area (P = 0.03) and percentage (P = 0.09), and insulin and HOMA-IR were associated with visceral fat area (P = 0.001) and percentage (P = 0.005). Introduction The prevalence of childhood obesity has become a major public health issue around the world [1,2] Obesity is often associated with metabolic syndrome, which confers an increased risk of cardiovascular events in adulthood [3–5]. Previous studies have shown that central obesity, an indicator of visceral adiposity, is correlated with all the components of metabolic syndrome, namely insulin resistance, dyslipidemia, and hypertension [6,7]. WC (waist circumference) is a good predictor of abdominal adiposity; however, it does not allow for quantification of adipose tissue nor can it distinguish between visceral and subcuta- neous fat. The accurate assessment of visceral fat is of utmost clinical importance, given its association with metabolic syndrome components, which in turn, contribute to increased morbidity and mortality. Several techniques are available to measure central adiposity [8–10]. While computed tomography (CT) is the most commonly used imaging modality to measure abdominal fat, Magnetic resonance imaging (MRI) has a similar accuracy [11]. An advantage of MRI is the absence of exposure to ionizing radiation, a limitation that restricts the use of CT in children and adolescents. In addition, the MRI approach to quantifying abdominal adipos- ity is efficient, allowing for image aqcuisition within 5 minutes. One aspect that has prevented the use of both CT and MRI is the need for expensive image analysis software in the quantification of abdominal fat [12,13]. However, Irving et al [14] have shown that a free software, NIH Image J, can reliably measure adipose tissue. Even though that study was focused on CT, tone can expect similar results from the analysis of magnetic reso- nance images obtained using the same software. Regarding MRI studies of abdominal fat, most have employed equipment with field strength of 1.5 Tesla (T) [11,12,15]. Studies using MRI 3T in adults have shown good accuracy for the quantification of abdominal fat [11]. However, it is important to determine whether 3T MRI has good performance for evaluation of abdominal fat of obese adolescents. Thus, the aim of this study was to quantify subcutaneous and visceral abdominal fat of ado- lescents using 3T MRI with the free software and to correlate these findings with anthropomet- ric variables and laboratory parameters that are reflective of metabolic dysregulation. Results In both groups, waist circumference (WC) was associated with visceral fat area (P = 0.001 and P = 0.01 respectively), and triglycerides were associated with fat percentage (P = 0.046 and P = 0.071 respectively). In obese individuals, total cholesterol/HDL ratio was associated Competing Interests: The authors have declared that no competing interests exist. 1 / 12 PLOS ONE \| DOI:10.1371/journal.pone.0167625 January 27, 2017 Laboratory Tests Laboratory tests were performed following a 12-hour fast on the same day of MRI examina- tions. Lipid profile was determined based on total cholesterol, low-density lipoprotein (LDL) and high-density lipoprotein (HDL) cholesterol, and triglyceride levels using a colorimetric enzymatic method (Mindray-BS 380 Chemistry Analyzer). Fasting glucose was determined using a glucose-oxidase enzymatic method and a Mindray-BS 380 Chemistry Analyzer. Fasting insulin was determined by chemiluminescence. Insulin resistance (IR) was quantified by homeostasis model assessment (HOMA) using the formula: HOMA-IR = fasting insulin (μUI/mL) x fasting glucose (mmol/L)/22.5. Anthropometric Data Participants were classified according to BMI for age as healthy (Z score -2 and < 1), over- weight (Z score 1 and < 2), or obese (Z score 2), following World Health Organization (WHO) guidelines [16]. Participants were weighed wearing light clothes and no shoes, in an upright position, using an anthropometer coupled to a scale (Filizola1) certified by the National Institute for Metrology, Quality, and Technology (INMETRO). Body mass index (BMI) was calculated along with Z scores and percentiles, using WHO software AnthroPlus (http://www.who.int/growthref/tools/en/). WC was measured using non-stretchable plastic tape at the midpoint between the iliac crest and the lowest rib. Waist to height ratio (WHR) was calculated, with a ratio of 0.5 used as cut- point to indicate cardiovascular risk [17,18]. Body surface area was calculated using the DuBois method. [19] Tanner pubertal stage was determined as well [20,21]. Arterial Pressure Blood pressure was measured on the day of anthropometric assessment. Two measurements were performed with the subjects in the sitting position after 1 and 5-minute rest periods fol- lowing their arrival at the medical office. An INMETRO-certified aneroid sphygmomanometer was used. Maximum systolic and diastolic pressures were recorded and categorized according to international guidelines for age, sex, and height as normal, upper limit of normal, and hypertension [22]. 3T MRI to Measure Abdominal Fat in Adolescents and a minimum of correlation coefficient 0.5 in BMI association with visceral fat as Kelly et al study. and a minimum of correlation coefficient 0.5 in BMI association with visceral fat as Kelly et al study. Exclusion criteria were chronic diseases, hepatorenal disease, and use of hepatotoxic drugs, corticosteroids, or immune suppressants that could promote fat storage in the liver. The adolescents were randomly selected from a cohort participating in a population study assessing longevity, which was performed in computer system. An informed consent form was signed by all participants or by their guardians. Included subjects were subsequently divided into two groups: Group A included 33 healthy participants and Group B included 24 over- weight/obese participants (2 overweight, 22 obese). All underwent anthropometric measure- ments, laboratory testing, and abdominal MRI measurements. Design and Participants This prospective, cross-sectional study was carried out between March 2013 and October 2014 and included 57 Caucasian adolescents aged 16–18 years. The study was approved the Research Ethics Committee at our university hospital. Sample size was calculated as 23 partici- pants in each group using the PEPI 4.0 software for a significance level of 5%, power of 90%, 2 / 12 PLOS ONE \| DOI:10.1371/journal.pone.0167625 January 27, 2017 Abdominal Magnetic Resonance Imaging All MRI exams were performed at Brain Institute of PUCRS. Images were acquired in a Signa HDxt 3.0T RM scanner (General Electric, Milwaukee, USA) and an eight-element phased array abdominal coil (8-channel coil). Patients were imaged in the supine position and axial T1-weighted fast-spin echo images (FOV 440 mm, matrix 512x512, TR 230, TE 4.40, slice 3 / 12 PLOS ONE \| DOI:10.1371/journal.pone.0167625 January 27, 2017 3T MRI to Measure Abdominal Fat in Adolescents Fig 1. Magnetic resonance image of obese male (BMI 32.59). A) .jpg MRI image from L3-L4. B) Binary ImageJ image for measurement of fat: fat appears in black. Image shows intra-abdominal (visceral) and subcutaneous fat. Fig 1. Magnetic resonance image of obese male (BMI 32.59). A) .jpg MRI image from L3-L4. B) Binary ImageJ image for measurement of fat: fat appears in black. Image shows intra-abdominal (visceral) and subcutaneous fat. doi:10.1371/journal.pone.0167625.g001 thickness 5.0 mm, gap 1.0 mm, NEX 1) were obtained. Each scan lasted approximately 5 min- utes. According to previous studies, a 5 mm thickness slice at the level of L3-L4 discs was selected for the quantification of fat, as it is thought to represent the limit of the upper abdo- men and is not influenced by liver or adipose tissue from the buttocks [23,24]. The selected image was saved in .TIFF format. Imaging Analysis The TIFF images (matrix 512 x 512) were analyzed using ImageJ software (rsbweb.nih.gov/ij) with auto threshold plugin which converts automatically grayscale pixels into binary images, based in a global histogram-derived method. Black pixels represent adipose tissue and white pixels the remaining soft tissue (muscle, solid organs, intestinal loops, and vessels) [25]. Adi- pose tissue was subsequently categorized into visceral and subcutaneous fat through manual division, which was accomplished by drawing a line following the abdominal wall to separate intra and extra abdominal compartments. Visceral and subcutaneous fat areas (cm2) were measured separately (Fig 1) [26]. Results Group A included 16 girls (48.5%) and 17 boys (51.5%), vs. 13 girls (54.2%) and 11 boys (45.8%) in Group B (Table 1). Mean age was 16.8±0.7 and 16.3±0.7 in Groups A and B respec- tively. There were no significant differences between the groups in terms of gender and age distribution. WC was significantly higher in Group B (96,4 vs 75,2 cm P < 0.001). WHR was also signifi- cantly higher in Group B (P < 0.001). WHR was higher than 0.5 in only one participant in Group A (3%), vs. 22 (91.7%) participants in Group B. BMI and body surface area were signifi- cantly higher in Group B (P < 0.001). All study participants were classified as Tanner stage V. Lipid profile and fasting glucose levels were similar between the two groups. However, fast- ing insulin and HOMA-IR were higher in group B than Group A (P < 0.001). Transaminases and alkaline phosphatase levels were not different between the groups. Elevated transaminase levels (above 22 for grils and 26 for boys) were seen in only one subject in Group B (Table 2). As expected, visceral and subcutaneous fat area and percentage, measured by MRI, were significantly higher in Group B (Table 2). Females had higher subcutaneous fat percentage than males in both groups; however, there were no gender differences in terms of visceral fat percentage. Subcutaneous fat area was also higher in girls. Total abdominal area and visceral fat area were significantly higher in Group A boys vs. girls. Boys and girls in Group B differed only in regarding subcutaneous fat percentage, which was higher in girls (P = 0.006) (Table 3). In Group B, both WC and WHR correlated with subcutaneous and visceral fat area (P < 0.001 and P < 0.01 respectively). In Group A, only WC correlated with visceral fat area (P < 0.01); WHR was correlated with subcutaneous fat area in this group (P < 0.01). BMI did not correlate with visceral fat (area or percentage) in either group. However, BMI correlated with subcutaneous fat area in Group B (P < 0.001) (Table 4) and total abdominal area in both groups (P < 0.01 for Group A and P < 0.001 for Group B). Total/HDL cholesterol ratio was associated with visceral fat area and percentage in Group B, but not in Group A. Statistical Analysis Quantitative variables were expressed as means and standard deviation or median and inter- quartile range. Qualitative variables were expressed as absolute and relative frequencies. Student’s t test was used to compare group means except in the presence of asymmetric dis- tribution, in which case Mann-Whitney’s test was applied. Pearson’s chi-square test was used to compare proportions. To determine associations between variables in each group, Pearson’s linear correlation (symmetric distribution) or Spearman’s correlation (asymmetric distribu- tion) were used. The level of significance was set at 5% (P 0.05). All analyses were carried out in SPSS v. 21.0. The level of significance was set at 5% (P 0.05). All analyses were carried out in SPSS v. 21.0. The level of significance was set at 5% (P 0.05). All analyses were carried out in SPSS v. 21.0. 4 / 12 PLOS ONE \| DOI:10.1371/journal.pone.0167625 January 27, 2017 3T MRI to Measure Abdominal Fat in Adolescents Table 1. Characteristics of the sample. Variable (mean ± SD unless indicated) Healthy (n = 33) Overweight/obese (n = 24) P Age (years) 16.8±0.7 16.3±0.7 0.013* Gendera 0.877** Male 17 (51.5) 11 (45.8) Female 16 (48.5) 13 (54.2) Systolic pressure (mmHg) 116.9±10.6 121.3±12.8 0.170* Diastolic pressure (mmHg) 78.7±11.4 84.0±13.1 0.105* WHR 0.57±0.06 0.44±0.04 < 0.001* WHR 0.5a 1 (3.0) 22 (91.7) < 0.001* Waist circumference (cm) 75.2±6.5 96.4±13.1 < 0.001* BMI (Z score) -0.11±0.53 2.45±0.54 < 0.001* Body surface (m2) 1.72±0.16 1.99±0.21 <0.001* BMI, body mass index; WHR, waist to height ratio. an (%). Student’s t test Pearson’s chi-square test. Healthy: Z score -2 and < 1; Obese: Z score 2 (group includes two overweight participants Z score 1 and < 2) Table 1. Characteristics of the sample. Pearson’s chi-square test. Pearson s chi square test. Healthy: Z score -2 and < 1; Obese: Z score 2 (group includes two overweight participants, Z score 1 and < 2). q Healthy: Z score -2 and < 1; Obese: Z score 2 (group includes two overweight participants, Z score 1 and < 2). PLOS ONE \| DOI:10.1371/journal.pone.0167625 January 27, 2017 Results Similarly, fasting insulin and HOMA-IR were associated with visceral fat area and percentage in Group B. In both groups, there was an association between 5 / 12 PLOS ONE \| DOI:10.1371/journal.pone.0167625 January 27, 2017 3T MRI to Measure Abdominal Fat in Adolescents Table 2. Laboratory and MRI findings in healthy and obese adolescents. Variable (mean ± SD unless indicated) Healthy (n = 33) Overweight/obese (n = 24) P Lipid proﬁle Total cholesterol (mg/dL) 159.2±28.0 153.5±31.6 0.478 HDL cholesterol (mg/dL) 57.9±10.1 50.3±8.0 0.003* Total cholesterol/HDL ratio 2.84±0.49 3.16±0.52 0.022* Triglycerides (mg/dL) 70.8±27.4 80.0±44.5 0.339* Glycemic proﬁle Glucose (mg/dL) 80.5±6.1 81.3±7.8 0.653* Insulin (μUI/mL)a 5.1 (4.5–7.0) 9.7 (5.7–12.4) < 0.001 HOMA-IRa 0.94 (0.79–1.31) 1.73 (1.03–2.16) < 0.001 Magnetic resonance imaging Total abdominal area (cm2) 415±63.5 692±152 < 0.001* Visceral fat (cm2)a 16.5 (12.9–21) 57.8 (40–84.5) < 0.001 Subcutaneous fat (cm2)a 54.4 (42–88) 250 (174–347) < 0.001 % Abdominal fat (%) 20.9±9.0 44.6±9.5 < 0.001* % Visceral fat (%) 4.10±1.36 9.11±4.05 <0.001* % Subcutaneous fat (%) 16.8±8.5 35.4±8.3 <0.001* Table 2. Laboratory and MRI findings in healthy and obese adolescents. Mann Whitney s test. Healthy: Z score -2 and < 1; Obese: Z score 2 (group includes two overweight participants, Z score 1 and < 2). Mann Whitney s test. Healthy: Z score -2 and < 1; Obese: Z score 2 (group includes two overweight participants, Z score 1 and < 2). increased visceral fat percentage and elevated triglyceride levels; however, this reached signifi- cance only in Group B (P = 0.046) (Table 5). increased visceral fat percentage and elevated triglyceride levels; however, this reached signifi- cance only in Group B (P = 0.046) (Table 5). increased visceral fat percentage and elevated triglyceride levels; however, this reached signifi- cance only in Group B (P = 0.046) (Table 5). Discussion Mann-Whitney’s test. Healthy: Z score -2 and < 1; Obese: Z score 2 (group includes two overweight participants, Z score 1 and < 2). 3T MRI to Measure Abdominal Fat in Adolescents Table 4. Correlation between anthropometric data and MRI findings. MRI Healthy (n = 33) Overweight/obese (n = 24) WC BMI (Z score) Body surface WHR WC BMI (Z score) Body surface WHR Total abdominal area (cm2) 0.474 0.451 0.712* 0.133 0.907* 0.875* 0.791* 0.862* Visceral fat (cm2) 0.456** -0.067 0.268 0.294 0.426* 0.387 0.326 0.602 Subcutaneous fat (cm2) 0.14 0.344 -0.211 0.455 0.709* 0.821* 0.490* 0.849*** % Abdominal fat 0.145 0.173 -0.188 0.332 0.287 0.464* 0.031 0.457* % Visceral fat (%) 0.234 -0.256 0.006 0.205 0.099 0.097 0.018 0.118 % Subcutaneous fat (%) 0.115 0.223 -0.198 0.316 0.282 0.486* 0.027 0.468* Table 4. Correlation between anthropometric data and MRI findings. ation between anthropometric data and MRI findings. doi:10.1371/journal.pone.0167625.t004 subcutaneous and visceral fat in a cohort of lean and overweight adolescents. Once again it was shown that BMI did not correlate with measures of abdominal adiposity, whereas WC cor- related with both visceral and subcutaneous fat tissue. The importance of quantifying visceral fat particularly of overweight and obese subjects was shown again in this study, as visceral adi- posity correlated with markers of insulin resistance and dyslipidemia. To our knowledge there have only been two other studies reporting on the use of MRI 3.0 Tesla in the evaluation of abdominal adipose tissue, both in adults [11,15]. Klopfenstein et al. compared images obtained by MRI 3.0 Tesla to images obtained using CT, which was consid- ered the gold standard [11]. Participants were young adults with a mean BMI of 37 kg/m2. This study demonstrated that MRI provides accurate measurements of visceral and subcutane- ous adipose tissue. Li et al. reported similar results [15]. In the present study we were able to show that 3T MRI allows clinicians to obtain good quality images in obese adolescents. Table 5. Association between metabolic variables and visceral fat. Discussion In this study we were able to show that the use of MRI 3.0 Tesla with the free software package Image J allows for simple, efficient and semi-automatic quantification of abdominal Table 3. MRI findings in adolescent boys and girls. Table 3. MRI findings in adolescent boys and girls. MRI variables (mean ± SD unless indicated) Healthy (n = 33) P Obese/overweight (n = 24) P Boys Girls Boys Girls Total abdominal area (cm2) 450±55.8 379±50.3 0.001 741±166 650±131 0.148* Visceral fat area (cm2)a 17 (15–21) 15 (12–19) 0.053 60 (42–95) 56 (38–83) 0.776 Subcutaneous fat area (cm2)a 44 (33–55) 80 (54–109) 0.004 219 (146–351) 254 (182–360) 0.424 % Abdominal fat 16.9±9.4 25.0±6.2 0.007 39.3±9.1 49.1±7.5 0.008* % Visceral fat 4.10±1.4 4.07±1.3 0.952 8.6±3.4 9.6±4.6 0.545* % Subcutaneous fat 12.8±8.6 20.9±6.1 0.004 30.7±7.3 39.5±6.9 0.006* aMedian (25–75 percentile). Student’s t test; Mann-Whitney’s test. Healthy: Z score -2 and < 1; Obese: Z score 2 (group includes two overweight participants, Z score 1 and < 2). doi:10.1371/journal.pone.0167625.t003 *Mann-Whitney s test. Healthy: Z score -2 and < 1; Obese: Z score 2 (group includes two overweight participants, Z score 1 and < 2). PLOS ONE \| DOI:10.1371/journal.pone.0167625 January 27, 2017 PLOS ONE \| DOI:10.1371/journal.pone.0167625 January 27, 2017 6 / 12 doi:10.1371/journal.pone.0167625.t005 doi:10.1371/journal.pone.0167625.t004 PLOS ONE \| DOI:10.1371/journal.pone.0167625 January 27, 2017 C, waist circumference; WHR, waist to height ratio. 3T MRI to Measure Abdominal Fat in Adolescents The availability of free software Image J suggests that the use of this technology is generaliz- able. Image J has been previously shown to provide reliable measurements of adipose tissue, with similar accuracy as Slice-O-Matic version 4.3 software (Tomovision) [14]. In addition, Image J features an “eraser” tool that allows for deletion of bowel contents, which can other- wise introduce an overestimate of fat measurements [10,11]. In comparison to other anthropometric parameters, WC correlated best with visceral adi- posity area in both groups. Using MRI, Brambilla et al [27]. previously showed that WC is a good predictor of visceral adiposity, whereas BMI predicts subcutaneous adiposity. In our study BMI did not correlate with visceral or subcutaneous fat. Other studies have underscored the superiority of WC to BMI in reflecting visceral adiposity [28–30]. The limitations of using BMI in this clinical setting are numerous. For example, depending on the definition of obesity used, the prevalence of overweight and obesity using the same BMI values can vary widely [31]. Furthermore, it has been shown that BMI fails to identify excess adiposity in over one quarter of children [32], which in turn means that clinicians may fail to identify the need to screen patients at risk for metabolic dysregulation. These data, along with the fact that because of its association with visceral adiposity, an elevated WC is associated with increased future cardiovascular risk, support the inclusion of WC measurements to the routine medical assess- ment of adolescents. While WC correlates with markers of abdominal adiposity, it is limited by the fact that it cannot distinguish between visceral and subcutaneous adipose tissue. This is a key distinction when determining the cardiometabolic risk of patients [33]. Access to an efficient and cheap imaging modality, such as the one described in this study, that can distinguish between visceral and subcutaneous fat can, hence, be complementary to the baseline assessment of patients who may be found to have an elevated WC. The additional benefit of this technology is that accu- rate images can be obtained without the risk of exposing children to ionizing radiation. Laslty, evidence of increasing visceral adiposity can be used as an additional clinical tool to convey to the families the need to be compliant with lifestyle changes aimed at improving their body composition and ultimately decreasing the patients’ future cardiometabolic risk. PLOS ONE \| DOI:10.1371/journal.pone.0167625 January 27, 2017 Variable Visceral fat area (cm2) % Visceral fat (%) Total cholesterol/HDL ratio Healthy r = -0.019; P = 0.918 r = -0.004; P = 0.981 Obese/overweight r = 0.586; P = 0.003 r = 0.522; P = 0.009 Insulin levels (μU/mL) Healthy rs = 0.019; P = 0.915 rs = 0.051; P = 0.780 Obese/overweight rs = 0.625; P = 0.001 rs = 0.553; P = 0.005 HOMA Healthy rs = 0.100; P = 0.581 rs = 0.065; P = 0.720 Obese/overweight rs = 0.625; P = 0.001 rs = 0.556; P = 0.005 Triglyceride levels Healthy rs = 0.054; P = 0.767 rs = 0.318; P = 0.071 Obese/overweight rs = 0.264; P = 0.213 rs = 0.412; P = 0.046 HOMA, homeostasis model assessment; r, Pearson correlation coefﬁcient; rs, Spearman correlation coefﬁcient. Healthy: Z score -2 and < 1; Obese: Z score 2 (group includes two overweight participants, Z score 1 and < 2 doi:10.1371/journal.pone.0167625.t005 Table 5. Association between metabolic variables and visceral fat. 7 / 12 PLOS ONE \| DOI:10.1371/journal.pone.0167625 January 27, 2017 Supporting Information S1 Data. (XLSX) S1 Table. Characteristics of the sample. BMI, body mass index; WHR, waist to height ratio. an (%). Student’s t test Pearson’s chi-square test. Healthy: Z score -2 and < 1; Obese: Z score 2 (group includes two overweight participants, Z score 1 and < 2). (DOCX) S2 Table. Laboratory and MRI findings in healthy and obese adolescents. HDL, high-den- sity lipoprotein cholesterol; HOMA-IR, homeostatic model assessment—insulin resistance index. aMedian (P25 –P75). Student’s t test; Mann-Whitney’s test. Healthy: Z score -2 and < 1; Obese: Z score 2 (group includes two overweight participants, Z score 1 and < 2). (DOCX) S3 Table. S3 Title: MRI findings in adolescent boys and girls. aMedian (25–75 percentile). Student’s t test; Mann-Whitney’s test. Healthy: Z score -2 and < 1; Obese: Z score 2 (group includes two overweight participants, Z score 1 and < 2). (DOCX) S4 Table. Correlation between anthropometric data and MRI findings. BMI, body mass index; WC, waist circumference; WHR, waist to height ratio. P < 0.05; P < 0.01; P < 0.001. Healthy: Z score -2 and < 1; Obese: Z score 2 (group includes two over- weight participants, Z score 1 and < 2). (DOCX) S5 Table. Association between metabolic variables and visceral fat. HOMA, homeostasis model assessment; r, Pearson correlation coefficient; rs, Spearman correlation coefficient. Healthy: Z score -2 and < 1; Obese: Z score 2 (group includes two overweight partici- pants, Z score 1 and < 2. (DOCX) Author Contributions Conceptualization: MB JCE ME JCBS RBS. PLOS ONE \| DOI:10.1371/journal.pone.0167625 January 27, 2017 We did not observe elevations in the fasting glucose of the adolescents included in this study. However, fasting insulin, HOMA-IR and triglycerides were significantly higher in over- weight and obese participants, suggesting the presence of insulin resistance. In addition, HOMA-IR was strongly correlated with visceral adiposity area in obese participants. This find- ing is in agreement with previously published studies that show a causative relationship between visceral adiposity and insulin resistance [34,35]. In both groups, there was a trend towards increased triglyceride levels and increased vis- ceral fat percentage. It should be noted that only three healthy (9%) and five obese (20.8%) par- ticipants had triglycerides above 100 mg/dL. We found a strong correlation of total to HDL cholesterol with visceral fat percentage. This is in accordance with other investigators, who have also shown a strong association between central obesity and dyslipidemia [36,37]. Limitations of the present study include the fact that only Caucasian adolescents of advanced pubertal stage were included. Another limitation is that we did not assess the accu- racy of 3T MRI in measuring subcutaneous and visceral abdominal fat but extrapolated data from the adult literature that suggests that this technology is accurate. A study assessing the accuracy of this MRI technology in adolescents would have required exposure to unnecessary radiation, as CT scans are considered the gold standard for these types of measurements. In conclusion, we show that 3T MRI can provide good quality images using a free software package that allows fast and accurate quantification of visceral and subcutaneous fat in over- weight and obese adolescents. The abdominal fat segmentation results demonstrate that WC is 8 / 12 3T MRI to Measure Abdominal Fat in Adolescents a good estimate of visceral and subcutaneous fat and the visceral fat area is associated with total cholesterol / HDL cholesterol, insulin and HOMA-IR. a good estimate of visceral and subcutaneous fat and the visceral fat area is associated with total cholesterol / HDL cholesterol, insulin and HOMA-IR. References 1. Ogden CL, Carroll MD, Kit BK, Flegal KM. Prevalence of childhood and adult obesity in the United States, 2011–2012. JAMA 2014; 311: 806–814. doi: 10.1001/jama.2014.732 PMID: 24570244 2. de Onis M, Blossner M, Borghi E. Global prevalence and trends of overweight and obesity among pre- school children. Am J Clin Nutr 2010; 92: 1257–1264. doi: 10.3945/ajcn.2010.29786 PMID: 20861173 3. Berenson GS, Srinivasan SR, Bao W, Newman WP 3rd, Tracy RE, Wattigney WA. Association between multiple cardiovascular risk factors and atherosclerosis in children and young adults. The Bogalusa Heart Study. N Engl J Med 1998; 338: 1650–1656. doi: 10.1056/NEJM199806043382302 PMID: 9614255 4. Juonala M, Magnussen CG, Berenson GS, Venn A, Burns TL, Sabin MA, et al. Childhood adiposity, adult adiposity, and cardiovascular risk factors. N Engl J Med 2011; 365: 1876–1885. doi: 10.1056/ NEJMoa1010112 PMID: 22087679 5. Tirosh A, Shai I, Afek A, Dubnov-Raz G, Ayalon N, Gordon B, et al. Adolescent BMI trajectory and risk of diabetes versus coronary disease. N Engl J Med 2011; 364: 1315–1325. doi: 10.1056/ NEJMoa1006992 PMID: 21470009 6. Agredo-Zu´ñiga RA, Aguilar-de Plata C, Sua´rez-Ortego´n MF. Waist:height ratio, waist circumference and metabolic syndrome abnormalities in Colombian schooled adolescents: a multivariate analysis con- sidering located adiposity. Br J Nutr. 2015; 114(5):700–5. doi: 10.1017/S0007114515002275 PMID: 26279413 7. Guntsche Z, Guntsche EM, Saravı´ FD, Gonzalez LM, Lopez Avellaneda C, Ayub E, et al. Umbilical waist-to-height ratio and trunk fat mass index (DXA) as markers of central adiposity and insulin resis- tance in Argentinean children with a family history of metabolic syndrome. J Pediatr Endocrinol Metab. 2010; 23(3):245–56. PMID: 20480723 8. Kullberg J, Brandberg J, Angelhed JE, Frimmel H, Bergelin E, Strid L, et al. Whole-body adipose tissue analysis: comparison of MRI, CT and dual energy X-ray absorptiometry. Br J Radiol 2009; 82: 123– 130. doi: 10.1259/bjr/80083156 PMID: 19168691 9. Karlsson AK, Kullberg J, Stokland E, Allvin K, Gronowitz E, Svensson PA, et al. Measurements of total and regional body composition in preschool children: A comparison of MRI, DXA, and anthropometric data. Obesity (Silver Spring) 2013; 21: 1018–1024. 10. Mook-Kanamori DO, Holzhauer S, Hollestein LM, Durmus B, Manniesing R, Koek M, et al. Abdominal fat in children measured by ultrasound and computed tomography. Ultrasound Med Biol 2009; 35: 1938–1946. doi: 10.1016/j.ultrasmedbio.2009.07.002 PMID: 19800165 11. Klopfenstein BJ, Kim MS, Krisky CM, Szumowski J, Rooney WD, Purnell JQ. Conceptualization: MB JCE ME JCBS RBS. Conceptualization: MB JCE ME JCBS RBS. Data curation: JCE MMG AP PFGV HBD NB MM. Formal analysis: MB RBS ME JCE. Funding acquisition: MB JCE. Investigation: JCE MMG AP PFGV HBD NB MM. Methodology: MB JCE ME JCBS RBS. Project administration: MB JCBS JCE. Resources: MMG AP PFGV HBD NB MM. Software: MB RBS JCE. 9 / 12 PLOS ONE \| DOI:10.1371/journal.pone.0167625 January 27, 2017 9 / 12 3T MRI to Measure Abdominal Fat in Adolescents Supervision: MB JCBS. Validation: MB RBS JCE. Visualization: MMG AP PFGV HBD NB MM. Writing – original draft: MB ME MM RBS. Writing – review & editing: MB ME MM RBS. PLOS ONE \| DOI:10.1371/journal.pone.0167625 January 27, 2017 References Pediatrics 2004; 114: 555–576. PMID: 15286277 23. Sottier D, Petit JM, Guiu S, Hamza S, Benhamiche H, Hillon P, et al. Quantification of the visceral and subcutaneous fat by computed tomography: interobserver correlation of a single slice technique. Diagn Interv Imaging. 2013; 94(9):879–84. doi: 10.1016/j.diii.2013.04.006 PMID: 23725783 24. Tong Y, Udupa JK, Torigian DA. Optimization of abdominal fat quantification on CT imaging through use of standardized anatomic space: a novel approach. Med Phys. 2014; 41(6):063501. doi: 10.1118/ 1.4876275 PMID: 24877839 25. Sezgin M & Sankur B. Survey over Image Thresholding Techniques and Quantitative Performance Evaluation. Journal of Electronic Imaging 13(1): 146–165. 26. Hu HH, Nayak KS, Goran MI. Assessment of Abdominal Adipose Tissue and Organ Fat Content by Magnetic Resonance Imaging. Obes Rev. 2011; 12(501): e504–e515. 27. Brambilla P, Bedogni G, Moreno LA, Goran MI, Gutin B, Fox KR, et al. Crossvalidation of anthropometry against magnetic resonance imaging for the assessment of visceral and subcutaneous adipose tissue in children. Int J Obes (Lond) 2006; 30: 23–30. 28. Schro¨der H, Ribas L, Koebnick C, Funtikova A, Gomez SF, Fı´to M, et al. Prevalence of abdominal obe- sity in Spanish children and adolescents. Do we need waist circumference measurements in pediatric practice? PLoS One. 2014; 9(1):e87549. doi: 10.1371/journal.pone.0087549 PMID: 24475305 29. Gro¨ber-Gra¨tz D, Widhalm K, de Zwaan M, Reinehr T, Blu¨her S, Schwab KO, et al. Body mass index or waist circumference: which is the better predictor for hypertension and dyslipidemia in overweight/ obese children and adolescents? Association of cardiovascular risk related to body mass index or waist circumference. Horm Res Paediatr. 2013; 80(3):170–8. doi: 10.1159/000354224 PMID: 24021483 30. Savva SC, Tornaritis M, Savva ME, Kourides Y, Panagi A, Silikiotou N, et al. Waist circumference and waist-to-height ratio are better predictors of cardiovascular disease risk factors in children than body mass index. Int J Obes Relat Metab Disord. 2000; 24(11):1453–8. PMID: 11126342 31. Gonzalez-Casanova I, Sarmiento OL, Gazmararian JA, Cunningham SA, Martorell R, Pratt M, Stein AD. Comparing three body mass index classification systems to assess overweight and obesity in chil- dren andadolescents. Rev Panam Salud Publica. 2013; 33(5):349–55. PMID: 23764666 32. Javed A, Jumean M, Murad MH, Okorodudu D, Kumar S, Somers VK, et al. Diagnostic performance of body mass index to identify obesity as defined by body adiposity in children andadolescents: a system- atic review and meta-analysis. Pediatr Obes. 2015; 10(3):234–44. doi: 10.1111/ijpo.242 PMID: 24961794 33. References Comparison of 3 T MRI and CT for the measurement of visceral and subcutaneous adipose tissue in humans. Br J Radiol 2012; 85: e826–830. doi: 10.1259/bjr/57987644 PMID: 22514099 12. Benfield LL, Fox KR, Peters DM, Blake H, Rogers I, Grant C, et al. Magnetic resonance imaging of abdominal adiposity in a large cohort of British children. Int J Obes (Lond) 2008; 32: 91–99. 13. Abate N, Burns D, Peshock RM, Garg A, Grundy SM. Estimation of adipose tissue mass by magnetic resonance imaging: validation against dissection in human cadavers. J Lipid Res 1994; 35: 1490– 1496. PMID: 7989873 14. Irving BA, Weltman JY, Brock DW, Davis CK, Gaesser Ga, Weltman A. NIH ImageJ and Slice-O-Matic computed tomography imaging software to quantify soft tissue. Obesity 2007; 15(2):370–6. doi: 10. 1038/oby.2007.573 PMID: 17299110 15. Li X, Youngren JF, Hyun B, Sakkas GK, Mulligan K, Majumdar S, et al. Technical evaluation of in vivo abdominal fat and IMCL quantification using MRI and MRSI at 3 T. Magn Reson Imaging 2008; 26: 188–197. doi: 10.1016/j.mri.2007.06.006 PMID: 17683890 10 / 12 PLOS ONE \| DOI:10.1371/journal.pone.0167625 January 27, 2017 3T MRI to Measure Abdominal Fat in Adolescents 16. de Onis M, Onyango AW, Borghi E, Siyam A, Nishida C, Siekmann J. Development of a WHO growth reference for school-aged children and adolescents. Bull World Health Organ 2007; 85: 660–667. doi: 10.2471/BLT.07.043497 PMID: 18026621 17. Nambiar S, Truby H, Davies PS, Baxter K. Use of the waist-height ratio to predict metabolic syndrome in obese children and adolescents. J Paediatr Child Health 2013; 49: E281–287. doi: 10.1111/jpc. 12147 PMID: 23521181 18. Savva SC, Lamnisos D, Kafatos AG. Predicting cardiometabolic risk: waist-to-height ratio or BMI. A meta-analysis. Diabetes Metab Syndr Obes 2013; 6: 403–419. doi: 10.2147/DMSO.S34220 PMID: 24179379 19. Du Bois D, Du Bois EF. A formula to estimate the approximate surface area if height and weight be known. 1916. Nutrition 1989; 5: 303–311; discussion 312–303. PMID: 2520314 20. Marshall WA, Tanner JM. Variations in the pattern of pubertal changes in boys. Arch Dis Child 1970; 45: 13–23. PMID: 5440182 21. Marshall WA, Tanner JM. Variations in pattern of pubertal changes in girls. Arch Dis Child 1969; 44: 291–303. PMID: 5785179 22. National high blood pressure education program working group on high blood pressure in children and adolescents. The fourth report on the diagnosis, evaluation, and treatment of high blood pressure in chil- dren and adolescents. PLOS ONE \| DOI:10.1371/journal.pone.0167625 January 27, 2017 36. Caprio S, Hyman LD, McCarthy S, Lange R, Bronson M, Tamborlane WV. Fat distribution and cardio- vascular risk factors in obese adolescent girls: importance of the intraabdominal fat depot. Am J Clin Nutr 1996; 64: 12–17. PMID: 8669407 37. Daniels SR, Morrison JA, Sprecher DL, Khoury P, Kimball TR. Association of body fat distribution and cardiovascular risk factors in children and adolescents. Circulation 1999; 99: 541–545. PMID: 9927401 PLOS ONE \| DOI:10.1371/journal.pone.0167625 January 27, 2017 3T MRI to Measure Abdominal Fat in Adolescents References Clemente G, Mancini M, Giacco R, Tornatore A, Ragucci M, Riccardi G. Visceral adiposity and subclini- cal atherosclerosis in healthy young men. Int J Food Sci Nutr. 2015; 66(4):466–70. doi: 10.3109/ 09637486.2015.1042845 PMID: 26017320 34. Frayn KN. Visceral fat and insulin resistance-causative or correlative? Br J Nutr. 2000; 83 Suppl 1: S71–7. 35. Kabir M, Catalano KJ, Ananthnarayan S, Kim SP, Van Citters GW, Dea MK, et al. Molecular evidence supporting the portal theory: a causative link between visceral adiposity and hepatic insulin resistance. Am J Physiol Endocrinol Metab. 2005; 288(2):E454–61. doi: 10.1152/ajpendo.00203.2004 PMID: 15522994 11 / 12 PLOS ONE \| DOI:10.1371/journal.pone.0167625 January 27, 2017 3T MRI to Measure Abdominal Fat in Adolescents 3T MRI to Measure Abdominal Fat in Adolescents 12 / 12
https://openalex.org/W2789946978	https://www.repository.cam.ac.uk/bitstream/1810/274083/3/Varela_et_al-2018-Angewandte_Chemie_International_Edition.pdf	English	null	Optical Structural Analysis of Individual α‐Synuclein Oligomers	Angewandte Chemie	2,018	cc-by	5,318	Protein Aggregates Optical Structural Analysis of Individual a-Synuclein Oligomers Juan A. Varela, Margarida Rodrigues, Suman De, Patrick Flagmeier, Sonia Gandhi, Christopher M. Dobson, David Klenerman,* and Steven F. Lee* Abstract: Small aggregates of misfolded proteins play a key role in neurodegenerative disorders. Such species have proved difficult to study due to the lack of suitable methods capable of resolving these heterogeneous aggregates, which are smaller than the optical diffraction limit. We demonstrate here an all- optical fluorescence microscopy method to characterise the structure of individual protein aggregates based on the fluorescence anisotropy of dyes such as thioflavin-T, and show that this technology is capable of studying oligomers in human biofluids such as cerebrospinal fluid. We first inves- tigated in vitro the structural changes in individual oligomers formed during the aggregation of recombinant a-synuclein. By studying the diffraction-limited aggregates we directly evalu- ated their structural conversion and correlated this with the potential of aggregates to disrupt lipid bilayers. We finally characterised the structural features of aggregates present in cerebrospinal fluid of ParkinsonQs disease patients and age- matched healthy controls. filament[5,6] is recognised to drive pathogenesis in a-synuclei- nopathies. Furthermore, aggregates of a-synuclein exhibit cell–cell transfer, leading to seeding and recruitment of more protein molecules to form additional aggregates that can generate new seeds in an exponential way,[7] leading to the region–region spread of disease. The distinct structure of a- synuclein aggregates has a role in its pathogenic properties, in particular, the toxicity of the aggregate, the cell type affected, seed competency, and the regional transfer of pathology.[8,9] This dramatic effect of the structure of the ordered assembly on the pathogenic pathway in the brain underpins the importance of understanding the order/structure of a-synu- clein aggregates. Previous studies have shown aggregates to be very diverse in terms of their mechanisms of formation, size and struc- ture.[10–13] Bulk measurements obtained with conventional, ensemble-based, biophysical techniques are able to character- ise many features of these heterogeneous aggregates,[14,15] but new quantitative tools are needed to specifically characterise in greater detail the structural features of individual aggre- gates, particularly in human tissue and biological fluids. 2018 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2018, 57, 4886 –4890 Protein Aggregates The formation and spread of small aggregates of proteins such as a-synuclein, b-amyloid or tau is reported in a wide range of neurodegenerative diseases.[1,2] Of these, ParkinsonQs disease (PD) is characterised by the accumulation of a mis- folded and aggregated protein called a-synuclein within neurons to form Lewy neurites and Lewy bodies.[3] Genetic and pathological evidence suggests that the protein a- synuclein is central to neurodegeneration in PD.[4] Specifi- cally, the transition from an intrinsically disordered a- synuclein monomer through a series of oligomeric intermedi- ates (with varying structures and size) to a highly structured By means of single-molecule fluorescence resonance energy transfer (smFRET) experiments, a subpopulation of aggregates formed with fluorescently labelled a-synuclein has previously been shown to undergo a slow structural rear- rangement before growing into fibrils.[12] This conversion can generate more cross-b structure and the resulting aggregates were reported to be both, more resistant to proteinase-K and more toxic to cells. Unlabelled fibrils of amyloid-b or a- synuclein can be imaged with total internal fluorescence (TIRF) microscopy and structurally specific dyes such as thioflavin T (ThT),[16,17] opening up the possibilities of study- ing aggregates in human biofluids.[18] At a single fibril level, dyes such as ThT or Congo Red have been shown to bind insulin fibrils in an ordered way, and by monitoring the intensity as a function of the polarisation angle, these dye classes can be provide information on the structure of fibrils.[19] [] Dr. J. A. Varela, Dr. M. Rodrigues, Dr. S. De, P. Flagmeier, Prof. C. M. Dobson, Prof. D. Klenerman, Dr. S. F. Lee Department of Chemistry, University of Cambridge Lensfield Road, Cambridge, CB2 1EW (UK) E-mail: dk10012@cam.ac.uk sl591@cam.ac.uk Prof. D. Klenerman UK Dementia Research Institute, University of Cambridge Cambridge CB2 0XY (UK) Dr. S. Gandhi Department of Molecular Neuroscience, Institute of Neurology, University College London Queen Square, London WC1N 3BG (UK) Supporting information and the ORCID identification number(s) for the author(s) of this article can be found under https://doi.org/10. 1002/anie.201710779. T 2018 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. [] Dr. J. A. Varela, Dr. M. Rodrigues, Dr. S. De, P. Flagmeier, Prof. C. M. Dobson, Prof. D. Klenerman, Dr. S. F. German Edition: DOI: 10.1002/ange.201710779 International Edition: DOI: 10.1002/anie.201710779 Protein Aggregates Angewandte Chemie Angewandte Chemie Communications German Edition: DOI: 10.1002/ange.201710779 International Edition: DOI: 10.1002/anie.201710779 Protein Aggregates Lee Department of Chemistry, University of Cambridge Lensfield Road, Cambridge, CB2 1EW (UK) E-mail: dk10012@cam.ac.uk sl591@cam.ac.uk Prof. D. Klenerman UK Dementia Research Institute, University of Cambridge Cambridge CB2 0XY (UK) Dr. S. Gandhi Department of Molecular Neuroscience, Institute of Neurology, University College London Queen Square, London WC1N 3BG (UK) Supporting information and the ORCID identification number(s) for the author(s) of this article can be found under https://doi.org/10. 1002/anie.201710779. T 2018 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. In this work, we have characterised structural features of aggregates formed by an amyloidogenic protein, by imple- menting a highly sensitive fluorescence anisotropy setup. Fluorophores absorb light with a probability proportional to the square of the dot product of the local optical electric field and the molecular transition dipole moment. Thus when a dye binds in a defined orientation it emits highly polarised anisotropic fluorescence. We therefore built a bespoke aniso- tropy instrument to study the structure of spatially isolated amyloid aggregates by placing a polariser in the detection path which rotates continuously during image acquisition (Figure 1a). ThT is a widely-used benzothiazole dye that increases its fluorescence quantum yield by several orders of T 2018 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. T 2018 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. 4886 T 2018 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim An Angewandte Chemie Angewandte Chemie Communications Figure 1. Procedure to monitor anisotropy: a) Fluorescence anisotropy was measured on a total internal fluorescence microscopy (TIRF) arrangement in which a rotating polariser is mounted between the dichroic and the tube lens of the microscope. b) ThT binds specifically to the cross-b architecture of amyloid fibrils in a well-defined manner. Protein Aggregates Procedure to monitor anisotropy: a) Fluorescence anisotropy was measured on a total internal fluorescence microscopy (TIRF) arrangement in which a rotating polariser is mounted between the dichroic and the tube lens of the microscope. b) ThT binds specifically to the cross-b architecture of amyloid fibrils in a well-defined manner. c) The detected fluorescence of ThT for a-synuclein fibrils is maximum when the polariser is aligned with the fibril, and minimum when it is orthogonal to the fibril (scale bar=5 mm). d) Montage of the modulat- ing fluorescence of an a-synuclein fibril as the polariser rotates (each frame corresponds to the intensity averaged during a rotation of approximately 3.68). e) This intensity can be fitted to a sinusoidal curve to quantify the degree of modulation and hence structural order. Figure 1. Procedure to monitor anisotropy: a) Fluorescence anisotropy was measured on a total internal fluorescence microscopy (TIRF) arrangement in which a rotating polariser is mounted between the dichroic and the tube lens of the microscope. b) ThT binds specifically to the cross-b architecture of amyloid fibrils in a well-defined manner. c) The detected fluorescence of ThT for a-synuclein fibrils is maximum when the polariser is aligned with the fibril, and minimum when it is orthogonal to the fibril (scale bar=5 mm). d) Montage of the modulat- ing fluorescence of an a-synuclein fibril as the polariser rotates (each frame corresponds to the intensity averaged during a rotation of approximately 3.68). e) This intensity can be fitted to a sinusoidal curve to quantify the degree of modulation and hence structural order. magnitude upon binding extended cross-b structures. Experi- ments suggest that the dye preferentially binds with its long axis parallel to the long axis of fibrils,[20–22] but depending on the protein under study and the structure of the fibril other binding sites may exist. When single a-synuclein fibrils are imaged as a function of the angle of the axis of polarisation, their fluorescence modulates sinusoidally between a maxi- mum when the axis is aligned with the fibril and a minimum if the axis is perpendicular to the fibril (Figure 1b–d). This further confirms that the dominant binding site (or possibly sites) of ThT is aligned with the axis of the a-synuclein fibrils. Figure 2. Angew. Chem. Int. Ed. 2018, 57, 4886 –4890 T 2018 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Protein Aggregates c) The detected fluorescence of ThT for a-synuclein fibrils is maximum when the polariser is aligned with the fibril, and minimum when it is orthogonal to the fibril (scale bar=5 mm). d) Montage of the modulat- ing fluorescence of an a-synuclein fibril as the polariser rotates (each frame corresponds to the intensity averaged during a rotation of approximately 3.68). e) This intensity can be fitted to a sinusoidal curve to quantify the degree of modulation and hence structural order. signal, b the constant angular frequency (user-defined rota- tion velocity of the polariser), c the phase and d an offset (Figure 1e). The response to the anisotropy measurement is defined by: modulation ¼ 2a=ða þ dÞ. A small modulation value implies disordered b-sheet content in the aggregate, whereas a larger modulation value implies spatially aligned b- sheet content. We used a single-molecule sensitive TIRF imaging mode to measure the structural arrangement of individual spatially isolated diffraction-limited aggregates which we have pre- viously characterised using super-resolution techniques.[23] We performed an aggregation reaction for recombinant a-synu- clein and focused on the kinetics of the lag phase of the aggregation.[24] The reaction was done at a concentration of 70 mm under agitation at 200 rpm in 25 mm Tris buffer (pH 7.4) supplemented with 0.1m NaCl and 0.01% NaN3 at 378C. The aggregation reaction was performed in low binding polypropylene tubes to minimise protein adhering to the tubes. More specifically we analysed samples taken from the aggregation reaction at times between 1 h and 4 h. At longer times aggregates larger than the diffraction limit of optical light (i.e. & 170 nm) start to form. A histogram of modulation Figure 2. Structural analysis of the species observed during an a- synuclein aggregation reaction: a) Modulation measurements of detected aggregates during an aggregation reaction at 70 mm show an evolution from relatively disordered aggregates (1 h and 2 h) to a mixture of relatively disordered and “fibril-like” aggregates (3 h and 4 h). Each histogram pools data from 3 independent aggregations. Representative fluorescence images are shown in (b) (scale bar=5 mm) with magnified sub-images (2.5x); number of aggregates analysed in each time point: 1 h n=121, 2 h n=385, 3 h n=581, 4 h n=719. Figure 1. Protein Aggregates Structural analysis of the species observed during an a- synuclein aggregation reaction: a) Modulation measurements of detected aggregates during an aggregation reaction at 70 mm show an evolution from relatively disordered aggregates (1 h and 2 h) to a mixture of relatively disordered and “fibril-like” aggregates (3 h and 4 h). Each histogram pools data from 3 independent aggregations. Representative fluorescence images are shown in (b) (scale bar=5 mm) with magnified sub-images (2.5x); number of aggregates analysed in each time point: 1 h n=121, 2 h n=385, 3 h n=581, 4 h n=719. A measure of the degree of extended ordered cross-b structure in an aggregate is the amplitude of the fluorescence signal of ThT as the polariser rotates. The normalised fluorescence intensity as a function of polariser angular displacement of each individual protein aggregate was fitted to a sinusoid (see Methods section and Figure S1 in the Supporting Information (SI)). The fluorescence was fitted to y ¼ acos bx þ c ð Þ þ d, where a represents the amplitude of the 4887 www.angewandte.org Angewandte Chem Communications be therefore fitted to three Gaussian distributions (Fig- ure 3c). The population corresponding to low modulation values have some cross-b content (as ThT binds to them), and behaves in a similar way to fluorescent beads (SI, Figure S5), meaning that these aggregates are not structurally aligned. The scatter plot of modulation vs. the mean intensity of all measured aggregates suggested once again that disordered aggregates convert to fibril-like aggregates without an increase in integrated fluorescence intensity (Figure 3d). We observed that fibrils are characterised by high modulation and intensity values (Figure 3d). This strong fluorescence aniso- tropy response of fibrils can be achieved by labelling with other fluorescent dyes such as pentameric formyl thiophene acetic acid (pFTAA) which has also been shown to bind to cross-b structures[25] (SI, Figure S6). values for each time point revealed that oligomers present at 1 h and 2 h have low modulation values (typically lower than 0.5), while at 3 h we found a distinct population of oligomers that respond with high modulation values (Figure 2). These data are indicative of the structural rearrangement from a relatively amorphous oligomer into a “fibril-like” periodic structure. Protein Aggregates The modulation measurement does not correlate with the fluorescence intensity of the aggregate (SI, Figure S2 and S3), suggesting that the number of ThT binding sites can remain constant during a structural re-arrangement. Although there is a variability associated with the stochas- ticity of the nucleation process during the lag phase of the aggregation, independent experiments show the presence of the two populations of aggregates and the same trend for the evolution of the species (SI, Figure S4). In order to understand further the evolution of aggregates we globally fitted two Gaussian functions to each time point, pooling data obtained in three independent aggregation reactions to describe better the landscape of aggregates (Figure 3a). The integrated areas of the Gaussians give an estimate of the number of aggregates in each population (Figure 4a). The results show that non-modulating aggregates appear before modulating ones, and at a slower rate, Combining all the detected aggregates, the overall dis- tribution of the degree of modulation can be fitted to two Gaussian distributions (Figure 3a). When compared to the modulation of long fibrils (for example formed after 24 h of aggregation, yielding fibrils that are several mm long), we observe that the fibrils typically have higher modulation values (Figure 3b). The histogram of oligomers and fibrils can Figure 3. Modulation landscape of small aggregates and fibrils of a- synuclein: a) Modulation measurements of all species detected in three independent aggregation reactions fitted to two Gaussian functions (1149 species in total). b) Long fibrils (formed after 24 h aggregation, typically several mm in length) respond with high modu- lation values, which can be fitted adding a third Gaussian (511 species considered). c) The complete landscape of aggregates and fibrils can be fitted with three Gaussian functions. d) Scatter plot of mean intensity of each aggregate (green dots) and fibril (magenta dots) and its corresponding modulation values. Figure 3. Modulation landscape of small aggregates and fibrils of a- synuclein: a) Modulation measurements of all species detected in three independent aggregation reactions fitted to two Gaussian functions (1149 species in total). b) Long fibrils (formed after 24 h aggregation, typically several mm in length) respond with high modu- lation values, which can be fitted adding a third Gaussian (511 species considered). c) The complete landscape of aggregates and fibrils can be fitted with three Gaussian functions. Protein Aggregates d) Scatter plot of mean intensity of each aggregate (green dots) and fibril (magenta dots) and its corresponding modulation values. Figure 4. Modulation and toxicity of a-synuclein during aggregation: a) Number of modulating and non-modulating a-synuclein aggregates obtained from the time evolution of the two populations shown in Figure 2a. (error bars correspond to error in population fits multiplied by the bin size). b) Ca2+ influx kinetics obtained with a-synuclein aggregates (average of the three independent aggregation reactions, error bars are standard deviation between three replicates). c) Example of liposome assay, showing liposomes with disrupted membranes as fluorescent puncta. A higher number of fluorescent liposomes can be seen when incubating with 4 h aggregates as compared with 1 h aggregates (scale bars=5 mm). www angewandte org T 2018 The Authors Published by Wiley VCH Verlag GmbH & Co KGaA Weinheim Angew Chem Int Ed 2018 57 4886 4890 Figure 4. Modulation and toxicity of a-synuclein during aggregation: a) Number of modulating and non-modulating a-synuclein aggregates obtained from the time evolution of the two populations shown in Figure 2a. (error bars correspond to error in population fits multiplied by the bin size). b) Ca2+ influx kinetics obtained with a-synuclein aggregates (average of the three independent aggregation reactions, error bars are standard deviation between three replicates). c) Example of liposome assay, showing liposomes with disrupted membranes as fluorescent puncta. A higher number of fluorescent liposomes can be seen when incubating with 4 h aggregates as compared with 1 h aggregates (scale bars=5 mm). Figure 4. Modulation and toxicity of a-synuclein during aggregation: a) Number of modulating and non-modulating a-synuclein aggregates obtained from the time evolution of the two populations shown in Figure 2a. (error bars correspond to error in population fits multiplied by the bin size). b) Ca2+ influx kinetics obtained with a-synuclein aggregates (average of the three independent aggregation reactions, error bars are standard deviation between three replicates). c) Example of liposome assay, showing liposomes with disrupted membranes as fluorescent puncta. A higher number of fluorescent liposomes can be seen when incubating with 4 h aggregates as compared with 1 h aggregates (scale bars=5 mm). Figure 4. Modulation and toxicity of a-synuclein during aggregation: a) Number of modulating and non-modulating a-synuclein aggregates obtained from the time evolution of the two populations shown in Figure 2a. (error bars correspond to error in population fits multiplied by the bin size). Protein Aggregates In contrast to these well-defined populations, aggregates formed during an aggregation reac- tion of a-synuclein at low monomer concentration (1 mm for 1 month) display a broader distribution of modulation responses, suggesting that a wider variety of species are formed over long periods of time (SI, Figure S7). To correlate the structural information of the aggregated species with the ability of these aggregates to generate toxic effects in cells we evaluated their capability to disrupt membranes with a technique that uses vesicles filled with a Ca2+ sensitive dye.[26] Upon the interaction of a protein aggregate with the vesicleQs membrane, Ca2+ ions enter the vesicle from the surrounding solution and hence becomes fluorescent. This change in fluorescence can be detected using TIRF microscopy. We imaged individual liposomes in the presence of Ca2+ buffer (blank), followed by the addition of an aliquot of a-synuclein aggregates at a concentration of 50 nm and subsequent addition of ionomycin. In the presence of only Ca2+ buffer, the fluorescence intensity of the vesicles was low and comparable to that of background noise due to minimal Ca2+ presence within the vesicle.[26] After incubation (for 10 minutes) with a-synuclein samples, we detected an increase in the localised fluorescence intensity of the vesicles showing that Ca2+ ions could enter the vesicles as a conse- quence of the aggregatesQ induced membrane permeability. Subsequent addition of ionomycin, an ionophore enabling Ca2+ to enter the vesicles, results in the saturation of all vesicles with Ca2+ ions, allowing us to quantify the extent of membrane disruption (Figure 4c). Figure 5. Anisotropy of species observed in human cerebrospinal fluid: Fluorescence anisotropy histogram of species in a) HC CSF and b) PD CSF obtained by pooling aggregates from four different individuals in each case (HC: n=1461 species, mean=0.3112, standard devia- tion=0.1015; PD: n=1069 species, mean=0.2929, standard devia- tion=0.1136). c) Cumulative distribution of species from the same data set showing higher number of modulating species (modula- tion>0.45) in PD. The permeabilisation assay showed a linear increase of the Ca2+ influx after incubation with aggregates (previously aggregated for 1 h to 4 h), suggesting that both disordered and fibril-like aggregates induce calcium influx in the liposomes (Figure 4b). Given that the trend of modulating species dominates the later time points while the abundance of non- modulating aggregates increases at a slower rate (Figure 4a), our results suggest that modulating species have a higher ability to disrupt membranes. Protein Aggregates b) Ca2+ influx kinetics obtained with a-synuclein aggregates (average of the three independent aggregation reactions, error bars are standard deviation between three replicates). c) Example of liposome assay, showing liposomes with disrupted membranes as fluorescent puncta. A higher number of fluorescent liposomes can be seen when incubating with 4 h aggregates as compared with 1 h aggregates (scale bars=5 mm). Figure 3. Modulation landscape of small aggregates and fibrils of a- synuclein: a) Modulation measurements of all species detected in three independent aggregation reactions fitted to two Gaussian functions (1149 species in total). b) Long fibrils (formed after 24 h aggregation, typically several mm in length) respond with high modu- lation values, which can be fitted adding a third Gaussian (511 species considered). c) The complete landscape of aggregates and fibrils can be fitted with three Gaussian functions. d) Scatter plot of mean intensity of each aggregate (green dots) and fibril (magenta dots) and its corresponding modulation values. Figure 3. Modulation landscape of small aggregates and fibrils of a- synuclein: a) Modulation measurements of all species detected in three independent aggregation reactions fitted to two Gaussian functions (1149 species in total). b) Long fibrils (formed after 24 h aggregation, typically several mm in length) respond with high modu- lation values, which can be fitted adding a third Gaussian (511 species considered). c) The complete landscape of aggregates and fibrils can be fitted with three Gaussian functions. d) Scatter plot of mean intensity of each aggregate (green dots) and fibril (magenta dots) and its corresponding modulation values. 4888 T 2018 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2018, 57, 4886 –4890 www.angewandte.org T 2018 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Ch www.angewandte.org Figure 5. Anisotropy of species observed in human cerebrospinal fluid: Fluorescence anisotropy histogram of species in a) HC CSF and b) PD CSF obtained by pooling aggregates from four different individuals in each case (HC: n=1461 species, mean=0.3112, standard devia- tion=0.1015; PD: n=1069 species, mean=0.2929, standard devia- tion=0.1136). c) Cumulative distribution of species from the same data set showing higher number of modulating species (modula- tion>0.45) in PD. Angewan nications Communications suggesting that there is a conversion of non-modulating to modulating aggregates. Angew. Chem. Int. Ed. 2018, 57, 4886 –4890 T 2018 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim The authors declare no conflict of interest. The authors declare no conflict of interest. [19] J. Duboisset, P. Ferrand, W. He, X. Wang, H. Rigneault, S. Brasselet, J. Phys. Chem. B 2013, 117, 784 – 788. Keywords: amyloid fibrils · fluorescence anisotropy · neurodegeneration · Parkinson’s disease · protein aggregation [20] M. Biancalana, K. Makabe, A. Koide, S. Koide, J. Mol. Biol. 2009, 385, 1052 – 1063. How to cite: Angew. Chem. Int. Ed. 2018, 57, 4886–4890 Angew. Chem. 2018, 130, 4980–4984 How to cite: Angew. Chem. Int. Ed. 2018, 57, 4886–4890 How to cite: Angew. Chem. Int. Ed. 2018, 57, 4886–4890 [21] M. Biancalana, K. Makabe, S. Koide, Proc. Natl. Acad. Sci. USA 2010, 107, 3469 – 3474. Angew. Chem. 2018, 130, 4980–4984 [22] M. R. H. Krebs, E. H. C. Bromley, A. M. Donald, J. Struct. Biol. 2005, 149, 30 – 37. Chiti, C. M. Dobson, Annu. Rev. Biochem. 2017, 86, 27 – 68 [1] F. Chiti, C. M. Dobson, Annu. Rev. Biochem. 2017, 86, 27 – 68 [1] F. Chiti, C. M. Dobson, Annu. Rev. Biochem. 2017, 86, 27 – 68. F. Chiti, C. M. Dobson, Annu. Rev. Biochem. 2006, 75, 333 – 3 [23] M. N. Bongiovanni, J. Godet, M. H. Horrocks, L. Tosatto, A. R. Carr, D. C. Wirthensohn, R. T. Ranasinghe, J.-E. Lee, A. Ponjavic, J. V. Fritz, et al., Nat. Commun. 2016, 7, 13544. [3] M. G. Spillantini, M. L. Schmidt, V. M. Lee, J. Q. Trojanowski, R. Jakes, M. Goedert, Nature 1997, 388, 839 – 840. [4] W. Poewe, K. Seppi, C. M. Tanner, G. M. Halliday, P. Brundin, J. Volkmann, A.-E. Schrag, A. E. Lang, Nat. Rev. Dis. Prim. 2017, 3, 17013. [24] M. Iljina, G. A. Garcia, M. H. Horrocks, L. Tosatto, M. L. Choi, K. A. Ganzinger, A. Y. Abramov, S. Gandhi, N. W. Wood, N. Cremades, et al., Proc. Natl. Acad. Sci. 2016, 113, E1206 – E1215. c [5] J. A. Rodriguez, M. I. Ivanova, M. R. Sawaya, D. Cascio, F. E. Reyes, D. Shi, S. Sangwan, E. L. Guenther, L. M. Johnson, M. Zhang, et al., Nature 2015, 525, 486 – 490. [25] T. Klingstedt, H. Shirani, K. O. A. cslund, N. J. Cairns, C. J. Sigurdson, M. Goedert, K. P. R. Nilsson, Chem. Eur. J. 2013, 19, 10179 – 10192. g [6] M. Vilar, H.-T. Chou, T. Lghrs, S. K. Maji, D. Riek-Loher, R. Verel, G. Manning, H. Stahlberg, R. Riek, Proc. Natl. Acknowledgements T. P. J. Knowles, Proc. Natl. Acad. Sci. USA 2013, 110, 9758 – 9763. [11] P. Flagmeier, G. Meisl, M. Vendruscolo, T. P. J. Knowles, C. M. Dobson, A. K. Buell, C. Galvagnion, Proc. Natl. Acad. Sci. USA 2016, 113, 10328 – 10333. ] P. Flagmeier, G. Meisl, M. Vendruscolo, T. P. J. Knowles, C This study is supported by the Michael J. Fox Foundation (10200); The Royal Society with a University Research Fellowship (UF120277) (S.F.L.); a Marie-Curie Individual Fellowship (S.D.); the Boehringer Ingelheim Fonds (P.F.), the Studienstiftung des Deutschen Volkes (P.F.); the UK Bio- technology and Biological Sciences Research Council (C.M.D.); the Wellcome Trust (C.M.D.); the Cambridge Centre for Misfolding Diseases (P.F. and C.M.D.), and the Royal Society and the European Research Council with an ERC Advanced Grant (669237) (D.K.). We thank Ewa Klimont and Swapan Preet for a-synuclein expression. We thank as well Dr. Alexandre Fgrstenberg, Dr. Mathew Horrocks and Dr. Daniel Whiten for critical and insightful comments on this manuscript. [12] N. Cremades, S. I. A. Cohen, E. Deas, A. Y. Abramov, A. Y. Chen, A. Orte, M. Sandal, R. W. Clarke, P. Dunne, F. A. Aprile, et al., Cell 2012, 149, 1048 – 1059. [13] A. Orte, N. R. Birkett, R. W. Clarke, G. L. Devlin, C. M. Dobson, D. Klenerman, Proc. Natl. Acad. Sci. USA 2008, 105, 14424 – 14429. [14] S. W. Chen, S. Drakulic, E. Deas, M. Ouberai, F. A. Aprile, R. Arranz, S. Ness, C. Roodveldt, T. Guilliams, E. J. De-Genst, [14] S. W. Chen, S. Drakulic, E. Deas, M. Ouberai, F. A. Aprile, R. Arranz, S. Ness, C. Roodveldt, T. Guilliams, E. J. De-Genst, et al., Proc. Natl. Acad. Sci. USA 2015, 112, E1994 – 2003. , , , , , et al., Proc. Natl. Acad. Sci. USA 2015, 112, E1994 – 2003. [15] G. Fusco, S. W. Chen, P. T. F. Williamson, R. Cascella, M. Perni, J. A. Jarvis, C. Cecchi, M. Vendruscolo, F. Chiti, N. Cremades, et al., Science 2017, 358, 1440 – 1443. [16] T. Ban, D. Hamada, K. Hasegawall, H. Naiki, Y. Goto, J. Biol. Chem. 2003, 278, 16462 – 16465. [17] M. M. Wçrdehoff, O. Bannach, H. Shaykhalishahi, A. Kulawik, S. Schiefer, D. Willbold, W. Hoyer, E. Birkmann, J. Mol. Biol. 2015, 427, 1428 – 1435. Communications Communications Conflict of interest [18] M. H. Horrocks, S. F. Lee, S. Gandhi, N. K. Magdalinou, S. W. Chen, M. J. Devine, L. Tosatto, M. Kjaergaard, J. S. Beckwith, H. Zetterberg, et al., ACS Chem. Neurosci. 2016, 7, 399 – 406. Manuscript received: October 19, 2017 Revised manuscript received: December 21, 2017 Accepted manuscript online: January 17, 2018 Version of record online: March 22, 2018 www.angewandte.org T 2018 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Protein Aggregates In summary, we have demonstrated that by combining sensitive TIRF microscopy with anisotropy measurements, one can directly characterise the structural features of individual oligomers. This method is highly flexible as it does not require protein labelling, but rather a dye that recognises cross-b motifs. We have shown the conversion from disordered aggregates of a-synuclein to fibrillar aggre- gates, in agreement with previously reported smFRET measurements. Furthermore, our experiments suggest that modulating aggregates have a higher capacity to disrupt lipid membranes. Our results provide clear evidence that most ThT active species in CSF are disordered, but do, however, contain cross-b sheet structure. Our ability to analyse single aggre- gates individually allowed us to detect an ultra-low abun- dance of fibril-like species in human CSF. This methodology is compatible with other proteins whose aggregation has been associated with human disorders such as amyloid-b, tau, lysozyme or insulin. Overall this approach provides a new method to characterise the degree of fibrillation in individual protein aggregates, contributing to the set of biophysical methods needed to understand some of the most fundamental mechanisms in neurodegeneration. To demonstrate the broad applicability of our technique, we also applied the method to samples of human cerebrospi- nal fluid (CSF), comparing aged-matched healthy controls (HC) to ParkinsonQs disease (PD) patients. By analysing the modulation of individual species in each group (obtained from 4 HC and 4 PD samples), we found that the large majority of species show non-modulating behaviour, (Fig- ure 5a,b) meaning that they are disordered. Only a small fraction ( & 1%) of species showed modulating behaviour in both HC and PD groups (Figure 5c). Species with higher modulation values (modulation over 0.45) in PD patients have a mean of 0.57 compared to HC with a mean of 0.48 (inset Figure 5c). The abundance of these modulating species in CSF is very low and therefore prevents us from a robust statistical analysis. Further studies need to be done to characterise these ordered species, as they are candidates to be involved in toxicity and spreading mechanisms. However, this does demonstrate the technique is capable of making structural measurements in human CSF. 4889 www.angewandte.org 886 –4890 T 2018 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angewandte Chem T 2018 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim An The authors declare no conflict of interest. Acad. Sci. USA 2008, 105, 8637 – 8642. [26] P. Flagmeier, S. De, D. C. Wirthensohn, S. F. Lee, C. Vincke, S. Muyldermans, T. P. J. Knowles, S. Gandhi, C. M. Dobson, D. Klenerman, Angew. Chem. Int. Ed. Angew. Chem. Int. Ed. Engl. 2017, 56, 7750 – 7754; Angew. Chem. 2017, 129, 7858 – 7862. [7] M. Goedert, Science 2015, 349, 1255555. [8] D. Pinotsi, C. H. Michel, A. K. Buell, R. F. Laine, P. Mahou, C. M. Dobson, C. F. Kaminski, G. S. Kaminski Schierle, Proc. Natl. Acad. Sci. USA 2016, 113, 3815 – 3819. [9] W. Peelaerts, L. Bousset, A. Van der Perren, A. Moskalyuk, R. Pulizzi, M. Giugliano, C. Van den Haute, R. Melki, V. Baeke- landt, Nature 2015, 522, 340 – 344. [10] S. I. A. Cohen, S. Linse, L. M. Luheshi, E. Hellstrand, D. A. White, L. Rajah, D. E. Otzen, M. Vendruscolo, C. M. Dobson, 4890 www.angewandte.org
https://openalex.org/W4255872606	http://www.scielo.br/pdf/spmj/v137n5/1806-9460-spmj-137-05-475.pdf	English	null	ERRATUM	São Paulo medical journal	2,019	cc-by	120	DOI: 10.1590/1516-3180.2019.ast.0458051119erratum ERRATUM At the request of the authors, we report that in the paper published in the Sao Paulo Medical Journal, volume 137, issue number 4, DOI: 10.1590/1516-3180.2018.0458220719, pages 349-55 (in the title): Where it read: “Procalcitonin levels among patients with fever secondary to severe intracerebral infection. A cross-sectional study” © 2019 by Associação Paulista de Medicina This is an open access article distributed under the terms of the Creative Commons license. It should read: “Procalcitonin levels among patients with fever secondary to intracerebral hemorrhage and severe infection. A cross-sectional study” “Procalcitonin levels among patients with fever secondary to intracerebral hemorrhage and severe infection. A Sao Paulo Med J. 2019; 137(5):475 475 Sao Paulo Med J. 2019; 137(5):475 475
https://openalex.org/W2778336362	https://europepmc.org/articles/pmc5742104?pdf=render	English	null	Tracing the Origins of IgE, Mast Cells, and Allergies by Studies of Wild Animals	Frontiers in immunology	2,017	cc-by	18,730	Keywords: IgE, Fc receptor, mast cell, IgE homeostasis, allergy, dermatitis, asthma Review published: 19 December 2017 doi: 10.3389/fimmu.2017.01749 Tracing the Origins of IgE, Mast Cells, and Allergies by Studies of Wild Animals Lars Torkel Hellman, Srinivas Akula, Michael Thorpe and Zhirong Fu Department of Cell and Molecular Biology, Uppsala University, Uppsala, Sweden In most industrialized countries, allergies have increased in frequency quite dramatically during the past 50 years. Estimates show that 20–30% of the populations are affected. Allergies have thereby become one of the major medical challenges of the twenty-first century. Despite several theories including the hygiene hypothesis, there are still very few solid clues concerning the causes of this increase. To trace the origins of allergies, we have studied cells and molecules of importance for the development of IgE-mediated allergies, including the repertoire of immunoglobulin genes. These studies have shown that IgE and IgG most likely appeared by a gene duplication of IgY in an early mammal, possibly 220–300 million years ago. Receptors specific for IgE and IgG subsequently appeared in parallel with the increase in Ig isotypes from a subfamily of the recently iden- tified Fc receptor-like molecules. Circulating IgE levels are generally very low in humans and laboratory rodents. However, when dogs and Scandinavian wolfs were analyzed, IgE levels were found to be 100–200 times higher compared to humans, indicating a generally much more active IgE synthesis in free-living animals, most likely connected to intestinal parasite infections. One of the major effector molecules released upon IgE- mediated activation by mast cells are serine proteases. These proteases, which belong to the large family of hematopoietic serine proteases, are extremely abundant and can account for up to 35% of the total cellular protein. Recent studies show that several of these enzymes, including the chymases and tryptases, are old. Ancestors for these enzymes were most likely present in an early mammal more than 200 million years ago before the separation of the three extant mammalian lineages; monotremes, marsupials, and placental mammals. The aim is now to continue these studies of mast cell biology and IgE to obtain additional clues to their evolutionary conserved functions. A focus concerns why the humoral immune response involving IgE and mast cells have become so dysregulated in humans as well as several of our domestic companion animals. Edited by: Greg Woods, Edited by: Greg Woods, University of Tasmania, Australia Reviewed by: Axel Lorentz, University of Hohenheim, Germany Pierre Boudinot, Institut National de la Recherche Agronomique, France Correspondence: Lars Torkel Hellman lars.hellman@icm.uu.se Specialty section: This article was submitted to Comparative Immunology, a section of the journal Frontiers in Immunology Specialty section: This article was submitted to Comparative Immunology, a section of the journal Frontiers in Immunology Received: 03 October 2017 Accepted: 24 November 2017 Published: 19 December 2017 INTRODUCTION Hellman LT, Akula S, Thorpe M and Fu Z (2017) Tracing the Origins of IgE, Mast Cells, and Allergies by Studies of Wild Animals. Front. Immunol. 8:1749. doi: 10.3389/fimmu.2017.01749 During the past 50 years, allergies has increased in prevalence quite dramatically and in most indus- trialized countries 20–30% of the population are affected. In some school classes, the percentage of affected children can reach as high as 50. Allergies have thereby become one of the major medical challenges of the twenty-first century. Although relatively few die from an anaphylactic shock, the December 2017 \| Volume 8 \| Article 1749 Frontiers in Immunology \| www.frontiersin.org 1 Origins of IgE, Mast Cells, and Allergies Hellman et al. severest form of allergic reaction, there are extensive burdens for sufferers, which in turn can also cause major economic loss. Asthma in children has in the USA been estimated to cost 56 billion $ per year and asthma in adults to cost 19 billion € per year in the EU so allergies involves large costs for the society (1, 2). Allergies not only affect humans but also our close domestic companions, including dogs, cats, and horses. In dogs, 3–15%, depending on the breed, suffer from atopic dermatitis, a type of allergic skin disease (3–5). By contrast, cats and horses suffer primarily from asthma. This indicates that domestication may be one contributing factor in this process. Rodents are the most frequently used animal models in allergy research and the numerous inbred strains of mice and rats may be seen as a form of domestication. However, neither rats nor mice can be considered naturally allergic. To obtain allergic mice or rats, these animals needs to be triggered by relatively high allergen doses and often complimented with additional immune stimulators to show allergy-like symptoms (6–8). These symptoms also disappear as soon as the sensitization protocol is terminated. Therefore, a life under controlled conditions with low pathogen load does not necessarily result in the induction of hypersensitivity. However, western lifestyle, with high hygiene levels, has been indicated as a potential contributing factor to the increase in allergy prevalence and been termed the “hygiene hypothesis” (9–11). The contact with high levels of bacteria, viruses, and parasites has likely been the normal condition for our immune system during the past millions of years of evolution (12). INTRODUCTION These B cells are then triggered by local IL-4 and/or IL-13 to switch to IgE and IgG1 producing cells in mice, and IgE and IgG4 in humans (Figure 1) (16–18). The locally produced IgE can then bind to high-affinity receptors on cells in the tissue. The cells that express the high-affinity receptor for IgE are primarily mast cells and basophilic granulocytes (Figure 1). Both of these cells store histamine and can also produce potent lipid mediators including prostaglandins and leukotrienes, which when released from the cells give the characteristic symptoms of allergies including tissue swelling, broncho constriction, and drop in blood pressure. The latter effects occur if mast cell activation involves larger regions of the body. This entire process, from the first antigen contact to the development of IgE and mast cell bound IgE, is termed sensitization and this can take weeks to months or even years to develop (13). In most cases, the immune system regulates itself so that more IgG than IgE is produced meaning no hypersensitivity appears and the individual does not become allergic (Figure 1). However, if this is dysregulated and more IgE is produced, the tis- sue becomes overly sensitive to contact with the allergen. When a sensitized person comes in contact with the same subsequent allergen, the mast cells release a number of vasoactive substances, which results in tissue swelling due to the influx of liquid into the tissue from the blood. The activation of mast cells also results in the recruitment of eosinophils to the tissue, a process at least partly dependent on locally produced IL-5 (Figure 1). INTRODUCTION This has been an important component for the immune system to become educated and if we remove the normal triggers in our daily life, we may be more likely to develop allergies. There is evidence supporting this hypothesis but there is also data that do not fit into this model. One previously mentioned example is rodents living under controlled conditions in animal houses. There are certainly other contributing factors of major importance. Therefore in order to look deeper into these questions, we have turned our interest to the components of our immune system that are involved in an allergic immune response. This approach can be used to see if by studying of an array of different wild, domestic, and non-domestic animals, it is possible to trace the origins of these components and the factors that have resulted in the massive increase in allergies in humans and other domestic animals. sites where allergens generally enter the body, for example, the skin, the lung, or the intestinal mucosa. These cells then process some of the protein into peptides of variable length in endosomal compartments using lysosomal proteases. The peptides, gener- ally between 12 and 18 amino acids, are then presented onto major histocompatibility complex (MHC) class II molecules on the cell surface of the dendritic cell (Figure 1). A T helper cell may then recognize this peptide MHC complex with its specific T cell receptor (TCR) and when supported by several additional receptor ligand interactions, including CD28-B7:1 or B7:2, CD40L–CD40, CD48R–CD22, CD2–LFA3, the T cell becomes activated and starts to proliferate. Furthermore, the activated cell produces cytokines and also to upregulate receptors and receptor ligands that can support and trigger other immune cells (Figure 1). If there is no intracellular parasite or other potent danger signal in the area of allergen entry, the T cells are generally becoming T-helper cells of type 2, so called TH2 cells. TH2 cells promote humoral immunity, consisting primarily of soluble fac- tors, involving B cells and Igs and not in cell-mediated immunity where cytotoxic T cells (CTLs) and NK cells are involved. The TH2 cells produce cytokines, including IL-4, IL-13, IL-5, IL-10, and sometimes also TNF-α (Figure 1) (14, 15). The B cells in the area of allergen contact bind with their surface IgM to the native protein antigen and thereby receive signal 1 to become activated. THE INDUCTION OF AN ALLERGIC IMMUNE RESPONSE The absolute majority of allergies in humans belong to the immunoglobulin (Ig) E-mediated allergies, which also are named atopic allergies. The major focus of this review will be on this type of allergy, where dendritic cells, IgE, and IgE-binding cells, primarily mast cells, and basophils are central players (13).hi Frontiers in Immunology \| www.frontiersin.org MAST CELLS AND BASOPHILS The first time we are exposed to an allergen there will be no visible response. However, the immune system may start to recognize the target molecule via uptake of the protein antigen, the allergen, by local dendritic cells. Generally, allergens are proteins of a relatively low molecular weight and are often also relatively stable proteins, indicating they can more easily enter the mucosa in a native conformation. Dendritic cells are abundant at Mast cells and basophils are the only two cells that express the complete high-affinity receptor for IgE (FcεRI) with all four polypeptide subunits (19) (Figures 2A–C). The FcεRI receptor belongs to a larger family of receptors binding to the constant domains of Igs, including IgG, IgA, and IgM receptors. The December 2017 \| Volume 8 \| Article 1749 Frontiers in Immunology \| www.frontiersin.org 2 Hellman et al. Origins of IgE, Mast Cells, and Allergies Figure 1 \| An allergic immune response. The figure presents a schematic overview of an allergic immune response starting with the allergens first contact when it enters through the skin, lungs, or intestinal mucosa. Antigen-presenting cells, primarily dendritic cells, take up the antigen and process it into peptides, which are subsequently presented onto major histocompatibility complex class II molecules to naïve T cells. In an environment lacking major danger signals and in the presence of certain TH2-promoting cytokines, the T cell becomes a TH2 type of T cell. These T cells produce IL-4 and IL-13, which stimulates B cells to switch to IgE production. The IgE produced by these local B cells binds tissue mast cells, which have now become sensitized and can respond by degranulation as well as prostaglandin and leukotriene synthesis, which provides all the symptoms of an allergic reaction. As a next step, locally produced IL-5 results in eosinophil influx and the induction of a late phase response. Figure 1 \| An allergic immune response. The figure presents a schematic overview of an allergic immune response starting with the allergens first contact when it enters through the skin, lungs, or intestinal mucosa. Antigen-presenting cells, primarily dendritic cells, take up the antigen and process it into peptides, which are subsequently presented onto major histocompatibility complex class II molecules to naïve T cells. In an environment lacking major danger signals and in the presence of certain TH2-promoting cytokines, the T cell becomes a TH2 type of T cell. MAST CELLS AND BASOPHILS These T cells produce IL-4 and IL-13, which stimulates B cells to switch to IgE production. The IgE produced by these local B cells binds tissue mast cells, which have now become sensitized and can respond by degranulation as well as prostaglandin and leukotriene synthesis, which provides all the symptoms of an allergic reaction. As a next step, locally produced IL-5 results in eosinophil influx and the induction of a late phase response. complete IgE receptor consists of an α chain, which is primarily located outside of the cell where it binds IgE; a β chain, which is a membrane protein with four transmembrane regions; and two identical γ chains. The γ chain has the majority of the protein on the cytoplasmic side of the membrane where it acts as the primary signal transducing subunit (Figure 2C). Other cells, such as dendritic cells and monocytes can in humans also express low levels of the IgE receptor; however in these cases, it is usually a three polypeptide variant, which lacks the β chain. This expression on dendritic cells and monocytes seems to be involved in IgE internalization and degradation, possibly also in antigen presentation (19). IgE binds to the FcεRI with high affinity, in the range of 1010 (20). Monomeric non-cross-linked IgE does not activate the mast cell or basophil. However, when two IgE- molecules bind the same allergen molecule they become cross- linked and if both molecules sit on the receptors, these receptors also become cross-linked. This receptor cross-linking changes the environment on the inside of the cell, favoring kinase activation at the expense of access of phosphatases to the cytoplasmic parts of the receptor subunits. This results in the phosphorylation of a number of cytoplasmic motifs on the β and the γ chains of the receptor (21). These activation motifs are called immune receptor tyrosine-based activation motifs, commonly referred to as ITAMs (Figure 2C). The phosphorylation of these short motifs result in the binding of signaling molecules including Lyn and Syk, which in turn result in a cascade of reactions including Ca2+ release from complete IgE receptor consists of an α chain, which is primarily located outside of the cell where it binds IgE; a β chain, which is a membrane protein with four transmembrane regions; and two identical γ chains. Frontiers in Immunology \| www.frontiersin.org MAST CELLS AND BASOPHILS The γ chain has the majority of the protein on the cytoplasmic side of the membrane where it acts as the primary signal transducing subunit (Figure 2C). Other cells, such as dendritic cells and monocytes can in humans also express low levels of the IgE receptor; however in these cases, it is usually a three polypeptide variant, which lacks the β chain. This expression on dendritic cells and monocytes seems to be involved in IgE internalization and degradation, possibly also in antigen presentation (19). IgE binds to the FcεRI with high affinity, in the range of 1010 (20). Monomeric non-cross-linked IgE does not activate the mast cell or basophil. However, when two IgE- molecules bind the same allergen molecule they become cross- linked and if both molecules sit on the receptors, these receptors also become cross-linked. This receptor cross-linking changes the environment on the inside of the cell, favoring kinase activation at the expense of access of phosphatases to the cytoplasmic parts of the receptor subunits. This results in the phosphorylation of a number of cytoplasmic motifs on the β and the γ chains of the receptor (21). These activation motifs are called immune receptor tyrosine-based activation motifs, commonly referred to as ITAMs (Figure 2C). The phosphorylation of these short motifs result in the binding of signaling molecules including Lyn and Syk, which in turn result in a cascade of reactions including Ca2+ release from intracellular stores, activation of phospholipase A2 and thereby the release of arachidonic acid from membrane phospholipids and the generation of prostaglandins and leukotrienes by two differ- ent enzyme pathways. The receptor phosphorylation also results in the production of PIP3 and a few other intracellular signaling molecules as well as in the activation of the cells to release their pre-stored granule material, including histamine, heparin, and the very abundant granule proteases (Figures 2A,B). Mast cells or mast cell-like cells have been described in most vertebrate lineages including mammals, birds, reptiles, amphib- ians, and bony fishes (29–32). Mast cell-like cells have also been described in an early ancestor of the vertebrates, the tunicate, or sea squirt (33, 34). Interestingly, these mast cell-like cells contain both histamine and heparin as well as some kind of trypsin-like enzyme indicating a relatively close resemblance to mammalian mast cells (33, 34). IgE Interestingly, the cartilaginous fish have a different organization of the Ig locus compared to all other vertebrates in that there are multiple genes for each Ig class and there are large numbers of such small clusters compared to single genes, which are present in most other vertebrates. The genes are organized as small rearranging units consisting of a single V, one or two D segments, followed by a single J and subsequent constant region (V-D-D-J-C). This is in marked contrast to the translocon model used by almost all other vertebrates with multiple V regions followed by a number of D segments, often over 20, and then 3-7J segments followed by constant region gene segments (35). When looking at the bony fishes, which represent the largest single group of vertebrates, there are varying numbers of Ig classes and isotypes. As one example the gar, which represent an early branch of the bony fishes, seems to primarily depend on one Ig class IgM whereas the zebra fish and the rainbow trout expresses three Ig classes, each with only one isotype, IgM, IgD, and IgT/Z (T for teleost) (Figure 3) (36, 37). Amphibians as exemplified by the clawed frog Xenopus laevis or tropicalis have five Ig classes, also here with only one isotype each, i.e., IgM, IgD, IgY, IgX, and IgF (38). IgX appears to be the functional equivalent of IgA in birds and mammals, and IgY is the ancestor of mammalian IgG and IgE (Figure 3) (38). Reptiles have similar to the bony fishes very varying numbers of Ig classes and isotypes. For example, the anolis lizard (Anolis carolinensis) having only three Ig classes also here with one isotype for each class, IgM, IgD, and IgY whereas the American and Chinese alligators have 4 Ig classes and 10 isotypes (Figure 3) (39–41). The alligators have experienced a series of gene duplications resulting in several functional as well as a few pseudogenes for IgM, IgA, and IgY (Figure 3). Birds have relatively few Ig classes and isotypes, probably due to later gene losses. For example, chickens have only three Ig genes, one each for IgM, IgA, and IgY (42).i In general, mammals have five Ig classes IgM, IgD, IgG, IgE, and IgA, and a varying number of isotypes. Quite big differences have been observed between different mammalian species. IgE Immunoglobulins and TCRs first appeared with jawed ver- tebrates. There has been a parallel increase in the different vertebrate lineages in the complexity of Ig classes and isotypes during vertebrate evolution through gene duplications, which have sometimes been followed by gene losses. In general, car- tilaginous fish have three Ig classes IgM and IgW, which are of the classical type with both heavy and light chains and a third IgNAR, which lacks the light chain. The entire antigen-binding December 2017 \| Volume 8 \| Article 1749 Frontiers in Immunology \| www.frontiersin.org 3 Hellman et al. Origins of IgE, Mast Cells, and Allergies Figure 2 \| Schematic representation of mast cells, basophils and the high-affinity IgE receptor. Below the schematic pictures of the cells is in (A,B) a listing of the most important granule components of the two cell types. The human basophil expresses the mast cell tryptase whereas the mouse basophil expresses mMCP-8, the first basophil-specific protein to be identified (22–24). (C) The high-affinity receptor for IgE consists of four subunits; the α chain, which binds the CH3 domain of IgE with its membrane proximal α2 domain, the β chain that spans the membrane four times and has one immunoreceptor tyrosine-based activation motif (ITAM) in its cytoplasmic region (marked by a gray box), and two identical γ chains, which are mostly intracellular, serving as the major signal transducing subunit (25). The γ chains contain two ITAMS, one in each subunit (26–28). Figure 2 \| Schematic representation of mast cells, basophils and the high-affinity IgE receptor. Below the schematic pictures of the cells is in (A,B) a listing of the most important granule components of the two cell types. The human basophil expresses the mast cell tryptase whereas the mouse basophil expresses mMCP-8, the first basophil-specific protein to be identified (22–24). (C) The high-affinity receptor for IgE consists of four subunits; the α chain, which binds the CH3 domain of IgE with its membrane proximal α2 domain, the β chain that spans the membrane four times and has one immunoreceptor tyrosine-based activation motif (ITAM) in its cytoplasmic region (marked by a gray box), and two identical γ chains, which are mostly intracellular, serving as the major signal transducing subunit (25). The γ chains contain two ITAMS, one in each subunit (26–28). site of the IgNAR thereby resides in the variable region of the heavy chain (35). IgE For example, humans have nine isotypes, due to four different IgG isotypes and two IgA isotypes. Mice have 8 due to that they have only 1 IgA isotype, and rabbits have as many as 17 because of having 13 different copies of the IgA gene (Figure 3).i f Interestingly, much of the functional diversification of the different Ig classes already came with the first tetrapods as the ancestors of IgA (IgX), IgG, and IgE (IgY) appeared with frogs and other amphibians. The most recent addition to the com- plexity came with early mammals from a gene duplication of IgY. One of the copies lost the second, CH2 encoding domain, and became a four domain Ig class with V, CH1, CH2, and CH3 domains. This gene was named IgG whereas the second duplicate, which became IgE, maintained all domains of IgY and thereby containing V, CH1, CH2, CH3, and CH4 domains (Figure 4) (43–46). Similarly to IgG, IgA also lost one domain in mammals, the CH2 domain, and instead gained a flexible hinge region (Figure 4). IgG, IgE, and IgA are found in all three extant mammalian lineages in very similar forms (43–47). The gene duplication of IgY made it possible to separate the function of the mast cell acti- vation via IgE, and the major plasma antibody, IgG, with func- tions in complement activation, immune complex clearance, and December 2017 \| Volume 8 \| Article 1749 Frontiers in Immunology \| www.frontiersin.org 4 Origins of IgE, Mast Cells, and Allergies Hellman et al. Figure 3 \| The immunoglobulin heavy chain locus from a panel of different jawed vertebrates. The different genes are color coded with IgG in light blue, IgE in purple, IgM in black, IgA in light green, IgD in dark green, IgO in greenish yellow, IgZ in yellow, IgX in red, IgY in purple, and IgF in orange. The arrows show local duplications, and the direction of the arrow represents the transcription orientation of the genes within the duplicate. Pseudo genes are marked with black boxes with a white inner oval and a pseudo sign. The figure is not to scale. Figure 3 \| The immunoglobulin heavy chain locus from a panel of different jawed vertebrates. Frontiers in Immunology \| www.frontiersin.org IgE Panel (A) shows the different steps in the evolution of IgE and gan with the duplication of IgY after which IgG lost one domain, the original CH2 domain, and became an Ig with three constant domains with one hinge 6). Panel (B) shows the evolution of IgA. A four constant domain IgA is present in birds and reptiles. Sometime during early mammalian evolution IgA also domain, the CH2 domain, and in its place gained a hinge region (47). This hinge region is very different in size between human IgA1 and IgA2 making the e 4 \| The evolution of IgE IgG and IgA and the distribution of N-linked carbohydrates in IgE Panel (A) shows the different steps in the evolution of IgE and Figure 4 \| The evolution of IgE, IgG, and IgA and the distribution of N-linked carbohydrates in IgE. Panel (A) shows the different steps in the evolution of IgE and IgG. It began with the duplication of IgY after which IgG lost one domain, the original CH2 domain, and became an Ig with three constant domains with one hinge region (46). Panel (B) shows the evolution of IgA. A four constant domain IgA is present in birds and reptiles. Sometime during early mammalian evolution IgA also lost one domain, the CH2 domain, and in its place gained a hinge region (47). This hinge region is very different in size between human IgA1 and IgA2 making the former more resistant to proteases and the later more flexible, facilitating binding of several antigens simultaneously. Panel (C) shows the positions of the different N-linked carbohydrates in a large panel of different IgE molecules from different species. Only one carbohydrate is conserved in a position shared by all of the different IgEs, marked in red, which is positioned in the middle of the CH3 domain where it has an important function in the folding of this domain. regions (54). The selective inactivation of the duplicate gene for IgE but not IgG or IgA strengthens the indication that there is an advantage of having only one functional gene for IgE (Figure 3). It should also be noted that within eutherians, IgE has been found in all of the major branches including the orders of Primates, Rodentia, Cetartiodactyla, Lagomorpha, Carnivora, Chiroptera, Afrotherians, and Xenarthra. IgE The different genes are color coded with IgG in light blue, IgE in purple, IgM in black, IgA in light green, IgD in dark green, IgO in greenish yellow, IgZ in yellow, IgX in red, IgY in purple, and IgF in orange. The arrows show local duplications, and the direction of the arrow represents the transcription orientation of the genes within the duplicate. Pseudo genes are marked with black boxes with a white inner oval and a pseudo sign. The figure is not to scale. a non-functional remnant of the duplication process (51). The platypus has eight isotypes, IgM, IgD, IgO, IgG1, IgG2, IgE, IgA1, and IgA2. One interesting observation in mammals is that despite the very varying number of isotypes across the species there always seems to be only one functional gene for IgE (46). Currently, the only exception to this rule is within the horse, which may have two genes. This indicates that IgE needs to be kept under very stringent control, most likely due to its potent mast cell-activating properties. Interestingly, there are three cop- ies of the IgE gene in the human genome; two of them are a result of what appears to be a large duplication involving genes for IgG, IgE, and IgA (Figure 3) (52, 53). This duplication resulted in a antibody-dependent cellular cytotoxicity (ADCC). Interestingly the marsupials, exemplified by the American opossum only have four Ig classes, IgM, IgG, IgE, and IgA, and only one isotype for each Ig class (44, 48). This is likely due to a secondary loss of IgD. The monotremes, which are an early branch on the mammalian tree with only three extant surviving members of egg-laying mammals, the platypus and two variants of the ant- eating echidnas, the short and the long nosed, have unlike the marsupials all five mammalian Ig classes as well as one additional Ig class IgO (45, 47, 49–51). The gene for IgO is most likely a remaining extra duplicate of IgY, which seems to be expressed only in the spleen at very low levels, indicating that it is primarily December 2017 \| Volume 8 \| Article 1749 Frontiers in Immunology \| www.frontiersin.org 5 Origins of IgE, Mast Cells, and Allergies Hellman et al. 4 \| The evolution of IgE, IgG, and IgA and the distribution of N-linked carbohydrates in IgE. IgE similar fold to the Ig constant or variable domains and belong to the large family of Ig-like domains. The Ig domains have been classified into four basic types depending on the number of antiparallel beta-sheets and the positions of cysteine and other conserved amino acids. These are described as the V, C1, C2, and I type of domains (64, 65). V domains are generally found in variable regions of Igs and TCRs as well as in cluster of dif- ferentiation markers including CD2, CD4, CD80, and CD86. C1 domains are found in the constant regions of Igs, TCRs, and in MHC class I and II. C2 domains are found in CD2, CD4, CD80, VCAM, and ICAM, and I domains are found in VCAM, ICAM, NCAM, MADCAM, and numerous other diverse protein families (EMBL-EBI InterPro). All Ig domains of the FcRL and classical FcRs are classed as C2 domains, whereas the Ig domains of the PIGRs, IgM receptors (FcμRs), and IgA/IgM receptors (FcαμRs) are V type domains (64, 65). In a phylogenetic analysis of the individual domains of these receptors, the C2 and V type domains separate into clear individual branches. This is also true for the individual domains within the different receptors (Figure 5C) (60, 61). The important signaling molecule for the classical FcRs, the common γ chain, is a member of a small family of non Ig-domain-containing molecules including the TCR zeta chain, DAP10, and DAP12 (60, 66–70). The latter two proteins serve as signaling components of NK cell receptors and as well as the related Ig-domain containing receptors (70). Based on a phylogenetic tree (shown in Figure 6), we have found strong indications that the classical receptors for IgG and IgE likely appeared as a separate subfamily of the FcRL molecules during early mammalian evolution (60). Related genes are also found in the Western clawed frog (Xenopus tropicalis) and the Chinese alligator, indicating that the processes forming the subfamily of receptors that later became the classical IgG and IgE receptors may have started already during early tetrapod evolution. However, this subfamily probably did not appear as a distinct subfamily until the appearance of the mammals (60). In the Western clawed frog, these receptors have a similar struc- ture to the human high affinity IgG receptor, FcγRI, with three extracellular Ig domains of the C2 type. IgE doubling of the genes for IgG from two to four and a doubling of the genes for IgA into two. However, following the duplication of the genes for IgE one of them has suffered a large deletion involving part of the intron and the CH1 exon to render the gene non-functional (Figure 3) (52, 53). The third IgE gene is a non-functional intron-less copy on another chromosome, which appears as an mRNA copy inserted randomly without regulatory December 2017 \| Volume 8 \| Article 1749 Frontiers in Immunology \| www.frontiersin.org 6 Origins of IgE, Mast Cells, and Allergies Hellman et al. One noteworthy finding concerning IgE is in the pattern of glycosylation. Almost all of the Igs are glycosylated. Most often on asparagines, so called N-linked glycosylation. The glycosyla- tion pattern does often differ between different species and the amount of carbohydrate also correlates with the amount of positive charge (46). Therefore, if the polypeptide backbone of an Ig is highly positively charged, then there is generally a higher number of attached, N-linked carbohydrates. These car- bohydrates also tend to have several negatively charged sialic acid moieties. A neutral or acidic charge makes the antibodies “less sticky,” a characteristic that may be essential for soluble effector molecules aimed to travel easily in the blood and through tis- sues. Of all the different carbohydrate chains of different IgGs and IgEs from different species, there is only one that is always conserved, which is in the middle of the CH3 domain of IgE (46). A carbohydrate chain is also found in the same position in the corresponding domain of IgG, the CH2 domain (46). This carbohydrate chain seems to be essential for proper domain fold- ing and thereby for the interaction with the α chain of the IgE, and IgG receptors, which bind specifically to this domain (25, 26, 46). This indicates that with only one conserved carbohydrate position the function is to primarily to neutralize a high positive charge, where the position of the carbohydrate on the structure is of lower importance (46). Therefore, it is not the exact posi- tion rather the charge neutralization that is of importance. It also indicates that the interaction between IgE and its high-affinity receptor is a central characteristic feature of IgE, and this interac- tion has apparently been conserved for more than 200 years of mammalian evolution (46). IgE In the platypus there are both two- and three-domain receptors, which are similar to the human three-domain FcγRI as well as the low-affinity IgG receptors FcγRII and III, both of which have two domains. This indicates that the development of high and low affinity receptors also took place during early mammalian evolution (60). Despite this knowledge, currently none of these amphibian, reptile, or non-placental mammalian receptors have been studied for their isotype specificities and affinities. Fc Receptors (FcRs)h p ( ) The IgE receptor α chain is one member of a complex set of pro- teins interacting with the constant domains of the Igs (55–59). These molecules, named FcRs, due to their interaction with the constant domain of the Igs, have a number of important functions in vertebrates including facilitating phagocytosis by opsoniza- tion, constituting key components in ADCC as well as activating cells to release their granular content. In placental mammals there are FcRs for all Ig classes, including four major types of classical FcRs for IgG as well as one high-affinity receptor for IgE, one for both IgM and IgA, one for IgM, and one for IgA (Figures 5A,B) (60, 61). Additionally, there is the transport receptor for IgA and IgM across epithelial layers the polymeric Ig receptor or PIGR (Figures 5A,B) (61). All of these receptors are related in structure, and they all contain Ig-like domains. Furthermore they all, with the exception of the IgA receptor, are found on chromosome 1 in humans, indicating that they originate from one or a few common ancestors via successive local gene dupli- cations (Figure 5A) (60, 61). A new family of receptors, called FcR-like (FcRL), are related in structure to the classical IgG and IgE receptors, and were discovered upon the completion of full genome sequences from a number of mammalian species (62, 63). Eight different such FcRL genes have been identified in the human genome: FcRL1–FcRL6 as well as FcRLA and FcRLB (Figures 5A,B). Interestingly, no gene related to any of the FcRs, including the PIGRs is found in cartilaginous fish. They first seem to appear with the bony fish. In bony fish, there are the PIGRs and genes closely related to the mammalian FcRL genes. There is also a gene for the signaling molecule for the classical FcRs the common γ chain, which is lacking in cartilaginous fish. This indicates a major step in evolution of the FcRs at the base of bony fish with the appearance of the PIGR, the FcR γ chain, and the FcRL molecules (60, 61). Subsequently, the second major step was the appearance of the classical receptors for IgG and IgE as a subfamily of the All of the FcRs, including the PIGRs and the FcRL molecules, contain one or several Ig type domains. Frontiers in Immunology \| www.frontiersin.org Fc Receptors (FcRs)h These domains have a December 2017 \| Volume 8 \| Article 1749 Frontiers in Immunology \| www.frontiersin.org 7 Origins of IgE, Mast Cells, and Allergies Hellman et al. Figure 5 \| Fc receptor (FcR) genes and proteins. Panel (A) shows the regions within chromosome 1 in the human genome that encodes all the different FcR genes except for the specific receptor for IgA, the FcαRI, which is located together with the NK cell receptors on chromosome 19 in the humans (60, 61). Panel (B) shows the domain structure of the different FcRs with their different subunits. All of these receptors except FcRLA and FcRLB are transmembrane proteins. FcRLA and FcRLB are both cytoplasmic proteins. Panel (C) shows a phylogenetic tree of the individual domains of a panel of the FcRs, which are color coded to reflect the same subunits shown in panel (B). The tree is strongly reduced in size. However, even in its reduced format, it shows the striking separation of the different domains into clearly separate branches. Only the filled circles represent the individual Ig-like domains that are included in the phylogenetic tree. The extracellular regions, the transmembrane regions, and cytoplasmic tails are not to scale in order to show the positions of potential signaling motifs such as immunoreceptor tyrosine-based activation motifs (ITAMs) (green boxes) and immunoreceptor tyrosine-based inhibitory motifs (ITIMs) (red boxes), which regulate the biological function the FcRs. Some of the intracellular proteins contain C-terminal mucin-like regions, which are depicted as blue triangles. The PIGR domains are depicted in gray. Figure 5 \| Fc receptor (FcR) genes and proteins. Panel (A) shows the regions within chromosome 1 in the human genome that encodes all the different FcR genes except for the specific receptor for IgA, the FcαRI, which is located together with the NK cell receptors on chromosome 19 in the humans (60, 61). Panel (B) shows the domain structure of the different FcRs with their different subunits. All of these receptors except FcRLA and FcRLB are transmembrane proteins. FcRLA and FcRLB are both cytoplasmic proteins. Panel (C) shows a phylogenetic tree of the individual domains of a panel of the FcRs, which are color coded to reflect the same subunits shown in panel (B). The tree is strongly reduced in size. However, even in its reduced format, it shows the striking separation of the different domains into clearly separate branches. Fc Receptors (FcRs)h Only the filled circles represent the individual Ig-like domains that are included in the phylogenetic tree. The extracellular regions, the transmembrane regions, and cytoplasmic tails are not to scale in order to show the positions of potential signaling motifs such as immunoreceptor tyrosine-based activation motifs (ITAMs) (green boxes) and immunoreceptor tyrosine-based inhibitory motifs (ITIMs) (red boxes), which regulate the biological function the FcRs. Some of the intracellular proteins contain C-terminal mucin-like regions, which are depicted as blue triangles. The PIGR domains are depicted in gray. FcRL molecules, most likely in parallel with the duplication of IgY and the slowly emerging separation of the functions of IgG and IgE. It therefore appears to be a close evolution of the new Ig classes and their receptors. present in most, if not all, jawed vertebrates. Therefore, one central question is when mast cells and the adaptive immunity as represented by antibodies did connect through isotype-specific FcRs. This step in the evolution is a very important step where innate immunity, which is the core of the immune system in all multicellular organisms connected with adaptive immunity. Adaptive immunity has most likely appeared a number of times and in very different shapes, which has later become an essential part of immunity in complex multicellular organisms such as the vertebrates. Although studies of the mammalian Ig repertoire and the FcR repertoires have not conclusively shown it, the results suggest that it is likely that all three extant mammalian lineages, IgE and the Connection between Innate and Adaptive Immunity Although it is most likely that this receptor binds IgE, it has not yet been proven. The situation in the monotremes, represented by the platypus, is even less clear, although there are receptors in the platypus genome, which are closely related to the IgG and IgE receptors in placental mammals (Figure 6). The two receptors that are most closely related to the IgG and IgE receptors in placental mammals appear as a separate branch in-between the IgG and IgE receptors in the phylogenetic three (Figure 6) (60). The isotype specificities of these two receptors and the single receptor, which is located just outside of the IgG and IgE receptors in the tree have not yet been studied but it is reasonable to think that one of them is IgE specific and the other is IgG specific. There are no direct homologs to the IgG and IgE receptors in birds and reptiles. However in the clawed frog, Xenopus laevis, there are three genes, which may be a very early ancestor to the IgG and IgE receptors (Figure 6) (60). However in a similar manner to the platypus, there is not yet any information on their isotype specificity, therefore we do not know if they bind IgY, IgX, IgM, or IgF or perhaps none of them. One of the major questions in the field is the function of these very abundant proteases. To trace the origin of these proteases and thereby get additional clues to their conserved primary functions, we have performed several evolutionary studies of the different loci encoding the different mast cell and basophil proteases (85, 90). These proteases belong to a larger subfamily of related proteases that are expressed by a number of hematopoietic cells and have therefore been named hematopoi- etic serine proteases. They are expressed by mammalian mast cells, basophils, neutrophils, CTLs, and NK cells. In mammals, these serine proteases are encoded from four loci (85, 90–92). One additional locus with related proteases is found in reptiles, indicating a loss of one locus in mammals (90). In mammals, the four loci include the chymase locus, which in humans encodes the mast cell chymotryptic enzyme, the chymase; the neutrophil cathepsin G; and a few T cell- and NK cell-expressed granzymes (Figure 7). In humans, only two granzymes are present in this locus, granzyme B and H (Figure 7). IgE and the Connection between Innate and Adaptive Immunity IgE and IgG are only found in mammals and we know that in all eutherian mammals studied mast cells interact with IgE and often also IgG with isotype-specific receptors (43–45). However, Igs are found in all jawed vertebrates and mast cells are also December 2017 \| Volume 8 \| Article 1749 Frontiers in Immunology \| www.frontiersin.org 8 Hellman et al. Origins of IgE, Mast Cells, and Allergies Figure 6 \| A phylogenetic tree of different Fc receptor (FcR) and FcR-like (FcRL) protein sequences. A few sequences of particular interest for the early evolution specific IgG and IgE receptors, the discussed opossum, and platypus are marked with red and orange ovals. The Fc γ and ε sequences form a separate subfami within the FcRL sequences, indicating that they originate as a subfamily from the FcRL genes. Figure 6 \| A phylogenetic tree of different Fc receptor (FcR) and FcR-like (FcRL) protein sequences. A few sequences of particular interest for the early evolution of specific IgG and IgE receptors, the discussed opossum, and platypus are marked with red and orange ovals. The Fc γ and ε sequences form a separate subfamily within the FcRL sequences, indicating that they originate as a subfamily from the FcRL genes. December 2017 \| Volume 8 \| Article 1749 Frontiers in Immunology \| www.frontiersin.org 9 Origins of IgE, Mast Cells, and Allergies Hellman et al. (86). Most of these proteases are positively charged and therefore bind to the long, negatively charged proteoglycan polysaccharide chains. Tryptases are one subfamily of serine proteases, which are also dependent on heparin for maximal activity (87). The mast cell tryptases form tetramers where the dependency on heparin keeps these formations together, as well as increasing their proteolytic activity (88, 89). Basophils also express proteases but to a much lesser extent. Human basophils primarily express the tryptase whereas mouse basophils express the basophil-specific protease mMCP-8 (22–24).i that is, monotremes, marsupials, and placental mammals have mast cells, which are armed with high-affinity receptors for IgE (43–45, 60). However, the situation in reptiles, birds, and amphib- ians is much less clear. Do their mast cells have receptors for IgY? Are other isotypes involved or are there no Ig receptors on mast cells in these tetrapods? In marsupials, represented by the American opossum, there is, based on phylogenetics, a direct ortholog of the human and mouse IgE receptor α chain (Figure 6) (60). IgE and the Connection between Innate and Adaptive Immunity A massive expansion of the chymase locus both in size and number of functional genes has occurred in rodents. The mouse chymase locus has, for example, 15 functional genes, including two new classes of genes, the β-chymases and the mMCP-8 gene and also several additional granzyme genes (Figure 7) (85, 90). The rat locus is fifteen times larger than the dog chymase locus and contains 28 functional genes (85). In ruminants, i.e., cows and sheep, an additional subfamily called the duodenases has appeared via gene duplications, most likely from the granzymes or from cathepsin G (85, 90). The duodenases in cow have changed tissue specificity and are now not expressed in hematopoietic cells but in secretory cells in the duodenum, where they most likely take part in food digestion (93, 94). The second locus is the metase locus, which encodes most of the neutrophil pro- teases; N-elastase, proteinase-3, neutrophil serine protease-4, and an inactive variant, azurocidine that acts as an antibacterial substance without protease activity. This locus also encodes a complement component, complement component D (Figure 7) (85, 90). The third locus encodes two T-cell expressed proteases, granzyme A and K, which are both tryptic enzymes (Figure 7). The fourth locus is the mast cell tryptase locus, which is divided into two regions, one expressing primarily the mast cell-expressed soluble tryptases and the second, which primar- ily encodes membrane bound tryptases that show a relatively broad pattern of expression (Figure 7) (95–97). The fifth locus has completely different bordering genes and is only found in reptiles, birds, and amphibians, where it encodes a few proteases that are distantly related to the mammalian chymase locus genes (Figure 7) (85, 90). By looking into the genomes of a panel of Frontiers in Immunology \| www.frontiersin.org Mast Cell and Basophil Granule Components These chymase locus related genes are marked in orange. A number of bordering genes are also included to show the similarity in the surrounding regions between the different species. The loci or genes marked with red stars in the alligator and frog genes (under chymase locus genes) are the genes closely related to the mammalian chymase locus genes and thereby may represent an early variant of the mammalian chymase locus. A more detailed analysis of these protease genes and their evolution containing a much larger number of species is found in Akula et al. (90). or NK-like cells and one from a macrophage-like cell line. One of these proteases, catfish granzyme-like I is a highly specific protease, probably the serine proteases with the highest specific- ity characterized so far. The cleavage specificity of this protease, which is expressed by fish NK-like cells, has an ability in vitro to cleave a sequence within catfish caspase 6, indicating it may have similar function as mammalian granzyme B, thereby inducing apoptosis in target cells (108). The second catfish protease is a highly specific tryptase, as of yet an unknown function, which is expressed by fish macrophage-like cells (unpublished results). The third of these three catfish proteases is not yet characterized at all and we do not know its specificity (90, 107). Several other fish proteases have also been produced as recombinant proteins, including ones from gar, zebrafish, and platyfish (Figure 8). However, no information concerning their tissue specificities, cleavage specificities, or potential targets are known yet; therefore currently, we cannot say if these fish proteases show similarities to any of the mast cell proteases in mammals. These studies are in their infancy and hopefully a more detailed picture will emerge within a few years’ time. vertebrates, it has been possible to trace the origin of these genes during vertebrate evolution. There is evidence for the existence of the T cell tryptase locus, encoding granzymes A and K, from cartilaginous fish to humans but not in jawless fish, including the lamprey and hagfish. Similarly, this locus is not found in tunicates or echinoderms, indicating an appearance with early jawed vertebrates. This locus is the only one of the five loci that is found in cartilaginous fish, indicating that it is the oldest of them all (Figure 7) (85, 90). Mast Cell and Basophil Granule Components In bony fish, there is evidence for the presence of the metase locus, and in frogs, the first gene that can be directly seen as an ancestor of the mammalian chy- mase locus is found. In frogs, a gene for the fifth locus exists, which is present in reptiles and birds but not in mammals. In this manner, we can start to see a gradual and probably parallel appearance of these proteases in the different vertebrate lineages during vertebrate evolution (85, 90). In order to obtain additional functional information concern- ing these proteases, we have used an unbiased technique with a very large number of potential target sequences. This consists of a phage library with 50 million different 9 amino acid long random sequences to study the extended cleavage specificity of a number of these hematopoietic serine proteases (85, 98). Despite numerous studies concerning the major functions and major targets of mammalian hematopoietic serine proteases, the picture concerning these proteases is still relatively incomplete. An array of potential targets has been described, some more and some less likely to represent evolutionary conserved functions of these proteases.hi Using a combination of genomic analyses and functional analyses of the extended cleavage specificities of a selected panel of the proteases, we have started to get a more detailed view of their emergence, evolution, cleavage specificities, and the basic functions performed by these proteases (22, 23, 80, 82, 85, 90, 98–106). The proteases we have focused on are the most interest- ing ones from an evolutionary perspective, i.e., ones that represent major branches on the phylogenetic tree and that can be most easily studied for their in vivo function based on available model systems (Figure 8) (85, 90). The five most interesting and in our minds most logical roles for the mast cell proteases include the following: they most likely have a central role in the defense against various snake, scorpion, and bee venoms (109–111). There is sufficient evidence that they probably also take an active part in connective tissue remod- eling by cleaving connective tissue components, as for example, fibronectin, and/or by activating other proteases, for example, matrix metalloproteases (MMPs), which can degrade several components including collagen (112–117). In addition, they are likely involved in both activation and degradation of cytokines, primarily inflammatory cytokines to dampen inflammation (106, 118). Mast Cell and Basophil Granule Components Both mast cells and basophils contain a large number of cytoplas- mic granules (Figures 2A,B). These granules that are functionally related to lysosomes, store a number of substances of low and high molecular weight substances. Histamine, one of these low molecular weight compounds, is stored by both mast cells and basophils (71). It is based on the amino acid histidine where the carboxyl acid, i.e., COOH group, has been removed by the enzyme histidine decarboxylase (72). The removal of the acid group results in a positively charged molecule of a size smaller than that of an amino acid. Histamine is a highly potent inflam- matory mediator due to its interaction with four different recep- tors termed H1, H2, H3, and H4 (73, 74). Binding of histamine to these receptors induces a number of processes including vascular leakage and itching (73, 74). The mast cell granules also contain large and heavily sulfated, negatively charged polysaccharides; heparin in the case of mast cells, and chondroitin sulfate in basophils (75–78). One function of these charged proteoglycans is most likely to counteract the positive charge of histamine. The cell would otherwise not be able to store such large amounts of a positively charged molecule (77). Heparin is also a potent anti- coagulant via its binding and activation of anti-thrombin (78). Mast cells also store massive amounts of proteases primarily chymotrypsin/trypsin-related serine proteases but also the mast cell-specific carboxypeptidase A3 (79–85). These proteases can make up to 35% of the total protein, thereby constituting the absolute majority of the protein content of the mast cell granules December 2017 \| Volume 8 \| Article 1749 10 Origins of IgE, Mast Cells, and Allergies Hellman et al. 11 Frontiers in Immunology \| www frontiersin org December 2017 \| Volume 8 \| Article 1749 Figure 7 \| Continued Figure 7 \| Continued December 2017 \| Volume 8 \| Article 1749 11 Origins of IgE, Mast Cells, and Allergies Hellman et al. Figure 7 \| Chromosomal loci encoding hematopoietic serine proteases. A selection of such loci representing the five different loci encoding hematopoietic serine proteases is shown; the chymase locus, the metase locus, the granzyme A/K locus, the mast cell tryptase locus, and the new chymase locus related locus found in amphibians, reptiles, and birds. The section of the figure showing the chymase loci includes this new amphibian, bird, and reptile locus. Mast Cell and Basophil Granule Components The mast cell chymase is also a potent activator of angiotensin from Ang I to Ang II thereby with the potential to increase blood pressure, which may be needed after a systemic mast cell activation where blood pressure drops due to the out- flow of liquid from the blood into the surrounding tissue (117, 119–121). During the opening of blood vessels to enhance the flow of blood components including antibodies and complement, there is a major risk that the inflow is blocked by coagulation. Here, mast cell proteases and heparin may act cooperatively as an anticoagulant by cleaving thrombin and other coagulation components (114, 122, 123). Although not to be ignored, there are many other suggested functions. However, given the evidence, Both chymotryptic and tryptic enzymes appear to have been a central component of mast cell granules from early mam- mals. This is based solely on the presence of the protease genes in marsupials and monotremes and not on a direct analysis of their mast cells. However, based on the presence of proteases with close structural similarities and with very similar cleavage specificities to the well-characterized ones in several placental animals, we can with reasonable certainty claim that they are also likely to be part of the mast cell phenotype (90, 103). The picture is less clear when we look at reptiles, birds, and amphibians. In birds, none of the classical chymase locus genes are present and in frogs, there is one such gene, which has structural similarity of its active site pocket that closely matches one of the T cell enzymes, granzyme B (Figure 7) (90). We have now also started to study the related fish proteases (Figure 8, brown colored branch) (90, 107), and have so far produced recombinant protein for three catfish proteases. Two of them are expressed in CTLs December 2017 \| Volume 8 \| Article 1749 Frontiers in Immunology \| www.frontiersin.org 12 Hellman et al. Origins of IgE, Mast Cells, and Allergies Figure 8 \| A phylogenetic analysis of a large panel of hematopoietic serine proteases. Three coagulation and complement components are used as outgroup. IgE Levels under Parasite-Free and Parasite-Rich Conditions IgE responses have also been seen to occur primarily against low levels of antigen. This would indicate that the IgE system is focused on early responses to low levels of antigen, possibly to sample the environment. The antigen-specific IgE, often locally produced, can then bind mast cells and prime the immune system for a second encounter with this antigen possi- bly in the form of a parasite, a virus, or a bacteria. The individual is subsequently already primed for a relatively strong response involving the majority of immune components. The IgE covered mast cell is an extremely potent amplifier of an inflammatory response, where cross-linking of less than 100 IgE molecules on the cell surface is sufficient for full activation/degranulation of the cell. Such a sensitive and massive response may be necessary to manage a massive infection of an intestinal worm parasite. Non-allergic persons have generally very low levels of IgE in their circulation, ranging from 20 to 400 ng/ml in blood (125). In com- parison to the IgG levels, which range from 8 to 16 mg/ml, IgE levels are between 100,000 times to a million times lower. Persons with the relatively mild allergies, rhinitis, and conjunctivitis, often have slightly elevated IgE levels, where asthmatics have even higher ranging from 400 ng to 1 or 2 µg per ml. The patients with the highest IgE levels are generally persons with severe atopic dermatitis where IgE levels may reach as high as 10 µg/ml. g y g µg An interesting question here is if these very low levels of IgE are reflected in the general situation in both domestic and wild animal populations, and how this affects our view of the function of IgE. Most inbred mouse and rat strains have similar low IgE levels to non-allergic humans, below 200 ng/ml, with many strains having less than 50 ng/ml (7, 126). However, there are some notable exceptions with strains such as the Balb/c mice, which may reach 100 µg of IgE/ml as well as the Brown Norway rats, which can have IgE in the range of several micrograms per milliliter (7). Both of these rodent strains are considered so called TH2 type of strains, with a dominance of humoral immunity compared to many strains with low IgE levels that are more TH1 prone, thereby having a stronger ten- dency to use cell-mediated immunity. IgE Levels under Parasite-Free and Parasite-Rich Conditions y g g One of the major questions in the field of IgE biology is why this gene that on face value causes us so much trouble has been maintained for several hundred million years of mammalian evolution. The indications for a coevolution of IgE and its recep- tors on mast cells and basophils strongly support an evolutionary selective advantage of the system. During the writing of my PhD thesis in 1985 I (LH) proposed a theory that IgE together with mast cells may actually function as a door or gate keeper, stopping the antigen at the site of entry (124). Much of the evidence that has accumulated on the subject now strengthens this hypothesis. By triggering a rapid release of histamine, prostaglandins, leu- kotrienes, proteoglycans, proteases, and cytokines mast cells can activate and recruit immune cells to the site of entry. Histamine and the arachidonic acid metabolites (leukotrienes and pros- taglandins) open blood vessels, which facilitates the entry of antibodies, complement and immune cells. The proteases and heparin that are also being released by the mast cells can limit coagulation, which would otherwise inhibit the movement of inflammatory cells. The proteases can also function by loosen- ing up the connective tissue allowing the entry of the immune cells and other molecules. The location of mast cells in regions where most pathogens enter, for example, the intestinal and lung mucosa, the skin as well as around organs and blood vessels, also provides strong support for the role of IgE and mast cells as part of a door keeper or sentry function. At these sites mast cells can also exert an important role in the process of venom inactivation. Likewise at these locations, the processing of Ang I into Ang II by the mast cell chymase can potentially counteract the resulting effect of the blood pressure drop after mast cell degranulation. Release of Ang II can result in a rapid increase in blood pressure. The tissue remodeling function of mast cells by activation of MMPs and cleavage of fibronectin, collagen, and other connective tissue components is probably relatively IgE independent, whereas cytokine activation or inactivation may be important for limiting the inflammation initiated by the IgE- dependent mast cell activation. We have also observed a peculiar early IgE response already at days 3 and 4 after antigen/allergen contact that seems to precede the rise in IgG (unpublished obser- vations). Mast Cell and Basophil Granule Components The proteases of the different genetic loci cluster together in separate branches of the tree and are color coded in similar fashion to Figure 7: the metase locus genes marked in green, the granzyme A/K locus in dark blue, the new reptile and amphibian locus in orange, the chymase locus marked in red in one branch, and the majority of the fish proteases in brown. The proteases we are currently analyzing for their extended cleavage specificities and tissue or cell type expression patterns are marked with small filled circles. A more detailed analysis of these protease genes and their evolution is found in Akula et al. (90). Figure 8 \| A phylogenetic analysis of a large panel of hematopoietic serine proteases. Three coagulation and complement components are used as outgroup. The proteases of the different genetic loci cluster together in separate branches of the tree and are color coded in similar fashion to Figure 7: the metase locus genes marked in green, the granzyme A/K locus in dark blue, the new reptile and amphibian locus in orange, the chymase locus marked in red in one branch, and the majority of the fish proteases in brown. The proteases we are currently analyzing for their extended cleavage specificities and tissue or cell type expression patterns are marked with small filled circles. A more detailed analysis of these protease genes and their evolution is found in Akula et al. (90). Figure 8 \| A phylogenetic analysis of a large panel of hematopoietic serine proteases. Three coagulation and complement components are used as outgroup. The proteases of the different genetic loci cluster together in separate branches of the tree and are color coded in similar fashion to Figure 7: the metase locus genes marked in green, the granzyme A/K locus in dark blue, the new reptile and amphibian locus in orange, the chymase locus marked in red in one branch, and the majority of the fish proteases in brown. The proteases we are currently analyzing for their extended cleavage specificities and tissue or cell type expression patterns are marked with small filled circles. A more detailed analysis of these protease genes and their evolution is found in Akula et al. (90). December 2017 \| Volume 8 \| Article 1749 Frontiers in Immunology \| www.frontiersin.org 13 Origins of IgE, Mast Cells, and Allergies Hellman et al. Mast Cell and Basophil Granule Components Therefore, in our minds many aspects related to IgE and mast cells, including their location and homeostasis, favor a role of these components in an early, door keeper function. we feel that these five are among the most likely. These proteases may have adapted a relatively broad array of functions, which may also explain their relatively broad specificities. Frontiers in Immunology \| www.frontiersin.org IgE Levels under Parasite-Free and Parasite-Rich Conditions Therefore, genetic factors are clearly important for the levels of circulating IgE.i However, other factors are also very significant. Both humans and rodents living under laboratory conditions are generally free from worm infections, which are known to be potent inducers of IgE production (127). By contrast, most wild animal populations have massive amounts of intestinal worm parasites. A few years ago we developed a reliable assay testing for dog IgE based on monoclonals raised against recombinant dog IgE. This made it possible to study IgE levels in both dogs and wolves, with high accuracy, which had not been possible with previously existing reagents. Analysis of a panel of 76 adult dogs showed that adult dogs have IgE levels that are between 10 and 40 µg/ ml, which is almost 100 times higher than non-allergic humans, but that young dogs started with low with levels often below 1 µg/ml of IgE (Figure 9) (128). A collaboration with Professors Jon Arnemo and Olof Liberg, which involved a large interdis- ciplinary study on the Scandinavian wolves, Scandulv, made it possible to obtain serum samples from approximately 30% of the total Scandinavian wolf population, around 65 individuals. By analyzing their IgE levels, we could see that wolves in general have twice as high IgE levels compared to domestic dogs, having a median value of 67 µg/ml (Figure 9) (129). The relatively high levels seen in domestic dogs (compared to non-allergic humans) were somewhat surprising as they are a domestic population. However, dogs are frequently parasite infected, then undergo treatment for this, but often get re-infected. This indicates that after infection levels of IgE tends to stay high for relatively long periods of time. December 2017 \| Volume 8 \| Article 1749 Frontiers in Immunology \| www.frontiersin.org 14 Hellman et al. Origins of IgE, Mast Cells, and Allergies 9 \| IgE levels in a panel of domestic dogs and wild wolves. The IgE levels, in micrograms per milliliter, are marked in red after each animal or group of Values from 33 young dogs and 5 young wolves are also included in the figure. The young wolves were from a zoo due to the difficulty obtaining samples wolf pups. Figure 9 \| IgE levels in a panel of domestic dogs and wild wolves. The IgE levels, in micrograms per milliliter, are marked in red after each animal or group of animals. Frontiers in Immunology \| www.frontiersin.org IgE Levels under Parasite-Free and Parasite-Rich Conditions Values from 33 young dogs and 5 young wolves are also included in the figure. The young wolves were from a zoo due to the difficulty obtaining samples from wild wolf pups. Figure 9 \| IgE levels in a panel of domestic dogs and wild wolves. The IgE levels, in micrograms per milliliter, are marked in red after each animal or group of animals. Values from 33 young dogs and 5 young wolves are also included in the figure. The young wolves were from a zoo due to the difficulty obtaining samples from wild wolf pups. A similar situation is seen with horses. Domestic horses have been found to have very high levels of IgE, twice the levels seen in wolves somewhere between 30 and 180 µg/ml (130). Horses also often get re-infected with worm parasites when grazing, which means they tend to need repeated treatment for intestinal worm infections (131, 132). Analysis of the young dogs previously described showed relatively low IgE levels, below 1 µg, indicating that parasite infections may be the major cause of the high IgE levels seen in adult dogs (Figure 9) (128). This also suggests that the puppies stay uninfected until they start to go outside their homes. Interestingly by contrast, the few young wolves coming from a zoo that were analyzed showed even higher IgE levels than the adult wolves, possibly indicating an early parasite infection or a genetic difference in IgE regulation between dogs and wolves (Figure 9). During the cloning and analysis of Ig isotypes in the platypus, we also found that transcript levels for IgE in the spleen were only six times lower than the IgG levels of these free-living animals, which further support the finding that IgE levels are much higher in wild compared to domestic animals (45). A very interesting study of Ethiopian Jews has also been performed, which provides indications further in line with these animal findings. When IgE levels were analyzed on people that had moved from Ethiopia to Israel at an adult age their IgE levels were high and stayed high during the entire study, whereas their children who were born and raised in Israel had similar low levels of IgE as other children born in Israel (125). The situation seen December 2017 \| Volume 8 \| Article 1749 Frontiers in Immunology \| www.frontiersin.org 15 Origins of IgE, Mast Cells, and Allergies Hellman et al. IgE Levels under Parasite-Free and Parasite-Rich Conditions a lack of fighting for territory, and an almost unlimited supply of food and water. One interesting finding relating to this idea is that persons lacking IgE seem to live a relatively normal life in spite of the fact that the connection between IgE and mast cells has been evolutionary conserved for what it seems at least 200 million years (134). By studying wild animals under natural conditions, we may get a better picture of the factors that have shaped our immune system. Recent data from the currently rapidly evolving field concerning the role of the microbiome show that as we are eating the same foods with low amount of fiber and high fat or high sugar, this can reduce the complexity of the intestinal flora (135, 136). A similar effect may also come from a massive overuse of antibiotics (137). A diverse microbiome obtained from eating different foods and not using antibiotics may be one factor in this picture (126, 138, 139). We know that the microbiome is important for stimulating the immune system, and that microbes are important for the production of vitamins and for the degradation of hormones and other substances. This may be suggestive as to why a reduced food complexity may be one factor that limits the development and functional diversity of our immune system. We are bombarded by the idea that lactobacilli, for example, are beneficial for our health. However, numerous tri- als with probiotics, including eating live lactobacilli in the form of yogurt or fermented foods have not shown any significant effects [reviewed in Ref. (140)]. Here the problems may be partly related to that these added lactobacilli only appearing to live for a short period of time in the intestine, thereby only marginally affecting the intestinal mucosa and the immune cells residing in the area just under the epithelial cells. In the industrialized world, we are almost completely free of both ecto and intestinal parasites. This is a factor that we as species are not so well adapted for. For example, studying the intestines of wild mice reveals that they are typically full of intestinal parasites, yet interestingly they are otherwise in reasonable health. Similarly, they are usually also infested with ecto-parasites including lice, fleas, and ticks factors that may markedly affect the immune system by triggering inflammatory cells, thereby affecting the cytokine environment. IgE Levels under Parasite-Free and Parasite-Rich Conditions Although these factors may be of importance for the development of our immune system we do not perceive that anyone of us would like to return to such conditions. However, the lesson here is that factors such as parasite loads and complexity of the intestinal flora may be of importance for many diseases including allergy, autoimmunity, diabetes, and possibly even certain cancers (127, 141, 142). Here studies of wild animals can shed light upon how the immune system reacts to, and handles, all of these parasites and also how a diverse intestinal flora may affect these processes. Respiratory virus infections are most likely also an important factor where respiratory syncytial virus and rhino virus are of prime interest for asthma development (143). There are other, more discrete factors, for example, there is an ever increasing amount of hormone-like substances in our environment, such as estrogens from contracep- tives that enter the sewage system. These may also be other factors to take into account, as it is well known that such substances can markedly affect reproduction of local fish and amphibian species. in dogs, horses, platypus and the migrating Ethiopian Jews show that IgE levels tend to stay high for long periods of time even after being free from parasite infection. The reasons for this are not known but indicates that the cytokine environment may change more permanently after a long or repeated exposure to intestinal worm parasites (133). Domestication and the Appearance of Allergies A number of factors have been indicated to be of importance for the high incidence of allergies in industrialized countries. One factor that seems to have a major impact is general domestication, as it is primarily among ourselves and among domestic animals that we find allergic individuals. To our knowledge, allergies have not been described in wild animals. One potential factor could be a genetic drift due to strong selection for phenotypic character- istics like coat color, long or short noses, running fast, or wanted social behaviors. Such strong selections are seen in the breeding programs for dogs, horses, and cats, but a questionable cause for human allergies. However, it is possible that we constantly need to be selecting against hypersensitivities, which may occur due to minor shift in immune functions caused by spontaneous point mutations. A strong such selection process most likely exists in wild animals under tough environmental conditions but not in domestic animals and in humans. Another factor could be parasite infections. However, the presence or absence of parasites cannot be the only explanation as we see allergies both in ourselves, who are generally free from intestinal parasites, at least in the industrialized world, whereas dogs and horses are often parasite infected.h t There could also be numerous other contributing factors, which are only partly dependent on domestication such as hygiene includ- ing a reduced complexity of intestinal microbiome. For humans, one factor could be the reduced use of fermented food. The list has become relatively long for potential causes of this increase and there are studies, which favor and disfavor almost all of them, making the situation very complex. It is clear that not only one single factor is involved but a combination of many and often also diverse factors, which together provides the imbalance of the immune system to overreact against often harmless substances such as pollen, nuts, animal fur, cow milk proteins, eggs, fish, and shellfish. Frontiers in Immunology \| www.frontiersin.org t We Can Learn from Wild Animals What We Can Learn from Wild Animals By studying only a few species, which live under very similar conditions we most likely get a fairly limited view of the regula- tion of the immune system. For example, many of the knockout strains of mice have been shown to display a very mild phenotype, indicating that many cell types and molecules are dispensable and not essential. However, this may be the situation under low pathogen loads and under conditions of good access to nutrients and clean water. A more pronounced phenotype may be seen under stress, under conditions of limited water and food supply, when individuals have to fight for territory and mating partners and under heavy parasite loads. Similarly, we may not see the factors regulating a normal immune response by only looking at a few species, which are essentially living under very similar and limited environmental conditions, including absence of parasites, December 2017 \| Volume 8 \| Article 1749 Frontiers in Immunology \| www.frontiersin.org 16 Origins of IgE, Mast Cells, and Allergies Hellman et al. Studies of such animals in the wild can provide information on how such pollutants affect their immune functions and thereby give us clues to as to how these substances will also affect us. Such substances may affect us more subtly by shifting the balance of the immune system. The question is very complex, and here by look- ing at wild populations may present us with a more multifaceted substances may affect us more subtly by shifting the balance of the immune system. The question is very complex, and here by look- ing at wild populations may present us with a more multifaceted Figure 10 \| Therapeutic vaccine proteins where parts of the vaccine antigen originate from a wild animal. Panel (A) shows a vaccine antigen consisting of a fusion protein between the CH2 and CH4 domains of opossum IgE together with the target region for the vaccine: the CH3 domain from the target animal, a human, dog, or rat. The figure shows the human variant of the vaccine. The rat variant of the vaccine antigen has been shown to induce a strong anti-self-IgE response in rats of several strains and to reduce circulating IgE levels in these animals (7). t We Can Learn from Wild Animals Panel (B) shows a multimeric vaccine component generated by the use of the C terminal tail of the lamprey antigen-specific receptor variable leukocyte receptor B (VLR-B). VLR-B is the functional but not the structural equivalent of human IgM, a pentameric antigen-binding molecule. Using the C terminal 30–40 amino acids from VLR-B fused to the C terminal of any soluble vaccine antigen, it is possible to obtain a multimeric vaccine antigen that serves as a very potent antigen due to their similarity to virus particles or bacterial surfaces with multiple identical epitopes. Using three cancer vaccine antigens as test antigens, we have shown that they are soluble when produced in bacteria (E. coli) and form stable cysteine-bridged multimers with from 5 to at least 15 monomers, in the multimeric structures (150). Figure 10 \| Therapeutic vaccine proteins where parts of the vaccine antigen originate from a wild animal. Panel (A) shows a vaccine antigen consisting of a fusion protein between the CH2 and CH4 domains of opossum IgE together with the target region for the vaccine: the CH3 domain from the target animal, a human, dog, or rat. The figure shows the human variant of the vaccine. The rat variant of the vaccine antigen has been shown to induce a strong anti-self-IgE response in rats of several strains and to reduce circulating IgE levels in these animals (7). Panel (B) shows a multimeric vaccine component generated by the use of the C terminal tail of the lamprey antigen-specific receptor variable leukocyte receptor B (VLR-B). VLR-B is the functional but not the structural equivalent of human IgM, a pentameric antigen-binding molecule. Using the C terminal 30–40 amino acids from VLR-B fused to the C terminal of any soluble vaccine antigen, it is possible to obtain a multimeric vaccine antigen that serves as a very potent antigen due to their similarity to virus particles or bacterial surfaces with multiple identical epitopes. Using three cancer vaccine antigens as test antigens, we have shown that they are soluble when produced in bacteria (E. coli) and form stable cysteine-bridged multimers with from 5 to at least 15 monomers, in the multimeric structures (150). Figure 10 \| Therapeutic vaccine proteins where parts of the vaccine antigen originate from a wild animal. t We Can Learn from Wild Animals Panel (A) shows a vaccine antigen consisting of a fusion protein between the CH2 and CH4 domains of opossum IgE together with the target region for the vaccine: the CH3 domain from the target animal, a human, dog, or rat. The figure shows the human variant of the vaccine. The rat variant of the vaccine antigen has been shown to induce a strong anti-self-IgE response in rats of several strains and to reduce circulating IgE levels in these animals (7). Panel (B) shows a multimeric vaccine component generated by the use of the C terminal tail of the lamprey antigen-specific receptor variable leukocyte receptor B (VLR-B). VLR-B is the functional but not the structural equivalent of human IgM, a pentameric antigen-binding molecule. Using the C terminal 30–40 amino acids from VLR-B fused to the C terminal of any soluble vaccine antigen, it is possible to obtain a multimeric vaccine antigen that serves as a very potent antigen due to their similarity to virus particles or bacterial surfaces with multiple identical epitopes. Using three cancer vaccine antigens as test antigens, we have shown that they are soluble when produced in bacteria (E. coli) and form stable cysteine-bridged multimers with from 5 to at least 15 monomers, in the multimeric structures (150). December 2017 \| Volume 8 \| Article 1749 Frontiers in Immunology \| www.frontiersin.org 17 Origins of IgE, Mast Cells, and Allergies Hellman et al. is the form the self-antigen is being presented (148). Multimeric antigens such as a virus particle or a bacterial surface are very potent antigens, probably due to their potent B cell-activating properties (148, 149). A multimeric antigen can crosslink IgM on the surface of the naïve B cell very efficiently and thereby giving a very strong activating signal 1 to the B cell. During the process of optimizing parameters to obtain potent therapeutic vaccines against allergies and different solid cancers, we identified a region of a molecule coming from a wild animal a jawless fish. The protein and region is the tail-piece of the variable leukocyte receptor B (VLR-B) that facilitates the pentamerization of the VLRB, which is the functional but not structural Ig equivalent in lamprey and hagfish, similar to pentameric human IgM. t We Can Learn from Wild Animals Using the 30 amino acid C-terminal region of lamprey VLR-B, this resulted in very efficient multimerization of the target antigen and in a marked enhancement of the anti-target immune response (Figure 10B) (150). In this context, wild animals can not only give us a more detailed view of the function and evolu- tion of our immune system but can also be a rich source of potential therapeutic proteins. view of the factors involved in shaping immunity and where these factors can go wrong. Our life style has changed dramatically, and the strong genetic selection acting upon us during early evolution is presently most likely not as efficient, which may be contribut- ing factors to the increase in allergies. However, what factors that dominate this increase are still not known, which may also vary from person to person as both genetic and environmental factors seem to be of importance (127, 140). CONCLUSION Wild animals can teach us a lot about our own immune system, including how it is regulated, how it has evolved, and which func- tions are essential for a potent immune defense. These reasons are but a few to consider, where non-domestic animals may facilitate new solutions to difficult therapeutic challenges. FUNDING Work related to the review was funded by The Swedish Research Council (VR-NT) (621-2011-5007). Wild Animals As a Source of Therapeutic Proteins In addition to giving us a more detailed view of the regulation and the evolution of our immune system, wild animals may also be a rich source of therapeutic proteins and other potential therapeutic molecules. As a separate line of research we have also been trying to develop new treatment strategies against atopic allergies. One such line of research has been the development of therapeutic vaccines targeting IgE and several of the early TH2 inducing cytokines including IL-33, IL-18, and TSLP (Figure 1) (6–8, 13, 144). Induction of an immune response against self-molecules, as is the case for all of these targets, is considerably more difficult than inducing an immune response against a foreign molecule. We are generally tolerant to self-molecules, and therefore, we need to use a number of tricks to overcome these tolerance mechanisms in order to induce an immune response strong enough to give a therapeutic effect. Here, adjuvants are very important and in order to obtain a strong anti-self-immune response potent adjuvants are a necessity (145, 146). On top of this issue are other factors including having to modify the self-protein by coupling it to a non-self carrier (6, 144, 147). This results in the recruitment of non-tolerized T cells to provide help to self-reactive B cells to expand and differentiate (144). In the design of a vaccine targeting IgE, we produced a fusion pro- tein between opossum IgE and human, rat or dog IgE (Figure 10A). The rat variant of this vaccine antigen induced a strong anti-rat IgE response in sensitized rats and resulted in a marked reduction in circulating IgE titers in these animals (7). Another important factor AUTHOR CONTRIBUTIONS The review was co-written by LH, SA, MT, and ZF. The review was co-written by LH, SA, MT, and ZF. REFERENCES 8. Lei Y, Boinapally V, Zoltowska A, Adner M, Hellman L, Nilsson G. Vaccination against IL-33 inhibits airway hyperresponsiveness and inflam- mation in a house dust mite model of asthma. PLoS One (2015) 10:e0133774. doi:10.1371/journal.pone.0133774 8. Lei Y, Boinapally V, Zoltowska A, Adner M, Hellman L, Nilsson G. Vaccination against IL-33 inhibits airway hyperresponsiveness and inflam- mation in a house dust mite model of asthma. PLoS One (2015) 10:e0133774. doi:10.1371/journal.pone.0133774 1. Pearson WS, Goates SA, Harrykissoon SD, Miller SA. State-based medicaid costs for pediatric asthma emergency department visits. Prev Chronic Dis (2014) 11:E108. doi:10.5888/pcd11.140139 j 9. Strachan DP. Hay fever, hygiene, and household size. BMJ (1989) 299:1259– 60. doi:10.1136/bmj.299.6710.1259 9. Strachan DP. Hay fever, hygiene, and household size. BMJ (1989) 299:1259– 60. doi:10.1136/bmj.299.6710.1259 2. Accordini S, Corsico AG, Braggion M, Gerbase MW, Gislason D, Gulsvik A, et al. The cost of persistent asthma in Europe: an international population- based study in adults. Int Arch Allergy Immunol (2013) 160:93–101. doi:10.1159/000338998 10. von Mutius E. Allergies, infections and the hygiene hypothesis – the epidemiological evidence. Immunobiology (2007) 212:433–9. doi:10.1016/ j.imbio.2007.03.002 h 11. Okada H, Kuhn C, Feillet H, Bach JF. The ‘hygiene hypothesis’ for autoim- mune and allergic diseases: an update. Clin Exp Immunol (2010) 160:1–9. doi:10.1111/j.1365-2249.2010.04139.x 3. Griffin C. Canine atopic disease. In: Griffin C, Kwochka K, McDonald J, editors. Current Veterinary Dermatology: The Science and Art of Practice. St Louis: Mosby Yearbook (1993). 1993 p. j 12. Stear MJ, Singleton D, Matthews L. An evolutionary perspective on gastro- intestinal nematodes of sheep. J Helminthol (2011) 85:113–20. doi:10.1017/ S0022149X11000058 y 4. Reedy LM, Miller WH, Willemse T. Allergic Skin Diseases of Cats and Dogs. Philadelphia: W.B. Saunders (1997). 12. Stear MJ, Singleton D, Matthews L. An evolutionary perspective on gastro- intestinal nematodes of sheep. J Helminthol (2011) 85:113–20. doi:10.1017/ S0022149X11000058 5. Lund EM, Armstrong PJ, Kirk CA, Kolar LM, Klausner JS. Health status and population characteristics of dogs and cats examined at private veterinary practices in the United States. J Am Vet Med Assoc (1999) 214:1336–41. 13. Hellman L. Regulation of IgE homeostasis, and the identification of potential targets for therapeutic intervention. Biomed Pharmacother (2007) 61:34–49. doi:10.1016/j.biopha.2006.10.001 13. Hellman L. Regulation of IgE homeostasis, and the identification of potential targets for therapeutic intervention. Biomed Pharmacother (2007) 61:34–49. doi:10.1016/j.biopha.2006.10.001 6. Hellman L. Profound reduction in allergen sensitivity following treatment with a novel allergy vaccine. Eur J Immunol (1994) 24:415–20. REFERENCES The mast cell binding site on human immunoglobulin E. Nature (1988) 331:180–3. doi:10.1038/331180a0 46. Vernersson M, Aveskogh M, Hellman L. Cloning of IgE from the echidna (Tachyglossus aculeatus) and a comparative analysis of epsilon chains from all three extant mammalian lineages. Dev Comp Immunol (2004) 28:61–75. doi:10.1016/S0145-305X(03)00084-3 27. Henry AJ, Cook JP, McDonnell JM, Mackay GA, Shi J, Sutton BJ, et al. Participation of the N-terminal region of Cepsilon3 in the binding of human IgE to its high-affinity receptor FcepsilonRI. Biochemistry (1997) 36:15568–78. doi:10.1021/bi971299+ 47. Vernersson M, Belov K, Aveskogh M, Hellman L. Cloning and structural analysis of two highly divergent IgA isotypes, IgA1 and IgA2 from the duck billed platypus, Ornithorhynchus anatinus. Mol Immunol (2010) 47:785–91. doi:10.1016/j.molimm.2009.10.007 28. Henry AJ, McDonnell JM, Ghirlando R, Sutton BJ, Gould HJ. Conformation of the isolated cepsilon3 domain of IgE and its complex with the high- affinity receptor, FcepsilonRI. Biochemistry (2000) 39:7406–13. doi:10.1021/ bi9928391 48. Aveskogh M, Pilstrom L, Hellman L. Cloning and structural analysis of IgM (mu chain) and the heavy chain V region repertoire in the marsupial Monodelphis domestica. Dev Comp Immunol (1999) 23:597–606. doi:10.1016/ S0145-305X(99)00050-6 29. Dobson JT, Seibert J, Teh EM, Da’as S, Fraser RB, Paw BH, et al. Carboxy peptidase A5 identifies a novel mast cell lineage in the zebrafish providing new insight into mast cell fate determination. Blood (2008) 112:2969–72. doi:10.1182/blood-2008-03-145011 49. Belov K, Hellman L. Immunoglobulin genetics of Ornithorhynchus anatinus (platypus) and Tachyglossus aculeatus (short-beaked echidna). Comp Biochem Physiol A Mol Integr Physiol (2003) 136:811–9. doi:10.1016/S1095- 6433(03)00165-X 30. Baccari GC, Pinelli C, Santillo A, Minucci S, Rastogi RK. Mast cells in non- mammalian vertebrates: an overview. Int Rev Cell Mol Biol (2011) 290:1–53. doi:10.1016/B978-0-12-386037-8.00006-5 hi 31. Crivellato E, Travan L, Ribatti D. The phylogenetic profile of mast cells. Methods Mol Biol (2015) 1220:11–27. doi:10.1007/978-1-4939-1568-2_2 li 50. Belov K, Hellman L, Cooper DW. Characterisation of echidna IgM provides insights into the time of divergence of extant mammals. Dev Comp Immunol (2002) 26:831–9. doi:10.1016/S0145-305X(02)00030-7 32. Reite OB, Evensen O. Inflammatory cells of teleostean fish: a review focusing on mast cells/eosinophilic granule cells and rodlet cells. Fish Shellfish Immu nol (2006) 20:192–208. doi:10.1016/j.fsi.2005.01.012 51. Zhao Y, Cui H, Whittington CM, Wei Z, Zhang X, Zhang Z, et al. Orni thorhynchus anatinus (platypus) links the evolution of immunoglobulin genes in eutherian mammals and nonmammalian tetrapods. J Immunol (2009) 183:3285–93. doi:10.4049/jimmunol.0900469 33. REFERENCES doi:10.1002/ eji.1830240222 j 14. Mosmann TR, Cherwinski H, Bond MW, Giedlin MA, Coffman RL. Two types of murine helper T cell clone. I. Definition according to profiles of lymphokine activities and secreted proteins. J Immunol (1986) 136:2348–57. 14. Mosmann TR, Cherwinski H, Bond MW, Giedlin MA, Coffman RL. Two types of murine helper T cell clone. I. Definition according to profiles of lymphokine activities and secreted proteins. J Immunol (1986) 136:2348–57. 7. Vernersson M, Ledin A, Johansson J, Hellman L. Generation of therapeutic antibody responses against IgE through vaccination. FASEB J (2002) 16:875–7. doi:10.1096/fj.01-0879fje 15. Zhu J, Paul WE. CD4 T cells: fates, functions, and faults. Blood (2008) 112:1557–69. doi:10.1182/blood-2008-05-078154 15. Zhu J, Paul WE. CD4 T cells: fates, functions, and faults. Blood (2008) 112:1557–69. doi:10.1182/blood-2008-05-078154 December 2017 \| Volume 8 \| Article 1749 Frontiers in Immunology \| www.frontiersin.org 18 Hellman et al. Origins of IgE, Mast Cells, and Allergies 36. Danilova N, Bussmann J, Jekosch K, Steiner LA. The immunoglobulin heavy-chain locus in zebrafish: identification and expression of a previously unknown isotype, immunoglobulin Z. Nat Immunol (2005) 6:295–302. doi:10.1038/ni1166 16. de Vries JE, Punnonen J, Cocks BG, de Waal Malefyt R, Aversa G. Regulation of the human IgE response by IL4 and IL13. Res Immunol (1993) 144:597–601. doi:10.1016/S0923-2494(05)80009-4 17. McKenzie GJ, Fallon PG, Emson CL, Grencis RK, McKenzie AN. Simulta neous disruption of interleukin (IL)-4 and IL-13 defines individual roles in T helper cell type 2-mediated responses. J Exp Med (1999) 189:1565–72. doi:10.1084/jem.189.10.1565 37. Castro R, Jouneau L, Pham HP, Bouchez O, Giudicelli V, Lefranc MP, et al. Teleost fish mount complex clonal IgM and IgT responses in spleen upon systemic viral infection. PLoS Pathog (2013) 9:e1003098. doi:10.1371/ journal.ppat.1003098 18. Stavnezer J, Amemiya CT. Evolution of isotype switching. Semin Immunol (2004) 16:257–75. doi:10.1016/j.smim.2004.08.005 f 38. Zhao Y, Pan-Hammarstrom Q, Yu S, Wertz N, Zhang X, Li N, et al. Iden tification of IgF, a hinge-region-containing Ig class, and IgD in Xenopus tropicalis. Proc Natl Acad Sci U S A (2006) 103:12087–92. doi:10.1073/pnas. 0600291103 19. Greer AM, Wu N, Putnam AL, Woodruff PG, Wolters P, Kinet JP, et al. Serum IgE clearance is facilitated by human FcepsilonRI internalization. J Clin Invest (2014) 124:1187–98. doi:10.1172/JCI68964 39. Gambon Deza F, Sanchez Espinel C, Magadan Mompo S. The immunoglob- ulin heavy chain locus in the reptile Anolis carolinensis. Mol Immunol (2009) 46:1679–87. doi:10.1016/j.molimm.2009.02.019 i 20. REFERENCES Ye H, Housden JE, Hunter MJ, Sabban S, Helm BA. The role of Cepsilon2, Cepsilon3, and Cepsilon4 domains in human and canine IgE and their contribution to FcepsilonRIalpha interaction. Mol Immunol (2014) 57:151–9. doi:10.1016/j.molimm.2013.08.004 40. Cheng G, Gao Y, Wang T, Sun Y, Wei Z, Li L, et al. Extensive diversification of IgH subclass-encoding genes and IgM subclass switching in crocodilians. Nat Commun (2013) 4:1337. doi:10.1038/ncomms2317 21. Barua D, Goldstein B. A mechanistic model of early FcepsilonRI signaling: lipid rafts and the question of protection from dephosphorylation. PLoS One (2012) 7:e51669. doi:10.1371/journal.pone.0051669 41. Magadan-Mompo S, Sanchez-Espinel C, Gambon-Deza F. IgH loci of American alligator and saltwater crocodile shed light on IgA evolution. Immunogenetics (2013) 65:531–41. doi:10.1007/s00251-013-0692-y 22. Lutzelschwab C, Huang MR, Kullberg MC, Aveskogh M, Hellman L. Characterization of mouse mast cell protease-8, the first member of a novel subfamily of mouse mast cell serine proteases, distinct from both the classical chymases and tryptases. Eur J Immunol (1998) 28:1022–33. doi:10.1002/ (SICI)1521-4141(199803)28:03<1022::AID-IMMU1022>3.0.CO;2-1 42. Zhao Y, Rabbani H, Shimizu A, Hammarstrom L. Mapping of the chicken immunoglobulin heavy-chain constant region gene locus reveals an inverted alpha gene upstream of a condensed upsilon gene. Immunology (2000) 101:348–53. doi:10.1046/j.1365-2567.2000.00106.x 23. Lutzelschwab C, Lunderius C, Enerback L, Hellman L. A kinetic analysis of the expression of mast cell protease mRNA in the intestines of Nippostrongylus brasiliensis-infected rats. Eur J Immunol (1998) 28:3730–7. doi:10.1002/ (SICI)1521-4141(199811)28:11<3730::AID-IMMU3730>3.0.CO;2-0 43. Warr GW, Magor KE, Higgins DA. IgY: clues to the origins of modern anti- bodies. Immunol Today (1995) 16:392–8. doi:10.1016/0167-5699(95)80008-5 44. Aveskogh M, Hellman L. Evidence for an early appearance of modern post- switch isotypes in mammalian evolution; cloning of IgE, IgG and IgA from the marsupial Monodelphis domestica. Eur J Immunol (1998) 28:2738–50. doi:10.1002/(SICI)1521-4141(199809)28:09<2738::AID-IMMU2738> 3.0.CO;2-I 24. Lunderius C, Hellman L. Characterization of the gene encoding mouse mast cell protease 8 (mMCP-8), and a comparative analysis of hematopoie tic serine protease genes. Immunogenetics (2001) 53:225–32. doi:10.1007/ s002510100316 25. Garman SC, Kinet JP, Jardetzky TS. Crystal structure of the human high-af- finity IgE receptor. Cell (1998) 95:951–61. doi:10.1016/S0092-8674(00) 81719-5 h 45. Vernersson M, Aveskogh M, Munday B, Hellman L. Evidence for an early appearance of modern post-switch immunoglobulin isotypes in mamma- lian evolution (II); cloning of IgE, IgG1 and IgG2 from a monotreme, the duck-billed platypus, Ornithorhynchus anatinus. Eur J Immunol (2002) 32:2145–55. doi:10.1002/1521-4141(200208)32:8<2145::AID-IMMU2145> 3.0.CO;2-I 26. Helm B, Marsh P, Vercelli D, Padlan E, Gould H, Geha R. REFERENCES doi:10.1134/S0006297911050038 h 81. Irani AM, Goldstein SM, Wintroub BU, Bradford T, Schwartz LB. Human mast cell carboxypeptidase. Selective localization to MCTC cells. J Immunol (1991) 147:247–53. 59. Bakema JE, van Egmond M. The human immunoglobulin A Fc receptor FcalphaRI: a multifaceted regulator of mucosal immunity. Mucosal Immunol (2011) 4:612–24. doi:10.1038/mi.2011.36 82. Lutzelschwab C, Pejler G, Aveskogh M, Hellman L. Secretory granule prote- ases in rat mast cells. Cloning of 10 different serine proteases and a carboxy- peptidase A from various rat mast cell populations. J Exp Med (1997) 185: 13–29. doi:10.1084/jem.185.1.13 60. Akula S, Mohammadamin S, Hellman L. Fc receptors for immunoglobulins and their appearance during vertebrate evolution. PLoS One (2014) 9:e96903. doi:10.1371/journal.pone.0096903 61. Akula S, Hellman L. The appearance and diversification of receptors for IgM during vertebrate evolution. Curr Top Microbiol Immunol (2017) 408:1–23. doi:10.1007/82_2017_22 83. Pejler G, Ronnberg E, Waern I, Wernersson S. Mast cell proteases: multi faceted regulators of inflammatory disease. Blood (2010) 115:4981–90. doi:10.1182/blood-2010-01-257287 l 62. Davis RS. Fc receptor-like molecules. Annu Rev Immunol (2007) 25:525–60. doi:10.1146/annurev.immunol.25.022106.141541 84. Caughey GH. Mast cell proteases as protective and inflammatory mediators. Adv Exp Med Biol (2011) 716:212–34. doi:10.1007/978-1-4419-9533-9_12 h 63. Ehrhardt GR, Cooper MD. Immunoregulatory roles for fc recep- tor-like molecules. Curr Top Microbiol Immunol (2011) 350:89–104. doi:10.1007/82_2010_88 85. Hellman L, Thorpe M. Granule proteases of hematopoietic cells, a family of versatile inflammatory mediators – an update on their cleavage specificity, in vivo substrates, and evolution. Biol Chem (2014) 395:15–49. doi:10.1515/ hsz-2013-0211 64. Nikolaidis N, Klein J, Nei M. Origin and evolution of the Ig-like domains present in mammalian leukocyte receptors: insights from chicken, frog, and fish homologues. Immunogenetics (2005) 57:151–7. doi:10.1007/s00251- 004-0764-0 86. Schwartz LB, Irani AM, Roller K, Castells MC, Schechter NM. Quantitation of histamine, tryptase, and chymase in dispersed human T and TC mast cells. J Immunol (1987) 138:2611–5. 65. Viertlboeck BC, Gobel TW. The chicken leukocyte receptor cluster. Vet Immunol Immunopathol (2011) 144:1–10. doi:10.1016/j.vetimm.2011. 07.001 87. Lindstedt KA, Kokkonen JO, Kovanen PT. Regulation of the activity of secreted human lung mast cell tryptase by mast cell proteoglycans. Biochim Biophys Acta (1998) 1425:617–27. doi:10.1016/S0304-4165(98)00115-9 66. Kaetzel CS. Coevolution of mucosal immunoglobulins and the polymeric immunoglobulin receptor: evidence that the commensal microbiota provided the driving force. ISRN Immunol (2014):1–20. doi:10.1155/2014/ 541537 88. Matsumoto R, Sali A, Ghildyal N, Karplus M, Stevens RL. Packaging of proteases and proteoglycans in the granules of mast cells and other hematopoietic cells. REFERENCES A cluster of histidines on mouse mast cell protease 7 regulates its binding to heparin serglycin proteoglycans. J Biol Chem (1995) 270:19524–31. doi:10.1074/jbc.270.33.19524 67. Blank U, Ra C, Miller L, White K, Metzger H, Kinet JP. Complete structure and expression in transfected cells of high affinity IgE receptor. Nature (1989) 337:187–9. doi:10.1038/337187a0 hfi 89. Hallgren J, Karlson U, Poorafshar M, Hellman L, Pejler G. Mechanism for activation of mouse mast cell tryptase: dependence on heparin and acidic pH for formation of active tetramers of mouse mast cell protease 6. Biochemistry (2000) 39:13068–77. doi:10.1021/bi000973b 68. Rodewald HR, Arulanandam AR, Koyasu S, Reinherz EL. The high affinity Fc epsilon receptor gamma subunit (Fc epsilon RI gamma) facilitates T cell receptor expression and antigen/major histocompatibility complex-driven signaling in the absence of CD3 zeta and CD3 eta. J Biol Chem (1991) 266:15974–8. 90. Akula S, Thorpe M, Boinapally V, Hellman L. Granule associated serine proteases of hematopoietic cells – an analysis of their appearance and diversification during vertebrate evolution. PLoS One (2015) 10:e0143091. doi:10.1371/journal.pone.0143091 ii 69. Weissman AM, Hou D, Orloff DG, Modi WS, Seuanez H, O’Brien SJ, et al. Molecular cloning and chromosomal localization of the human T-cell receptor zeta chain: distinction from the molecular CD3 complex. Proc Natl Acad Sci U S A (1988) 85:9709–13. doi:10.1073/pnas.85.24.9709 91. Gallwitz M, Hellman L. Rapid lineage-specific diversification of the mast cell chymase locus during mammalian evolution. Immunogenetics (2006) 58:641–54. doi:10.1007/s00251-006-0123-4 70. Lanier LL. DAP10- and DAP12-associated receptors in innate immunity. Immunol Rev (2009) 227:150–60. doi:10.1111/j.1600-065X.2008.00720.x 92. Gallwitz M, Reimer JM, Hellman L. Expansion of the mast cell chymase locus over the past 200 million years of mammalian evolution. Immunogenetics (2006) 58:655–69. doi:10.1007/s00251-006-0126-1 j 71. Riley JF, West GB. Histamine in tissue mast cells. J Physiol (1952) 117:72–3. 93. Zamolodchikova TS, Scherbakov IT, Khrennikov BN, Svirshchevskaya EV. Expression of duodenase-like protein in epitheliocytes of Brunner’s glands in human duodenal mucosa. Biochemistry (Mosc) (2013) 78:954–7. doi:10.1134/ S0006297913080130 72. Ohtsu H, Tanaka S, Terui T, Hori Y, Makabe-Kobayashi Y, Pejler G, et al. Mice lacking histidine decarboxylase exhibit abnormal mast cells. FEBS Lett (2001) 502:53–6. doi:10.1016/S0014-5793(01)02663-1 73. Parsons ME, Ganellin CR. Histamine and its receptors. Br J Pharmacol (2006) 147(Suppl 1):S127–35. doi:10.1038/sj.bjp.0706440 94. Zamolodchikova TS, Vorotyntseva TI, Antonov VK. Duodenase, a new serine protease of unusual specificity from bovine duodenal mucosa. Purification and properties. Eur J Biochem (1995) 227:866–72. doi:10.1111/ j.1432-1033.1995.tb20212.x 74. REFERENCES Cavalcante MC, de Andrade LR, Du Bocage Santos-Pinto C, Straus AH, Takahashi HK, Allodi S, et al. Colocalization of heparin and histamine in the intracellular granules of test cells from the invertebrate Styela plicata (Chordata-Tunicata). J Struct Biol (2002) 137:313–21. doi:10.1016/ S1047-8477(02)00007-2 52. Flanagan JG, Rabbitts TH. Arrangement of human immunoglobulin heavy chain constant region genes implies evolutionary duplication of a segment containing gamma, epsilon and alpha genes. Nature (1982) 300:709–13. doi:10.1038/300709a0 34. Wong GW, Zhuo L, Kimata K, Lam BK, Satoh N, Stevens RL. Ancient origin of mast cells. Biochem Biophys Res Commun (2014) 451:314–8. doi:10.1016/ j.bbrc.2014.07.124 53. Max EE, Battey J, Ney R, Kirsch IR, Leder P. Duplication and deletion in the human immunoglobulin epsilon genes. Cell (1982) 29:691–9. doi:10.1016/ 0092-8674(82)90185-4 35. Dooley H, Flajnik MF. Antibody repertoire development in cartilaginous fish. Dev Comp Immunol (2006) 30:43–56. doi:10.1016/j.dci.2005.06.022 December 2017 \| Volume 8 \| Article 1749 Frontiers in Immunology \| www.frontiersin.org 19 Hellman et al. Origins of IgE, Mast Cells, and Allergies 77. Forsberg E, Pejler G, Ringvall M, Lunderius C, Tomasini-Johansson B, Kusche- Gullberg M, et al. Abnormal mast cells in mice deficient in a heparin- synthesizing enzyme. Nature (1999) 400:773–6. doi:10.1038/23488 54. Battey J, Max EE, McBride WO, Swan D, Leder P. A processed human immu- noglobulin epsilon gene has moved to chromosome 9. Proc Natl Acad Sci U S A (1982) 79:5956–60. doi:10.1073/pnas.79.19.5956 78. Ronnberg E, Melo FR, Pejler G. Mast cell proteoglycans. J Histochem Cytochem (2012) 60:950–62. doi:10.1369/0022155412458927 55. Nimmerjahn F, Ravetch JV. Fcgamma receptors as regulators of immune responses. Nat Rev Immunol (2008) 8:34–47. doi:10.1038/nri2206 56. van der Poel CE, Spaapen RM, van de Winkel JG, Leusen JH. Functional characteristics of the high affinity IgG receptor, FcgammaRI. J Immunol (2011) 186:2699–704. doi:10.4049/jimmunol.1003526 79. Reynolds DS, Stevens RL, Lane WS, Carr MH, Austen KF, Serafin WE. Different mouse mast cell populations express various combinations of at least six distinct mast cell serine proteases. Proc Natl Acad Sci U S A (1990) 87:3230–4. doi:10.1073/pnas.87.8.3230 57. Rivera J, Fierro NA, Olivera A, Suzuki R. New insights on mast cell activa- tion via the high affinity receptor for IgE. Adv Immunol (2008) 98:85–120. doi:10.1016/S0065-2776(08)00403-3 80. Huang RY, Blom T, Hellman L. Cloning and structural analysis of MMCP-1, MMCP-4 and MMCP-5, three mouse mast cell-specific serine proteases. Eur J Immunol (1991) 21:1611–21. doi:10.1002/eji.1830210706 58. Klimovich VB. IgM and its receptors: structural and functional aspects. Biochemistry (Mosc) (2011) 76:534–49. Frontiers in Immunology \| www.frontiersin.org REFERENCES Khanfar MA, Affini A, Lutsenko K, Nikolic K, Butini S, Stark H. Multiple targeting approaches on histamine H3 receptor antagonists. Front Neurosci (2016) 10:201. doi:10.3389/fnins.2016.00201 75. Kjellen L, Pettersson I, Lillhager P, Steen ML, Pettersson U, Lehtonen P, et al. Primary structure of a mouse mastocytoma proteoglycan core protein. Biochem J (1989) 263:105–13. doi:10.1042/bj2630105 95. Caughey GH, Raymond WW, Blount JL, Hau LW, Pallaoro M, Wolters PJ, et al. Characterization of human gamma-tryptases, novel members of the chromosome 16p mast cell tryptase and prostasin gene families. J Immunol (2000) 164:6566–75. doi:10.4049/jimmunol.164.12.6566 76. Angerth T, Huang RY, Aveskogh M, Pettersson I, Kjellen L, Hellman L. Cloning and structural analysis of a gene encoding a mouse mastocytoma proteoglycan core protein; analysis of its evolutionary relation to three cross hybridizing regions in the mouse genome. Gene (1990) 93:235–40. doi:10.1016/0378-1119(90)90230-O 96. Trivedi NN, Tong Q, Raman K, Bhagwandin VJ, Caughey GH. Mast cell alpha and beta tryptases changed rapidly during primate speciation and evolved from gamma-like transmembrane peptidases in ancestral vertebrates. J Immunol (2007) 179:6072–9. doi:10.4049/jimmunol.179.9.6072 December 2017 \| Volume 8 \| Article 1749 Frontiers in Immunology \| www.frontiersin.org 20 Hellman et al. Origins of IgE, Mast Cells, and Allergies 97. Reimer JM, Samollow PB, Hellman L. High degree of conservation of the multigene tryptase locus over the past 150-200 million years of mammalian evolution. Immunogenetics (2010) 62:369–82. doi:10.1007/s00251-010- 0443-2 115. Okumura K, Takai S, Muramatsu M, Katayama S, Sakaguchi M, Kishi K, et al. Human chymase degrades human fibronectin. Clin Chim Acta (2004) 347:223–5. doi:10.1016/j.cccn.2004.04.019 116. Tchougounova E, Lundequist A, Fajardo I, Winberg JO, Abrink M, Pejler G. A key role for mast cell chymase in the activation of pro-matrix metallopro- tease-9 and pro-matrix metalloprotease-2. J Biol Chem (2005) 280:9291–6. doi:10.1074/jbc.M410396200 98. Karlson U, Pejler G, Froman G, Hellman L. Rat mast cell protease 4 is a beta-chymase with unusually stringent substrate recognition profile. J Biol Chem (2002) 277:18579–85. doi:10.1074/jbc.M110356200 99. Huang R, Hellman L. Genes for mast-cell serine protease and their molecular evolution. Immunogenetics (1994) 40:397–414. doi:10.1007/BF00177823 117. Thorpe M, Yu J, Boinapally V, Ahooghalandari P, Kervinen J, Garavilla LD, et al. Extended cleavage specificity of the mast cell chymase from the crab-eating macaque (Macaca fascicularis): an interesting animal model for the analysis of the function of the human mast cell chymase. Int Immunol (2012) 12:771–82. doi:10.1093/intimm/dxs081 100. Lunderius C, Xiang Z, Nilsson G, Hellman L. REFERENCES Iancovici Kidon M, Stein M, Geller-Bernstein C, Weisman Z, Steinberg S, Greenberg Z, et al. Serum immunoglobulin E levels in Israeli-Ethiopian children: environment and genetics. Isr Med Assoc J (2005) 7:799–802. 106. Fu Z, Thorpe M, Alemayehu R, Roy A, Kervinen J, de Garavilla L, et al. Highly selective cleavage of cytokines and chemokines by the human mast cell chymase and neutrophil cathepsin G. J Immunol (2017) 198:1474–83. doi:10.4049/jimmunol.1601223 126. Cahenzli J, Koller Y, Wyss M, Geuking MB, McCoy KD. Intestinal microbial diversity during early-life colonization shapes long-term IgE levels. Cell Host Microbe (2013) 14:559–70. doi:10.1016/j.chom.2013.10.004 h 107. Wernersson S, Reimer JM, Poorafshar M, Karlson U, Wermenstam N, Bengten E, et al. Granzyme-like sequences in bony fish shed light on the emergence of hematopoietic serine proteases during vertebrate evolution. Dev Comp Immunol (2006) 30:901–18. doi:10.1016/j.dci.2005.10.014 127. Yazdanbakhsh M, van den Biggelaar A, Maizels RM. Th2 responses without atopy: immunoregulation in chronic helminth infections and reduced allergic disease. Trends Immunol (2001) 22:372–7. doi:10.1016/S1471- 4906(01)01958-5 108. Thorpe M, Akula S, Hellman L. Channel catfish granzyme-like I is a highly specific serine protease with metase activity that is expressed by fish NK-like cells. Dev Comp Immunol (2016) 63:84–95. doi:10.1016/j.dci. 2016.05.013 128. Ledin A, Bergvall K, Hillbertz NS, Hansson H, Andersson G, Hedhammar A, et al. Generation of therapeutic antibody responses against IgE in dogs, an animal species with exceptionally high plasma IgE levels. Vaccine (2006) 24:66–74. doi:10.1016/j.vaccine.2005.07.052 109. Metz M, Piliponsky AM, Chen CC, Lammel V, Abrink M, Pejler G, et al. Mast cells can enhance resistance to snake and honeybee venoms. Science (2006) 313:526–30. doi:10.1126/science.1128877 129. Ledin A, Arnemo JM, Liberg O, Hellman L. High plasma IgE levels within the Scandinavian wolf population, and its implications for mammalian IgE homeostasis. Mol Immunol (2008) 45:1976–80. doi:10.1016/j.molimm. 2007.10.029 110. Akahoshi M, Song CH, Piliponsky AM, Metz M, Guzzetta A, Abrink M, et al. Mast cell chymase reduces the toxicity of Gila monster venom, scorpion venom, and vasoactive intestinal polypeptide in mice. J Clin Invest (2011) 121:4180–91. doi:10.1172/JCI46139 130. Wagner B, Radbruch A, Rohwer J, Leibold W. Monoclonal anti-equine IgE antibodies with specificity for different epitopes on the immunoglobulin heavy chain of native IgE. Vet Immunol Immunopathol (2003) 92:45–60. doi:10.1016/S0165-2427(03)00007-2 111. Mukai K, Tsai M, Starkl P, Marichal T, Galli SJ. IgE and mast cells in host defense against parasites and venoms. Semin Immunopathol (2016) 38(5):581–603. doi:10.1007/s00281-016-0565-1 131. REFERENCES Murine mast cell lines as indi- cators of early events in mast cell and basophil development. Eur J Immunol (2000) 30:3396–402. doi:10.1002/1521-4141(2000012)30:12<3396::AID- IMMU3396>3.0.CO;2-O 118. Piliponsky AM, Chen CC, Rios EJ, Treuting PM, Lahiri A, Abrink M, et al. The chymase mouse mast cell protease 4 degrades TNF, limits inflammation, and promotes survival in a model of sepsis. Am J Pathol (2012) 181:875–86. doi:10.1016/j.ajpath.2012.05.013 101. Poorafshar M, Helmby H, Troye-Blomberg M, Hellman L. MMCP-8, the first lineage-specific differentiation marker for mouse basophils. Elevated numbers of potent IL-4-producing and MMCP-8-positive cells in spleens of malaria-infected mice. Eur J Immunol (2000) 30:2660–8. doi:10.1002/1521- 4141(200009)30:9<2660::AID-IMMU2660>3.0.CO;2-I 119. Caughey GH, Raymond WW, Wolters PJ. Angiotensin II generation by mast cell alpha- and beta-chymases. Biochim Biophys Acta (2000) 1480:245–57. doi:10.1016/S0167-4838(00)00076-5 102. Karlson U, Pejler G, Tomasini-Johansson B, Hellman L. Extended substrate specificity of rat mast cell protease 5, a rodent alpha-chymase with elastase- like primary specificity. J Biol Chem (2003) 278:39625–31. doi:10.1074/jbc. M301512200 120. Wintroub BU, Schechter NB, Lazarus GS, Kaempfer CE, Schwartz LB. Angiotensin I conversion by human and rat chymotryptic proteinases. J Invest Dermatol (1984) 83:336–9. doi:10.1111/1523-1747.ep12264144 103. Reimer JM, Enoksson M, Samollow PB, Hellman L. Extended substrate specificity of opossum chymase – implications for the origin of mast cell chy- mases. Mol Immunol (2008) 45:2116–25. doi:10.1016/j.molimm.2007.10.015 h 121. Chandrasekharan UM, Sanker S, Glynias MJ, Karnik SS, Husain A. Angio tensin II-forming activity in a reconstructed ancestral chymase. Science (1996) 271:502–5. doi:10.1126/science.271.5248.502 h 104. Fu Z, Thorpe M, Hellman L. rMCP-2, the major rat mucosal mast cell protease, an analysis of its extended cleavage specificity and its potential role in regulating intestinal permeability by the cleavage of cell adhesion and junction proteins. PLoS One (2015) 10:e0131720. doi:10.1371/journal. pone.0131720 122. Pejler G, Karlstrom A. Thrombin is inactivated by mast cell secretory granule chymase. J Biol Chem (1993) 268:11817–22. 123. Pejler G, Soderstrom K, Karlstrom A. Inactivation of thrombin by a com- plex between rat mast-cell protease 1 and heparin proteoglycan. Biochem J (1994) 299(Pt 2):507–13. doi:10.1042/bj2990507 124. Hellman, L. Genes for Immunoglobulin E: Structure, Rearrangement and Expression [Thesis]. Uppsala: Acta Universitatis Upsaliensis (1985). 1–54 p. 105. Fu Z, Thorpe M, Akula S, Hellman L. Asp-ase activity of the opossum granzyme B supports the role of granzyme B as part of anti-viral immunity already during early mammalian evolution. PLoS One (2016) 11:e0154886. doi:10.1371/journal.pone.0154886 125. REFERENCES Lyons ET, Tolliver SC, Kuzmina TA, Collins SS. Critical tests evaluating efficacy of moxidectin against small strongyles in horses from a herd for which reduced activity had been found in field tests in Central Kentucky. Parasitol Res (2010) 107:1495–8. doi:10.1007/s00436-010-2025-5 112. Vartio T, Seppa H, Vaheri A. Susceptibility of soluble and matrix fibronec- tins to degradation by tissue proteinases, mast cell chymase and cathepsin G. J Biol Chem (1981) 256:471–7. 132. Matthews JB. Facing the threat of equine parasitic disease. Equine Vet J (2011) 43:126–32. doi:10.1111/j.2042-3306.2010.00356.x i 113. Tchougounova E, Forsberg E, Angelborg G, Kjellen L, Pejler G. Altered pro- cessing of fibronectin in mice lacking heparin. a role for heparin-dependent mast cell chymase in fibronectin degradation. J Biol Chem (2001) 276:3772–7. doi:10.1074/jbc.M008434200 133. Mitre E, Nutman TB. IgE memory: persistence of antigen-specific IgE responses years after treatment of human filarial infections. J Allergy Clin Immunol (2006) 117:939–45. doi:10.1016/j.jaci.2005.12.1341 114. Tchougounova E, Pejler G, Abrink M. The chymase, mouse mast cell protease 4, constitutes the major chymotrypsin-like activity in peritoneum and ear tissue. A role for mouse mast cell protease 4 in thrombin regula- tion and fibronectin turnover. J Exp Med (2003) 198:423–31. doi:10.1084/ jem.20030671 134. Levy DA, Chen J. Healthy IgE-deficient person. N Engl J Med (1970) 283:541. doi:10.1056/NEJM197009032831018 h 135. Thorburn AN, Macia L, Mackay CR. Diet, metabolites, and “western- lifestyle” inflammatory diseases. Immunity (2014) 40:833–42. doi:10.1016/ j.immuni.2014.05.014 December 2017 \| Volume 8 \| Article 1749 Frontiers in Immunology \| www.frontiersin.org 21 Origins of IgE, Mast Cells, and Allergies Hellman et al. efficiently break self-tolerance – a key issue in the development of therapeutic vaccines. Vaccine (2009) 28:48–52. doi:10.1016/j.vaccine.2009.09.122 efficiently break self-tolerance – a key issue in the development of therapeutic vaccines. Vaccine (2009) 28:48–52. doi:10.1016/j.vaccine.2009.09.122 136. Daien CI, Pinget GV, Tan JK, Macia L. Detrimental impact of microbiota- accessible carbohydrate-deprived diet on gut and immune homeostasis: an overview. Front Immunol (2017) 8:548. doi:10.3389/fimmu.2017.00548 147. Saupe F, Huijbers EJ, Hein T, Femel J, Cedervall J, Olsson AK, et al. Vaccines targeting self-antigens: mechanisms and efficacy-determining parameters. FASEB J (2015) 29:3253–62. doi:10.1096/fj.15-271502 i 137. Hill DA, Siracusa MC, Abt MC, Kim BS, Kobuley D, Kubo M, et al. Com mensal bacteria-derived signals regulate basophil hematopoiesis and allergic inflammation. Nat Med (2012) 18:538–46. doi:10.1038/nm.2657 j 148. Johansson J, Hellman L. REFERENCES Modifications increasing the efficacy of recombi- nant vaccines; marked increase in antibody titers with moderately repet- itive variants of a therapeutic allergy vaccine. Vaccine (2007) 25:1676–82. doi:10.1016/j.vaccine.2006.10.055 l 138. McCoy KD, Koller Y. New developments providing mechanistic insight into the impact of the microbiota on allergic disease. Clin Immunol (2015) 159:170–6. doi:10.1016/j.clim.2015.05.007 139. Hong SW, Kim KS, Surh CD. Beyond hygiene: commensal microbiota and allergic diseases. Immune Netw (2017) 17:48–59. doi:10.4110/in.2017.17.1.48 l 149. Lechner F, Jegerlehner A, Tissot AC, Maurer P, Sebbel P, Renner WA, et al. Virus-like particles as a modular system for novel vaccines. Intervirology (2002) 45:212–7. doi:10.1159/000067912 140. Jatzlauk G, Bartel S, Heine H, Schloter M, Krauss-Etschmann S. Influences of environmental bacteria and their metabolites on allergies, asthma, and host microbiota. Allergy (2017) 72(12):1859–67. doi:10.1111/all.13220 150. Saupe F, Reichel M, Huijbers EJ, Femel J, Markgren PO, Andersson CE, et al. Development of a novel therapeutic vaccine carrier that sustains high antibody titers against several targets simultaneously. FASEB J (2017) 31:1204–14. doi:10.1096/fj.201600820R 141. Abrahamsson TR, Jakobsson HE, Andersson AF, Bjorksten B, Engstrand L, Jenmalm MC. Low diversity of the gut microbiota in infants with atopic eczema. J Allergy Clin Immunol (2012) 129:434–40, 440.e1–2. doi:10.1016/ j.jaci.2011.10.025 Conflict of Interest Statement: The patents for the VLR-B sequence used to enhance immunogenicity to therapeutic vaccines targeting self-molecules is owned by a company owned by LH. All other authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. 142. Abrahamsson TR, Jakobsson HE, Andersson AF, Bjorksten B, Engstrand L, Jenmalm MC. Low gut microbiota diversity in early infancy precedes asthma at school age. Clin Exp Allergy (2014) 44:842–50. doi:10.1111/cea.12253 143. Dulek DE, Peebles RS Jr. Viruses and asthma. Biochim Biophys Acta (2011) 1810:1080–90. doi:10.1016/j.bbagen.2011.01.012 144. Hellman L. Therapeutic vaccines against IgE-mediated allergies. Expert Rev Vaccines (2008) 7:193–208. doi:10.1586/14760584.7.2.193 Copyright © 2017 Hellman, Akula, Thorpe and Fu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. 145. Johansson J, Ledin A, Vernersson M, Lovgren-Bengtsson K, Hellman L. December 2017 \| Volume 8 \| Article 1749 REFERENCES Identification of adjuvants that enhance the therapeutic antibody response to host IgE. Vaccine (2004) 22:2873–80. doi:10.1016/j.vaccine.2003.12.029 146. Ringvall M, Huijbers EJ, Ahooghalandari P, Alekseeva L, Andronova T, Olsson AK, et al. Identification of potent biodegradable adjuvants that December 2017 \| Volume 8 \| Article 1749 Frontiers in Immunology \| www.frontiersin.org 22
https://openalex.org/W1986692340	https://europepmc.org/articles/pmc4029939?pdf=render	English	null	Radiographic signs for detection of femoroacetabular impingement and hip dysplasia should be carefully used in patients with osteoarthritis of the hip	BMC musculoskeletal disorders	2,014	cc-by	5,326	RESEARCH ARTICLE Open Access * Correspondence: IngmarIpach@gmx.de 1Department of Orthopaedic Surgery, University Hospital of Tuebingen, Hoppe-Seyler-Straße 3, 72074 Tuebingen, Germany 2Department of Orthopedic Surgery, Hospital of Ingolstadt, Krumenauerstraße 25, 85021 Ingolstadt, Germany Full list of author information is available at the end of the article Ipach et al. BMC Musculoskeletal Disorders 2014, 15:150 http://www.biomedcentral.com/1471-2474/15/150 Ipach et al. BMC Musculoskeletal Disorders 2014, 15:150 http://www.biomedcentral.com/1471-2474/15/150 Radiographic signs for detection of femoroacetabular impingement and hip dysplasia should be carefully used in patients with osteoarthritis of the hip Ingmar Ipach1,2, Ina-Christine Rondak3, Saskia Sachsenmaier1, Elisabeth Buck1, Roland Syha4 and Falk Mittag1 Ingmar Ipach1,2, Ina-Christine Rondak3, Saskia Sachsenmaier1, Elisabeth Buck1, Roland Syha4 an 2*, Ina-Christine Rondak3, Saskia Sachsenmaier1, Elisabeth Buck1, Roland Syha4 and Falk Mittag1 Abstract Background: During the last years, terms like acetabular retroversion, excessive overcoverage, and abnormal head-neck-junction with the so called “pistol-grip-deformity” has been added to the classical description of hip dysplasia. These anatomical changes could lead to a femoroacetabular impingement (FAI). Both kinds of FAI has been indentified as a main reason for hip pain and progressive degenerative changes leading to early osteoarthritis of the hip. A lot of radiographic criteria on pelvic views have been established to detect classical dysplasia and FAI. The present study was initiated to assess the hypothesis that age and severity of osteoarthritis affect measurements of different radiographic parameters. Methods: The pelvic radiographs of 1614 patients were measured for head-ratio, CE-angle, roof obliquity, extrusion-index, depth-to-width ratio, CCD-angle, sharp’s angle. To evaluate the severity of osteoarthritis of the hip the classification by Kellgren and Lawrence was used. Associations between age and radiographic parameters or severity of osteoarthritis were assessed by Spearman’s (ρ) or Kendall’s (r) rank correlation coefficient, respectively. Methods: The pelvic radiographs of 1614 patients were measured for head-ratio, CE-angle, roof obliquity, extrusion-index, depth-to-width ratio, CCD-angle, sharp’s angle. To evaluate the severity of osteoarthritis of the hip the classification by Kellgren and Lawrence was used. Associations between age and radiographic parameters or severity of osteoarthritis were assessed by Spearman’s (ρ) or Kendall’s (r) rank correlation coefficient, respectively. Results: 366 (22.7%) patients presented no sign of osteoarthritis, 367 (22.7%) patients presented I° osteoarthritis, 460 (28.5%) patients presented II° osteoarthritis, 307 (19%) III° osteoarthritis and 114 (7.1%) IV° osteoarthritis of the hip. The mean head-ratio of all patients was 1.13 ± 0.26 (0.76 – 2.40), the mean CE-angle 40.05° ± 10.13° (0° - 70°), the mean roof obliquity was 35.27°± 4.96° (10° – 55°), the mean extrusion-index was 12.99 ± 9.21 (6.20 – 95.2), the mean depth-to-width ratio was 59.30 ± 8.90 (6.30 – 100), the mean CCD-angle was 127.68° ± 7.22° (123° – 162°) and the mean sharp’s angle was 9.75° ± 5.40° (1° - 34°) There was a weak association between age and the severity of osteoarthritis of the hips (left: r = 0.291; right: r = 0.275; both P < 0.001) with higher osteoarthritis levels observable for elderly patients). Results: 366 (22.7%) patients presented no sign of osteoarthritis, 367 (22.7%) patients presented I° osteoarthritis, 460 (28.5%) patients presented II° osteoarthritis, 307 (19%) III° osteoarthritis and 114 (7.1%) IV° osteoarthritis of the hip. Background the symphysis was 32 mm for men and 47 mm for women and the teardrop sign appeared symmetrical [1]. During the last years, terms like acetabular retroversion, excessive overcoverage, and abnormal head-neck-junction with the so called “pistol-grip-deformity” [1-4] has been added to the classical description of hip dysplasia with a lateral CEA-angle of less then 25° and a roof obliquity angle of more than 10° [5]. These anatomical changes could lead to a femoroacetabular impingement (FAI). Two kinds of FAI have been described. In Cam impingement a repetitive contact between an abnormal head-neck- junction and the acetabular rim causes cartilage damage in the anterosuperior area of the acetabulum. In Pincer impingement a direct contact between the femoral neck and a local/generalized overcovered acetabulum leads also to repetitive damage of the cartilage at the acetabu- lar rim [6-16]. The pelvic-views were measured for head-ratio, CE- angle, roof obliquity, extrusion-index, depth-to-width ra- tio, CCD-angle, sharp’s angle [8,17-20,22,26]. In cases with the presents of a THA, fracture or Dysplasia Crowne II/IV on one side, only the other was measured. To evaluate the severity of osteoarthritis of the hip the classification by Kellgren and Lawrence was used [25]. Methods All reported P values are two-tailed, with a P value of 0.05 indicating statistical significance and have not been adjusted for multiple testing. Ethical approval has been received by the ethic commit- tee of Tübingen (025/2014R). We analysed our data bank for all pelvic-views which has been performed in our in- stitution in the period between 1st January 2006 and 31st December 2011. Abstract The mean head-ratio of all patients was 1.13 ± 0.26 (0.76 – 2.40), the mean CE-angle 40.05° ± 10.13° (0° - 70°), the mean roof obliquity was 35.27°± 4.96° (10° – 55°), the mean extrusion-index was 12.99 ± 9.21 (6.20 – 95.2), the mean depth-to-width ratio was 59.30 ± 8.90 (6.30 – 100), the mean CCD-angle was 127.68° ± 7.22° (123° – 162°) and the mean sharp’s angle was 9.75° ± 5.40° (1° - 34°) There was a weak association between age and the severity of osteoarthritis of the hips (left: r = 0.291; right: r = 0.275; both P < 0.001) with higher osteoarthritis levels observable for elderly patients). Conclusion: Severity of osteoarthritis has a negative impact on measurements of different radiographic parameters. Therefore - in our opinion - epidemiological studies on prearthrotic deformities should only be performed in healthy adults with no signs of osteoarthritic changes. Keywords: Radiographic Signs, Impingement, Hip, Dysplasia, Osteoarthritis © 2014 Ipach et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Ipach et al. BMC Musculoskeletal Disorders 2014, 15:150 http://www.biomedcentral.com/1471-2474/15/150 Ipach et al. BMC Musculoskeletal Disorders 2014, 15:150 http://www.biomedcentral.com/1471-2474/15/150 Page 2 of 7 Statistical analysis Statistical analyses were conducted with the use of IBM SPSS Statistics 21 and R software, version 3.0.0 (R Core Team). In order to account for repeated measurements of both hips on patient level, analysis was conducted separately for left and right hips or using summarized values for both sides (i.e. mean or maximum values of both hips). Both kinds of FAI has been indentified as a main rea- son for hip pain and progressive degenerative changes leading to early osteoarthritis of the hip [1-4,6,7]. When analyzing the influence of age on severity of osteoarthritis different radiographic parameters, left and right hip were analyzed separately. The diagnosis of FAI should be based on detailed physical examination and appropriate imaging studies. A lot of radiographic criteria on pelvic views have been established to detect classical dysplasia and FAI [17-21]. The head-ratio has already described as reliable for the detection of “pistol-grip-deformity” [7,8,22]. It has also been shown that radiographic FAI findings are very common in a population of healthy young adults [23]. When analyzing the influence of severity of osteoarth- ritis on the different radiographic parameters, the most severe hip and corresponding measurements or in case of equal severity on both sides, averaged measurements over both hips were considered for each patient. Categorical variables are presented as frequencies, per- centages, and continuous variables as means and standard deviations, or medians and interquartile ranges (Box- plots) for variables with skewed distributions. Goodman et al. have shown that “pistol-grip-deformity” is due to a three-dimensional structural abnormality with no change in severity with age [24]. According to Kellgren and Lawrence [25] progressive de- generative changes lead to osteophytes, narrowing of the joint space, and deformity of the bone ends. These changes may have a negative impact on radiographic parameters. The present study was initiated to assess the influence of severity of osteoarthritis on radiographic parameters. Associations between age and radiographic parameters or severity of osteoarthritis were assessed by Spearman’s (ρ) or Kendall’s (τ) rank correlation coefficient, respect- ively. Group comparisons based on severity of osteoarth- ritis of continuous measurements were conducted using either the one-way analysis of variances (ANOVA) or the Kruskall-Wallis test, as appropriate. Results The pelvic radiographs of 1614 Patients were included into this study. In 1052 of all cases both sides were measured in 562 cases only the left (n = 311) or right side (n = 251) were measured for radiographic signs of FAI and dysplasia. To exclude the negative influence of pelvic tilt and rotation on radiographic parameters, only standardized pelvic radiographs were included in this study. The dis- tance between the tip of the coccyx and the middle of Table 1 Measurement values for the different radiographic parameters of both hips n= Sharps-angle (in degree) Head-ratio CE-angle (in degree) Roof obliquity (in degree) Extrusion-index (in %) Depth-to-width ratio (in %) CCD-angle (in degree) 1614 9.75° ± 5.40° (1° - 34°) 1.13 ± 0.26 (0.76 – 2.40) 40.05° ± 10.13° (0° - 70°) 35.27 ± 4.96 (10 – 55) 12.99 ± 9.21 (6.20 – 95.2) 59.30 ± 8.90 (6.30 – 100) 127.68° ± 7.22° (123° – 162°) The results are presented as means and standard deviations (Minimum – Maximum). Table 1 Measurement values for the different radiographic parameters of both hips n= Sharps-angle (in degree) Head-ratio CE-angle (in degree) Roof obliquity (in degree) Extrusion-index (in %) Depth-to-width ratio (in %) CCD-angle (in degree) 1614 9.75° ± 5.40° (1° - 34°) 1.13 ± 0.26 (0.76 – 2.40) 40.05° ± 10.13° (0° - 70°) 35.27 ± 4.96 (10 – 55) 12.99 ± 9.21 (6.20 – 95.2) 59.30 ± 8.90 (6.30 – 100) 127.68° ± 7.22° (123° – 162°) The results are presented as means and standard deviations (Minimum – Maximum). Table 1 Measurement values for the different radiographic parameters of both hips The results are presented as means and standard deviations (Minimum – Maximum). Page 3 of 7 Ipach et al. BMC Musculoskeletal Disorders 2014, 15:150 http://www.biomedcentral.com/1471-2474/15/150 Page 3 of 7 Figure 1 Association between severity of osteoarthritis and age. Y-axis: age in years, X-axis: severity of osteoarthritis according to Kellgren-Lawrence (Kendall’s correlation coefficient for left hips: r = 0.291; and right hips: r = 0.275; both p < 0.001). The mean age of all patients was 60.2 years ± 17.1. 44.4% of all patients were male, 55.6% were female. 366 (22.7%) patients presented no sign of osteoarth- ritis, 367 (22.7%) patients presented I° osteoarthritis, 460 (28.5%) patients presented II° osteoarthritis, 307 (19%) III° osteoarthritis and 114 (7.1%) patients presented IV° osteoarthritis of the hip. Results The mean values of all different radiographic parameters are demonstrated in Table 1. There was a weak association between age and the severity of osteoarthritis of the hips (left: τ = 0.291; right: τ = 0.275; both P < 0.001) with higher osteoarthritis levels observable for elderly patients (Figure 1). The correlation between age and head-ratio, CE-angle, roof obliquity, extrusion-index, depth-to-width ratio, CCD- angle, sharp’s angle is demonstrated in Figures 2 and 3. Weak monotone associations were also observed between age and mean CE-angle (ρ = 0.334) (p < 0.001), mean sharps angle (ρ = -0.299) (p < 0.001), mean extrusion-index (ρ = -0.218) (p < 0.001), and mean CCD-angle (ρ = -0.205) (p < 0.001). On the other hand no correlations were seen between age and depth-to-width ratio (ρ = 0.033) (p < 0.181), roof obliquity (ρ = 0.133) (p < 0.001) and head- ratio (ρ = 0.135) (p < 0.001). These results imply that age Figure 1 Association between severity of osteoarthritis and age. Y-axis: age in years, X-axis: severity of osteoarthritis according to Kellgren-Lawrence (Kendall’s correlation coefficient for left hips: r = 0.291; and right hips: r = 0.275; both p < 0.001). Figure 1 Association between severity of osteoarthritis and age. Y-axis: age in years, X-axis: severity of osteoarthritis according to Kellgren-Lawrence (Kendall’s correlation coefficient for left hips: r = 0.291; and right hips: r = 0.275; both p < 0.001). Figure 2 Correlation between CE-angle, roof-obliquity, depth-to-width-ratio, extrusion-index and age. a: Correlation between age and CE-angle. X-axis: age in years, Y-axis: CE-angle in degree. There was a weak monotone correlation between age and CE-angle (ρ = 0.334). b: Correlation between age and roof-obiquity. X-axis: age in years, Y-axis: roof obliquity in degree. There no correlation between age and roof obliquity (ρ = -0.133). c: Correlation between age and depth-to-width-ratio. X-axis: age in years, Y-axis: depth-to-width-ratio in %. There was no correlation between age and depth-to-width-ratio (ρ = 0.033) d: Correlation between age and extrusion-index. X-axis: age in years, Y-axis: extrusion-index in %. There was weak negative correlation between age and extrusion-index (ρ = -0.218). Figure 2 Correlation between CE-angle, roof-obliquity, depth-to-width-ratio, extrusion-index and age. a: Correlation between age and CE-angle. X-axis: age in years, Y-axis: CE-angle in degree. There was a weak monotone correlation between age and CE-angle (ρ = 0.334). b: Correlation between age and roof-obiquity. Results has only a weak or no impact on the different radiographic parameters for FAI and dysplasia. One-way ANOVA re- vealed statistical significant differences between patient groups based on severity (Kellgren-Lawrence 1°-4°) of osteoarthritis for the following measurements: CE-angle (p < 0.001), sharps-angle (p < 0.001), Extrusion-index (p < 0.001), and CCD-angle (p < 0.001). No statistical significant difference between patient groups based on severity (Kellgren-Lawrence 1°-4°) and depth-to-width ratio was seen (p = 0.535). Assuming a skewed distribu- tion of head-ratio and roof-obliquity measurements, the nonparametric Kruskall-Wallis-test revealed a statistical significant difference between osteoarthritis groups head- ratio (P < 0.001) but none for roof-obliquity (Figure 4) (P = 0.18). These results imply that there is an association between the different radiographic parameters (excluded roof-obliquity and depth-to-width ratio) and the severity of osteoarthritis of the hip. overcoverage or an acetabular retroversion or changes in the head-neck-region lead to the so-called Pincer- or Cam-impingement [6-18]. However, the “classic” definition of hip-dysplasia with an undercoverage of the femoral head (reduced CE-angle) or a steep acetabular roof (increased roof obliquity) is still playing an important role in the indication of total hip arthroplasty in young adults [26-30]. While watching different epidemiological studies, an abnormal hip morphology with acetabular dysplasia has been reported in about 51- 80% of all cases with OA of the hip [23]. Including radiographic findings for an ex- cessive overcoverage of the femoral head, acetabular retroversion or an abnormal head-neck-junction, as a dysplastic change, hip-dysplasia was seen in nearly 97% of all patients [26]. The prevalence of acetabular dysplasia has been re- ported to be about 3.8% in the British population [28] and 4.5% in Chinese men [29]. An abnormal head-neck- junction seems to be present in about 40%-50% of all hips with OA [26,30]. A recent study has shown a high preva- lence of radiographic finding for FAI in a cohort of 2081 healthy adults. The prevalence of CAM-impingement was up to 40%, the prevalence of Pincer-impingement up to 50% [23]. Results X-axis: age in years, Y-axis: roof obliquity in degree. There no correlation between age and roof obliquity (ρ = -0.133). c: Correlation between age and depth-to-width-ratio. X-axis: age in years, Y-axis: depth-to-width-ratio in %. There was no correlation between age and depth-to-width-ratio (ρ = 0.033) d: Correlation between age and extrusion-index. X-axis: age in years, Y-axis: extrusion-index in %. There was weak negative correlation between age and extrusion-index (ρ = -0.218). Ipach et al. BMC Musculoskeletal Disorders 2014, 15:150 http://www.biomedcentral.com/1471-2474/15/150 Page 4 of 7 Figure 3 Correlation between sharps-angle, CCD-angle, head-ratio and age. a: Correlation between age and sharps-angle. X-axis: age in years, Y-axis: sharps-angle in degree. There was weak negative correlation between age and sharps-angle (ρ = -0.299) b: Correlation between age and CCD-angle. X-axis: age in years, Y-axis: CCD-angle in degree. There was weak negative correlation between age and CCD-angle (ρ = -0.205) c: Correlation between age and head-ratio. X-axis: age in years, Y-axis: head-ratio There was no correlation between age and head-ratio (ρ = 0.135). Figure 3 Correlation between sharps-angle, CCD-angle, head-ratio and age. a: Correlation between age and sharps-angle. X-axis: age in years, Y-axis: sharps-angle in degree. There was weak negative correlation between age and sharps-angle (ρ = -0.299) b: Correlation between age and CCD-angle. X-axis: age in years, Y-axis: CCD-angle in degree. There was weak negative correlation between age and CCD-angle (ρ = -0.205) c: Correlation between age and head-ratio. X-axis: age in years, Y-axis: head-ratio There was no correlation between age and head-ratio (ρ = 0.135). has only a weak or no impact on the different radiographic parameters for FAI and dysplasia. One-way ANOVA re- vealed statistical significant differences between patient groups based on severity (Kellgren-Lawrence 1°-4°) of osteoarthritis for the following measurements: CE-angle (p < 0.001), sharps-angle (p < 0.001), Extrusion-index (p < 0.001), and CCD-angle (p < 0.001). No statistical significant difference between patient groups based on severity (Kellgren-Lawrence 1°-4°) and depth-to-width ratio was seen (p = 0.535). Assuming a skewed distribu- tion of head-ratio and roof-obliquity measurements, the nonparametric Kruskall-Wallis-test revealed a statistical significant difference between osteoarthritis groups head- ratio (P < 0.001) but none for roof-obliquity (Figure 4) (P = 0.18). These results imply that there is an association between the different radiographic parameters (excluded roof-obliquity and depth-to-width ratio) and the severity of osteoarthritis of the hip. Discussion During the last years a lot of studies have shown changes in the acetabular geometry and the head neck region as a main reason for an early osteoarthritis of the hip. Terms like femoroacetabular impingement have become more and more important in the explanation for hip pain and the early development of OA. These morphologic changes in the acetabular geometry such as an excessive The present study was performed to assess the impact of severity of osteoarthritis on different radiographic Page 5 of 7 Ipach et al. BMC Musculoskeletal Disorders 2014, 15:150 http://www.biomedcentral.com/1471-2474/15/150 Figure 4 Association between severity of osteoarthritis and radiographic parameters. a-g: Patient based measurement values of CE-angle, sharps-angle, Extrusion-index, depth-to-width ration, CCD-angle, head-ratio and roof-oblique by severity of osteoarthritis of the hip (Kellgren-Lawrence 0-4) [most severe hip and corresponding measurements or in case of equal severity on both sides, averaged measurements over both hips were considered]. a: X-axis: group 0-4 according to severity of osteoarthritis, Y-axis: CE-angle in degree. There was a statistical significant difference in the CE-angle between the groups (p < 0.001). b: X-axis: group 0-4 according to severity of osteoarthritis Y-axis: sharp’s angle in degree. There was a statistical significant difference in the sharp’s angle between the groups (p < 0.001) c: X-axis: group 0-4 according to severity of osteoarthritis Y-axis: depth-to-width-ratio in%. No statistical significant difference between the groups was seen (p = 0.535). d: X-axis: group 0-4 according to severity of osteoarthritis Y-axis: extrusion-index in%. There was a statistical significant difference in the extrusion-index between the groups (p < 0.001) e: X-axis: group 0-4 according to severity of osteoarthritis, Y-axis: roof-obliquity in degree. No statistical significant difference between the groups was seen (p = 0.18). f: X-axis: group 0-4 according to severity of osteoarthritis Y-axis: CCD-angle in degree. There was a statistical significant difference in the CCD-angle between the groups (p < 0.001). g: X-axis: group 0-4 according to severity of osteoarthritis Y-axis: head-ratio. There was a statistical significant difference in the head-ratio between the groups (p < 0.001). Figure 4 Association between severity of osteoarthritis and radiographic parameters. a-g: Patient based measurement values of CE-angle, sharps-angle, Extrusion-index, depth-to-width ration, CCD-angle, head-ratio and roof-oblique by severity of osteoarthritis of the hip (Kellgren-Lawrence 0-4) [most severe hip and corresponding measurements or in case of equal severity on both sides, averaged measurements over both hips were considered]. Discussion a: X-axis: group 0-4 according to severity of osteoarthritis, Y-axis: CE-angle in degree. There was a statistical significant difference in the CE-angle between the groups (p < 0.001). b: X-axis: group 0-4 according to severity of osteoarthritis Y-axis: sharp’s angle in degree. There was a statistical significant difference in the sharp’s angle between the groups (p < 0.001) c: X-axis: group 0-4 according to severity of osteoarthritis Y-axis: depth-to-width-ratio in%. No statistical significant difference between the groups was seen (p = 0.535). d: X-axis: group 0-4 according to severity of osteoarthritis Y-axis: extrusion-index in%. There was a statistical significant difference in the extrusion-index between the groups (p < 0.001) e: X-axis: group 0-4 according to severity of osteoarthritis, Y-axis: roof-obliquity in degree. No statistical significant difference between the groups was seen (p = 0.18). f: X-axis: group 0-4 according to severity of osteoarthritis Y-axis: CCD-angle in degree. There was a statistical significant difference in the CCD-angle between the groups (p < 0.001). g: X-axis: group 0-4 according to severity of osteoarthritis Y-axis: head-ratio. There was a statistical significant difference in the head-ratio between the groups (p < 0.001). measurement values. We were able to show that degen- erative changes has an negative impact on radiographic parameters for FAI and hip dysplasia. To our knowledge this was the first study focusing on this topic. with the severity of osteoarthritis. On the other hand it was shown that this deformity did not change with age and therefore it could be seen as a prearthrotic deformity and not as secondary to degeneration [24]. The posterior head tilt in osteoarthritic hips has been discussed as being an acquired deformity created by the formation of osteophytes [8]. The present study support these findings, as the head-ratio demonstrated an increase We not several limitations in this study. A testing on intra- and interobserver reliability hasn’t be performed in this study, but previous studies already focussed on this topic [31]. Clohisy et al. [32] and Gosving et al. [33] Ipach et al. BMC Musculoskeletal Disorders 2014, 15:150 http://www.biomedcentral.com/1471-2474/15/150 Ipach et al. BMC Musculoskeletal Disorders 2014, 15:150 http://www.biomedcentral.com/1471-2474/15/150 Page 6 of 7 reported about poor results for inter- and intra-observer reliability of measurements of radiographic signs for ace- tabular dysplasia, head-neck offset and pelvic tilt. Other studies were able to show a high inter- and intraobserver reliability for different radiographic parameters by using the Balnd-Altman-method [26,31,34,35]. Discussion The discrepancy between these studies could be explained by the using of different statistical methods. The kappa-coefficient by Cohen [34] is the first choice for testing inter- and intraobserver quality of categorical variables, it should not be used for data on a continuous level [34,35]. Correlation coefficients for testing reliability between two observers should not be used at all [34]. Therefore the Bland- Altman-method is a better choice for assessing agreement of continuous data [34,35]. 3. Tannast M, Kubiak-Langer M, Langlotz F, Puls M, Murphy SB, Siebenrock KA: Non invasive threedimensional assessment of femoroacetabular impingement. J Orthop Res 2007, 25:122–131. 4. Siebenrock KA, Schoeninger R, Ganz R: Anterior femoro-acetabular impingement due to acetabular retroversion: treatment with periacetabular osteotomy. J Bone Joint Surg Am 2003, 5(A):278–286. 5. Miller MD: Review of Orthopaedics 5th edition. Philadelphia: Saunders Elsevier; 2008. 6. Ganz R, Parvizi J, Beck M, Leunig M, Nötzli H, Siebenrock KA: Femoroacetabular impingement: a cause for osteoarthritis of the hip. Clin Orthop Relat Res 2003, 417:112–120. Femoroacetabular impingement: a cause for osteoarthritis of the hip. Clin Orthop Relat Res 2003, 417:112–120. 7. Nötzli HP, Wyss TF, Stoecklin CH, Schmid MR, Treiber K, Hodler J: The contour of the femoral head-neck junction as a predictor for the risk of anterior impingement. J Bone Joint Surg (Br) 2002, 84:556–560. 8. Murray RO: The aetiology of primary osteoarthritis of the hip. Br J Radiol 7. Nötzli HP, Wyss TF, Stoecklin CH, Schmid MR, Treiber K, Hodler J: The contour of the femoral head-neck junction as a predictor for the risk of anterior impingement. J Bone Joint Surg (Br) 2002, 84:556–560. 8. Murray RO: The aetiology of primary osteoarthritis of the hip. Br J Radiol 1965, 38:810–824. 9. Jäger M, Wild A, Westhoff B: Femoroacetabular impingement caused by a femoral osseous head-neck bump deformity: clinical, radiological, and experimental results. J Orthop Sci 2004, 9:256–263. 10. Koegh MJ, Batt ME: A review of femoroacetabular impingement in athletes. Sports Med 2008, 38:863–878. 11. Stafford G, Witt J: The anatomy, diagnosis and pathology of femoroacetabular impingement. Br J Hosp Med 2009, 70:72–77. 10. Koegh MJ, Batt ME: A review of femoroacetabular impingement in athletes. Sports Med 2008, 38:863–878. 10. Koegh MJ, Batt ME: A review of femoroacetabular impingement in athletes. Sports Med 2008, 38:863–878. Discussion Nevertheless the inter- and intrabserver quality of radiographic findings in the diagnostic of dysplasia, FAI and excessive overcoverage is one of the main problem in the clinical routine. Therefore the use of a three- dimensional imaging might be helpful in unclear cases. 11. Stafford G, Witt J: The anatomy, diagnosis and pathology of femoroacetabular impingement. Br J Hosp Med 2009, 70:72–77. 11. Stafford G, Witt J: The anatomy, diagnosis and pathology of femoroacetabular impingement. Br J Hosp Med 2009, 70:72–77. 12. Anda S, Terjesen T, Kvistad KA, Svenningsen S: Acetabular angles and femoral anteversion in dysplastic hips in adults: CT-investigation. J Comput Assist Tomogr 1991, 15(1):115–120. 12. Anda S, Terjesen T, Kvistad KA, Svenningsen S: Acetabular angles and femoral anteversion in dysplastic hips in adults: CT-investigation. J Comput Assist Tomogr 1991, 15(1):115–120. 13. Anderson LA, Peters CL, Park BB, Stoddard GJ, Erickson JA, Crim JR: Acetabular cartilage delamination in femoroacetabular impingement. Risk factors and Magnetic Resonance imaging diagnosis. J Bone Joint Surg Am 2009, 91:305–313. Competing interests Competing interests The authors declare that they have no competing interests. 20. Tönnis D, Heinecke A: Acetabular and femoral anteversion: relationship with osteoarthritis of the hip. J Bone Joint Surg Am 1999, 81(12):1747–1770. Author details 1 1Department of Orthopaedic Surgery, University Hospital of Tuebingen, Hoppe-Seyler-Straße 3, 72074 Tuebingen, Germany. 2Department of Orthopedic Surgery, Hospital of Ingolstadt, Krumenauerstraße 25, 85021 Ingolstadt, Germany. 3Department of Medical Statistics and Epidemiology, Technische Universität München, Ismaninger Str. 22, Munich 81675, Germany. 4Department of Radiology, University Hospital of Tuebingen, 23. Laborie LB, Lehmann TG, Engeæter I, Eastwood DM, Engeæter LB, Rosendahl K: Prevalence or radiographic findings thought to be associated with femoroacetabular impingement in a population-based cohort of 2081 healthy young adults. Radiology 2011, 260:494–502. 24. Goodman DA, Feighan JE, Smith AD, Latimer B, Buly RL, Cooperman DR: Subclinical slipped capital femoral epiphysis. Relationship to osteoarthrosis of the hip. J Bone Joint Surg Am 1997, 79(10):1489–1497. 25. Kellgren JH, Lawrence JS: Radiological assessment of osteo-arthrosis. Ann Rheum Dis 1957, 16:494–502. Received: 10 January 2014 Accepted: 30 April 2014 Published: 8 May 2014 Received: 10 January 2014 Accepted: 30 April 2014 Published: 8 May 2014 26. Ipach I, Mittag F, Syha R, Kunze B, Wolf P, Kluba T: Indications for total hip arthroplasty in young adults - idiopathic osteoarthritis seems to be overestimated. Röfo 2012, 184(3):239–247. Authors’ contributions f h 21. Wiberg G: Studies on dysplastic acetabula and congenital subluxation of the hip joint: With special reference to the complications of osteoarthritis. Acta Chir Scan 1939, 58:7–38. II: Drafting the manuscript IR: statistical analysis SS: graphical design, recruiting x-rays EB: performing measurements RS: controlling accuracy of measurements and quality of x-rays FM: Study design. All authors read and approved the final manuscript. 22. Ipach I, Mittag F, Sachsenmaier S, Heinrich P, Kluba T: A new classification for “pistol grip deformity”- correlation between the severity of the deformity and the grade of osteoarthritis of the hip. Fortschr Röntgenstrahl 2010, 183(4):365–371. Conclusion There seems to be a negative impact of severity of osteo- arthritis on different radiographic parameters for the detec- tion of femoroacetabular impingement and hip dysplasia. Therefore - in our opinion - epidemiological studies on prearthrotic deformities and radiographic paramters for FAI and hip dysplasia should only be performed in adults with no signs of osteoarthritic changes. 14. Fadul DA, Carrino JA: Imaging of femoroacetabular impingement. J Bone Joint Surg Am 2009, 91:138–143. 14. Fadul DA, Carrino JA: Imaging of femoroacetabular impingement. J Bone Joint Surg Am 2009, 91:138–143. 15. Murphy SB, Ganz R, Müller ME: The prognosis in untreated dysplasia of the hip. A study of radiographic factors that predict the outcome. J Bone Joint Surg Am 1995, 77(7):985–989. 15. Murphy SB, Ganz R, Müller ME: The prognosis in untreated dysplasia of the hip. A study of radiographic factors that predict the outcome. J Bone Joint Surg Am 1995, 77(7):985–989. 16. Laude F, Boyer T, Nogier A: Anterior femoroacetabular impingement. Joint Bone Spine 2007, 74:127–132. 16. Laude F, Boyer T, Nogier A: Anterior femoroacetabular impingement. Joint Bone Spine 2007, 74:127–132. 17. Lequesne M: Coxometry measurement of the basic angles of the adult radiographic hip by a combined protractor. Rev Rhum Mal Osteoartic 1963, 30:479–485. 17. Lequesne M: Coxometry measurement of the basic angles of the adult radiographic hip by a combined protractor. Rev Rhum Mal Osteoartic 1963, 30:479–485. Abbreviations 18. Tannast M, Siebenrock KA, Anderson SE: Femoroacetabular Impingement: radiographic diagnosis – what the radiologist should know. Am J Roentgenol 2007, 188(6):1540–1552. 18. Tannast M, Siebenrock KA, Anderson SE: Femoroacetabular Impingement: radiographic diagnosis – what the radiologist should know. Am J Roentgenol 2007, 188(6):1540–1552. CE-angle: Centre-edge-angle; CCD-angle: Caput-collum-diaphysis-angle; FAI: Femoroacetabular impingement; OA: Osteoarthritis; THA: Total hip arthroplasty. 19. Steppacher SD, Tannast M, Werlen S, Siebenrock KA: Femoral morphology differs between deficient and excessive acetabular coverage. Clin Orthop Relat Res 2008, 466(4):782–790. 2. James SL, Ali K, Malara F, Young D, O’Donnell J, Connell DA: MRI findings of femoroacetabular impingement. AJR 2006, 187:1412–1419. References 27. Harris WH: Etiology of osteoarthritis of the hip. Clin Orthop 1986, 213:20–33. 1. Beall DP, Sweet CF, Martin HD: Imaging findings of femoroacetabular impingement syndrome. Skeletal Radiol 2005, 34:91–701. 28. Cooperman DR, Wallensten R, Stulberg SD: Post-reduction avascular necrosis in congenital dislocation of the hip. J Bone Joint Surg Am 1980, 62(2):247–258. Page 7 of 7 Page 7 of 7 Ipach et al. BMC Musculoskeletal Disorders 2014, 15:150 http://www.biomedcentral.com/1471-2474/15/150 29. Lau EM, Lin F, Lam D, Silman A, Croft P: Hip osteoarthritis and dysplasia in Chinese men. Ann Rheum Dis 1995, 54(12):965–969. 30. Stulberg SD, Cordell LD, MacEwen GD: Unrecognized childhood hip disease: a major cause of idiopathic osteoarthritis of the hip. In The Hip Procs Third Open Scientific Meeting of the Hip Society St Louis. Edited by Amstutz HC. St. Louis: C V Mosby; 2010:212–218. y 31. Ipach I, Arlt EM, Mittag F, Kunze B, Wolf P, Kluba T: A classification-system improves the intra- and interobserver reliability of radiographic diagnosis of “pistol-grip-deformity”. Hip Int 2011, 21(6):732–739. 32. Clohisy JC, Carlisle JC, Trousdale R, Kim YJ, Beaule PE, Morgan P, Steger-May K, Schoenecker PL, Millis M: Radiographic evaluation of the hip has limited reliability. Clin Orthop Relat Res 2009, 467(3):666–675. 33. Gosvig KK, Jacobsen S, Sonne-Holm S, Palm H, Troelsen A: Prevalence of malformations of the hip joint and their relationship to sex, groin pain, and risk of osteoarthritis: a population-based survey. J Bone Joint Surg Am 2010, 92(5):1162–1169. 34. Bland JM, Altman DG: Measuring agreement in method comparison studies. Stat Methods Med Res 1999, 8:135–160. 35. Bland JM, Altman DG: A note on the use of the intracalss correlation coefficient in the evaluation of agreement between two methods of measurement. Comput Biol Med 1990, 20:337–340. coefficient in the evaluation of agreement between two methods of measurement. Comput Biol Med 1990, 20:337–340. doi:10.1186/1471-2474-15-150 Cite this article as: Ipach et al.: Radiographic signs for detection of femoroacetabular impingement and hip dysplasia should be carefully used in patients with osteoarthritis of the hip. BMC Musculoskeletal Disorders 2014 15:150. Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color ﬁgure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color ﬁgure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit
W2037219219.txt	https://tos.org/oceanography/assets/docs/27-1_pilson.pdf	en		Changing pH in the Surface Ocean	Oceanography	2,014	cc-by	3,575	Oceanography THE OFFICIAL MAGAZINE OF THE OCEANOGRAPHY SOCIETY CITATION Pilson, M.E.Q. 2014. Changing pH in the surface ocean. Oceanography 27(1):120–125, http://dx.doi.org/10.5670/oceanog.2014.15. DOI http://dx.doi.org/10.5670/oceanog.2014.15 COPYRIGHT This article has been published in Oceanography, Volume 27, Number 1, a quarterly journal of The Oceanography Society. Copyright 2014 by The Oceanography Society. All rights reserved. USAGE Permission is granted to copy this article for use in teaching and research. Republication, systematic reproduction, or collective redistribution of any portion of this article by photocopy machine, reposting, or other means is permitted only with the approval of The Oceanography Society. Send all correspondence to: info@tos.org or The Oceanography Society, PO Box 1931, Rockville, MD 20849-1931, USA. DOWNLOADED FROM HTTP://WWW.TOS.ORG/OCEANOGRAPHY SPECIAL ISSUE ON CHANGING O CE AN CH E MIS T RY » A N T H RO P O C E N E : T H E F U T U R E … S O FA R Changing pH in the Surface Ocean BY MICHAEL E .Q. PIL SON In 1896, Arrhenius provided the first roughly quantitative sense of the plausible magnitude of human-induced changes in the concentration of CO2 in the atmosphere. Since then, all chemists could be aware that increasing CO2 in the atmosphere must lead to increasing amounts in the ocean and a corresponding increase in acidity. For a long time, however, no one appears to have thought much about this latter consequence, probably because the likely effects were small and were, in any case, rather troublesome to calculate. It was only in 1909 that Sørensen proposed the concept of the pH scale. The negligible level of concern about the effect of CO2 on Earth’s heat balance was not much affected by the work of Callendar (1938), who argued that the increasing concentrations in the atmosphere could be observed and would be significant. The radiative balance calculations of Plass (1956), published in Tellus, began to influence those who read such journals, and the beginnings of the Keeling curve brought widespread attention to the increasing atmospheric CO2 concentration. Bolin and Eeriksson (1959) quantitatively worked out the uptake of CO2 by the ocean corresponding to increases 120 Oceanography \| Vol. 27, No. 1 in atmospheric concentrations. They showed that a change in the partial acidification, resulting in the publication of a major report by The Royal Society pressure of CO2 also changes the pH of seawater and, consequently, its carbonate ion concentration. While it was certainly pointed out by Smith and Buddemeier (1992, and references therein) that increasing CO2 would lead to decreasing saturation of surface seawater with calcium carbonate, and that this could cause problems for calcifying organisms, the large degree to which such seawater was supersaturated then (and still today) suggested that the situation might not be serious. Indeed, Smith and Buddemeier calculated that even at 600 ppmv CO2, corals would still be bathed in water 2.4 times saturated with respect to aragonite. During the 1990s, increasing evidence of a decrease in calcification by most calcifying organisms with a reduction in the level of supersaturation (Gattuso et al., 1998; Kleypas et al., 1999, and references therein; Langdon et al., 2000) led to widespread concern about the possible effects of decreasing pH and the associated decrease in carbonate ion concentration. Plausible pH changes over the next 3,000 years, explored by Caldeira and Wickett (2003), led to even more widespread recognition of the potential consequences of ocean (2005); a workshop report sponsored by the National Science Foundation, the National Oceanic and Atmospheric Administration (NOAA), and the US Geological Survey (Kleypas et al. 2006); and numerous publications since. Not only will biological and chemical processes change in response to changing pH, but even the absorption of sound within the ocean will decrease (Hester et al., 2008). It should be noted, perhaps, that multiple chemical changes are associated with increasing CO2: concentrations (and activities) of hydrogen ions and bicarbonate ions increase, while concentrations of carbonate ion and borate ion decrease. It is generally believed that the decreasing rates of calcification are due to decreasing carbonate ion concentrations, but there is little information on the biological effects of changing hydrogen ion or bicarbonate concentrations or changing partial pressures of CO2. Certainly, these changes affect many other chemical equilibria. Figure 1 shows pH changes in surface seawater in equilibrium with the atmosphere, as atmospheric CO2 concentrations change from a minimum during the last glaciation up to and beyond “ DURING THE 1990s, INCREASING EVIDENCE OF A DECREASE IN CALCIFICATION BY MOST CALCIFYING ORGANISMS WITH A REDUCTION IN THE LEVEL OF SUPERSATURATION LED TO WIDESPREAD CONCERN ABOUT THE POSSIBLE EFFECTS OF DECREASING pH AND THE ASSOCIATED DECREASE IN CARBONATE ION CONCENTRATION. . 8.5 8.4 Surface Seawater pH values that could plausibly be reached before the end of the present century. The calculation is straightforward, the only assumptions being that the surface ocean is in equilibrium with the CO2 in the atmosphere and that variations of salinity and salinity-normalized alkalinity of the ocean are too small to make a significant difference during the time span considered. Ice core data provide an estimate of about 180 ppmv of CO2 in the atmosphere 20,000 years ago, and data from Keeling and NOAA (Tans and Keeling, 2013) show concentrations from 1959 until the present. The vertical lines indicate appropriate values for four dates. No dates are listed for times in the future, because the trajectory of changing atmospheric concentrations will depend on the amount of fossil fuel burned, release of CO2 from other sources, possible changes in ocean circulation and temperature, and exchange processes between the atmosphere and the land and the sea. None of these uncertainties (except changes in the amount of fossil fuel burned) will likely have a major impact on the concentration of CO2 in the atmosphere, so plausible projections suggest that atmospheric concentrations could approach 600 ppmv or more by the end of this ” 8.3 8.2 8.1 8.0 7.9 7.8 200 300 400 500 600 Atmospheric CO2 (ppmv) 700 800 Figure 1. Calculated pH of surface seawater in equilibrium with CO2 in the atmosphere, updated from figures published in Pilson (1998, 2013). The atmospheric concentration is expressed as ppmv: parts per million by volume in a dry atmosphere. This unit is the same as the mole fraction: moles of CO2 per total moles of all gases except water. Table 1 provides some details of the calculation. Results for three temperatures are plotted: those for 18°C because that is approximately the area-weighted average for the world ocean (Levitus, 1982), and for 0°C and 30°C to span nearly the range in the world ocean. The value of 180 ppmv of CO2 corresponds approximately to the low value at the height of the last glaciation. Dates are entered to show the values for the year 1880 (from ice core data) and for 1960 and for 2014, predicted from the Keeling curve and NOAA data (http://www.esrl.noaa.gov/gmd/ccgg/trends). NOAA also publishes global monthly averages near sea level on that same website; the Mauna Loa data were used here because the series is longer and the average values differ by less than 1 ppmv. At the height of glaciation, the average salinity of the world ocean may have been greater by about 3.5%; the resulting pH at 18°C would accordingly have been about 0.005 greater. No adjustment for such changes is attempted here. Oceanography \| March 2014 121 century. In due course, over several hundred to several thousand years, the total alkalinity of the ocean must slowly increase (due to reduced calcification in the ocean and increased dissolution of carbonates on land and in the ocean), and thereby influence the pH, but such considerations are beyond the scope of this paper. The pH of seawater at equilibrium with the atmosphere depends on its salinity, temperature, and total alkalinity, and the partial pressure (properly the fugacity) of atmospheric CO2. The usual unit for expressing the concentration of CO2 in the atmosphere (commonly plotted over time as the famous Keeling Curve; Figure 2) is parts per million by volume (ppmv), equivalent to the mixing ratio or the mole fraction. For quick calculation, the numerical ppmv values are sometimes treated as the partial pressure at sea level, but they are not exactly the same. The concentration is expressed on the basis of a dry atmosphere, but the atmosphere exactly at the ocean’s surface must be saturated with water vapor. The water vapor pressure must be taken into account in considering the equilibrium with seawater. A sample calculation in Table 1 includes this effect, and also a small correction to yield the fugacity. Table 1 also provides brief estimates of the sensitivity of the calculation to variations in salinity and atmospheric pressure. The calculations shown in Figure 1 were all carried out for a salinity of 35, as a common convention, close to the world average surface salinity of 34.63 reported by Levitus (1982). The uptake of atmospheric CO2 decreases the pH of seawater (Figure 1), and this has had and will have many consequences. Figure 3 illustrates one of them. As seawater absorbs CO2, its capacity to take up more of this gas diminishes. Seawater at 18°C in equilibrium with an atmosphere containing 420 Atmospheric CO2 (ppmv) 400 380 360 340 320 300 1960 1970 1980 1990 Year 2000 2010 2020 Figure 2. The concentration of CO2 in the atmosphere at a measurement station located near the top of Mauna Loa, Hawaii. Concentrations are measured many times per day; the monthly average values (see http://www.esrl.noaa.gov/gmd/ccgg/trends) are plotted here starting from the time in 1958 that Charles David Keeling began this set of observations. NOAA has continued recording these measurements since 1974. From ice core data, the value in 1880 would have been about 280 ppmv. 122 Oceanography \| Vol. 27, No. 1 about 200 ppmv of CO2 will take up 1.17 µmol of CO2 into each kilogram of seawater if the atmospheric concentration increases to 201 ppmv. This year, with an atmospheric concentration of 400 ppmv, an increase to 401 ppmv will lead to an uptake of less than 0.5 µmol of CO2 into each kilogram. The effect is not trivial in the global mass balance, or as an entry in the list of processes associated with the absorption of CO2 into the ocean. For a quantitative perspective, consider the top 100 m of the world ocean; this has a mass of about 35 × 1018 kg. In the atmosphere, 1 ppmv amounts to 2.128 Gt of carbon. At 18°C and an atmospheric concentration of 200 ppmv, an increase of 1 ppmv leads to the absorption of 0.49 Gt of C into the top 100 m of the world ocean, or about 23% of the atmospheric increase. At 400 ppmv, the increase of 1 ppmv leads to the absorption of only 0.21 Gt C, or 9.7% of the atmospheric increase. The surface layers are, of course, gradually mixed with and replaced by subsurface waters, but the rates are locally variable and the global average rate is not well known. It is likely that the most important consequence of CO2 uptake and decreasing pH is the consequent decrease in carbonate ion concentration (Figure 4). This is important because the saturation state of calcium carbonate in seawater depends largely on the concentration of carbonate ion (the concentration of calcium is always present in nearly constant proportion to the salinity). The solubility product constant, K sp = [Ca2+][CO 32–], has a Michael E.Q. Pilson (pilson@mail. uri.edu) is Professor of Oceanography, Emeritus, Graduate School of Oceanography, University of Rhode Island, Narragansett, RI, USA. substantial dependence on salinity, but a very weak dependence on temperature (Mucci, 1983). Although it may seem that the equilibrium concentration of carbonate ion is currently well above that required for the equilibrium solubility of aragonite in seawater, it is observed that many (though not all) organisms apparently will only secrete their carbonate shells or skeletons at a healthy rate if the carbonate ion concentration is well above that required for saturation (Kleypas et al., 1999, 2006; Ries et al., 2009). The total alkalinity (salinity normalized) does vary somewhat over the surface ocean, being maintained at slightly lower than average values in latitudes where CaCO3 is actively precipitated. The total alkalinity of seawater is likely to increase over time in response to elevated atmospheric CO2 (due to decreased precipitation of calcium carbonate in surface waters and increased dissolution at depth), ameliorating, to some extent, the effects of increasing atmospheric CO2 by increasing the carbonate ion concentration and increasing Table 1. Data needed for a sample calculation, and sensitivity to some variables. Item Concentration of CO2 18°C Salinity 35.00 SP, Practical salinity 2.320 mmol kg–1 Total alkalinity (TA) Close to global average Atmospheric pressure 101,325 Pa 1 standard atmosphere Water vapor pressure* 2,032 Pa 603 Pa @ 0°C; 4,179 @ 30°C Atm. Pressure, dry atmosphere 99,293 Pa Partial pressure (pp) CO2 39.717 Pa Fugacity adjustment to pp (–0.142 Pa)* 39.575 Pa = 390.6 µatm pH calculated with CO2 equations* for 390.6 µatm 8.189 Sensitivity to salinity at 18°C, [CO2] = 400 ppmv, and atmospheric pressure = 101,325 Pa Sensitivity to atmospheric pressure at 18°C, [CO2] = 400 ppmv, and S = 35 (due to consequent changes in the partial pressure of CO2) S = 33 pH = 8.175 S = 35 pH = 8.189 S = 37 pH = 8.201 Atm pressure = 98,000 Pa pH = 8.201 Atm pressure = 101,325 Pa pH = 8.189 Atm pressure = 105,000 Pa pH = 8.176 *Two factors are needed to enter the equations to calculate all components of the CO2 system. I used TA and fugacity of atmospheric CO2, ignoring small corrections to total alkalinity due to variations in nutrient concentrations. There are several sources for the data and equations used here. It was convenient to use those in Pilson (2013). The fugacity calculation is from Dickson et al. (2007). Atmospheric CO2 values were obtained from http://www.esrl.noaa.gov/gmd/ ccgg/trends. The pH values in Figure 1 are based on the widely familiar NBS (now NIST) scale. Conversions among the various pH scales, importantly the “total” scale common for open ocean calculations, are very conveniently done using the “CO2calc” program (Robbins et al., 2010). It really would be advisable for all CO2 partial pressures to be expressed in pascals, and the CO2 equations to be entered with that unit, but it is still common to use µatm. Figure 3. The uptake factor is the incremental increase in the concentration of total CO2 in seawater, in equilibrium with the atmosphere, corresponding to an incremental increase in the atmospheric concentration of CO2 (Pilson, 2013), and is expressed as 1.2 1880 Δµmol TCO2 kg –1 Δppmv CO2 The calculation was done over small increments and plotted at the mid-point of each increment. Conditions were: seawater with S = 35 and total alkalinity = 2.32 µmol kg –1, a standard atmospheric pressure of 101,325 Pa, and the fugacity at each temperature and atmospheric concentration. Values appropriate for the years 1880 and 2014 are indicated. 1.0 Utake Factor Notes in a dry atmosphere Temperature of water 1.4 18°C Value 400 ppmv UF = 0.8 2014 0.6 0.4 30°C 0.2 0°C 0.0 200 300 400 500 600 Atmospheric CO 2 (ppmv) 700 800 Oceanography \| March 2014 123 the capacity of seawater to take up CO2. Whether alkalinity will be measurably altered during the rest of this century is uncertain. Any change will not significantly affect the projected average values in the figures shown here. It is likely that no place in the surface ocean is ever exactly in equilibrium with the atmosphere with regard to CO2 except by chance. Temperature changes daily and seasonally, and CO2 exchanges more slowly than does heat, so the CO2 concentration changes required to reach equilibrium will commonly lag changes in temperature. More importantly, the balance between biological processes of photosynthesis and respiration, as phytoplankton blooms come and go, varies much faster than the exchange of CO2 with the atmosphere can keep up. In extreme examples, especially in inshore waters, the observed range of pH may be greater than the total range shown in Figure 1. Thus, the values shown here in all the figures (except Figure 2) are equilibrium targets that ocean water in any given place and time will usually be moving toward, at rates controlled by the chemical history of the water and physical conditions such as water turbulence and local wind strength. It is reasonable to assume that observed average values in the ocean will center around equilibrium values, allowing for some likelihood that there will be asymmetry in the rates of approach from undersaturation versus 400 350 [CO=3 ] (µmol kg–1) 300 250 200 150 100 50 0 200 300 400 500 600 Atmospheric CO2 (ppmv) 700 800 Figure 4. Concentration of carbonate ion in seawater characterized by S = 35 and TA = 2.032 µmol kg–1 at three temperatures and at equilibrium with a range of atmospheric concentrations of CO2. The two horizontal dotted lines show the carbonate ion concentration at saturation with aragonite, the upper one at 0°C and the lower one at 30°C. The one horizontal dashed line (there is not much temperature effect) shows the carbonate ion concentration at saturation with calcite (calculated from Mucci, 1983). The vertical lines indicate the equilibrium conditions for the years indicated. 124 Oceanography \| Vol. 27, No. 1 oversaturation. Measurements of pH provide sensitive and important assessments of current and recent past conditions in any body of water. ACKNOWLEDGEMENTS I thank Robert Byrne, Michael DeGrandpre, and Flip Froelich for helpful comments. REFERENCE S Arrhenius, S. 1896. On the influence of carbonic acid in the air upon the temperature of the ground. Philosophical Magazine and Journal of Science (Fifth Series) 41:237–276. Available online at: http://www.rsc.org/images/ Arrhenius1896_tcm18-173546.pdf (accessed November 23, 2013). Bolin, B., and E. Eriksson. 1959. Changes in the carbon dioxide content of the atmosphere and sea due to fossil fuel combustion. Pp. 130–142 in Rossby Memorial Volume. Rockefeller Institute Press, New York. Caldeira, K., and M.E. Wicket. 2003. Anthropogenic carbon and ocean pH. Nature 425:365, http://dx.doi.org/10.1038/425365a. Callendar, G.S. 1938. The artificial production of carbon dioxide and its influence on temperature. Quarterly Journal of the Royal Meteorological Society 64:223–240, http://dx.doi.org/10.1002/qj.49706427503. Dickson, A.G., C.L. Sabine, and J.R. Christian, eds. 2007. Guide to the Best Practices for Ocean CO2 Measurements. PICES Special Publication 3, 191 pp. Available online at: http://cdiac.ornl. gov/oceans/Handbook_2007.html (accessed November 23, 2013). Gattuso, J.P., M. Frankignoule, I. Bourge, S. Romaine, and R.W. Buddemeier. 1998. Effect of calcium carbonate saturation of seawater on coral calcification. Global and Planetary Change 18:37–46, http://dx.doi.org/10.1016/ S0921-8181(98)00035-6. Hester, K.C., E.T. Peltzer, W.J. Kirkwood, and P.G. Brewer. 2008. Unanticipated consequences of ocean acidification: A noisier ocean. Geophysical Research Letters 35, L19601, http:// dx.doi.org/10.1029/2008GL034913. Kleypas, J.A., R.W. Buddemeier, D. Archer, J.-P. Gattuso, C. Langdon, and B.N. Opdyke. 1999. Geochemical consequences of increased atmospheric carbon dioxide on coral reefs. Science 284:118–120, http://dx.doi.org/10.1126/ science.284.5411.118. Kleypas, J.A., R.A. Feely, V.J. Fabry, C. Langdon, C.L. Sabine and L.L. Robbins. 2006. Impacts of Ocean Acidification on Coral Reefs and Other Marine Calcifiers: A Guide for Future Research. Report of a workshop sponsored by NSF, NOAA, and USGS, April 18–20, 2005, St. Petersburg, FL, 88 pp. Available online at: http://www.ucar.edu/communications/Final_ acidification.pdf (accessed November 23, 2013). Langdon, C., T. Takahashi, C. Sweeney, D. Chipman, J. Goddard, F. Marubini, H. Aceves, H. Barnett, and M.J. Atkinson. 2000. Effect of calcium carbonate saturation state on the calcification rate of an experimental coral reef. Global Biogeochemical Cycles 14:639–654, http://dx.doi.org/10.1029/1999GB001195. Levitus, S. 1982. Climatological Atlas of the World Ocean. NOAA Professional Paper No. 13, US Government Printing Office, Washington, DC, 173 pp. Available online at: ftp://ftp.nodc.noaa.gov/pub/data.nodc/ woa/PUBLICATIONS/levitus_atlas_1982.pdf (accessed November 23, 2013). Mucci, A. 1983. The solubility of calcite and aragonite in seawater at various salinities, temperatures, and one atmosphere total pressure. American Journal of Science 283:780–799, http://dx.doi.org/10.2475/ajs.283.7.780. Pilson, M.E.Q. 1998. An Introduction to the Chemistry of the Sea. Prentice Hall, Upper Saddle River, NJ, 431 pp. Pilson, M.E.Q. 2013. An Introduction to the Chemistry of the Sea, 2nd ed. Cambridge University Press, Cambridge, UK, 524 pp. Plass, G.N. 1956. The carbon dioxide theory of climate change. Tellus 8:140–154, http:// dx.doi.org/10.1111/j.2153-3490.1956.tb01206.x. Ries, J.B., A.L. Cohen, and D.C. McCorkle. 2009. Marine calcifiers exhibit mixed responses to CO2-induced ocean acidification. Geology 37:1,131–1,134, http://dx.doi.org/ 10.1130/G30210A.1. Robbins, L.L., M.E. Hansen, J.A. Kleypas, and S.C. Meylan. 2010. CO2calc: A User-Friendly Seawater Carbon Calculator for Windows, Mac OS X, and iOS (iPhone). US Geological Survey Open-File Report, 2010–1280, 17 pp. Available online at http://pubs.usgs.gov/of/2010/1280 (accessed November 23, 2013). Smith, S.V., and R.W. Buddemeier. 1992. Global change and coral reef ecosystems. Annual Review of Ecology and Systematics 23:89–118, http://dx.doi.org/10.1146/annurev.es.23. 110192.000513. Sørensen, S.P.L. 1909. Études enzymatiques: II. Sur la mesure et l’importance de la concentration des ions hydrogen dans les reactions enzymatiques. Comptes rendus des travaux du laboratoire Carlsberg 8:1–168. (Cited from: Nørby, J.G. 2000. The origin and the meaning of the little p in pH. Trends in Biochemical Science 25:36–37, http://dx.doi.org/10.1016/ S0968-0004(99)01517-0.) Tans, P., and R. Keeling. 2013. Trends in atmospheric carbon dioxide, full Mauna Loa CO2 record. http://www.esrl.noaa.gov/gmd/ccgg/ trends. The Royal Society. 2005. Ocean Acidification due to Increasing Atmospheric Carbon Dioxide. Policy document, June 2005, 57 pp. Available online at: http://royalsociety.org/policy/ publications/2005/ocean-acidification (accessed November 23, 2013). CO MI NG IN JU NE 20 15 A Special Issue of Oceanography Ocean Acidification The Connection Between Science and Policy GUEST EDITORS: Sarah Cooley Ocean Conservancy Jeremy Mathis National Oceanic and Atmospheric Administration Kim Yates US Geological Survey Oceanography \| March 2014 125
https://openalex.org/W2112224171	https://www.maxwellsci.com/announce/RJASET/6-1606-1613.pdf	English	null	Selenium Alleviates Coleus from Oxidative Damage under Pb Stress by Resource Allocation and Antioxidant Defense System	Research journal of applied sciences, engineering and technology	2,013	cc-by	5,669	Corresponding Author: Hu Mian-Hao, Department of Resource Environment, School of Tourism and Urban Management, Jiangxi University of Finance and Economics, Nanchang, 330032, China, Tel.: +086-791-83881172; Fax: +086-791- 83842424 This work is licensed under a Creative Commons Attribution 4.0 International License (URL: http://creativecommons.org/licenses/by/4.0/). 1606 Keywords: Antioxidant defense system, Coleus blumei benth, Lead, resource allocation, selenium Keywords: Antioxidant defense system, Coleus blumei benth, Lead, resource allocation, seleniu Published: July 15, 2013 Published: July 15, 2013 Submitted: November 12, 2012 Submitted: November 12, 2012 Accepted: January 07, 2013 Research Article Selenium Alleviates Coleus from Oxidative Damage under Pb Stress by Resource Allocation and Antioxidant Defense System 1Yuan Ju-Hong, 2Hu Mian-Hao and 2Zhou Zao-Hong 1Department of Landscape, College of Arts, 2Department of Resource Environment, School of Tourism and Urban Management, Jiangxi University of Finance and Economics, Nanchang, 330032, China Yuan Ju-Hong, Hu Mian-Hao and Zhou Zao-Hong 1Department of Landscape, College of Arts, 2Department of Resource Environment, School of Tourism and Urban Management, Jiangxi University of Finance and Economics, Nanchang, 330032, China Abstract: The role of selenium (Se) against lead (Pb) stress in Coleus (Coleus blumei benth.) was investigated by evaluating the allocation plasticity and antioxidant properties of plants cultured hydroponically with lead (1.0 mM Pb) and selenite (0～5.0 mM Se) in this study. It was showed that Pb reduced the root growth and caused serious damage in the roots, which was accompanied by biomass changes and Pb accumulation in the organ. When Se application up to 1.0 mM improved the biomass allocation and Pb accumulation of Coleus organs and significantly decreased Thiobarbituric Acid Reactive Substances (TBARS) accumulation in Coleus treated with 1.0 mM Pb. However, above 1.0 mM, Se induced stress in Coleus grown with Pb. These results indicated that at this concentration, Se can protect Coleus from Pb stress by expressing primarily biomass allocation and metal partitioning. Significant changes of antioxidant defense system in Coleus exposed to Pb stress were also found as a result of the added Se. Se supplied up to 1.0 mM greatly decreased root Superoxide Dismutase (SOD) and Peroxidase (POD) activity in Pb-stressed Coleus (p<0.05). In addition, Catalase (CAT) and glutathione peroxidase (GSH-Px) activity was increased with increasing Se concentration (p<0.05). The effect of Se on glutathione (GSH) in Coleus Pb-treated was dose-dependent. The GSH content increased up to a concentration of 2.5 mM Se and then decreased. These changes in enzymatic and non-enzymatic antioxidants showed that Se supplied had a significant disturbance on Coleus under Pb stress and affected the biochemical and physiological processes. Furthermore, the obtained results also showed that appropriate Se supplementation may ameliorate Pb-induced oxidative stress by decreasing lipid peroxidation and altering antioxidant defense system and that Se detoxification and accumulation in Coleus might be associated closely with the efficiency of these mechanisms. INTRODUCTION et al., 2009). Nevertheless, several studies reported the beneficial effects of Se, because it increased the antioxidant activity in plants, leading to better plant yield (Lyons et al., 2009; Cartes et al., 2010). Recent publications indicated that Se addition may alter the total content of heavy metals in plant tissues (Pedrero et al., 2008; Feng et al., 2009). In Brassica napus seedlings, Se was found to reverse the Cd-induced decrease in fresh mass and changes in lipid peroxidation as well as changes in the DNA methylation pattern (Filek et al., 2008). Studies on animals had also shown that Se was one of the potential antagonists to Pb, Cr, Hg and Cd and limited the toxic effects of heavy metals (Cerklewski and Forbes, 1976; Ikemoto et al., 2004; Soudani et al., 2010). Coleus (Coleus blumei Benth.) is a salinity and humidity- resistance ornamental plant which is widely planted in arid and semiarid urban regions. Although Coleus can remove nitrogen and phosphorus in eutrophication Lead (Pb), a heavy metal with characteristic toxic action, has attracted considerable attention for its widespread distribution and potential risk to the environment. The plant processes such as the biosynthesis of nitrogenous compounds, carbohydrate metabolism and water absorption are adversely affected by increasing Pb levels in soil and even at every low concentration (John et al., 2009; Hamid et al., 2010). Furthermore, the metals are also accumulated in different plant parts and thereby enter into the food chain. However, the plant response to Pb contamination is a key research problem and a special effort is undertaken in seeking factors which affect the reduction of Pb absorption or toxicity in plants. Selenium (Se) is an important element for human and animal nutrition based on its presence in antioxidative defence systems, but was not considered essential for plants (Kápolna 1606 Res. J. Appl. Sci. Eng. Technol., 6(9): 1606-1613, 2013 water (Liu et al., 2004) and had strong tolerance and accumulation capacity of aluminum (Panizza de León et al., 2011), however, its growth and productivity are frequently threatened by different abiotic stresses such as drought, salinity or heavy metal. To cope with these stresses, Coleus has developed arrays of physiological and biochemical strategies to adapt to the adverse conditions, so it is important to under the mechanisms that confer tolerance to heavy metal environments. In addition, the facts of Se is intriguing, enigmatic and challenging (even capricious) for researcher. INTRODUCTION Thus, in this study, we were to investigate that: Determination of Pb and Se content: Root, shoot and leaves samples of 0.25 g dry weight were put into 100 mL digestion tubes. Next, a 5 mL acid mixture of HNO3 and HC1O4 (4:1, v/v) was added and digested at 25°C overnight. Samples were then completely digested at 150～165°C in a digestion oven until the solution became clear. After cooling, a 2.5 mL aliquot of 6 M HCl was added and heated to 100°C until gaseous brown fumes ceased and the solution became clear. After cooling, the digested samples were diluted to 25 ml with Milli-Q water. Pb and Se contents were analyzed by the ICP-AES (Inductively Coupled Plasma-Atomic Emission Spectroscopy) method in the Ministry of Education Key Laboratory of Environmental Remediation and Ecological Health, College of Natural Resources and Environmental Science, Zhejiang University. Calibration curves were made using Pb and Se standard aqueous solutions “Suprapur” pure (Merck chemical). • The appropriate concentration of exogenous Se alleviating Pb stress by allocation plasticity and plant-metal partitioning, which were relevant to phytoremediation and representative of distinctive growth strategies • The occurrence of any antagonistic or synergistic interaction between Pb and Se by membrane stability and antioxidant enzyme activities in Coleus. • The occurrence of any antagonistic or synergistic interaction between Pb and Se by membrane stability and antioxidant enzyme activities in Coleus. Antioxidant enzyme extraction and assays: Plant material (about 0.5g) was homogenized in 5 ml solutions containing: 100 mM of potassium phosphate buffer (pH7.0) containing 0.1 mM of EDTA-Na2 and 1% polyvinylpyrrolidone (W/V) at 4°C. The homogenate was filtered through four layers of cheesecloth and centrifuged at 15,000×g for 15 min at 4°C. Enzyme activity was measured in the supernatant solution. Protein contents were determined according to Bradford (1976), using bovine serum albumin as a standard. SOD, POD and CAT activities were determined using the method of Gajewska et al. (2006) and GPX activities were determined using the methods of Mishra et al. (2006). MATERIALS AND METHODS Plant material and growth conditions: Coleus (Coleus blumei benth.) used in this study was from the clone continually propagated in the Botanical Garden, Nanchang, Jiangxi. The plants were pretreated with tap water for two weeks, to adapt to the water environment. Plant material and growth conditions: Coleus (Coleus blumei benth.) used in this study was from the clone continually propagated in the Botanical Garden, y p p g , Nanchang, Jiangxi. The plants were pretreated with tap water for two weeks, to adapt to the water environment. Next, the seedlings of uniform size were transplanted in 15L pots (16 plants per one pot) with modified full- strength Hoagland’s medium (Zhao et al., 2007) containing the following mineral components (mM): Ca(NO3)2 3.0; KNO3 4.0; KH2PO4 1.0; MgSO4 1.0; MnCl2 3.6×10-3; H2BO3 4.5×10-2; CuCl2 8×10-4; ZnCl2 1.5×10-3; (NH4)6Mo7O24 1.4×10-5; Fe-EDTANa2 9.0×10- 2。To determine the effect of Pb and Se, the growth medium was supplemented with 1.0 mM Pb in the form of Pb(NO3)2 (Sigma) and the concentrations of Se in the form of Na2SeO3 (Sigma) are 0 (control), 0.1, 0.5, 1.0, 2.5 and 5.0 mM. The Pb and Se concentrations were chosen on the basis of preliminary experiments and literature data. Four-week cultivation was performed in the greenhouse and the photon flux density was c. 450/µmol photon s-1 m-2 at the height of the upper leaves. The temperatures rage were 1518°C (night) and 22-26°C (day), relative humidity was between 45 and 70%. During the experiment, the nutrient solution was continuously aerated and its losses were supplemented daily with dH2O. The medium were renewed once a week and its pH was kept at 6.0-6.5. Each treatment contained six individual single pots. At the end of experiment, the fresh weights of the seedlings (g/pot) were recorded. The dry weights of seedlings were measured by drying the seedlings at 75°C, to give a constant weight. Nonenzymatic antioxidant and Lipid peroxidation measurement: Reduced (GSH) and oxidized (GSSG) glutathione contents were determined by the recycling method described by Anderson (1985). Lipid peroxidation was determined by estimation of the Melondialodehyde (MDA) content following Fu and Huang (2001). Statistical analysis: All data were analyzed in four replications using one-way analysis of variance (ANOVA) and the LSD test to determine significant differences between treatments. The graphs were done on OriginPro Version 8.5 (OriginLab Corporation, Northampton, USA). RESULTS Influence of selenium on resource allocation under Pb stress: Influence of selenium on resource allocation under Pb stress: Influence of selenium on resource allocation under Pb stress: Biomass allocation to root, stem and leaf of Coleus: Biomass is a key factor for phytoremediation practices and it is also an overall measurement of plant health. 1607 Res. J. Appl. Sci. Eng. Technol., 6(9): 1606-1613, 2013 Fig. 1: Influence of Se on biomass allocation of Coleus exposed to Pb stress. The bars indicate mean±SE (n = 6). Values with different letters differ significantly from each other at p<0.05 Fig. 1: Influence of Se on biomass allocation of Coleus exposed to Pb stress. The bars indicate mean±SE (n = 6). Values with different letters differ significantly from each other at p<0.05 Table 1: Influence of Se on Pb and Se distribution in the different organs of Coleus exposed to Pb stress [Se]solution (mM) Pb accumulation (mg/g DW) ----------------------------------------------------------------------- Se accumulation (mg/g DW) ----------------------------------------------------------------- Root Stem Leaf Root Stem Leaf 0 58.49±1.94a 2.79±0.46a 3.27±0.06b ND ND ND 0.1 69.63±8.99b 4.69±0.12a 2.91±0.07b 0.07±0.001a 0.005±0.000a 0.003±0.000a 0.5 103.99±3.13c 7.69±0.53b 5.20±0.03c 0.20±0.009b 0.01±0.008b 0.01±0.002b 1.0 112.89±9.97c 8.71±0.69b 2.69±0.07b 0.39±0.008b 0.02±0.009b 0.02±0.006b 2.5 107.49±2.96c 4.14±0.31a 2.59±0.09b 0.98±0.004c 0.01±0.004b 0.007±0.00a 5.0 71.86±5.59b 2.69±0.58a 0.47±0.07a 0.80±0.003c 0.01±0.003b 0.003±0.00a Each value represents means±SE (n = 6). Different letters within columns indicate significant differences between treatments at p<0.05, the same letters are not significantly different between treatments at p>0.05. ND not detectable in Pb and Se-containing nutrient solutions, Se concentrations increased in Coleus organs (p<0.05). The concentration of Se in different organs differed depending on the Se concentrations in the nutrient solution (Table 1). Concentrations of Se in control Coleus, which were grown without Se, were below the limits of ICP-AES detection. There was significantly difference in Se accumulation in the root organ among treatments and the highest accumulation of Se was found at 2.5 mM Se treatment, but in the stem and leaf was 1.0 mM Se treatment (Table 1). in Pb and Se-containing nutrient solutions, Se concentrations increased in Coleus organs (p<0.05). The concentration of Se in different organs differed depending on the Se concentrations in the nutrient solution (Table 1). Concentrations of Se in control Coleus, which were grown without Se, were below the limits of ICP-AES detection. There was significantly difference in Se accumulation in the root organ among treatments and the highest accumulation of Se was found at 2.5 mM Se treatment, but in the stem and leaf was 1.0 mM Se treatment (Table 1). Influence of selenium on resource allocation under Pb stress: Coleus organ biomass varied greatly by different Se concentration exposed to Pb (Fig. 1). The enhancement of Coleus biomass under Pb stress by Se treatments was observed in this study up to 1.0 mM selenium, the root, stem and leaf biomass reached 13.8, 82.9, 64.8 g Fresh Weight (FW), respectively; and then reduced with increasing of Se concentrations, but still enhanced in comparison to the control (Fig. 1). As compared to the control, the root, stem and leaf biomass of Coleus increased by 93.7, 49.6 and 28.9%, respectively (Fig. 1). Pb and Se distribution in Coleus organs: Pb accumulation varied among Coleus organs and was affected by different Se concentration. The highest Pb contents in the leaf, stem and root of Coleus under Pb stress were found with a low concentration of Se (0.5 and 1.0 mM) (Table 1). A significant increase in Pb accumulation was observed in both root and stem organ (p<0.05) at 1.0mM Se treatment, but in leaf for 0.5 mM Se treatment (Table 1). As expected, a high concentration of Se decreased Pb accumulation in the leaf, stem and root of Coleus, but was higher than those of control treatment. These findings have great implications for optimizing phytoextraction of environmental Pb pollution. When Coleus were grown Influence of selenium on lipid peroxidation and antioxidant defense system under Pb stress: y Lipid peroxidation: The MDA content, one of the major TBARS reactive metabolites, is indicator of the lipid peroxidation of plasma membrane of plant cells and its accumulation is indicative of enhanced production of reactive oxygen species. In this study, the MDA contents under Pb stress in the 2.5 mM Se-treated Coleus root decreased by 32% in comparison to the control (Fig. 2). Moreover, 0.5～2.5 mM Se decreased significantly the MDA level, while 5 mM Se increased the lipid peroxidation by 29.4% in comparison to 2.5mM under Pb stress, but was lower than that of the control (Fig. 2). 1608 Res. J. Appl. Sci. Eng. Technol., 6(9): 1606-1613, 2013 Fig. 2: Influence of Se on lipid peroxidation of Coleus root exposed to Pb stress. The bars indicate mean±SE (n = 6). Values with different letters differ significantly from each other at p<0.05 Fig. 2: Influence of Se on lipid peroxidation of Coleus root exposed to Pb stress. The bars indicate mean±SE (n = 6). Values with different letters differ significantly from each other at p<0.05 Fig. 2: Influence of Se on lipid peroxidation of Coleus root exposed to Pb stress. The bars indicate mean±SE (n = 6). Values with different letters differ significantly from each other at p<0.05 Fig. 3: Influence of Se on antioxidant enzyme of Coleus root exposed to Pb stress. The bars indicate mean±SE (n = 6). Values with different little letters differ significantly from each other at p<0.05. The Capital letters of A, B, C and D are SOD, CAT, GSH-PX and POD, respectively Fig. 3: Influence of Se on antioxidant enzyme of Coleus root exposed to Pb stress. The bars indicate mean±SE (n = 6). Values with different little letters differ significantly from each other at p<0.05. The Capital letters of A, B, C and D are SOD, CAT, GSH-PX and POD, respectively Antioxidant defense system: The activities of CAT, SOD, APX and POD enzymes in Coleus root by Se treatment exposed to Pb stress were given in Fig. 3. The SOD activity decreased significantly at a concentration of 1.0 mM Se treatment and then increased with increasing of Se concentration, but was lower than that of control. Moreover, there was no marked change in SOD activity between 2.5 mM and 5.0 mM treatment (Fig. 3A). Influence of selenium on lipid peroxidation and antioxidant defense system under Pb stress: In Se treatment, Coleus exposed to Pb showed a significant increase in the activity of CAT compared to the control, but there was no significant different among low concentrations of Se. However, the CAT activity was increased by 58.4% at 5.0 mM Se treatment (Fig. 3B). All treatments containing Se had higher GSH-Px activity than the control treatment with the highest activity in the treatment with 5.0 Mm 1609 Res. J. Appl. Sci. Eng. Technol., 6(9): 1606-1613, 2013 Table 2: Influence of Se on levels of reduced and oxidized glutathione, reduced and oxidized glutathione ratios in Coleus root exposed to Pb stress [Se]solution (mM) GSH (µmoL/g FW) GSSH (µmoL/g FW) GSH+GSSG (µmoL/g FW) GSH/GSSG 0 2.19±0.43a 1.58±0.02b 4.18a 1.39a 0.1 4.51±0.70b 1.29±0.07b 5.97a 3.50b 0.5 5.13±0.76b 1.08±0.03a 7.13b 4.75b 1.0 7.04±0.58c 0.93±0.02a 9.97b 7.57c 2.5 9.10±0.72c 0.81±0.05a 11.02c 11.23d 5.0 8.65±0.24c 0.97±0.01a 9.59b 8.95c Each value represents means±SE (n = 6). Different letters within columns indicate significant differences between treatments at p<0.05, the same letters are not significantly different between treatments at p>0.05 uced and oxidized glutathione, reduced and oxidized glutathione ratios in Coleus root exposed to Pb the combined effect of Pb addition and high Se supply levels, compared with no Se addition exposed to Pb. This result indicated that Se had either stimulating or toxic effects on Coleus depending on the Se concentration in the culture media. This was in agreement with a number of recent reports on plants such as Lolium perenne L. (Cartes et al., 2010) and Vicia faba L. (Mroczek-Zdyrska and Wójcik, 2012). According to the results obtained in this study, Coleus could accumulate Pb efficiently during the cultivation and the accumulated amount increased with the concentrations of Se treatments (≤2.5 mM) (Table 1). The higher Se concentrations, such as ≤1.0 mM, led to high Pb accumulation and a significant increase in Pb accumulation was observed in both root and stem (Table 1), which may suggest a greater demand of Se in the root to counteract the toxic effects induced by Pb. Nevertheless, the molecular and physiological mechanisms responsible of this behaviour will require further research. These results for Coleus were similar to those reported by Zembala et al. (2010) who found that Se addition significantly decreased the Cd concentration of rape and wheat seedlings exposed to Cd stress. Influence of selenium on lipid peroxidation and antioxidant defense system under Pb stress: The Coleus Se contents in all treatments differed significantly from each other and roughly proportionally to the Se addition (Table 1). As for the allocation of Se into various organs of Coleus increased effectively with increasing Se dosages (≤2.5 mM) (Table 1). Nevertheless, the effects in root and stem can be different, the highest accumulation of Se in the root was found at 2.5 mM Se treatments, but in the stem was 1.0 mM Se treatment (Table 1). This result was in good agreement with Krystofova et al. (2010) who had mentioned that a higher amount of Se was determined in the root of Urtica dioica L. Our findings indicated that Coleus with Se treatments shifted their biomass and metals distribution more to roots than shoots possibly to circumvent the challenges of Pb conditions. (Fig. 3C). Addition of 0.1mM Se Coleus exposed to Pb stress did not alter the GSH-Px activity in the root while 0.5mM Se increased the activity of this enzyme by 34.4%. Whereas GSH-Px activity was approximately 2.4 fold increases than the control at the end of experiment. In the root exposed to Pb stress with Se treatment for 0.1 mM, 0.5 mM and 1.0 mM Se, the POD activity was decreased by 9.1, 24.1 and 42.7%, respectively (Fig. 3D). After addition of Se concentration was above 1.0 mM, a increase in POD activity was found, but still decreased in comparison to the control (Fig. 3D). The results presented in Table 2 showed that GSH level was significantly increased in Coleus root with increasing of Se concentration under Pb stress than the control. The maximum GSH content in the root was observed at 2.5 mM Se treatment, which was 75.9% higher than in the control (Table 2). A decrease was recorded for GSSG and the GSH/GSSG ratio increased in the root under Se treatment, but no significant difference was observed among low Se concentrations (≤1.0 mM) (Table 2), indicating that glutathione pool appeared to be more reduced after Se treatments for Pb stress as compared to the control. DISCUSSION The Se-induced decrease in SOD and POD activities indicated that lower amounts of superoxide anion radicals were produced in cells due to the higher activity of GPX and CAT. On one hand, it can be presumed that the increase in CAT and GPX, which were scavengers of H2O2 and lipid hydroperoxides, resulted in reduced formation of superoxide anion radicals through the dynamic inter-transformation among oxygen species. On the other hand, Se increased GPX activities and enhanced the spontaneous disproportion of superoxide radicals, consequently and reduced the need for their scavenger SOD. These results indicated that the prevention of damage to cell membrane of Coleus can be achieved by co-operative effects of the whole system of antioxidant enzymes. Interestingly, we found that Coleus exposed to Pb with high Se concentration (≥5.0 mM) exhibited the increase of lipid peroxidation, but was lower than that of the control (Fig. 2). At simultaneous Pb and Se addition the maximal decrease in TBARS was observed at 0.98 mg Se g-1 DW in the root. These findings demonstrated that Coleus not only take up more Se as a consequence of root Pb injury, but also suggested that at low Se addition an extra amount of Se was taken up by Coleus to alleviate the Pb-induced oxidative stress. Indeed, plants possess a complex ROS-scavenging system that includes several antioxidant enzymes and low- molecular-weight antioxidants such as ascorbate, glutathione and phenolic compounds (Noctor and Foyer, 1998). Thus, at low dosages, Se possibly triggered the co-operation of various antioxidative systems to help counteract the oxidative damage induced by Pb in the roots. In contrast, the increase in TBARS accumulation at high selenite levels (5 mM Se) indicated that Se may act as a pro-oxidant in Coleus as reported earlier by Mroczek-Zdyrska and Wójcik (2012). ( ) The fluctuations in the level of lipid peroxidation by effect of the added Se were accompanied by significant changes in the activity of the antioxidant enzymes SOD, CAT, GSH-Px and POD in Coleus exposed to Pb stress (Fig. 3). SOD acts as the first line of defense against ROS by catalysing the disproportionation of superoxide radicals (O2 -.) to H2O2 and molecular oxygen (Gratão et al., 2008). In the present study, it was remarkable that selenite additions up to 1.0 mM Se greatly decreased SOD activity in the Coleus root to lower levels compared to those of control (Fig. DISCUSSION 3A) and at low selenite levels, the decrease of SOD activity (Fig. 3A) coincided with a reduction of damage of cell membranes (Fig. 2). This fact suggested that, at these concentrations, Se was able to diminish the need of SOD by reducing the level of toxic O2 -. in the root of Pb-stressed Coleus. CAT is tetrameric heme-containing enzymes, which directly dismutates H2O2 into H2O and O2 and is indispensable for ROS detoxification during stress (Garg and Manchanda, 2009). Application of Se, combined with Pb stress, significantly increased the activity of CAT in this study, especially at 5.0 mM Se treatment (Fig. 3B), which indicated a protective role for Se in scavenging H2O2 in Coleus root under Pb stress. Similar to our results, increased CAT activity in Se-supplemented plants under cadmium, high temperature and salt stresses were described by other researchers (Filek et al., 2008; Djanaguiraman et al., 2010; Hasanuzzaman et al., 2011). GSH-Px is another enzyme that uses GSH to reduce H2O2 and therefore, protects plant cells from damage due to oxidative stress (Gill and Tuteja, 2010). In our study, in comparison to Pb stress alone, the combination of Pb stress and moderate Se resulted in a significant increase in the activities of GSH-Px, but its increase had no significant y y GSH can react chemically with single oxygen, superoxide and hydroxyl radicals and function directly as a free radical scavenger. GSH and its oxidized form, GSSG, maintain a redox balance in the cellular compartments. The conversion of GSSG to GSH by the GR enzyme was correlated with the change in GSH/GSSG ratios, which played an important role in the signal transduction of several transcription and metabolic processes (Namjooyan et al., 2012). Our results demonstrated that the Pb-stressed Coleus with Se supplement showed a higher increase in the level of GSH than did the Coleus subjected to Pb stress alone (without Se). This increase of GSH in Se-treated Coleus root might be due to Se boosting GSH synthesis. This was supported by Anderson and McMahon (2001) who described that Se accelerated efficient recycling of GSH and reported the relationship between Se and GSH synthesis. In our study, the Se and Pb-stressed Coleus showed lower GSSG level than the Coleus treated with Pb alone. DISCUSSION Influence of selenium on plant resource allocation under Pb stress: Plants adjust their relative biomass allocation and heavy metal distribution to organ systems (e.g. roots or shoots) when subjected to environmental stress conditions (Gedroc et al., 1996). Although Se was not yet confirmed to be required by higher plants (Terry et al., 2000), several studies demonstrated that at low concentration it may exert diverse beneficial effects, including growth-promoting activities (Djanaguiraman et al., 2005; Cartes et al., 2010). Our results showed that exposed to Pb, Se promoted the each organ biomass of Coleus at concentrations up to 1.0 mM, whereas each organ biomass was greatly depressed when 2.5 or 5.0 µM Se was added (Fig. 1). This was considered to be Se toxicity because retardation of Coleus growth was one of the symptoms noted when Coleus was grown with high levels of Se. However, each organ biomass was reduced by the simultaneous addition of 1.0 mM Pb and 2.5 or 5.0 µM Se (Fig. 1), but it was still increased by Influence of selenium on lipid peroxidation and antioxidant defense system under exposed to Pb stress: The benefits of Se on plant growth had been often related with an improvement of plant antioxidant ability. According to our results, the ability of Se to reduce lipid peroxidation was noticeable in Coleus root and the application of 2.5 mM Se decreased the TBARS level by 32% in Coleus root exposed to Pb stress. 1610 Res. J. Appl. Sci. Eng. Technol., 6(9): 1606-1613, 2013 different from higher Se concentration (Fig. 3C).This brought the levels down of H2O2 and lipid peroxidation in Coleus root (Fig. 3C and 2). Similar increases in GSH-Px activity after Se supplement during stress was observed by other researchers (Filek et al., 2008). Synergistic effect of Se on GSH-Px activity had been demonstrated before (Lobanov et al., 2008). POD is also an important enzyme, able to scavenge H2O2, which is a major substance degraded by SOD. In the present study, the POD activity in the root exposed to Pb stress was decreased with increasing of Se concentration (≤1.0mM), but the POD activity for 2.5 or 5.0 mM Se increased noteworthy, which seemed to reflect an increased hydrogen peroxide (H2O2) production at higher Se supply levels. In a word, opposite variation trends between SOD-POD activities and CAT-GPX activities were observed during the cultivation. REFERENCES Anderson, J.W. and P.J. McMahon, 2001. The Role of Glutathione in the Uptake and Metabolism of Sulfur and Selenium. In: Grill, D., M.M. Tausz and L.J. de Kok (Eds.), Significance of Glutathione to Plant Adaptation to the Environment. Kluwer Academic, the Netherlands, pp: 57-99. Hamid, N., N. Bukhari and F. Jawaid, 2010. Physiological responses of Phaseolus vulgaris to different lead concentrations. Pak. J. Bot., 42: 239-246. Hasanuzzaman, M., A.M. Hossain and M. Fujita, 2011. Selenium-induced up-regulation of the antioxidant defense and methylglyoxal detoxification system reduces salinity-induced damage in rapeseed seedlings. Biol. Trace Elem. Res., 143: 1704-1721. Anderson, M.E., 1985. Determination of glutathione and glutathione disulfide in biological samples. Meth. Enzymol., 113: 548-554. Bradford, M., 1976. A rapid and sensitive method for the quantitation and sensitivity of microgram quantities of protein utilising the principle of protein-dye binding. Anal. Biochem., 72: 248-254. Ikemoto, T., T. Kunito, H. Tanaka, N. Baba, N. Miyazaki and S. Tanabe, 2004. Detoxification mechanism of heavy metals in marine mammals and seabirds: Interaction of selenium with mercury, silver, copper, zinc and cadmium in liver. Arch. Environ. Con. Tox., 47: 402-413. Cartes, P., A. Jara, L. Pinilla, A. Rosas and M. Mora, 2010. Selenium improves the antioxidant ability against aluminium-induced oxidative stress in ryegrass roots. Ann. Appl. Biol., 156: 297-307. John, R., P. Ahmad, K. Gadgil and S. Sharma, 2009. Heavy metal toxicity: Effect on plant growth, biochemical parameters and metal accumulation by Brassica juncea. Int. J. Plant Prod., 3: 65-76. Cerklewski, F.L. and R.M. Forbes, 1976. Influence of dietary selenium on lead toxicity in the rat. J. Nutr., 106: 778-783. Djanaguiraman, M., D.D. Devi, A.K. Shanker, A. Sheeba and U. Bangarusamy, 2005. Selenium an antioxidative protectant in soybean during senescence. Plant Soil, 272: 77-86. Kápolna, E., P.R. Hillestrom, K.H. Laursen, S. Husted and E.H. Larsen, 2009. Effect of foliar application of selenium on its uptake and speciation in carrot. Food Chem., 115: 1357-1363. Djanaguiraman, M., P.V.V. Prasad and M. Seppanen, 2010. Selenium protects sorghum leaves from oxidative damage under high temperature stress by enhancing antioxidant defense system. Plant Physiol. Biochem., 48: 999-1007. Krystofova, O., V. Adam, P. Babula, J. Zehnalek, M. Beklova, L. Havel and R. Kizek, 2010. Effects of various doses of selenite on stinging nettle (Urtica dioica L.). Int. J. Environ. Res. Public Health, 7: 3804-3815. Feng, R., C. Wei, S. Tu and X. Sun, 2009. ACKNOWLEDGMENT Gill, S.S. and N. Tuteja, 2010. Reactive oxygen species and antioxidant machinery in abiotic stress tolerance in crop plants. Plant Physiol. Biochem., 48: 909-930. This study in part has been founded through a Major Research Grant by the Jiangxi Natural Science foundation (Project no. 2009GQH0027) and the Science and Technology Projects of Education Bureau of Jiangxi Province (Project no.GJJ10115). Gratão, P.L., C.C. Monteiro, A.M. Antunes, L.E.P. Peres and R.A. Azevedo, 2008. Acquired tolerance of tomato (Lycopersicon esculentum cv. Micro- Tom) plants to cadmium-induced stress. Ann. Appl. Biol., 153: 321-333. DISCUSSION Therefore, an increased GSH/GSSG ratio appeared to be an “overcompensation” resulting from intensified recycling of GSH with the aim to keep it in its active, reduced form. This increase in the GSH/GSSG ratio in Se-supplemented Pb stressed Coleus also provided a clear demonstration of the role of Se towards Pb tolerance. Similar results were 1611 Res. J. Appl. Sci. Eng. Technol., 6(9): 1606-1613, 2013 obtained by Hasanuzzaman et al. (2011) on Brassica napus cv. Bina under salt stress supplied with selenium. Based on the above results, Coleus exposed to Pb stress with Se treatment could protect itself from toxicity of Pb by altering various metabolic processes with the involvement of the whole antioxidant systems. Gajewska, E., M. Skłodowska, M. Słaba and J. Mazur, 2006. Effect of nickel on antioxidative enzyme activities, proline and chlorophyll contents in wheat shoots. Biol. Plantarum., 50: 653-659. Garg, N. and G. Manchanda, 2009. ROS generation in plants: Boon or bane? Plant Biosyst., 143: 81-96. Gedroc, J.J., K.D.M. McConnaughay and J.S. Coleman, 1996. Plasticity in root/shoot partitioning: Optimal, ontogenetic, or both? Funct. Ecol., 10: 44-50. REFERENCES Interactive effects of selenium and arsenic on their uptake by Pteris vittata L. under hydroponic conditions. Environ. Exp. Bot., 65: 363-368. Liu, S.Z., D.J. Lin and S.J. Tang, 2004. Purification of eutrophic wastewater by Cyperus alternifolius, Coleus blumei and Jasminum sambac planted in a floating phytoremediation system. Chin. J. Appl. Ecol., 15: 1261-1265. Filek, M., R. Keskinen, H. Hartikainen, I. Szarejko, A. Janiak, Z. Miszalski and A. Golda, 2008. The protective role of selenium in rape seedlings subjected to cadmium stress. J. Plant Physiol., 165: 833-844. Lobanov, A.V., D.L. Hatfield and V.N. Gladyshev, 2008. Reduced reliance on the trace element selenium during evolution of mammals. Genome Biol., 9: R62. Fu, J.M. and B.R. Huang, 2001. Involvement of antioxidants and lipid peroxidation in the adaptation of two cool-season grasses to localized drought stress. Environ. Exp. Bot., 45: 105-114. Lyons, G.H., Y. Genc, K. Soole, J.C.R. Stangoulis, F. Liu and R.D. Graham, 2009. Selenium increases seed production in Brassica. Plant Soil, 318: 73-80. 1612 Res. J. Appl. Sci. Eng. Technol., 6(9): 1606-1613, 2013 Pedrero, Z., Y. Madrid, H. Hartikainen and C. Cámara, 2008. Protective effect of selenium in broccoli (Brassica oleracea) plants subjected to cadmium exposure. J. Agric. Food Chem., 54: 2412-2417. Mishra, S., S. Srivastava, R.D. Tripathi, R. Kumar, C.S. Seth and D.K. Gupta, 2006. Lead detoxification by coontail (Ceratophyllumdemersum L.) involves induction of phytochelatins and antioxidant system in response to its accumulation. Chemosphere, 65: 1027-1039. Soudani, N., M. Sefi, I. Ben Amara, T. Boudawara and N. Zeghal, 2010. Protective effects of selenium (Se) on chromium (VI) induced nephrotoxicity in adult rats. Ecotox. Environ. Safe, 73: 671-678. Mroczek-Zdyrska, M. and M. Wójcik, 2012. The influence of selenium on root growth and oxidative stress induced by lead in Vicia faba L. minor Plants. Biol. Trace Elem. Res., 147: 320-328. Terry, N., A.M. Zayed, M.P. de Souza and A.S. Tarun, 2000. Selenium in higher plants. Ann. Rev. Plant Physiol. Plant Mol. Biol., 51: 401-432. Namjooyan, S., R. Khavari-Nejad, F. Bernard, S. Namdjoyan and H. Piri, 2012. The effect of cadmium on growth and antioxidant responses in the safflower (Carthamus tinctorius L.) callus. Turk. J. Agr. Forest., 36: 145-152. Zembala, M., M. Filek, S. Walas, H. Mrowiec, A. Kornaś, Z. Miszalske and H. Hartikainen, 2010. Effect of selenium on macro- and microelement distribution and physiological parameters of rape and wheat seedlings exposed to cadmium stress. Plant Soil, 329: 457-468. Noctor, G. and C.H. REFERENCES Foyer, 1998. Ascorbate and glutathione: Keeping active oxygen under control. Ann. Rev. Plant Physiol. Plant Mol. Biol., 49: 249-279. Zhao, L.Z., D. Mao and Z.Y. Lin, 2007. Efects of diferent nutrient solution on pigment content and photosynthesis of Coleus blume. Guangdong Agric. Sci., 6: 30-32. (In Chinese) Panizza de León, A., R. Carrillo González, M. Bernal González, M. Vaca Mier and C. Durán- Domínquez-de-Bazúa, 2011. Exploration of the ability of Coleus blumei to accumulate aluminum. Int. J. Phytoremed., 13: 421-33. 1613
W2970401288.txt	http://www.journals.vu.lt/nonlinear-analysis/article/download/14024/12940	en		Phenomenological model of bacterial aerotaxis with a negative feedback	Nonlinear analysis	2,013	cc-by	9,460	Nonlinear Analysis: Modelling and Control, 2013, Vol. 18, No. 2, 227–249 227 Phenomenological model of bacterial aerotaxis with a negative feedback∗ Vladas Skakauskasa,1 , Pranas Katauskisa , Remigijus Šimkusb , Feliksas Ivanauskasa a Faculty of Mathematics and Informatics, Vilnius University LT-03225 Vilnius, Lithuania vladas.skakauskas@maf.vu.lt; pranas.katauskis@mif.vu.lt; feliksas.ivanauskas@maf.vu.lt b Vilnius University Institute of Biochemistry LT-08662 Vilnius, Lithuania simkus@bchi.lt Received: 25 June 2012 / Revised: 26 February 2013 / Published online: 18 April 2013 Abstract. A phenomenological model for the suspension of the aerotactic swimming microorganisms placed in a chamber with its upper surface open to air is presented. The model was constructed to embody some complexity of the aerotaxis phenomenon, especially, changes in the average bacteria drift velocity under changing environmental conditions. It was assumed that effective forces applied to the cell (gravitational, drag, and thrust) should be essential for the overall system dynamics; and that bacterial propulsion force, but not their swimming velocity, is proportional to the gradient of the oxygen concentration. Mathematically, the model consists of three coupled equations for the oxygen dynamics; for the cell conservation; and for the balance of forces acting on bacteria. An analytical steady-state solution is given for the shallow and deep layers and numerical results are given for the steady-state and initial value problems which are compared with corresponding ones to the Keller–Segel model. Keywords: bioconvection, thermo-bioconvection, swimming microorganisms, oxytactic bacteria. 1 Introduction The term aerotaxis (or oxytaxis) refers to the situation where bacterium moves towards or away from air or oxygen. Aerotaxis can be regarded as a kind of the more general process, chemotaxis, which is a motion of bacteria towards a favorable chemical field. The basic mathematical model in chemotaxis was introduced by Keller and Segel (KS) [1, 2]. In its original form this model consists of four coupled reaction–advection–diffusion equations. Under the quasi-steady-state assumptions this model can be reduced to two coupled parabolic equations for the concentration of microorganisms and the attracting ∗ This work was supported by the Research Council of Lithuania (project No. MIP-052/2012). author. 1 Corresponding c Vilnius University, 2013 228 V. Skakauskas et al. species (attractants). Mathematical modeling of chemotaxis on the basis of the KS model has developed into a large and diverse discipline [3, 4]. KS type equations were used to describe the oxytactic motion of bacteria in a water column as well [5–7]. Hillesdon, Pedley and Kessler (HPK) [6] and Hillesdon and Pedley (HP) [7] considered dynamics of an oxytactic bacteria Bacillus subtilis suspension placed in a chamber with its upper surface open to air. In the HPK model the phenomena of gravitational sedimentation, bulk fluid motion, and diffusion of inactive cells were assumed to be negligible. The assumption that the sedimentation rate is much smaller than the typical cell swimming speed is inaccurate in the case of inactive cells [7]. The typical form of the KS model for the oxytactic bacteria is [5–7]:  ∂C   = div(κ1 ∇C) − κ2 (C)B,  ∂t    ∂B = div −κ3 (C)B∇C + κ4 (C)∇B . ∂t (1) Here C and B are the oxygen and cells concentrations, κ1 and κ4 (C) are the diffusivity of oxygen and bacterial cells, κ2 (C) is the oxygen consumption rate by cells, v = κ3 (C)∇C is the oxytactic bacteria drift velocity, κ3 (C) is the oxytactic sensitivity, ∇ and div are the gradient and divergence operators. In the steady-state, wide chamber, and constant κ1 , . . . , κ4 case, HPK gave an analytic solution of the model for both shallow and deep chambers. In the deep chamber case, the authors neglected diffusion of inactive cells and therefore one constant was not determined. This constant was obtained by numerically solving the initial value problem. In the time-dependent one-dimensional case with constant κ1 and depending on C coefficients κ2 , κ3 , and κ4 , HPK solved model (1) numerically using the method of lines. HP examined the stability of the steadystate solution. To describe the convective chemotaxis Dombrowski et al. [8], Tuval et al. [9] generalized the KS model by including the bulk fluid motion. This model describes the collective behavior (bioconvection) of a suspension of oxytactic bacteria in an incompressible fluid under assumptions [10] that the contribution of bacteria to the bacteria–fluid suspension is sufficiently small and that more detailed cell–cell interactions (e.g., of hydrodynamic type) are neglected. In [10], this model was studied numerically in detail. The solvability of the model was examined by Lorz [11], Duan et al. [12], and Di Francesco et al. [13]. Becker et al. [14] and Kuznetsov [15, 16] generalized the KS model and studied the bioconvection of oxytactic cells in a fluid saturated porous medium. Kuznetsov also investigated models for the thermo-bioconvection of oxytactic cells in a fluid layer [17–19] and in a fluid saturated porous layer [15, 16] and carried out the stability analysis of their steady-state solutions. In all models of bioconvection and thermo-bioconvection, the oxytactic bacteria drift velocity, as in the KS model, is proportional to the gradient of oxygen concentration while the gravitational force is approximated by the buoyancy term. Papers of Alloui et al. [20,21] are devoted to numerical study of the development of gravitactic bioconvection and thermo-bioconvection of swimming microorganisms which are little denser than water and move randomly, but on the average, upwardly against www.mii.lt/NA Phenomenological model of bacterial aerotaxis with a negative feedback 229 gravity. Alloui et al. [22] also carried out the linear stability analysis of the thermobioconvection of swimming against gravity microorganisms. We note three problems that arise in mathematical modeling of bacterial aerotaxis. Firstly, it should be noted, that an important aspect of KS chemotaxis models is the expected onset of chemotactic collapse [3, 4, 23]. This term refers to the fact that, under suitable circumstances, the whole population should concentrate in a single point in finite time. However, it seems likely, that real bacterium is searching for the optimal place to be. Therefore, the overcrowding of bacteria in small space domains is unrealistic due to lack of nutrition for population in this domain. A number of modifications have been made to the minimal KS model of auto-aggregation that allows preventing such unrealistic singularities [4]. In general, this means, that in real systems there are certain dispersal mechanisms, which can be regarded as a negative chemotaxis and/or the suppression of the positive chemotaxis. The corresponding dispersal mechanisms were not taken into account in the KS type models of aerotaxis. Secondly, in real systems, the response of microorganisms to oxygen is much more complicated than the simplest KS model suggests [24–27]. It is known, that dependently on the type of bacteria and local environmental conditions, oxygen can act as an attractant or as a repellent ( [24,25] and references therein). The positive aerotaxis (oxygen is attractant) results in aggregation of cells at the oxygen-exposed surfaces and the negative aerotaxis (oxygen is repellent) imply dispersal of the cell. Thus, again, a certain mechanism of the suppression of the positive aerotaxis should be included to the overall dynamic system. Thirdly, a common feature of many chemotaxis models based on the approximation of the cells drift velocity by the gradient of the chemoattractant or chemorepellent is to incorporate some complexity of the chemotaxis into the equations through a chemotactic sensitivity function (see system (1)). But different microorganisms detect spatial gradients of the chemical signal through distinct mechanisms. Certain cells [28], such as Dystyostelium discoideum, fibroblasts and leukocytes, can detect and respond to a small gradient in the chemical signal across the length of their body using a process of internal amplification and polarisation. Smaller cells [29], such as E. coli, detect a gradient by sampling the concentration at different time points and modifying their movement accordingly. The observation that the gradient sensed by bacteria is temporal means that bacteria possess a memory, which compares past information with present information to make a decision. This memory is long enough so that the bacteria can make an accurate comparison between two points more distal than the bacterial body length. In both cases, the signal detected by the cell is intrinsically non-local and it may therefore be appropriate to consider movement based on non-local gradient by the integration of the signal by the cell over some region. For cells which detect a gradient in the chemical signal across the length of their body, Othmer and Hillen R [30] approximated their drift velocity by the formula χCn/(ωρ) S n−1 σB(t, x + ρσ) dσ where ω = \|S n−1 \|, S n−1 denotes the (n − 1)-dimensional unit sphere in Rn , and ρ is the radius of a sphere which enclose the cell. Analytical and numerical study of this model is given in [31]. Studies of smaller cells revealed (see [28] and literature therein) that, like many other sensory systems, the chemotactic response involves two processes: excitation and adaptation. When bacteria are stimulated, their swimming mode are changed instantaneously. This initial process, termed excitation, is very fast. Later on, Nonlinear Anal. Model. Control, 2013, Vol. 18, No. 2, 227–249 230 V. Skakauskas et al. bacteria resume their prestimulus behaviour, even though the stimulus is still present. This process, termed adaptation, is relatively slow (in the range of seconds or minutes). Adaptation thus enable bacteria to adjust to changes in the stimulus intensity and respond to new stimuli. Models of KS type are based on the quasi-steady-state approximation of the cells drift velocity and therefore cannot describe correctly the adaptation period. The aim of this paper is study of the dynamics of the dilute suspensions of aerotactic bacteria by the modified KS model. To describe some complexity of the aerotaxis phenomenon (especially, the bacteria adaptation period under changes in environmental conditions), we replace the quasi-steady-state equation v = κ3 (C)∇C by the momentum equation for cells drift velocity. In the modified model it is assumed that 1) gravitational, drag, and thrust forces applied to the cell are not negligible and 2) that bacterial thrust force, but not its average drift velocity, is proportional to the gradient of the oxygen concentration. The model consists of three coupled equations: 1) the equations for oxygen dynamics; 2) cell conservation equation; and 3) momentum equation for the cells drift velocity. This original model of oxytaxis is termed as FB (Feedback/Force balance) model. In the wide chamber case, we give an analytic solution of the model for both shallow and deep chambers and discuss the numerical results of the initial value and steadystate problems. We also solved the KS model and using numerical results demonstrate difference between the FB and KS models. The paper is organized as follows. In the auxiliary Section 2, we introduce the KS model and give a detail derivation of its steady-state solution found by HPK. In Section 3, we present the FB model and demonstrate its steady-state solution. In Section 4, we compare numerical results for the KS and FB models. Some remarks in Section 5 conclude the paper. 2 The one-dimensional Keller–Segel (KS) model In this auxiliary section, we introduce the KS model and give a detail derivation of its steady-state solution which was found by HPK [6]. Let x be the vertical coordinate. In the one-dimensional case, Eqs. (1) can be written in the form  ∂ ∂C ∂C   = κ1 − κ2 (C)B, x ∈ (0, h),  ∂t ∂x ∂x  ∂B ∂ ∂C ∂B   = −κ3 (C)B + κ4 (C) , x ∈ (0, h), ∂t ∂x ∂x ∂x (2)   C(0, x) = C0 , B(0, x) = B0 , C(t, h) = C0 ,      ∂C ∂B = = 0, ∂x ∂x x=0  x=0     ∂C ∂B   κ3 (C)B − κ4 (C) = 0. ∂x ∂x x=h (3) www.mii.lt/NA Phenomenological model of bacterial aerotaxis with a negative feedback 231 Here x = 0 and x = h correspond to the bottom and surface of the chamber, the constant C0 is the initial and atmosphere oxygen concentration, B0 is the initial cells concentration. The total number of cells is equal to the hB0 . Now we introduce the dimensionless variables. Set Θ = (C − Cmin )/(C0 − Cmin ), where Cmin is the minimal value of the oxygen concentration necessary for cells to be active. Let κ2 (C) = κ20 w2 (Θ), κ3 (C) = κ30 w3 (Θ), κ4 (C) = κ40 w4 (Θ), with constants κ20 , κ30 , and κ40 . Assuming that κ1 is also constant HPK considered the case where w2 = w3 , w4 = w22 , with w2 the Heaviside step function. Set x = hx̄, B = B0 B̄, t = (h2 /κ40 )t̄. Omitting the bar, we rewrite (2), (3) in the dimensionless form  2 ∂ Θ ∂Θ  2   − β w (Θ)B , x ∈ (0, 1), = δ 2   ∂t ∂x2      ∂B ∂ ∂B ∂Θ   = w (Θ) − αw (Θ)B , x ∈ (0, 1), 4 3   ∂t ∂x ∂x ∂x     ∂Θ ∂B = = 0, (4) ∂x ∂x x=0  x=0     ∂B ∂Θ   = 0, w (Θ) − αw (Θ)B  4 3   ∂x ∂x x=1     Θ(t, 1) = 1,    B(0, x) = Θ(0, x) = 1, where δ = κ1 /κ40 , β 2 = B0 κ20 h2 /(κ1 (C0 − Cmin )), α = κ30 (C0 − Cmin )/κ40 . Note that α is swimming upwards parameter. Integrating over (0, 1) Eq. (4)2 and using R1 conditions (4)3,4 , we get that 0 B(t, x) dx = 1 is preserved. In the steady-state case, system (4) reads  Θ00 = β 2 w2 (Θ)B,     0 0 0    w4 (Θ)B − αw3 (Θ)BΘ = 0, (5) Θ(1) = 1,    0 0 w4 (1)B (1) − αw3 (1)B(1)Θ (1) = 0,     Θ0 (0) = B 0 (0) = 0. In addition, we formulate the condition Z1 B(x) dx = 1. (6) 0 Now we derive the HPK steady-state solution of system (5) and (6). 2.1 The shallow layer case (Θ > 0 ∀x ∈ (0, 1]) Integrating (5)2 and using condition (5)4 we get the equation w4 (Θ)B 0 = αw3 (Θ)BΘ0 . Nonlinear Anal. Model. Control, 2013, Vol. 18, No. 2, 227–249 (7) 232 V. Skakauskas et al. In this section, we consider the case w2 = w3 = w4 = 1. From (5)1 and (7) we get the equation B 0 = αβ −2 θ0 θ00 which has the solution B = (α/2)β −2 (θ0 )2 +B(0). Hence, r 2p 0 B − B(0). (8) Θ =β α From (7) and (8) we derive the equation r B 0 = αβB 2 B − B(0) α which has the solution r −2 B = B(0) cos (ξx), ξ=β αB(0) ∈ 2 π 0, . 2 (9) Then combining (8) and (9) we get Θ0 = 2 ξ tan(ξx), α Hence Θ =1− Θ(1) = 1. 2 cos(ξx) ln . α cos ξ (10) Since Θ must be nonnegative the condition cos(ξx)/ cos ξ 6 eα/2 has to be satisfied for all x ∈ [0, 1]. Hence cos ξ > e−α/2 or cos−2 ξ 6 eα and 0 6 ξ 6 arccos e−α/2 . From condition (6) and Eq. (9) we get the equation q = tan ξ, ξ (11) where q = αβ 2 /2. Eq. (11) has a unique solution ξ = ξ(q) ∈ (0, arccos e−α/2 ) growing together with q since dξ/dq > 0. Then, from Eqs. (9)2 and (11) it follows that B(0) = 2 −2 2 2 α −α/2 ξ 2 (q)/q and > √ cos ξ(q) = 1 + √tan ξ(q) = 1 + q /ξ (q) 6 e . Thus arccos e ξ(q) > q/ eα − 1 = αβ 2 /(2 eα − 1) and, hence, β 2 6 β∗2 = 2√ α e − 1 arccos e−α/2 . α (12) Thus θ(0) > 0 if 0 < β < β∗ , and θ(0) = 0 if β = β∗ . Values of β that satisfy the inequality β 6 β∗ correspond to a shallow chamber. All others correspond to a deep chamber. Note that ∂Θ/∂q = (2/α)(− tan ξ + x tan xξ) dξ/dq < 0 ∀x ∈ (0, 1). 2.2 The deep layer case In this case β > β∗ and positive Θ determined by Eq. (10) does not exist for all x ∈ [0, 1]. Θ is positive in a layer (x∗ , 1], Θ(x∗ ) = 0, where x∗ > 0 is unknown a priori. There is not www.mii.lt/NA Phenomenological model of bacterial aerotaxis with a negative feedback 233 enough oxygen available in the layer [0, x∗ ] and cells are inactive. Therefore w2 (Θ) = 0 and we have to consider the task Θ00 = 0, Θ0 (0) = Θ(x∗ ) = 0. Hence, Θ = 0 for all x ∈ [0, x∗ ]. Independently of w3 and w4 from (7) it follows that B(x) = B(x∗ ) ∀x ∈ [0, x∗ ], where B(x∗ ) is also unknown a priori. HPK neglected diffusion of inactive cells and therefore B(x∗ ) was not determined by the conditions of the steady-state case. HPK determined it by solving the initial value problem. Differently from HPK, we postulate the continuity of B and B 0 at x∗ . We also use the continuity condition for Θ and Θ0 at x∗ . In (x∗ , 1] we have to solve Eqs. (5)1,2,3,4 and (6). Let w2 = w3 = w4 = 1. By the argument used for the shallow chamber and by the continuity conditions at x∗ we get r   −2 B = B(x∗ ) cos ˜ − x∗ ) , ξ˜ = β αB(x∗ ) , ξ(x  2 (13) ˜ − x∗ )) 2 cos( ξ(x   >0 Θ = 1 − ln ˜ − x∗ )) α cos(ξ(1 for x ∈ (x∗ , 1] and (1 − x∗ )ξ˜ ∈ (0, π/2). Due to the condition Θ(x∗ ) = 0 it follows that ˜ − x∗ )). Hence −α/2 = ln cos(ξ(1 √ ˜ − x∗ ) = eα − 1. tan ξ(1 (14) At last, from condition (6) we get Z1 1 = B(x∗ )x∗ + B(x) dx = B(x∗ ) x∗ + x∗ 1 . ˜ − x∗ )) ξ˜ tan(ξ(1 (15) Combining the last two equations and using the definition of ξ˜ we get the equation √ α 2ξ˜2 e −1 1= x∗ + αβ 2 ξ˜ which has the solution ξ˜ = p eα − 1 + 2x∗ αβ 2 − 2x∗ √ eα − 1 . (16) Now from (14) it follows that (1 − x∗ )ξ˜ = η(α) := arccos e−α/2 . Hence, √ 1 − x∗ p α e − 1 + 2x∗ αβ 2 − eα − 1 = η(α). 2x∗ This equation has the solution q √ √ (η − eα − 1 )2 + 2αβ 2 − (η − eα − 1 ) x∗ = 1 − η ∈ (0, 1) αβ 2 Nonlinear Anal. Model. Control, 2013, Vol. 18, No. 2, 227–249 (17) 234 V. Skakauskas et al. such that dx∗ /dβ > 0. Combining (16) and (17) we derive an equation for ξ˜ and then from (15) get an equation for B(x∗ ):  αβ 2   > 0, ξ˜ = q  √ √   (η − eα − 1 )2 + 2αβ 2 − (η − eα − 1 ) (18) 2αβ 2   q B(x ) = .  ∗  √ √  [ (η − eα − 1 )2 + 2αβ 2 − (η − eα − 1 )]2 We see that the steady-state solution does not depend on the positive diffusivity of inactive cells. In Section 4, we give the numerical solution of the initial value problem (4) determined by using the finite-difference scheme. 3 The feedback/force balance (FB) model Keller and Segel neglected the bulk fluid velocity and approximated the average velocity of swimming upwards oxytactic bacteria by the quasi steady-state formula v = κ3 (C)∇C. We also neglect the bulk fluid motion but, to describe some complexity of the aerotaxis phenomenon (especially, the bacteria adaptation period under changes in environmental conditions), we use the momentum equation for cells which includes the gravitational sedimentation of cells, swimming upwardly strength (force which arises from chemotaxis and enables cells to swim), and resistance force to movement of cells through the fluid. We postulate that swimming upwards strength of cells, but not their average velocity, is parallel and proportional to ∇C. For simplicity, we neglect the bacteria to bacteria communication and convective acceleration, (v · ∇)v, terms in the momentum equation for cells. In the one-dimensional case, the model consists of the equations  ∂C ∂ ∂C   = κ − κ2 (C)B, x ∈ (0, h), 1   ∂t ∂x ∂x    ∂B ∂ ∂B (19) = −Bv + κ (C) , x ∈ (0, h), 4  ∂t ∂x ∂x      B ∂v = −Bg̃ − κ (C)v + κ̃ (C) ∂C , x ∈ (0, h), 5 3 ∂t ∂x subject to the conditions  C(0, x) = C0 , B(0, x) = B0 , v(0, x) = v0 ,      C(t, h) = C , ∂C = 0, 0 (20) ∂x x=0    ∂B   = 0.  Bv − κ4 (C) ∂x x=0;h Here κ1 = const, g̃ = g(1 − ρw /ρB ) where g is the acceleration due to the gravity, ρw and ρB are the water (fluid) and a cell density. Note that the dimensions of κ3 in Section 2 www.mii.lt/NA Phenomenological model of bacterial aerotaxis with a negative feedback 235 and κ̃3 in Section 3 are different. Integrating Eq. (19)3 we get a non-local in time equation for average cells drift velocity v, Zt ∂C(ξ, x) −B(ξ, x)g̃ + κ̃3 C(ξ, x) v(t, x) = v0 (x)Π(0, t, x) + Π(ξ, t, x) dξ ∂x 0 Rt where Π(ξ, t, x) = exp{− ξ κ5 (C(τ, x)) dτ }. Let κ2 (C) = κ20 w2 (Θ), κ̃3 (C) = κ̃30 w3 (Θ), κ4 (C) = κ40 w4 (Θ), κ5 (C) = κ50 w5 (Θ), Θ = (C − Cmin )/(C0 − Cmin ), x = x̄h, B = B0 B̄, t = (h2 /κ40 )t̄, v = v∗ v̄. Omitting the bar we rewrite system (19) and (20) in the dimensionless form 2  ∂ Θ ∂Θ 2   − β w2 (Θ)B , x ∈ (0, 1), =δ   ∂t ∂x2    ∂B ∂ ∂B (21) = − γBv , x ∈ (0, 1), w4 (Θ)  ∂t ∂x ∂x      B ∂v = −ρ B − ρ w (Θ)v + ρ w (Θ) ∂Θ , x ∈ (0, 1), 1 3 5 2 3 ∂t ∂x   Θ(0, x) = 1, B(0, x) = 1, v(0, x) = v0 ,        ∂Θ = 0, Θ\|x=1 = 1, (22) ∂x x=0     ∂B    w4 (Θ) = 0. − γvB  ∂x x=0;1 Here δ = κ1 /κ40 , γ = v∗ h/κ40 , β 2 = B0 κ20 h2 /(κ1 (C0 −Cmin )), ρ1 = g̃h2 /(κ40 v∗ ) = g̃h3 /(γκ240 ), ρ2 = κ̃30 h(C0 − Cmin )/(v∗ B0 κ40 ), ρ3 = h2 κ50 /(B0 κ40 ). R1 Note, that model (21)–(22) preserves condition 0 B(t, x) dx = 1. Determining a1 = ρ1 γ/ρ3 and a2 = ρ2 γ/ρ3 we rewrite Eq. (21)3 in the form ∂v ρ3 ∂Θ B = −a1 B − γω5 (Θ)v + a2 ω3 (Θ) . ∂t γ ∂x Now we consider the steady-state case. 3.1 The shallow chamber case (Θ > 0 ∀x ∈ (0, 1]) We study the case where w2 (Θ) = w3 (Θ) = w4 (Θ) = w5 (Θ) = 1. From Eqs. (21) and (22) we get  Θ00 = β 2 B,     γvB = B 0 , a2 a1 1 (23) v = (ρ2 Θ0 − ρ1 B) = Θ0 − B,    ρ3 γ γ    0 Θ (0) = 0, Θ(1) = 1. Nonlinear Anal. Model. Control, 2013, Vol. 18, No. 2, 227–249 236 V. Skakauskas et al. In addition, we formulate the condition Z1 B(x) dx = 1. (24) 0 Integrating (23)1 and using (24) we get Θ0 (1) = β 2 . (25) Then from Eqs. (23)1,2,3 it follows that γv = B0 dB = a2 Θ0 − a1 B. = β2 B dΘ0 (26) The last equation of (26) is linear and integrates to get a2 a1 a2 a2 B(x) = B̂ Θ0 , B(0) = B(0) + 2 β 2 exp − 2 Θ0 + Θ0 − 2 β 2 . a1 β a1 a1 (27) From (26) we can see that v = 0 and dB/dΘ0 = 0 at B = (ρ2 /ρ1 )Θ0 . Now, inserting this value of B into (27), we get β2 B(0)a21 0 Θ = . (28) ln 1 + a1 a2 β 2 Then from (26) and (23)1 it follows that γ 00 0 γ 0 00 0 ρ B \|B= ρ2 Θ0 = B (ρ2 Θ − ρ1 B) + B (ρ2 Θ − ρ1 B ) 1 ρ3 ρ3 ρ2 = B 2 γβ 2 > 0. ρ3 B=(ρ2 /ρ1 )Θ 0 Hence, (28) is the point of a minimum of B and ρ2 β 2 a2 B(0) min B = > 0. ln 1 + 1 2 ρ1 a1 a2 β We integrate Eqs. (23)1 and (27) and use condition (23)4 to get β 2 x = x̂ Θ0 , B(0) := ZΘ 0 0 dy B̂(y, B(0)) . (29) From here and by (25) we get the equation for B(0), β2 = Zβ 0 2 dy B̂(y, B(0)) . (30) www.mii.lt/NA Phenomenological model of bacterial aerotaxis with a negative feedback 237 After this equation is solved for B(0), we determine B = B̂(Θ0 , B(0)) by Eq. (27), x = x̂(Θ0 , B(0)) by Eq. (29), and v by (23)3 for Θ0 ∈ [0, β 2 ]. At last, using (29) we integrate the equation dΘ = Θ0 dx to get 1 Θ =1− 2 β Zβ 2 Θ0 y dy . (31) . (32) B̂(y, B(0)) This formula shows that Θ > 0 for all x ∈ [0, 1] if 2 β 2 6 β̃∗2 := Zβ̃∗ 0 y dy B̂(y, B(0)) Values β that satisfy condition β 6 β̃∗ correspond to shallow chamber. Substituting y = β̃∗2 z, B(0) = β̃∗2 B̃(0) we rewrite Eqs. (30) and (31) in the form β̃∗2 Z1 = 0 dz , B̃(z, B̃(0)) Z1 1= 0 z dz , B̃(z, B̃(0)) (33) where B̃(z, B̃(0)) = (B̃(0) + a2 /a21 ) exp(−a1 z) + a2 z/a1 − a2 /a21 . From (33)2 we determine B̃(0) and then by (33)1 find β̃∗2 . Clearly, β̃∗2 > 1. 3.2 The deep chamber case (β > β̃∗ ) Since function (31) is not positive for all x > 0, we divide (0, 1] into two intervals (0, x̃∗ ] and (x̃∗ , 1], Θ(x̃∗ ) = 0. In (x̃∗ , 1], Θ > 0, and cells consume oxygen and are active. Since Θ(x̃∗ ) = 0, there is not enough oxygen available in (0, x̃∗ ). Hence, cells are inactive and therefore κ2 (Θ) = 0 ∀x ∈ [0, x̃∗ ]. Thus in (0, x̃∗ ) we have the system   Θ00 = 0, Θ(x̃∗ ) = 0, Θ0 (0) = 0,    γvB = w4 (Θ)B 0 ,  1   ρ2 ω3 (Θ)Θ0 − ρ1 B . v = ρ3 w5 (Θ) It is easy to see that Θ = 0, v = −ρ1 B/(ρ3 w5 (Θ)) < 0, B 0 = −qB 2 , q = γρ1 / (ρ3 w4 (0)w5 (0)) ∀x ∈ (0, x̃∗ ). Therefore B(x) = B(x̃∗ ) B(0) = . 1 − qB(x̃∗ )(x̃∗ − x) 1 + qxB(0) (34) In (x̃∗ , 1), we consider the case w2 = w3 = w4 = w5 = 1, and solve Eqs. (23) with conditions Θ(1) = 1, Θ(x̃∗ ) = Θ0 (x̃∗ ) = 0. By the argument used for the shallow Nonlinear Anal. Model. Control, 2013, Vol. 18, No. 2, 227–249 238 V. Skakauskas et al. chamber we get  0 ZΘ   dy 1    x − x̃∗ = β 2 B̂(y, B(x̃ )) ,  ∗   0   0 Θ Z (1) 1 y dy    Θ =1− 2 ,   β  B̂(y, B(x̃ )) ∗   Θ0    B(x) = B̂ Θ0 , B(x̃∗ ) , (35) where B̂ Θ0 , B(x̃∗ ) = B(x̃∗ ) + a2 2 a1 0 a2 a2 β exp − Θ + Θ0 − 2 β 2 . 2 2 a1 β a1 a1 From condition (24) by Eqs. (23)1 and (34) we derive the equation Zx̃∗ 1= B dx + 1 β2 0 Z1 Θ00 dx = 1 0 Θ (1) + B(0) β2 Zx∗ dx 1 + a1 xB(0) 0 x̃∗ 1 1 1 1 = 2 Θ0 (1) + ln 1 + a1 x̃∗ B(0) = 2 Θ0 (1) − ln 1 − a1 x̃∗ B(x̃∗ ) . β a1 β a1 Hence, Θ0 (1) = β 2 u x̃∗ , B(x̃∗ ) , 1 u x̃∗ , B(x̃∗ ) = 1 + ln 1 − a1 x̃∗ B(x̃∗ ) > 0 a1 (36) provided that B(x̃∗ ) < (1 − e−a1 )/(a1 x̃∗ ). Inserting x = 1, Θ0 (x̃∗ ) determined by (36) into Eq. (35)1 and Θ(x̃∗ ) = 0, Θ0 (x̃∗ ) from (36) into Eq. (35)2 we get β2 = β 2 u(x̃Z∗ ,B(x̃∗ )) 0 y dy B̂(y, B(x̃∗ )) , 1 1 − x̃∗ = 2 β β 2 u(x̃Z∗ ,B(x̃∗ )) 0 dy B̂(y, B(x̃∗ )) , 2 which by substitution y = β u(x̃∗ , B(x̃∗ ))ξ can be transformed into  Z1   u(x̃∗ , B(x̃∗ )) dξ    1 − x̃∗ = B̂(β 2 u(x̃ , B(x̃ ))ξ, B(x̃ )) ,  0      1 = β2   Z1 0 ∗ ∗ ∗ u2 (x̃∗ , B(x̃∗ )) ξ dξ B̂(β 2 u(x̃∗ , B(x̃∗ ))ξ, B(x̃∗ )) (37) . In Section 4, we solve initial value problem (21)–(22) by using the finite-difference scheme. www.mii.lt/NA Phenomenological model of bacterial aerotaxis with a negative feedback 4 239 Numerical results 4.1 The initial-value problem for KS and FB models Problems (4) and (21)–(22) are nonlinear and no their analytical solutions could be derived. To solve these problems the finite-difference technique [32] was used. An implicit finite-difference schemes (see Appendix) were applied. The approximation resulted in the systems of linear algebraic equations with tridiagonal matrix which are solved effectively by using the elimination method [32]. To find a numerical solution the uniform discrete grids in space and time directions were introduced. The constant dimensionless steps of size 0.001 in space and time directions were used. The difference schemes preserve a disR1 crete analogue of condition 0 B(t, x) dx = 1. Digital experiments show that numerical solutions of both models for large time practically coincide with their exact steady-state solutions. The numerical experiments for different values of space and time steps show that the difference schemes are stable. Numerical experiments performed for system (21)–(22) with w2 (Θ) = 1, x ∈ (0, 1), show that, in the case of deep and some shallow layers, Θ becomes negative near the bottom of the chamber for small time, but later it becomes positive and tends to correct steady-state value as time grows. To preserve the positivity of Θ we used w2 (Θ) = 0 for x ∈ (0, x∗ ) and w2 (Θ) = 1 for x ∈ (x∗ , 1). The similar scheme was used to solve system (4). 4.2 The steady-state Keller–Segel model In the case of the shallow chamber, for fixed α and β, we determine q = αβ 2 /2 and by Eq. (12) calculate β∗ . Then, for β < β∗ , to find ξ we solve Eq. (11) by splitting the interval in half. Knowing q and ξ(q) we determine B(0) = 2αξ 2 /β 2 and from Eqs. (9) and (10) find B and Θ for x ∈ [0, 1]. Numerical results are demonstrated in Figs. 1–4 and Table 1. In the deep layer case for fixed α and β, we calculate η(α), then, by Eqs. (17) and ˜ and B(x∗ ). Finally, by Eqs. (13) find B(x) and Θ(x). Numerical (18), determine x∗ , ξ, results are exhibited in Figs. 5, 6 and Table 1. 4.3 The steady-state FB model In the shallow chamber case, to determine B̃(0), for fixed a1 and a2 , we solve (33)2 by splitting the interval in half. Knowing B̃(0) we determine β̃∗ from Eq. (33)1 . Then by formula B(0) = β̃∗ B̃(0) we calculate B(0). Now, from Eqs. (29) and (31) for given β and B(0), we determine x(Θ0 ) and Θ(Θ0 ) for Θ0 ∈ [0, β 2 ]. Results are exhibited in Figs. 1–4 and Table 2. In the deep layer case, in order to determine x̃∗ and B(x̃∗ ), for fixed β, a1 , and a2 , we solve system (37). This system is solved by determining numerically the point of the intersection of curves (37)1 and (37)2 . Each curve we draw by determining B(x̃∗ ) of argument x̃∗ . To calculate B(x̃∗ ) for a given x̃∗ we use a method of splitting the interval in half. Then, knowing x̃∗ , β, B(x̃∗ ), from Eq. (35) we determine x(Θ0 ) and Θ(Θ0 ) for Θ0 ∈ [0, β 2 ]. At last, by Eq. (34), we calculate B(x) for x ∈ [0, x̃∗ ] and, for given γ, find v by equation v = (1/γ)(a2 Θ0 − a1 B), x ∈ [0, 1]. Results are exhibited in Figs. 5, 6 and Table 2. Nonlinear Anal. Model. Control, 2013, Vol. 18, No. 2, 227–249 240 V. Skakauskas et al. 4.4 Analysis of the numerical results The selection of parameters for numerical simulation approximately corresponds to the practically relevant values reported in the literature with the significant ranges in variation to allow demonstration of different regimes of microorganisms transport. We employ parameters that were used in the most calculations in [6,26,27]: κ1 = 2.12 · 10−3 mm2 s−3 , κ20 = 105 s−1 , κ40 ∈ (1.3 · 10−4 , 1.3 · 10−3 ) mm2 s−1 , B0 = 105 mm−3 , C0 = 1.5 · 1014 mm−3 , v∗ = 7.4 · 10−3 mm s−1 . In all calculations we used δ = 16.3 and γ = 460. Plots in Figs. 1–4 and 5, 6 depict the behaviour of bacteria and oxygen concentrations B and Θ for shallow and deep chambers, respectively. 1,0 2,5 0,8 2,0 0,6 B 1,5 Θ 0,4 1,0 0,2 0,0 0,0 0,2 0,4 0,6 0,8 0,5 0,0 1,0 0,2 0,4 0,6 0,8 1,0 x x (a) (b) Fig. 1. The steady-state concentrations of the oxygen Θ (a) and bacterial cells B (b) versus x for the shallow layer in the case a1 = 2 and β 2 = 1.6: a2 = α = 0.5 (O FB model, × KS model), a2 = α = 1 (• FB model, — KS model), a2 = α = 2, β∗2 = 3.01 (◦ FB model, – – KS model). 1,0 1,6 0,8 1,4 0,6 1,2 Θ B 0,4 1,0 0,2 0,8 0,0 0,0 0,2 0,4 0,6 0,8 1,0 0,0 0,2 0,4 x (a) 0,6 0,8 1,0 x (b) Fig. 2. Dependence of the steady-state Θ (a) and B (b) for the shallow layer on a2 in the case a1 = 2, α = 1, β 2 = 1.6 at three values of a2 : 0.5 (O), 1 (•), 2 (◦)– a2 = 2 FB model. Solid line for KS model. www.mii.lt/NA Phenomenological model of bacterial aerotaxis with a negative feedback 241 1,0 2,0 0,8 0,6 1,5 B Θ 0,4 1,0 0,2 0,0 0,0 0,2 0,4 0,6 0,8 0,0 1,0 0,2 0,4 0,6 0,8 1,0 x x (a) (b) Fig. 3. Plot of steady-state Θ (a) and B (b) versus x for the shallow layer in the case a2 = α = 1. β 2 = 1.6 and a1 : 1 (•), 2 (O) for FB model and solid line for KS model. β 2 = 1.2 and a1 : 1 (◦), 2 (×) for FB model and dashed line for KS model. 1,0 1,6 0,8 1,4 0,6 B Θ 1,2 0,4 1,0 0,2 0,8 0,0 0,0 0,2 0,4 0,6 0,8 1,0 0,0 0,2 0,4 (a) 0,6 0,8 1,0 x x (b) Fig. 4. The steady-state concentrations of the oxygen Θ (a) and bacterial cells B (b) versus x for the shallow layer in the case a1 = a2 = α = 1 at three values of β 2 : 1.2 (• FB model, — KS model), 1.5 (◦ FB model, - - - KS model), 1.8 (O FB model, × KS model). Figs. 1(a) and 1(b) demonstrate a monotonic dependence of Θ and B for both KS and FB models on parameter a2 = α for a1 = 2, β 2 = 1.6. In the steady-state case, parameter a2 = α characterizes a swimming upward velocity of microorganisms. Figs. 1(a) and 1(b) also illustrate a natural increase of Θ as a2 grows and different behaviour of B near the bottom and the open surface. Concentration B decreases near the bottom and grows in a region of the open surface as a2 increases. Moreover, B calculated from the FB model possesses a minimum value in the case where a1 = a2 = α = 2 and β 2 = 1.6. B determined by KS model near the bottom is smaller than that calculated by FB model but near the open surface it behaves vice-versa. Oxygen concentration determined by KS model is greater than that calculated by FB one. Nonlinear Anal. Model. Control, 2013, Vol. 18, No. 2, 227–249 242 V. Skakauskas et al. 2,2 1,0 2,0 0,8 1,8 1,6 0,6 Θ B 0,4 1,4 1,2 1,0 0,2 0,8 0,0 0,0 0,2 0,4 0,6 0,8 0,6 0,0 1,0 0,2 0,4 0,6 0,8 1,0 x x (a) (b) Fig. 5. The steady-state concentrations of the oxygen Θ (a) and bacterial cells B (b) versus x for the deep layer in the case a1 = a2 = α = 1 at three values of β 2 : 2.5 (• FB model, — KS model), 3.5 (◦ FB model, - - - KS model), 4.5 (O FB model, × KS model). 1,0 2,5 0,8 2,0 0,6 Θ B 0,4 1,0 0,2 0,0 0,0 1,5 0,5 0,2 0,4 0,6 0,8 1,0 0,0 0,2 0,4 (a) 0,6 0,8 1,0 x x (b) Fig. 6. Plots of the steady-state Θ (a) and B (b) versus x for the deep layer in the case a1 = 2 and β 2 = 3.5: a2 = α = 0.5 (• FB model, — KS model), a2 = α = 1 (◦ FB model, - - KS model), a2 = α = 2 (O FB model, × KS model). Figs. 2(a) and 2(b) illustrate the dependence of Θ and B on the parameter a2 in the case a1 = α = 1 and β 2 = 1.6. Solid line depict Θ and B calculated by KS model. Curves of B corresponding to a2 = 1 or a2 = 2 possesses a minimum value which increases as a2 decreases. Concentration Θ grows as a2 increases but Θ determined by FB model is less than the oxygen concentration corresponding to KS one. Plots in Figs. 3(a) and 3(b) depict the dependence of Θ and B on parameters a1 and β 2 for FB model and on parameter β 2 for KS model. Note that parameters a1 and β 2 characterize an influence of gravity and a chamber depth. Curves demonstrate the different behaviour of B in a region of the bottom and open surface. Near the bottom B grows www.mii.lt/NA Phenomenological model of bacterial aerotaxis with a negative feedback 243 as a1 increases but far from the bottom it behaves vice-versa. Figure also demonstrates the different behaviour of B determined by both KS and FB models. Oxygen concentration Θ decreases for all x as a1 or β 2 increases. Plots in Figs. 4(a) and 4(b) depict the dependence of Θ and B on parameter β 2 for both KS and FB models and demonstrate their monotonic behaviour as β 2 grows. Values of B that correspond to KS and FB models near the bottom are larger than corresponding ones near the open surface. Values of Θ determined by KS model for all x are larger that those corresponding with FB model. Figs. 5(a) and 5(b), 6(a) and 6(b) demonstrate the dependence of Θ and B on parameters β 2 and a2 , respectively, for a deep chamber. We observe a monotonic behaviour of B for both KS and FB models with respect to parameter β 2 . But this behaviour is similar near the bottom and the open surface and is different in the intermediate region (see Figs. 5(a) and 5(b)). The qualitative behaviour of B and Θ with respect to a2 is similar to that of the shallow chamber (see Figs. 6(a) and 6(b)). Table 1 demonstrates the dependence of β̃∗2 on parameters a1 , a2 and x̃∗ , B(x̃∗ ) on β 2 from steady-state FB model. Parameter β̃∗2 decreases as a1 grows and increases with a2 increasing. For the deep chamber case x̃∗ , B(x̃∗ ) grow as β 2 increases. Table 2 demonstrates the similar behaviour of β∗2 , x∗ and B(x∗ ) for steady state case of KS model. Fig. 7 demonstrates the comparison of the behaviour of B determined by the KS and FB models for the shallow chamber in the time-dependent case. For small time and v0 = 0.01, we observe near the top and bottom a non-monotonic behaviour of B determined by the FB model. For fixed x, B reaches a maximum value and then tends to an asymptotic one as time increases. We also observe a small maximum value of B determined by the KS mode l for small time at x = 1. In the top region, Figs. 7(a) and 7(b) demonstrate a different behaviour of B determined by KS and FB models as the depth of the chamber, β 2 , grows. From Fig. 7(a) for small time and β 2 = 5, we can see that B determined by the FB model in the top region is larger than that determined by the KS one, while Fig. 7(b) depicts the vice-versa behaviour of B for β 2 = 2. Numerical experiments reveal a similar qualitative behaviour of B for the deep chamber as well. Table 1. Parameters of the KS model. α 0.5 β∗2 2.18 1 2.41 2 3.02 5 7.23 β2 2.5 3.5 4.5 2.5 3.5 4.5 3.5 4.5 7.5 8.5 10 x∗ 0.071 0.224 0.321 0.021 0.195 0.302 0.095 0.233 0.032 0.133 0.248 Nonlinear Anal. Model. Control, 2013, Vol. 18, No. 2, 227–249 B(x∗ ) 0.851 0.873 0.887 0.706 0.745 0.771 0.497 0.539 0.126 0.139 0.157 244 V. Skakauskas et al. 5 1 4 1 4 3 2 2 3 3 B 3 B 2 2 4 4 1 1 5 5 0 0 0 1 2 3 4 0 5 1 2 3 4 5 t t (a) (b) Fig. 7. Comparison of B for time-dependent KS (Eq. (4)) and FB (Eqs. (21), (22) with v0 = 0.01) models for fixed values of x: 1 (1), 0.95 (2), 0.9 (3), 0.8 (4) 0.1 (5) in the case where a1 = 0.001, a2 = α = 3, β̃∗2 = 3.9195, β∗2 = 3.9186. (a) β 2 = 2, (b) β 2 = 3. — KS model, - - - FB model. Table 2. Parameters of the FB model. a1 0.01 a2 1 β̃∗2 2.40 0.01 5 7.21 0.1 1 2.36 0.1 5 7.02 1 0.5 1.85 1 1 2.01 1 2 2.44 2 2 2 0.5 1 2 1.62 1.73 2.04 β2 2.5 3.5 5 7.5 8.5 10 2.5 3.5 5 7.5 8.5 10 2.5 3.5 4.5 2.5 3.5 4.5 3.5 4.5 3.5 3.5 3.5 x̃∗ 0.022 0.195 0.342 0.035 0.135 0.249 0.032 0.199 0.343 0.056 0.154 0.264 0.110 0.224 0.302 0.089 0.214 0.297 0.170 0.271 0.199 0.197 0.186 B(x̃∗ ) 0.709 0.747 0.783 0.127 0.140 0.157 0.741 0.770 0.798 0.134 0.147 0.165 1.254 1.116 1.037 1.091 0.995 0.939 0.743 0.730 1.312 1.205 0.978 www.mii.lt/NA Phenomenological model of bacterial aerotaxis with a negative feedback 245 Fig. 8 demonstrates the influence of the initial value v0 on the behaviour of B determined by the FB model. B is monotonic in time for v0 < 0.007, but it is non-monotonic if v0 > 0.007. 1 4 3 2 B 2 3 1 4 0 0 1 2 3 4 5 t Fig. 8. Effect of v0 on cells concentration B determined by time-dependent FB model (21) and (22) at a1 = 0.001, a2 = 3, β̃∗2 = 3.9195, β 2 = 3 for fixed values of x: 1 (1), 0.9 (2), 0.8 (3), 0.1 (4). Values of v0 : — 0.01, · · · 0.007, - - - 0.001. 5 Concluding remarks In this paper we studied the steady-state and the initial value problems for oxytactic swimming upwardly bacteria using the modified Keller–Segel model termed FB model. It includes the conservation equations for the oxygen and cells concentrations and the additional one for the balance of forces acting on bacteria (momentum equation for cells). Differently from the KS model, the thrust force, but not the average cells velocity, is approximated by the term proportional to the oxygen gradient. The steady-state solution of the FB model is given analytically and studied numerically but the initial value problem is studied only numerically. In the deep layer case near the bottom the oxygen concentration is insufficient for cells to be active. In this region we only excluded the oxygen consumption and used the continuity conditions for concentrations and their derivatives at the point where Θ = 0. Above this point Θ is positive. We studied difference between solutions of the the KS and FB models and found that in the steady-state case because of gravity the FB model determines more slow swimming upwards of the oxytactic cells comparing with that determined by the KS one. This means that oxygen consumption time grows and, hence, the oxygen concentration decreases since its diffusion is a slow process. This leads to the increase of cells concentration near the bottom and its decrease at the top of the chamber. In the case of small difference of bacteria and fluid densities, asymptotic in time values of B for the KS and FB models practically coincide. In the time-dependent problem the transition period after the initial moment of the concentrations B and Θ (the cells adaptation period) determined by the KS model is shorter than the corresponding one determined by the FB model. Nonlinear Anal. Model. Control, 2013, Vol. 18, No. 2, 227–249 246 V. Skakauskas et al. We hope that the model of the FB type with an additional term of bacteria to bacteria communication in the differential equation for v could be regarded as phenomenological effective one to capture the possible variability of the oxytactic bacteria response to the environmental changes in real systems. Appendix To find the numerical solution of problem (21) and (22) the finite-difference scheme was used. Assume that tj = jτ , 0 6 j 6 M , τ = T /M , xi = ih, 0 6 i 6 N , h = 1/N . Set θij = θ(tj , xi ), bji = b(tj , xi ), vij = v(tj , xi ). Approximating the differential equations (21) with an implicit scheme the following finite difference equations are obtained:  j+1 j+1 j+1 θi−1 − 2θij+1 + θi+1 θi − θij  j 2   =δ − β bi ,   τ h2     j+1 j j+1 j j+1 j j j+1 j    bi − bi = bi−1 − 2bi + bi+1 − γ vi+1 bi+1 − vi−1 bi−1 , τ h2 2h   j j j+1 j  θi+1 − θi−1  ρ3 − vi j vi j j   bi = − γvi , −a1 bi + a2   τ γ 2h     i = 1, 2, . . . , N − 1; j = 0, 1, . . . , M − 1. (A.1) The boundary and initial conditions (22) are approximated by  j+1 b1 − bj+1 v1j bj1 + v0j bj0  0  − γ = 0,    h 2    j j   bN + vN bj+1 − bj+1 v j bj  N −1   N − γ N −1 N −1 = 0, h 2  θj+1 = θj+1 , θj+1 = 1,  1 0 N     0 0 0   θi = 1, bi = 1, vi = v0 ,    i = 0, 1, . . . , N ; j = 0, 1, . . . , M − 1. (A.2) To get the discrete form of Eq. (21)3 at x = 0 and 1 we use the difference equations  ρ3 v j+1 − v0j   bj0 0 = −a1 bj0 − γv0j ,   τ γ   j j j+1 j θN − θN (A.3) − vN ρ3 j vN j j −1  b = −a b + a − γv , 1 N 2  N N  τ γ h     j = 0, 1, . . . , M − 1. www.mii.lt/NA Phenomenological model of bacterial aerotaxis with a negative feedback 247 To write the difference scheme to KS model the following approximation of (4)2 and boundary conditions for function B instead of Eqs. (A.1)2 and (A.2)1,2 are used:  j+1 j+1 j+1 j+1 j   bi − bi = bi−1 − 2bi + bi+1    τ h2  j j j j j bji + bji−1 θij − θi−1 α bi+1 + bi θi+1 − θi − − ,   h 2 h 2 h     i = 1, 2, . . . , N − 1; j = 0, 1, . . . , M − 1, (A.4)  j+1 j+1  b1 = b0 ,    j+1 j j bj + bjN θN − θN bN − bjN −1 −1 − α N −1 = 0,   h 2 h    j = 0, 1, . . . , M − 1. (A.5) From (A.1) , (A.2)1,2 or (A.4), (A.5) the discrete analogue of condition (6) PN −1 j+1 2 PN −1 h i=1 bi = h i=1 bji , j = 0, 1, . . . , M − 1, follows. References 1. E.F. Keller, L.A. Segel, Model for chemotaxis, J. Theor. Biol., 30:225–234, 1971. 2. E.F. Keller, L.A. Segel, Travelling bands of chemotaxis bacteria, J. Theor. Biol., 30:235–248, 1971. 3. D. Horstmann, From 1970 until present: The Keller–Segel model in chemotaxis and its consequences. I, Jahresberichte DMV, 105(3):103–165, 2003. 4. T. Hillen, K.J. Painter, A user’s guide to PDE models for chemotaxis, J. Math. Biol., 58:183– 217, 2009. 5. N.A. Hill, T.J. Pedley, Bioconvection, Fluid Dyn. Res., 37:1–20, 2005. 6. A.J. Hillesdon, T.J. Pedley, J.O. Kessler, The development of concentration gradients in a suspension of chemotactic bacteria, Bull. Math. Biol., 57(2):299–344, 1995. 7. A.J. Hillesdon, T.J. Pedley, Bioconvection in suspensions of oxytactic bacteria: Linear theory, J. Fluid Mech., 324:223–259, 1996. 8. C. Dombrowski, L. Cisneros, S. Shatkaew, R. Goldstein, J. Kessler, Self-concentration and large-scale coherence in bacterial dynamics, Phys. Rev. Lett., 93, 098103, 2004. 9. I. Tuval, L. Cisneros, C. Dombrowski, C. Wolgemuth, J. Kessler, R. Goldstein, Bacterial swimming and oxygen transport near contact lines, Proc. Natl Acad. Sci., 102:2277–2282, 2005. 10. A. Chertock, K. Fellner, A. Kurganov, A. Lorz, P.A. Markowich, Sinking, merging and stationary plumes in a coupled chemotaxis-fluid model: A high-resolution numerical approach, J. Fluid Mech., 694:155–190, 2012. Nonlinear Anal. Model. Control, 2013, Vol. 18, No. 2, 227–249 248 V. Skakauskas et al. 11. A. Lorz, Coupled chenmotaxis fluid model, Math. Models Meth. Appl. Sci., 20:1–17, 2010. 12. R.-J. Duan, A. Lorz, P. Markovich, Global solutions to the coupled chemotaxis-fluid equations, Commun. Part. Diff. Equ., 35:1–39, 2010. 13. M. Di Francesco, A. Lorz, P. Markovich, Chemotaxis-fluid coupled model for swimming bacteria with nonlinear diffusion: Global existence and asymptotic behaviour, Discrete Contin. Dyn. Syst. A, 28(4):1437–1453, 2010. 14. S.M. Becker, A.V. Kuznetsov, A.A. Avramenko, Numerical modelling of a falling bioconvection plume in a porous medium, Fluid Dyn. Res., 33:323–339, 2004. 15. A.V. Kuznetsov, A.A. Avramenko, Analysis of stability of bioconvection of motile oxytactic bacteria in a horizontal fluid saturated porous layer, Int. Commun. Heat Mass Transfer, 30:593– 602, 2003. 16. A.V. Kuznetsov, A.A. Avramenko, P. Geng, Analytical investigation of a falling plume caused by bioconvection of oxytactic bacteria in a fluid saturated porous medium, Int. J. Eng. Sci., 42:557–569, 2004. 17. A.V. Kuznetsov, Thermo-bioconvection in a suspension of oxytactic bacteria, Int. Commun. Heat Mass Transfer, 32:991–999, 2005. 18. A.V. Kuznetsov, Investigation of the onset of thermo-bioconvection in a suspension of oxytactic microorganisms in a shallow fluid layer heated from below, Theor. Comput. Fluid. Dyn., 19(4):287–299, 2005. 19. A.V. Kuznetsov, The onset of thermo-bioconvection in a shallow fluid saturated porous layer heated from below in a suspension of oxytactic microorganisms, Eur. J. Mech. B Fluids, 25:223–233, 2006. 20. Z. Alloui, T.H. Nguyen, E. Bilgen, Bioconvection of gravitactic microorganisms in a vertical cylinder, Int. J. Heat Mass Transfer, 32:739–747, 2005. 21. Z. Alloui, T.H. Nguyen, E. Bilgen, Numerical investigation of thermo-bioconvection in a suspension of gravitactic microorganisms, Int. J. Heat Mass Transfer, 50:1435–1441, 2007. 22. Z. Alloui, T.H. Nguyen, E. Bilgen, Stability analysis of thermo-bioconvection in suspensions of gravitactic microorganisms in a fluid layer, Int. Commun. Heat Mass Transfer, 33:1198–1206, 2006. 23. M.A. Herrero, J.J.L. Velazquez, Chemotactic collapse for the Keller–Segel model, J. Math. Biol., 35:177–194, 1996. 24. B.L. Taylor, I.B. Zhulin, M.S. Johnson, Aerotaxis and other energy sensing behavior in bacteria, Ann. Rev. Microbiol., 53:103–128, 1999. 25. J. Shioi, C.V. Dang, B.L. Taylor, Oxygen as attractant and repellent in bacterial chemotaxis, J. Bacteriol., 169(7):3118–3123, 1987. 26. R. Baronas, R. Šimkus, Modeling the bacterial self-organization in a circular container along the contact line as detected by bioluminescence imaging, Nonlinear Anal. Model. Control, 16(3):270–282, 2011. www.mii.lt/NA Phenomenological model of bacterial aerotaxis with a negative feedback 249 27. R. Šimkus, R. Baronas, Metabolic self-organization of bioluminescent Escherichia coli, Luminescence, 26(6):716–721, 2011. 28. M. Eisenbach, Chemotaxis, Imperial College Press, London, 2004. 29. R.M. Macnab, D.E. Koshland, Jr., The gradient-sensing mechanism in bacerial chemotaxis, Proc. Nat. Acad. Sci. USA, 69(9):2509–2512, 1972. 30. H.G. Othmer, T. Hillen, The diffusion limit of transport equations II: Chemotaxis equations, SIAM J. Appl. Math., 62(4):1222–1250, 2002. 31. T. Hillen, K. Painter, C. Schmeiser, Global existence for chemotaxis with finite sampling radius, Discr. Cont. Dyn. Syst. B, 7(1):125–144, 2007. 32. A.A. Samarskii, The Theory of Difference Schemes, Marcel Dekker, New York, 2001. Nonlinear Anal. Model. Control, 2013, Vol. 18, No. 2, 227–249
https://openalex.org/W2746035652	https://europepmc.org/articles/pmc5573736?pdf=render	English	null	A Structural and Functional Elucidation of the Rumen Microbiome Influenced by Various Diets and Microenvironments	Frontiers in microbiology	2,017	cc-by	19,860	A Structural and Functional Elucidation of the Rumen Microbiome Inﬂuenced by Various Diets and Microenvironments Simon Deusch 1, Amélia Camarinha-Silva 1, Jürgen Conrad 2, Uwe Beifuss 2, Markus Rodehutscord 1 and Jana Seifert 1* 1 Department of Feed-Gut Microbiota Interaction, Institute of Animal Science, University of Hohenheim, Stuttgart, Germany, 2 Department of Bioorganic Chemistry, Institute of Chemistry, University of Hohenheim, Stuttgart, Germany The structure and function of the microbiome inhabiting the rumen are, amongst other factors, mainly shaped by the animal’s feed intake. Describing the inﬂuence of different diets on the inherent community arrangement and associated metabolic activities of the most active ruminal fractions (bacteria and archaea) is of great interest for animal nutrition, biotechnology, and climatology. Samples were obtained from three ﬁstulated Jersey cows rotationally fed with corn silage, grass silage or grass hay, each supplemented with a concentrate mixture. Samples were fractionated into ruminal ﬂuid, particle-associated rumen liquid, and solid matter. DNA, proteins and metabolites were analyzed subsequently. DNA extracts were used for Illumina sequencing of the 16S rRNA gene and the metabolomes of rumen ﬂuids were determined by 500 MHz-NMR spectroscopy. Tryptic peptides derived from protein extracts were measured by LC-ESI-MS/MS and spectra were processed by a two-step database search for quantitative metaproteome characterization. Data are available via ProteomeXchange with the identiﬁer PXD006070. Protein- and DNA-based datasets revealed signiﬁcant differences between sample fractions and diets and afﬁrmed similar trends concerning shifts in phylogenetic composition. Ribosomal genes and proteins belonging to the phylum of Proteobacteria, particularly Succinivibrionaceae, exhibited a higher abundance in corn silage-based samples while ﬁber-degraders of the Lachnospiraceae family emerged in great quantities throughout the solid phase fractions. The analysis of 8163 quantiﬁed bacterial proteins revealed the presence of 166 carbohydrate active enzymes in varying abundance. Cellulosome afﬁliated proteins were less expressed in the grass silage, glycoside hydrolases appeared in slightest numbers in the corn silage. Most expressed glycoside hydrolases belonged to families 57 and 2. Enzymes analogous to ABC transporters for amino acids and monosaccharides were more abundant in the corn silage whereas oligosaccharide transporters showed a higher abundance in the ﬁber-rich diets. Proteins involved in carbon metabolism were detected in high numbers and identiﬁcation of metabolites like short-chain fatty acids, methylamines and phenylpropionate by NMR enabled linkage between producers and products. This study forms a solid basis to retrieve deeper insight into the complex network of microbial adaptation in the rumen. Edited by: Itzhak Mizrahi, Ben-Gurion University of the Negev, Israel Edited by: Itzhak Mizrahi, Ben-Gurion University of the Negev, Israel Reviewed by: Shengguo Zhao, Institute of Animal Science (CAAS), China Steven Singer, Lawrence Berkeley National Laboratory, United States Renee Maxine Petri, Veterinärmedizinische Universität Wien, Austria Reviewed by: Shengguo Zhao, Institute of Animal Science (CAAS), China Steven Singer, Lawrence Berkeley National Laboratory, United States Renee Maxine Petri, Veterinärmedizinische Universität Wien, Austria Correspondence: Jana Seifert jseifert@uni-hohenheim.de Correspondence: Jana Seifert jseifert@uni-hohenheim.de Specialty section: This article was submitted to Systems Microbiology, a section of the journal Frontiers in Microbiology Received: 16 March 2017 Accepted: 07 August 2017 Published: 24 August 2017 Specialty section: This article was submitted to Systems Microbiology, a section of the journal Frontiers in Microbiology Received: 16 March 2017 Accepted: 07 August 2017 Published: 24 August 2017 ORIGINAL RESEARCH published: 24 August 2017 doi: 10.3389/fmicb.2017.01605 Keywords: rumen microbiome, dietary impact, metaproteomics, LC-ESI-MS/MS, 16S rRNA gene, NMR, CAZy INTRODUCTION Furthermore, investigations of diet induced shifts in microbial community composition of the rumen in diﬀerent contexts are numerous (Tajima et al., 2001; Fernando et al., 2010; Kong et al., 2010; Pitta et al., 2010; de Menezes et al., 2011; Ann Huws et al., 2012; Belanche et al., 2012; Carberry et al., 2012; Petri et al., 2013; Thoetkiattikul et al., 2013; Zhang et al., 2013; Lengowski et al., 2016) but generally rumen studies are restricted to nucleic acids-based approaches with limited functional insights. Ruminant livestock with about 3.6 billion farm animals globally (Hackmann and Spain, 2010) represents an important source of human food since these animals have the ability to convert plant- derived non-starch polysaccharides, indigestible for humans, to usable food products in form of milk and meat. This includes also an undesired side eﬀect as ruminants release a substantial portion of methane, a potential greenhouse gas, to the atmosphere (McMichael et al., 2007). The underlying metabolic processes are driven by a complex microbial network consisting of archaea, bacteria, fungi, and protists residing in the strictly anaerobic rumen (Hungate, 1966; Mackie, 2000). Composition and activity of the rumen microbiome are, among other factors, primarily shaped by the diet (Ley et al., 2008; Henderson et al., 2015) and play an important role regarding the animals’ health (Russell and Rychlik, 2001; Gressley et al., 2011) as well as feed eﬃciency and emission of environmentally harmful substances (Mizrahi, 2011; Shabat et al., 2016). Furthermore, the huge amount of ﬁber- degrading enzymes expressed in the rumen serves as a unique resource for the discovery of new lignocellulolytic enzymes useful for biofuel production (Brulc et al., 2009; Hess et al., 2011; Ferrer et al., 2012). Bioinformatic and technical progress in mass spectrometry as well as a growing availability of reference sequences facilitate metaproteomic studies yielding increased information about taxonomic diversity, actual functional proﬁles, and interactions of the most active fractions of the investigated microbiota (Hettich et al., 2013; Seifert et al., 2013; Muth et al., 2016; Tanca et al., 2016). However, investigations of the prokaryotic rumen metaproteome are challenged by the complexity of rumen samples, which requires speciﬁc sample preparation procedures to separate archaeal and bacterial cells from the residual matter prior to protein extraction. Moreover, humic compounds are present and interfere with the metaproteomic workﬂow (Chourey et al., 2010; Heyer et al., 2013). INTRODUCTION Nevertheless, LC-ESI- MS/MS-based rumen studies have already been implemented successfully (Deusch and Seifert, 2015). The combination of diﬀerent up to date Omics-technologies represents the most powerful tool to analyze the microbiome of complex ecosystems like the rumen (Lamendella et al., 2012; Deusch et al., 2015), but studies of the ruminal prokaryotic communities that include multiple approaches and state of the art methods are rare. Bacteria, the most abundant (Krause and Russell, 1996; Mackie, 2000), diverse (McSweeney et al., 2005) and metabolically active species in the rumen are mainly responsible for the degradation and fermentation of plant ﬁbers and proteins ingested by the animals (Hungate, 1966; Brulc et al., 2009). Bacterial species attached to feed particles constitute up to 75% of the total microbial population (McAllister et al., 1994; Koike et al., 2003). Others are free ﬂoating in the rumen ﬂuid or live associated to fungi, protists, and the rumen epithelium (McAllister et al., 1994; Miron et al., 2001). Besides, bacteria can be classiﬁed according to their functional potential as there are, amongst others, ﬁbrolytic, amylolytic, proteolytic, and saccharolytic species. Generally, starch and sugar degraders constitute the largest part of the ruminal bacterial population and are of great importance since diets for high-producing ruminants usually contain large amounts of readily fermentable starch and sugars. Despite their undoubtable signiﬁcance, bacteria specialized for ﬁber degradation are typically less present (Puniya et al., 2015). Degradation of the entire organic matter taken in by the host animals cannot be achieved by a single organism but requires the functional capacities and cooperation of a succession of many microorganisms (Bladen et al., 1961). Hence, to obtain energy, bacterial communities interact synergistically in building diverse ﬁbrolytic enzymes that ﬁnally yield in the production of short chain fatty acids and microbial protein which serve as the main energy and amino acid sources for the host (Hungate, 1966; Mackie, 2002). So far, to the best of our knowledge, there are no publications investigating the impact of the most common forages used as feed for dairy cows and fattening cattle on the metaproteome expressed by the entirety of archaeal and bacterial communities in the diﬀerent phases of the rumen ecosystem. Complementary, structural and functional information obtained from the mass spectrometry-based analysis of the rumen metaproteome targeting the most metabolically active prokaryotes was further supplemented with Illumina MiSeq sequencing of the 16S rRNA gene that includes all cells present. Citation: Deusch S, Camarinha-Silva A, Conrad J, Beifuss U, Rodehutscord M and Seifert J (2017) A Structural and Functional Elucidation of the Rumen Microbiome Inﬂuenced by Various Diets and Microenvironments. Front. Microbiol. 8:1605. doi: 10.3389/fmicb.2017.01605 August 2017 \| Volume 8 \| Article 1605 Frontiers in Microbiology \| www.frontiersin.org 1 Rumen Microbiome Inﬂuenced by Diet and Microenvironments Deusch et al. Ethics Statement The animals of this study were kept according to the German Animal Welfare legislation at the Agricultural Experiment Station Meiereihof of the University of Hohenheim in Stuttgart, Germany. The experimental procedures and treatments were authorized by the Regierungspräsidium Stuttgart in Germany as previously reported (Lengowski et al., 2016). Various methods have been employed to study the rumen microbiome ranging from classical cultivation (Bryant, 1959; Hungate et al., 1964) to molecular approaches including next generation sequencing (Edwards et al., 2004; Mackie and Cann, 2005; Creevey et al., 2014) and functional metagenomics (Brulc et al., 2009; Hess et al., 2011; Ferrer et al., 2012) as well as metabolomics (Ametaj et al., 2010; Saleem et al., 2013). INTRODUCTION Additionally, metabolome patterns were investigated by 500 MHz NMR spectroscopy. The aim of this investigation was to provide deeper insights into the complicated microbial network of the rumen ecosystem and its response to diﬀerent animal diets to improve eﬃciency in animal production. Frontiers in Microbiology \| www.frontiersin.org Animals and Diets To access the dietary and host-related impact on the rumen microbiome a Latin square design using three rumen cannulated August 2017 \| Volume 8 \| Article 1605 Frontiers in Microbiology \| www.frontiersin.org 2 Rumen Microbiome Inﬂuenced by Diet and Microenvironments Deusch et al. instruction with slight modiﬁcations as described in Burbach et al. (2016). The extraction protocol included a bead-beating step for improved mechanical disruption of Gram-positive bacteria as suggested by Henderson et al. (2013). Quality and purity of DNA extracts were analyzed using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientiﬁc, Waltham, MA, USA). Illumina library preparation by PCR ampliﬁcation of the V1-2 region of the 16S rRNA gene was performed as reported recently (Camarinha-Silva et al., 2014). The archaeal community was ampliﬁed using the previously described primers Arch349 and Arch806 (Lee et al., 2012). The forward primer contained a 6-nt barcode, a 2-nt linker and both primers comprised sequences complementary to the Illumina speciﬁc adaptors (Camarinha-Silva et al., 2014). The PCR mixture of a total volume of 20 µl contained PrimeSTAR HS DNA polymerase (2.5 U, Clontech Laboratories, Mountain View, CA, USA), 2.5 mM dNTP mixture, 0.2 µM primers and 1 µl of template DNA. An initial denaturation at 95◦C for 3 min was followed by 20 cycles of denaturation at 98◦C for 10 s, annealing at 59◦C for 10 s, extension at 72◦C for 45 s and a ﬁnal extension for 2 min at 72◦C. One microliter of the PCR product was used for a second PCR (15 cycles) under the same conditions with the reverse primer containing a sequence that integrated the Illumina multiplexing sequence and Illumina index primers (Camarinha-Silva et al., 2014). Integrity of amplicons was analyzed by gel electrophoresis, puriﬁed, and normalized using SequalPrep Normalization Kit (Invitrogen Inc., Carlsbad, CA, USA). Samples were pooled and sequenced using 250 bp paired-end sequencing chemistry on an Illumina MiSeq platform. Sequences were processed using the MOTHUR software pipeline (Kozich et al., 2013). Sequences were excluded if they had any primer or barcode mismatch or N character, aligned, checked for chimeras using UCHIME (Edgar et al., 2011) and clustered into operational taxonomic units (OTUs) at ≥97% similarity. Low abundance OTUs (<0.05% of total reads) were removed and a total of 1,484 bacterial and 626 archaeal phylotypes were taxonomically assigned using the naïve Bayesian RDP classiﬁer (Wang et al., 2007) and the RDP database (Cole et al., 2014). Animals and Diets Sequences were submitted to European Nucleotide Archive under the study accession number PRJEB19491. The mean number of sequence reads for bacteria and archaea was 41,410 ± 1,689 and 14,986 ± 1,713 respectively. Sample Preparation for Mass Spectrometry lactating Jersey cows was applied. Animals were fed rotationally with three diﬀerent diets for ad libitum consumption and free access to drinking water. Feed was given once daily at 7.30 a.m. Based on dry matter, diets consisted of 52% concentrate mixture and 48% of either corn silage, grass silage, or grass hay. The concentrate was composed of 19% wheat, 19% barley, 7% soybean meal, 6% molasses, and 1% vitamin mineral premix. The corn silage-based diet was supplemented with urea to obtain a balanced nitrogen content in comparison to the grass silage- and hay-based diets. The chemical characteristics of the experimental diets are shown in Table 1. Sampling Samples were taken 5 h after feeding at 12.30 p.m. with a preceding adaptation time of 20 days for each diet. A quantity of 200 g of rumen matter was taken from ﬁve diﬀerent positions (cranial, caudal, dorsal, ventral, medial) each and squeezed vigorously by hand using disposable polyethylene gloves to obtain the particle-associated liquid phase (LP) sample fraction, the remains constituted the solid phase (SP) sample fraction. Equal parts of the obtained LP and SP sample fractions were pooled across the ﬁve rumen positions. As a third sample fraction ventral rumen ﬂuid (RF) was collected using a vacuum pump. Two times 27 samples (of three cows, diets and sample fractions) with 40 ml of the RF and LP sample fractions and 20 g of the SP sample fractions were frozen immediately at −80◦C until further processing. The bacterial populations in the rumen can be subdivided into planktonic species, free-living in the RF and the ﬁber-adherent communities which can be further separated into groups of loosely and tightly attached species supposed to be present in the LP and the SP sample fractions, respectively (McAllister et al., 1994). Bioinformatic Data Analysis July 15, 2016) using hmmscan of the HMMER3 software package (Yin et al., 2012) and considering entirely the best e-value hits. The mass spectrometry data have been deposited to the ProteomeXchange Consortium via the o o at c ata a ys s To improve the false discovery rate of peptide identiﬁcations and enhance the conﬁdence of protein identiﬁcations, a two-step search approach was applied to create an artiﬁcial metagenome (Jagtap et al., 2013; Hansen et al., 2014). Therewith, the size of the search databases was reduced and simultaneously the sample- speciﬁcity was increased. First, all 54 raw data ﬁles were processed separately by Thermo Proteome Discoverer software (v. 1.4.1.14), Mascot engine (v. 2.4) in searching independently against the UniProtKB/TrEMBL databases (v. April 28, 2016) for bacteria (Taxonomy ID 2; 40,026,301 sequences) and archaea (Taxonomy ID: 2157; 1,200,545 sequences). Oxidation of methionine was set as variable modiﬁcation and carbamidomethylation of cysteine as ﬁxed modiﬁcation. Precursor ion tolerance was deﬁned at 10 ppm and fragment ion tolerance at 0.02 Da with two missed trypsin cleavages. Furthermore, all peaks besides the top 12 peaks per 100 Da in each MS/MS were removed to denoise spectra before identiﬁcation and the Percolator node was activated with a false discovery rate of 1%. Using Thermo Proteome Discoverer, protein grouping was enabled with a minimum PSM conﬁdence of medium and a delta Cn better than 0.15, strict maximum parsimony principle was applied. As a second step, the protein identiﬁcations inferred from the previous process were used to create sample-speciﬁc databases for label-free quantiﬁcation (LFQ) of proteins via MaxQuant (v. 1.5.3.8) as previously demonstrated (Cox et al., 2014). The ﬁnal in-house databases contained 22,331 bacterial and 818 archaeal protein sequences. The LFQ modality of MaxQuant was enabled with a minimum ratio count of two. Matching between runs with a match time window of 0.7 min and re-quantiﬁcation was applied. Technical duplicates were combined to one experiment. Oxidation of methionine was set as variable modiﬁcation with a maximum of ﬁve modiﬁcations per peptide and carbamidomethylation of cysteine was set as ﬁxed modiﬁcation. Besides, the default settings of MaxQuant were kept which included two missed trypsin cleavages, fully tryptic peptides, a peptide and protein false discovery rate below 1%, at least one peptide per protein (Gupta and Pevzner, 2009), a precursor mass tolerance of 4.5 ppm after mass recalibration and a fragment ion mass tolerance of 20 ppm. Illumina Amplicon Sequencing Using the FastDNATM SPIN Kit for Soil (MP Biomedical, Solon, OH, USA) 250 mg of defrosted and vortexed samples were used for DNA extraction according to the manufacturer’s TABLE 1 \| The measured chemical characteristics of the three forage sources and the thereof calculated properties of the ﬁnal total mixed rations fed to the cows. Components (% of dry matter) Basic forages Total mixed rations CS GS H CS diet GS diet H diet Dry matter (%) 38.3 61.8 97.0 63.4 60.6 71.4 Crude ash 4.1 11.9 8.8 5.3 8.9 7.4 Ether extract 3.2 3.7 2.5 2.8 3.2 2.6 Crude protein 7.8 12.8 13.1 14.2 15.3 15.4 Neutral detergent ﬁber (organic) 41.3 52.9 58.9 27.2 33.2 36.0 Acid detergent ﬁber (organic) 21.0 31.8 33.6 13.3 18.8 19.6 Acid detergent lignin 1.7 2.6 3.0 1.0 1.4 1.6 Non-ﬁber carbohydrates 43.7 18.7 16.7 50.5 39.5 38.6 CS, corn silage; GS, grass silage; H, grass hay. Frontiers in Microbiology \| www.frontiersin.org TABLE 1 \| The measured chemical characteristics of the three forage sources and the thereof calculated properties of the ﬁnal total mixed rations fed to the cows. Sample Preparation for Mass Spectrometry Samples were thawed and vortexed prior to sample preparation as described by Deusch and Seifert (2015). To detach ﬁrmly ﬁber-associated bacteria, 4 g of each SP sample were shaken horizontally for 2 h at 4◦C in 35 ml precooled 50 mM Tris- HCl (pH 8; 0.2 M NaCl; 0.1% methylcellulose 400cP). To dilute the liquid fractions, 5 ml of the respective buﬀer were added to 8 g of the RF and LP samples. All samples were sonicated brieﬂy for 1 min and pressed through two-layered sterile cheesecloth. Residues were rinsed again with 30 ml of the above mentioned buﬀer and pressed vigorously. Obtained ﬁltrates were centrifuged at 200 × g for 10 min at 4◦C and supernatants were further ﬁltered through sterile 40 µm PE ﬁlters. Cells were pelleted at 10,000 × g for 15 min at 4◦C and August 2017 \| Volume 8 \| Article 1605 3 Rumen Microbiome Inﬂuenced by Diet and Microenvironments Deusch et al. Bioinformatic Data Analysis The phylogenetic composition was inferred from the proteins quantiﬁed by MaxQuant as annotated in the UniProtKB/TrEMBL database. KEGG Orthology (KO) identiﬁers and Cluster of Orthologous Groups of proteins (COG) were assigned using WebMGA (Wu et al., 2011) with an e- value cutoﬀof 10−3 considering exclusively the best hits. In order to link the quantiﬁed proteins to pathway maps of carbohydrate metabolism, KO identiﬁers were translated and grouped manually to the respective KEGG REACTION numbers as deﬁned by the KO database. Carbohydrate-active enzymes (CAZymes) were annotated by searching the quantiﬁed bacterial proteins against the database for automated CAZyme annotation with hidden Markov models (dbCAN HMMs v. 5.0, based th CAZ DB J l 15 2016) i h f th Bioinformatic Data Analysis washed three times in 1 ml 50 mM Tris-HCl (pH 7.5; 0.1 mg/ml chloramphenicol; 1 mM PMSF). Subsequently, aliquoted cell pellets were stored at −20◦C. Protein extraction was performed as described previously (Deusch and Seifert, 2015). Cell pellets were resuspended by vortexing in 200 µl 50 mM Tris-HCl (pH 7.5; 0.1 mg/ml chloramphenicol; 1 mM PMSF) and 300 µl of 20 mM Tris-HCl (pH 7.5; 2% SDS) were added. After shaking in a Thermo-Mixer (Eppendorf) for 10 min at 60◦C and 1,200 rpm 1 ml of 20 mM Tris-HCl (pH 7.5; 0.1 mg/ml MgCl2; 1 mM PMSF; 1 µl/ml Benzonase, Novagen) was added. Cells were lysed by ultra-sonication on ice, four times 2 min (amplitude 60%; cycle 0.5) followed by shaking in a Thermo-Mixer for 10 min at 37◦C and 1,200 rpm. Samples were centrifuged at 10,000 × g for 10 min at 4◦C and proteins in the supernatant were precipitated for 30 min at 4◦C using 20% precooled trichloroacetic acid. Subsequently, precipitates were centrifuged at 12,000 × g for 15 min at 4◦C, protein pellets were washed twice in precooled acetone and dried by vacuum centrifugation. Protein pellets were resuspended in 35 µl Laemmli buﬀer by 5 min sonication bath and vortexing followed by incubation for 5 min at 95◦C to reduce disulﬁde bonds. Twenty microliters were puriﬁed with a short run on a one-dimensional sodium dodecyl sulfate polyacrylamide gel electrophoresis (1D-SDS- PAGE; 4% stacking gel, 20 mA; 12% running gel, 40 mA). Each gel lane of 0.5 cm length representing one sample was cut out and subjected to in-gel trypsin (Promega, Madison, USA) digestion overnight (Jehmlich et al., 2008). Obtained peptides were puriﬁed and desalted using Stage tips equipped with ﬁve layers of EmporeTM SPE Disks (C18; diameter 47 mm; thickness 0.5 mm) as described in detail by Rappsilber et al. (2007). Bioinformatic Data Analysis To improve the false discovery rate of peptide identiﬁcations and enhance the conﬁdence of protein identiﬁcations, a two-step search approach was applied to create an artiﬁcial metagenome (Jagtap et al., 2013; Hansen et al., 2014). Therewith, the size of the search databases was reduced and simultaneously the sample- speciﬁcity was increased. First, all 54 raw data ﬁles were processed separately by Thermo Proteome Discoverer software (v. 1.4.1.14), Mascot engine (v. 2.4) in searching independently against the UniProtKB/TrEMBL databases (v. Bioinformatic Data Analysis April 28, 2016) for bacteria (Taxonomy ID 2; 40,026,301 sequences) and archaea (Taxonomy ID: 2157; 1,200,545 sequences). Oxidation of methionine was set as variable modiﬁcation and carbamidomethylation of cysteine as ﬁxed modiﬁcation. Precursor ion tolerance was deﬁned at 10 ppm and fragment ion tolerance at 0.02 Da with two missed trypsin cleavages. Furthermore, all peaks besides the top 12 peaks per 100 Da in each MS/MS were removed to denoise spectra before identiﬁcation and the Percolator node was activated with a false discovery rate of 1%. Using Thermo Proteome Discoverer, protein grouping was enabled with a minimum PSM conﬁdence of medium and a delta Cn better than 0.15, strict maximum parsimony principle was applied. As a second step, the protein identiﬁcations inferred from the previous process were used to create sample-speciﬁc databases for label-free quantiﬁcation (LFQ) of proteins via MaxQuant (v. 1.5.3.8) as previously demonstrated (Cox et al., 2014). The ﬁnal in-house databases contained 22,331 bacterial and 818 archaeal protein sequences. The LFQ modality of MaxQuant was enabled with a minimum ratio count of two. Matching between runs with a match time window of 0.7 min and re-quantiﬁcation was applied. Technical duplicates were combined to one experiment. Oxidation of methionine was set as variable modiﬁcation with a maximum of ﬁve modiﬁcations per peptide and carbamidomethylation of cysteine was set as ﬁxed modiﬁcation. Besides, the default settings of MaxQuant were kept which included two missed trypsin cleavages, fully tryptic peptides, a peptide and protein false discovery rate below 1%, at least one peptide per protein (Gupta and Pevzner, 2009), a precursor mass tolerance of 4.5 ppm after mass recalibration and a fragment ion mass tolerance of 20 ppm. The phylogenetic composition was inferred from the proteins quantiﬁed by MaxQuant as annotated in the UniProtKB/TrEMBL database. KEGG Orthology (KO) identiﬁers and Cluster of Orthologous Groups of proteins (COG) were assigned using WebMGA (Wu et al., 2011) with an e- value cutoﬀof 10−3 considering exclusively the best hits. In order to link the quantiﬁed proteins to pathway maps of carbohydrate metabolism, KO identiﬁers were translated and grouped manually to the respective KEGG REACTION numbers as deﬁned by the KO database. Carbohydrate-active enzymes (CAZymes) were annotated by searching the quantiﬁed bacterial proteins against the database for automated CAZyme annotation with hidden Markov models (dbCAN HMMs v. 5.0, based on the CAZyDB v. Metaproteomics and Amplicon Sequencing of the Rumen Microbiome The RF samples were defrosted, vortexed, and ﬁltered through sterile 100 µm PE ﬁlters. Filtrates were centrifuged at 13,000 × g for 30 min at 4◦C. Obtained supernatants were sterilized by passing through a 0.22 µm syringe ﬁlter and 3 ml of each RF sample were dried completely by vacuum centrifugation at room temperature overnight. By vigorous vortexing and 5 min brief sonication, dehydrated RF samples were reconstituted in 1.5 ml 50 mM sodium phosphate buﬀer (pH 6.5) in 99.9% deuterium oxide (Sigma-Aldrich, Germany) containing 5 mM 3-(trimethylsilyl)propionic-2,2,3,3-d4 acid sodium salt (TSP; Sigma-Aldrich, Germany) as an internal chemical shift reference and quantiﬁcation standard. Subsequently, dissolved samples were centrifuged at 13,000 × g for 30 min at 4◦C and 1 ml supernatant was transferred to a 5 mm glass NMR tube for measuring at 500 MHz using a Varian INOVA NMR spectrometer (Agilent Technologies). All 1H- NMR spectra were acquired at 25◦C using the ﬁrst transient of the noesy presaturation pulse sequence (Saude et al., 2006). Each spectrum was collected with 32 transients using a 4 s acquisition time, 1 s recycle delay and a mixing time of 0.1 s at a spectral width of 6,490 Hz. Spectral assignments were performed by 2D homonuclear and heteronuclear NMR: DQFCOSY, gHSQCAD, gHSQCTOCSY, as well as gHMBCAD were run using CHEMPACK 7.2 pulse sequences implemented in VnmrJ 4.2 (Agilent Technologies Inc., Santa Clara, CA, USA). Additionally, 1H-NMR spectra were imported into the Chenomx NMR Suite 8.2 software (database available at pH 6.5, Chenomx Inc., Edmonton, AB, Canada) for quantiﬁcation (Weljie et al., 2006; Wishart, 2008) as described in Ametaj et al. (2010). Spectra were referenced to TSP (δ 0.0 ppm) for chemical shift and quantiﬁcation. Prior to spectral analysis, all free induction decays (FIDs) were automatically zero-ﬁlled to 64 k data points, corrected for phase and baseline distortions and a line broadening of 0.5 Hz was applied. Concentrations of identiﬁed metabolites were divided by a factor of two since dried RF samples were reconstituted in half of the initial volume. Over all samples 8,163 bacterial and 358 archaeal proteins were quantiﬁed by mass spectrometric measurements of the peptides and a two-step search identiﬁcation and quantiﬁcation process. Illumina MiSeq sequencing of the V1-2 region of 16S rRNA gene resulted in 1,484 bacterial and 626 archaeal OTUs assigned (Table S1). Metaproteomics and Amplicon Sequencing of the Rumen Microbiome Regarding the abundances of bacterial proteins and OTUs, NMDS plots based on the Bray Curtis similarity revealed diet induced shifts and variations between sample fractions with similar trends for both, the metaproteomic dataset (Figure 1A) and the DNA-based approach (Figure 1B). There were no dietary, sample fraction or host related eﬀects on the abundance of archaeal proteins and OTUs. ANOSIM veriﬁed signiﬁcant diﬀerences in bacterial protein abundances regarding the diets with an R-value of 0.600 (P = 0.0001) for the corn silage- and grass silage-based samples and 0.442 (P = 0.0001) for the corn silage- and hay-derived samples. Respectively, the numbers of assigned bacterial OTUs of the corn silage- and grass silage- based samples diﬀered with an R-value of 0.500 (P = 0.0003) and the corn silage- and hay-derived samples showed an R-value of 0.558 (P = 0.0001). Furthermore, ANOSIM conﬁrmed signiﬁcant variations in protein abundances between sample fractions with an R-value of 0.362 (P = 0.0001) for the LP and SP fractions. The RF and SP fractions diﬀered with an R-value of 0.321 (P = 0.0008). Concerning the sequencing data, the R-value for the LP and SP fractions was 0.596 (P = 0.0001) and 0.561 (P = 0.0002) for the RF and SP fractions. Complete statistical information of the bacterial protein and OTU abundances considering diets, sample fractions and the individual cows is shown in Table 2. LC-ESI-MS/MS Measurements LC-ESI-MS/MS Measurements LC-ESI-MS/MS analyses were performed in technical duplicates on an EasyLC 1000 nano-UHPLC (Thermo Scientiﬁc) coupled to a Q Exactive HF mass spectrometer (Thermo Scientiﬁc). Prior to LC-ESI-MS/MS measurements peptides were reconstituted in 20 µl 0.1% formic acid and 4 µl were injected by the autosampler. Separations of the peptide mixtures were done on a 20 cm fused silica emitter of 75 µm inner diameter (Proxeon Biosystems), in- house packed with reversed-phase ReproSil-Pur 120 C18-AQ 1.9 µm resin (Dr. Maisch GmbH). Peptide mixtures were injected onto the separation column in HPLC solvent A (0.1% formic acid) at a ﬂow rate of 500 nl/min and eluted with a solvent B (80% acetonitrile in 0.1% formic acid) gradient of 1–33% within the ﬁrst 73 min followed by an increase to 50% within 3 min plus an additional 3 min at 90%. The Q Exactive HF was operated in the positive ion mode. Full scan was acquired in the mass range from 300 to 1,650 m/z in the Orbitrap mass analyzer at a resolution of r = 120,000 followed by higher energy collisional dissociation (HCD) fragmentation of the twelve most intense precursor ions. High resolution MS/MS spectra were acquired with a resolution of r = 30,000. The target values were 3 × 106 charges for the MS scans and 1 × 105 charges for the MS/MS scans with a maximum ﬁll time of 25 and 45 ms, respectively. The dynamic precursor exclusion was set to 30 s and peptide match was enabled. August 2017 \| Volume 8 \| Article 1605 Frontiers in Microbiology \| www.frontiersin.org 4 Rumen Microbiome Inﬂuenced by Diet and Microenvironments Deusch et al. PRIDE partner repository (Vizcaino et al., 2016) with the dataset identiﬁer PXD006070. metabolite concentrations in diets (IBM SPSS Statistics, Version 20.0. Armonk, NY: IBM Corp). Dietary Impact on Community Structure and Variations in Fractions Taxonomic information was obtained from quantiﬁed proteins as annotated by the UniProtKB/TrEMBL database and from RDP retrieved OTU assignments of the Illumina amplicon sequencing. The taxonomic distribution of archaeal and bacterial proteins and OTUs in each sample at phyla, class, order, and family level as well as the corresponding numbers of proteins and OTUs are listed in Table S2. Overall, the bacterial community composition was dominated by the phylum of Bacteroidetes followed by the phylum of Firmicutes. Less abundant phyla were Actinobacteria, Elusimicrobia, Proteobacteria, Spirochaetes, Synergistetes, Tenericutes, and Verrucomicrobia (Table S2). Frontiers in Microbiology \| www.frontiersin.org Statistical Analyses The LFQ abundance values of proteins and the OTU counts were analyzed using the Primer 6 (v. 6.1.16) and Permanova+ (v. 1.0.6) statistical software package (PRIMER-E, Plymouth, UK). Non- metric multidimensional scaling (NMDS) was performed using the Bray–Curtis similarity matrix (Bray and Curtis, 1957). One- way analysis of similarities (ANOSIM) was used to determine statistical diﬀerences in protein and OTU abundance between diets, sample fractions and host animals (Clarke and Warwick, 2001). One-way analysis of variance (ANOVA) with post-hoc Tukey HSD (Honestly Signiﬁcant Diﬀerence) was used for pairwise comparisons of the abundance means of taxonomic groups in diets and sample fractions and to test diﬀerences of However, the bacterial community structure averaged over the three animals per treatment revealed concordant tendencies concerning the dietary inﬂuence and variations in diﬀerent microenvironments for both applied methods. Figure 2 shows the average abundance in diets and sample fractions of bacterial phyla, orders and families commonly identiﬁed by metaproteomics on the left and 16S rRNA gene sequencing on August 2017 \| Volume 8 \| Article 1605 Frontiers in Microbiology \| www.frontiersin.org 5 Rumen Microbiome Inﬂuenced by Diet and Microenvironments Deusch et al. TABLE 2 \| Analysis of similarity of the metaproteomic- and the 16S rRNA gene-based datasets. LFQ-values of 8,163 bacterial proteins Abundance of 1,484 bacterial OTUs R P R P Diets* 0.418 0.0001 0.366 0.0001 Fractions* 0.244 0.0004 0.379 0.0001 Cows 0.143 0.0090 0.176 0.0070 Diets GS:H 0.248 0.0100 0.078 0.1630 CS:GS* 0.600 0.0001 0.500 0.0003 CS:H* 0.442 0.0001 0.558 0.0001 Fractions RF:LP 0.053 0.1870 −0.045 0.6980 RF:SP* 0.321 0.0008 0.561 0.0002 LP:SP* 0.362 0.0001 0.596 0.0001 Cows 28:59 0.088 0.1090 0.133 0.0790 23:59 0.169 0.0280 0.207 0.0280 23:28 0.180 0.0150 0.202 0.0270 The upper part of the table shows the global R statistics and the respective probability values (P) for the main factors (diets, sample fractions and individual cows). The lower part shows the R statistics and corresponding probability values (P) for the pairwise group comparisons within the main factors. Signiﬁcantly different main factors and groups within are marked (). CS, corn silage-based diet; GS, grass silage-based diet; H, hay-based diet; RF, rumen ﬂuid; LP, liquid phase; SP, solid phase; cows are indicated by no. 23, 28, and 59. TABLE 2 \| Analysis of similarity of the metaproteomic- and the 16S rRNA gene-based datasets. FIGURE 1 \| Non-metric multidimensional scaling (NMDS) plots of the metaproteome (A) and the bacterial community structure (B). Statistical Analyses Yellow, corn silage-based diet; green, grass silage-based diet; brown, grass hay-based diet. Squares, solid phase; stars, liquid phase; circles, rumen ﬂuid. FIGURE 1 \| Non-metric multidimensional scaling (NMDS) plots of the metaproteome (A) and the bacterial community structure (B). Yellow, corn silage-based diet; green, grass silage-based diet; brown, grass hay-based diet. Squares, solid phase; stars, liquid phase; circles, rumen ﬂuid. major parts of the Firmicutes phylum (Figure 2) and revealed the highest LFQ-values and numbers of OTUs in the SP fractions of all diets when compared to the respective RF and LP fractions with P < 0.01 (LFQs) and P < 0.05 (OTUs). No OTUs were assigned to the Firmicutes order of Bacillales although highly abundant proteins were identiﬁed by mass spectrometry. Within diets, Bacillales proteins were more abundant in the RF fractions. The Erysipelotrichaceae family was more abundant in the hay-based diet when compared to the corn silage- and grass silage-based samples (Figure 2) emphasized primarily by the abundance of assigned OTUs (P < 0.01). Proteins and OTUs belonging to the order of Lactobacillales and to the family of Veillonellaceae constituted only a small part of the Firmicutes phylum as determined by both approaches (Table S2). Proteins of the Veillonellaceae family were more abundant in the corn silage-based diet when compared to the grass silage- and hay- based samples (P < 0.05). The abundance of OTUs belonging to the phylum of Fibrobacteres was higher when compared to the LFQ-values of the corresponding proteins (Figure 2). According to the 16S rRNA gene sequencing, the phylum of Fibrobacteres was less abundant in the RF and LP fractions of the corn silage- based diet when compared to the respective fractions of the grass silage- and hay-based diets. Within diets, the abundance of Fibrobacteres proteins was higher in the SP fractions of all diets when compared to the RF and LP samples (P < 0.05). Related to the overall bacterial community structure, OTU and LFQ- values assigned to the phylum of Elusimicrobia exhibited a low abundance across all samples (Table S2). Regardless the dietary treatments, the Prevotellaceae family dominated the phylum of Bacteroidetes and the overall bacterial community composition as determined by the protein- and DNA-based approaches FIGURE 1 \| Non-metric multidimensional scaling (NMDS) plots of the metaproteome (A) and the bacterial community structure (B). Yellow, corn silage-based diet; green, grass silage-based diet; brown, grass hay-based diet. Frontiers in Microbiology \| www.frontiersin.org Statistical Analyses In the same way, sequences assigned to the phylum of Bacteroidetes were more abundant than the LFQ-values of the corresponding proteins (Figure 2). In contrast, the family of Bacteroidaceae showed a higher abundance of proteins when compared to the abundance of the corresponding OTUs (Figure 2). OTU and LFQ-values of Biﬁdobacteriaceae were higher in the grass silage- and hay-based samples when compared to the corn silage-based fractions (Table S2). The abundance of OTUs and LFQs of the Actinobacteria phylum including the family of Biﬁdobacteriaceae was higher in the SP fractions of all diets when compared to the respective RF and LP fractions (Figure 2). Within diets, the phylum of Spirochaetes exhibited the highest abundance in the SP fractions as found by both methods. Similarly, the phylum of Synergistetes revealed the highest abundance of proteins and particularly OTUs in the SP fractions when compared to respective the RF and LP fractions (Figure 2). Proteins assigned to the phylum of Tenericutes were low abundant when compared to the corresponding OTU abundance. OTUs of the phylum of Tenericutes were more abundant in the RF and LP fractions when compared to the respective SP sample fractions whereas the LFQ-values of the corresponding proteins were higher in the SP fractions of all diets (Figure 2). In contrast, proteins of the Verrucomicrobia phylum were more abundant in comparison to the respective OTUs (Table S2). (Figure 2). Within diets, Prevotellaceae proteins showed the highest LFQ-values in the LP fractions (P < 0.05). The family of Porphyromonadaceae exhibited a higher abundance of OTUs when compared to the LFQ-values of the respective proteins. In the same way, sequences assigned to the phylum of Bacteroidetes were more abundant than the LFQ-values of the corresponding proteins (Figure 2). In contrast, the family of Bacteroidaceae showed a higher abundance of proteins when compared to the abundance of the corresponding OTUs (Figure 2). OTU and LFQ-values of Biﬁdobacteriaceae were higher in the grass silage- and hay-based samples when compared to the corn silage-based fractions (Table S2). The abundance of OTUs and LFQs of the Actinobacteria phylum including the family of Biﬁdobacteriaceae was higher in the SP fractions of all diets when compared to the respective RF and LP fractions (Figure 2). Within diets, the phylum of Spirochaetes exhibited the highest abundance in the SP fractions as found by both methods. Statistical Analyses Squares, solid phase; stars, liquid phase; circles, rumen ﬂuid. the right side. Proteins and OTUs belonging to the phylum of Proteobacteria including the family of Succinivibrionaceae were signiﬁcantly more abundant throughout the samples of the corn silage-based diet when compared to the grass silage- and hay- based samples with P < 0.01 (LFQs) and P < 0.05 (OTUs). Within the corn silage-derived samples the respective proteins and OTUs were more abundant in the SP fraction (Figure 2). The abundance of OTUs assigned to Succinivibrionaceae was higher when compared to the LFQ-values of the corresponding proteins. Likewise, the family of Acidaminococcaceae belonging to the phylum of Firmicutes showed a higher protein (P < 0.01) and OTU abundance in the corn silage-based diet when compared to the grass silage- and hay-based samples (Figure 2). The abundance of the respective OTUs was least in the SP fraction of all diets whereas the LFQ-values of the corresponding proteins were highest in the SP fractions (Figure 2). The Firmicutes family of Selenomonadaceae was identiﬁed exclusively in the protein- based dataset (Figure 2). Selenomonadaceae proteins exhibited a lower abundance in the grass silage-derived sample fractions when compared to the corn silage- and hay-based samples (P < 0.01). Members of the order of Clostridiales including the families of Lachnospiraceae and Ruminococcaceae constituted August 2017 \| Volume 8 \| Article 1605 Frontiers in Microbiology \| www.frontiersin.org 6 Rumen Microbiome Inﬂuenced by Diet and Microenvironments Deusch et al. FIGURE 2 \| Bacterial phylogenetic distribution in diets and sample fractions (n = 3) obtained from the metaproteome (8,163 proteins) on the left and the 16S rRNA gene sequencing (1,484 OTUs) on the right. The order of Bacillales and the family of Selenomonadaceae were found exclusively in the metaproteomic dataset (shaded coloring). FIGURE 2 \| Bacterial phylogenetic distribution in diets and sample fractions (n = 3) obtained from the metaproteome (8,163 proteins) on the left and the 16S rRNA gene sequencing (1,484 OTUs) on the right. The order of Bacillales and the family of Selenomonadaceae were found exclusively in the metaproteomic dataset (shaded coloring). (Figure 2). Within diets, Prevotellaceae proteins showed the highest LFQ-values in the LP fractions (P < 0.05). The family of Porphyromonadaceae exhibited a higher abundance of OTUs when compared to the LFQ-values of the respective proteins. Frontiers in Microbiology \| www.frontiersin.org Statistical Analyses Similarly, the phylum Figure 3 shows the average abundance in diets and sample fractions of archaeal phyla, classes and families identiﬁed by metaproteomics on the left and 16S rRNA gene sequencing on the right. The metaproteomics-based approach identiﬁed 358 archaeal proteins that distributed over all archaeal phyla including as well the family of Thermococcaceae whereas August 2017 \| Volume 8 \| Article 1605 7 Rumen Microbiome Inﬂuenced by Diet and Microenvironments Deusch et al. at phyla level of the ﬁve CAZy-categories in diets and sample fractions is depicted in Figure 4B. The ratio of the most abundant CAZy-categories of GH, mainly produced by Bacteroidetes species and SLH, almost exclusively derived from Firmicutes species varied between the diets (Figures 4A,B). Proteins of the SLH category were less abundant in the grass silage-based sample fractions when compared to the corn silage- and hay- based samples. Contrarily, proteins of the GH category were more abundant in the grass silage-based diet in comparison to the corn silage- and hay-based samples (Figure 4A). Proteobacteria- derived proteins belonging to the GT category were more abundant in the corn silage-based samples when compared to grass silage- and hay-based samples (Figure 4B). Proteins assigned to the CE category were more abundant in the LP and SP fractions of the corn silage-based diet when compared to the respective fractions of the grass silage and hay-based diets (Figure 4A). Moreover, the LP and SP fractions of the corn silage-based diet exhibited a higher abundance of proteobacterial proteins in the CE category when compared to the remaining samples (Figure 4B). CBM related proteins were more abundant in the sample fractions of the grass silage-based diet when compared to the corn silage- and hay-based diets (Figure 4A). Across all samples, the family GH57, based on seven proteins like alpha-amylases, was most abundant (Figure 4A). The most expressed GT family was GT35 including 18 proteins of diﬀerent glycogen and starch phosphorylases. Within the category of CE the most expressed family was CE1 with eight proteins. The the 626 OTUs were exclusively assigned to three families of methanogens: Methanobacteriaceae, Methanosarcinaceae, and Methanomassiliicoccaceae (Figure 3). Proteins belonging to the phyla of Thaumarchaeota and Crenarchaeota constituted a minor part of the total archaeal protein abundance (Figure 3). The LFQ-values of unclassiﬁed archaeal proteins were above 16% in each sample (Table S2). Statistical Analyses Within the phylum of Euryarchaeota, proteins belonging to the families of Methanobacteriaceae, Methanosarcinaceae, Thermococcaceae, and the class of Methanomicrobia were most abundant while OTUs of the family of Methanobacteriaceae prevailed with above 72% in each sample (Table S2). Carbohydrate-Active Enzymes Glycoside hydrolases = GH (91 proteins), glycosyltransferases = GT (38 proteins), carbohydrate esterases = CE (16 proteins), carbohydrate-binding modules = CBM (2 proteins) and S-layer homology domains = SLH (19 proteins) (A). Pie charts depict the phylogenetic origin of CAZymes of the respective categories in diets and sample fractions at phyla level: blue, Bacteroidetes; green, Firmicutes; red, Proteobacteria; yellow, environmental samples; gray, others (B). RF, rumen ﬂuid; LP, liquid phase; SP, solid phase. FIGURE 4 \| Abundance of 166 CAZymes in diets and sample fractions (n = 3). Glycoside hydrolases = GH (91 proteins), glycosyltransferases = GT (38 proteins), carbohydrate esterases = CE (16 proteins), carbohydrate-binding modules = CBM (2 proteins) and S-layer homology domains = SLH (19 proteins) (A). Pie charts depict the phylogenetic origin of CAZymes of the respective categories in diets and sample fractions at phyla level: blue, Bacteroidetes; green, Firmicutes; red, Proteobacteria; yellow, environmental samples; gray, others (B). RF, rumen ﬂuid; LP, liquid phase; SP, solid phase. similarity to ABC transporters (Table S5) as deﬁned by the KO database. Figure 5 depicts the average abundance of the respective membrane transporters in diets and sample fractions in relation to the maximum LFQ-values within each group of transporters and the origin at bacterial phyla level. Analyses aﬃrmed 106 proteins to the group of oligosaccharide, polyol, and lipid transporters that include several subunits of multiple sugar transport systems and of cellobiose, arabinose/lactose, maltose/maltodextrin, sorbitol/mannitol, and galactose oligomer CBM32 family was most abundant in the SP fraction of the grass silage-based diet (Figure 4A). Details about the proteins assigned to the diﬀerent CAZy-categories including the phylogenetic origin and the respective functions are listed in Table S3. Carbohydrate-Active Enzymes y y CAZyme annotation with hidden Markov models identiﬁed a total of 166 bacterial proteins in ﬁve CAZy-categories (Table S3). A majority of 91 proteins were assigned to 16 glycoside hydrolase (GH) families and 38 proteins belonged to seven glycosyltransferase (GT) families. Furthermore, 16 proteins were assigned to three families of carbohydrate esterases (CE) and two proteins fell into two families of carbohydrate- binding modules (CBM). In addition, 19 proteins with sequence similarity to S-layer homology domains (SLH) were present in high abundance throughout diets and sample fractions. The percentage abundance of the respective bacterial CAZymes in diets and sample fractions in relation to the total abundance of CAZymes is shown in Figure 4A. The phylogenetic origin FIGURE 3 \| Archaeal phylogenetic distribution in diets and sample fractions (n = 3) obtained from the metaproteome (358 proteins) on the left and the 16S rRNA gene sequencing (626 OTUs) on the right. FIGURE 3 \| Archaeal phylogenetic distribution in diets and sample fractions (n = 3) obtained from the metaproteome (358 proteins) on the left and the 16S rRNA gene sequencing (626 OTUs) on the right. August 2017 \| Volume 8 \| Article 1605 Frontiers in Microbiology \| www.frontiersin.org 8 Rumen Microbiome Inﬂuenced by Diet and Microenvironments Deusch et al. ance of 166 CAZymes in diets and sample fractions (n = 3). Glycoside hydrolases = GH (91 proteins), glycosyltransferases = GT (38 proteins), ses = CE (16 proteins), carbohydrate-binding modules = CBM (2 proteins) and S-layer homology domains = SLH (19 proteins) (A). Pie charts etic origin of CAZymes of the respective categories in diets and sample fractions at phyla level: blue, Bacteroidetes; green, Firmicutes; red, w, environmental samples; gray, others (B). RF, rumen ﬂuid; LP, liquid phase; SP, solid phase. dance of 166 CAZymes in diets and sample fractions (n = 3). Glycoside hydrolases = GH (91 proteins), glycosyltransferases = GT (38 proteins), rases = CE (16 proteins), carbohydrate-binding modules = CBM (2 proteins) and S-layer homology domains = SLH (19 proteins) (A). Pie charts netic origin of CAZymes of the respective categories in diets and sample fractions at phyla level: blue, Bacteroidetes; green, Firmicutes; red, low environmental samples; gray others (B) RF rumen ﬂuid; LP liquid phase; SP solid phase FIGURE 4 \| Abundance of 166 CAZymes in diets and sample fractions (n = 3). ABC Transporters Archaeal and Bacterial Enzymes Involved in Carbon Metabolism Varying abundance, in diets and sample fractions, of enzymes involved in the carbon metabolism are shown in Figure 6A. To visualize the abundance of the respective enzymes in the Frontiers in Microbiology \| www.frontiersin.org 1 pathway map, a total of 1121 bacterial proteins assigned to 80 KO identiﬁers carrying out 70 KEGG REACTIONS were arranged into 60 functional groups Table S6). A few KOs were assigned to more than one KEGG REACTION and thus, appear more than once in the grouping. Furthermore, 28 archaeal proteins of eight KOs involved in the methane metabolism were grouped to ﬁve KEGG REACTIONS (Table S6). Additionally, eight compounds identiﬁed by NMR are shown in Figure 6A. Twenty-four enzymes of the groups 8 and 12 involved in the phosphate acetyltransferase-acetate kinase pathway were present in all diets and sample fractions, whereas only one acetyl- CoA synthetase from an uncultured bacterium in group 9 was found in the SP fraction of the hay-based diet (Figure 6A). The abundance of seven proteins of group 6 carrying out the oxidation of pyruvate to acetyl-CoA was higher in the RF and SP fractions of the grass silage-based diet and higher in the LP fraction of the hay-based diet when comparing to the other fractions and diets. Conversion of acetyl-CoA to malonyl- CoA employing two proteins of group 20 produced by Bacillus and Prevotella species, exhibited abundance exclusively in the grass silage-derived RF fraction and regarding the corn silage- based diet, were present only in the SP fraction (Figure 6A). A Tenericutes-derived carbamate kinase in group 3 producing carbamoyl phosphate, a metabolite in nitrogen disposal through the urea cycle, was found exclusively in the fractions of the corn silage-based diet (Figure 6A). Summing up the LFQ-values of groups involved in the Embden-Meyerhof pathway (5, 19, 27, 30, transporters. The respective proteins were more abundant in the hay- and grass silage-based diets when compared to the corn silage-based diet (Figure 5). Within diets, the highest abundance was observed in the SP fractions. Oligosaccharide, polyol, and lipid transporters originated mainly from Firmicutes species. In contrast, the group of monosaccharide transporters, based on 20 proteins, was more abundant in the corn silage-based samples when compared to the grass silage- and hay-based diets (Figure 5). This group includes subunits of ribose, rhamnose, methyl-galactoside, and sn-glycerol 3- phosphate transport systems. ABC Transporters Proteobacterial proteins showed an increased contribution regarding monosaccharide transporters when compared to the respective phylogenetic origin of the group of oligosaccharide, polyol, and lipid transporters (Figure 5). Phosphate and amino acid transporters based on 27 proteins showed the highest abundance in corn silage-derived samples and were mainly produced by Proteobacteria species (Figure 5). Mineral and organic ion transporters including 17 proteins were almost exclusively produced by Firmicutes species and showed the highest LFQ-values in the corn silage-derived SP fraction (Figure 5). ABC Transporters WebMGA assigned 7,745 and 336 KO identiﬁers to 8,163 bacterial and 358 archaeal proteins respectively (Table S4), revealing a total of 170 bacterial proteins with sequence August 2017 \| Volume 8 \| Article 1605 Frontiers in Microbiology \| www.frontiersin.org 9 Rumen Microbiome Inﬂuenced by Diet and Microenvironments Deusch et al. FIGURE 5 \| Varying abundance of bacterial ABC transporters in diets and sample fractions (n = 3). Pie sizes indicate the abundance in relation to the sample with the maximum LFQ-values () in each group. Phylogenetic origin of proteins in the respective samples is shown at phyla level. RF, rumen ﬂuid; LP, liquid phase; SP, solid phase. FIGURE 5 \| Varying abundance of bacterial ABC transporters in diets and sample fractions (n = 3). Pie sizes indicate the abundance in relation to the sample with the maximum LFQ-values () in each group. Phylogenetic origin of proteins in the respective samples is shown at phyla level. RF, rumen ﬂuid; LP, liquid phase; SP, solid phase. FIGURE 5 \| Varying abundance of bacterial ABC transporters in diets and sample fractions (n = 3). Pie sizes indicate the abundance in relation to the sample with the maximum LFQ-values () in each group. Phylogenetic origin of proteins in the respective samples is shown at phyla level. RF, rumen ﬂuid; LP, liquid phase; SP, solid phase. transporters. The respective proteins were more abundant in the hay- and grass silage-based diets when compared to the corn silage-based diet (Figure 5). Within diets, the highest abundance was observed in the SP fractions. Oligosaccharide, polyol, and lipid transporters originated mainly from Firmicutes species. In contrast, the group of monosaccharide transporters, based on 20 proteins, was more abundant in the corn silage-based samples when compared to the grass silage- and hay-based diets (Figure 5). This group includes subunits of ribose, rhamnose, methyl-galactoside, and sn-glycerol 3- phosphate transport systems. Proteobacterial proteins showed an increased contribution regarding monosaccharide transporters when compared to the respective phylogenetic origin of the group of oligosaccharide, polyol, and lipid transporters (Figure 5). Phosphate and amino acid transporters based on 27 proteins showed the highest abundance in corn silage-derived samples and were mainly produced by Proteobacteria species (Figure 5). Mineral and organic ion transporters including 17 proteins were almost exclusively produced by Firmicutes species and showed the highest LFQ-values in the corn silage-derived SP fraction (Figure 5). Archaeal and Bacterial Enzymes Involved in Carbon Metabolism Varying abundance, in diets and sample fractions, of enzymes involved in the carbon metabolism are shown in Figure 6A. To visualize the abundance of the respective enzymes in the August 2017 \| Volume 8 \| Article 1605 Frontiers in Microbiology \| www.frontiersin.org 10 Rumen Microbiome Inﬂuenced by Diet and Microenvironments Deusch et al. FIGURE 6 \| (A) Abundance of archaeal (blue) and bacterial (red) enzymes (n = 3) involved in carbon metabolism, grouped and numbered according to the respective reactions (Table S6). Outer circle, solid phase; middle circle, liquid phase; inner circle, rumen ﬂuid. Purple, compounds identiﬁed by NMR. The percentage of total LFQ-values of groups involved in (B) the Embden-Meyerhof-Parnas pathway (5, 19, 27, 30, 32, 39, 41, 44, 48, 54), (C) citrate cycle (11, 14, 17, 33, 38, 47, 57) and (D) pentose phosphate pathway (29, 31, 45, 48). RF, rumen ﬂuid; LP, liquid phase; SP, solid phase. FIGURE 6 \| (A) Abundance of archaeal (blue) and bacterial (red) enzymes (n = 3) involved in carbon metabolism, grouped and numbered according to the respective reactions (Table S6). Outer circle, solid phase; middle circle, liquid phase; inner circle, rumen ﬂuid. Purple, compounds identiﬁed by NMR. The percentage of total LFQ-values of groups involved in (B) the Embden-Meyerhof-Parnas pathway (5, 19, 27, 30, 32, 39, 41, 44, 48, 54), (C) citrate cycle (11, 14, 17, 33, 38, 47, 57) and (D) pentose phosphate pathway (29, 31, 45, 48). RF, rumen ﬂuid; LP, liquid phase; SP, solid phase. August 2017 \| Volume 8 \| Article 1605 Frontiers in Microbiology \| www.frontiersin.org 11 Rumen Microbiome Inﬂuenced by Diet and Microenvironments Deusch et al. 32, 39, 41, 44, 48, 54) consisting of a total of 417 proteins, the abundance within diets was the highest in the LP fractions of the grass silage- and hay-based diets and the corn silage-derived SP fraction (Figure 6B). The sum of abundance of groups belonging to the citrate cycle (11, 14, 17, 33, 38, 47, 57) based on a total of 76 proteins exhibited lower LFQ-values in the SP fractions of the grass silage- and hay-based diets when compared to the respective RF and LP fractions and the corn silage-based samples (Figure 6C). Archaeal and Bacterial Enzymes Involved in Carbon Metabolism Looking at the groups 29, 31, 45, 48 of the pentose phosphate pathway based on 34 proteins, the sum of LFQ-values was higher in the LP fractions of the grass silage- and hay-based diets when compared to the remaining samples (Figure 6D). FIGURE 7 \| Average abundance and phylogenetic origin of bacterial enzymes involved in (A) butyrate, (B) propionate, (C) acetate, and (D) formic acid production in diets (n = 9). Numbers of assigned proteins are shown in the respective phylogenetic legends. For butyrate biosynthesis: COG4770 (acetyl/propionyl-CoA carboxylase), COG3426 (butyrate kinase), COG1250/COG1024 (3-hydroxyacyl-CoA dehydrogenase), COG0183 (acetyl-CoA acetyltransferase). For propionate biosynthesis: COG4799 (acetyl-CoA carboxylase), COG2185/COG1884 (methylmalonyl-CoA mutase). For acetate production: COG1012 (NAD-dependent aldehyde dehydrogenase), COG0282 (acetate kinase), COG0280 (phosphotransacetylase), and for formic acid COG1882 (formate acetyltransferase). Enzymes of Short-Chain Fatty Acid Production Bacterial enzymes involved in the production of acetate, butyrate, propionate and formic acid (Table S7) were retrieved using COG assignments of WebMGA (Table S4). The LFQ-values of proteins belonging to the respective COGs were summarized as described in Polansky et al. (2015). Figure 7 depicts the phylogenetic origin of the corresponding enzymes averaged over the diets. In total 45 enzymes involved in butyrate production were dominated by Firmicutes species constituting above 90% of the total abundance in all diets whereas enzymes from Actinobacteria appeared exclusively in the corn silage-based diet (Figure 7A). Five proteins of Bacteroidetes species showed a maximum of 3.9% of abundance in the grass silage-based samples and exhibited LFQ- values below 0.3% in the corn silage-based diet. Thirteen proteins of the Lachnospiraceae family exhibited the highest LFQ-values in the hay-based diet with 59% of the total abundance, decreasing in the grass (56%) and corn (31%) silage-based diets. Likewise, the abundance of two enzymes derived from Ruminococcaceae was higher in the grass silage- and hay-based diet (1.7%) when compared to the average in the corn silage-based diet (1.3%). In contrast, ﬁve enzymes of Eubacteriaceae showed the highest LFQ-values in the corn silage-based diet and decreased in the grass silage- and hay-derived samples. A 3-hydroxybutyryl-CoA dehydrogenase of Megasphaera elsdenii belonging to the family of Veillonellaceae was present in high abundance in the corn silage-based diet (24%) while the LFQ-values in the grass silage- and hay-based diets accounted for 0.4% and 0.1% respectively (Figure 7A). Propionate production based on 52 proteins was dominated by 25 enzymes from Bacteroidetes species, mainly Prevotellaceae, composing above 75% of the total abundance in all samples (Figure 7B). Correspondingly, nine enzymes of the Prevotellaceae family prevailed in acetate production constituting above 60% of abundance in the corn silage-based diet increasing to 77% in the hay-based diet and 84% in the grass silage-based diet (Figure 7C). Two enzymes of Eubacterium species showed an abundance of 20% while decreasing in the hay- and the grass silage-based diets with 10 and 7% respectively. Concerning the 23 enzymes involved in formic acid production, four proteins from Lachnospiraceae species constituted 39% of abundance in the corn silage-based diet whereas reaching 65% in the hay- and 67% in the grass silage-derived samples (Figure 7D). Frontiers in Microbiology \| www.frontiersin.org Enzymes of Short-Chain Fatty Acid Production A formate C-acetyltransferase of Succinatimonas FIGURE 7 \| Average abundance and phylogenetic origin of bacterial enzymes involved in (A) butyrate, (B) propionate, (C) acetate, and (D) formic acid production in diets (n = 9). Numbers of assigned proteins are shown in the respective phylogenetic legends. For butyrate biosynthesis: COG4770 (acetyl/propionyl-CoA carboxylase), COG3426 (butyrate kinase), COG1250/COG1024 (3-hydroxyacyl-CoA dehydrogenase), COG0183 (acetyl-CoA acetyltransferase). For propionate biosynthesis: COG4799 (acetyl-CoA carboxylase), COG2185/COG1884 (methylmalonyl-CoA mutase). For acetate production: COG1012 (NAD-dependent aldehyde dehydrogenase), COG0282 (acetate kinase), COG0280 (phosphotransacetylase), and for formic acid COG1882 (formate acetyltransferase). Frontiers in Microbiology \| www.frontiersin.org August 2017 \| Volume 8 \| Article 1605 12 Rumen Microbiome Inﬂuenced by Diet and Microenvironments Deusch et al. TABLE 3 \| Average (∅) concentration (mM) and standard error of mean (SEM) of compounds identiﬁed in the RF fractions (n = 3) are shown. Metabolite Average concentration (mM) and SEM (n = 3) Corn RF Grass RF Hay RF ∅ SEM ∅ SEM ∅ SEM 2-Phenylpropionate* 0.14 0.02 0.12 0.05 0.15 0.05 3-Phenylpropionate* 0.38 0.05 0.40 0.08 0.39 0.10 Acetate* 54.86 3.79 59.70 8.49 55.95 3.41 Adipate 1.78 0.67 1.66 0.34 1.45 0.58 Aspartate 0.23 0.07 0.39 0.15 0.31 0.12 Butyrate* 14.98 2.20 16.84 5.58 15.81 1.54 Dimethylamine* 0.10 0.07 0.29 0.08 0.28 0.05 Formate 0.14 0.03 0.10 0.02 0.11 0.02 Fumarate 0.02 0.00 0.01 0.00 0.02 0.00 Isobutyrate* 1.43 0.57 1.29 0.42 1.51 0.49 Isovalerate* 1.31 0.26 1.09 0.32 1.02 0.04 Lactate 1.18 1.11 0.84 0.45 5.25 4.35 Methylamine* 0.49 0.15 0.94 0.12 0.70 0.16 Phenylacetate* 0.13 0.05 0.17 0.11 0.19 0.09 Pimelate 4.54 0.53 3.13 0.70 3.73 0.19 Propionate* 15.22 3.17 15.94 3.34 13.86 1.30 Succinate 0.38 0.38 0.10 0.05 0.25 0.11 Trimethylamine* 0.17 0.13 0.59 0.35 0.62 0.49 Urea 2.74 1.21 2.68 1.54 2.98 0.86 Valerate* 5.27 0.95 2.99 0.42 4.15 0.31 Acetate : Propionate 3.60 3.75 4.04 Compounds validated by 2D-NMR spectroscopy. RF, rumen ﬂuid. There was no statistically signiﬁcant difference in metabolite abundance between the diets (P > 0.05). TABLE 3 \| Average (∅) concentration (mM) and standard error of mean (SEM) of compounds identiﬁed in the RF fractions (n = 3) are shown. TABLE 3 \| Average (∅) concentration (mM) and standard error of mean (SEM) of compounds identiﬁed in the RF fractions (n = 3) are shown. Metabolites in Rumen Fluid NMR spectroscopy allowed the quantiﬁcation of 20 diﬀerent compounds in all RF samples. Two-dimensional NMR spectroscopy further validated the presence of 12 compounds (∗) including the major short-chain fatty acids acetate, butyrate, propionate, and valerate (Table 3). There were no statistically signiﬁcant alterations of the metabolites regarding the diﬀerent diets as determined by one-way ANOVA (P > 0.05). Overall, acetate was detected in highest concentrations ranging from 54.86 mM in the corn silage-based samples to 59.70 mM in the grass silage-derived RF samples. Propionate concentration was lower in the RF samples of the hay-based diet (13.86 mM) when compared to 15.22 and 15.94 mM in the corn silage- and grass silage-based samples, respectively (Table 3). The ratio of acetate to propionate concentration was highest in the hay-based samples and lowest in the corn silage-based samples. Butyrate was more abundant in the grass silage- and hay-derived RF samples (16.84 and 15.81 mM) when compared to 14.98 mM in the corn silage-based samples (Table 3). Valerate concentration was higher in corn silage- and hay-derived RF fractions (5.27 and 4.15 mM) when compared to 2.99 mM in the grass silage-based diet. The corn silage-based samples exhibited the highest concentration of succinate (0.38 mM) in comparison to the grass silage- and hay-based samples with 0.1 and 0.25 mM respectively (Table 3). Higher amounts of lactate were found in in the hay diet-based samples (5.25 mM) when compared to the corn silage- (1.18 mM) and grass silage-based samples (0.84 mM). Methylamine, dimethylamine, and trimethylamine were less abundant in the corn silage-based samples when compared to the grass silage and hay-based RF fractions (Table 3). Several compounds identiﬁed in low concentrations like aspartate, formate, fumarate, methylamines, and succinate are involved in the carbon metabolism as indicated in Figure 6. Compounds validated by 2D-NMR spectroscopy. RF, rumen ﬂuid. There was no statistically signiﬁcant difference in metabolite abundance between the diets (P > 0.05). as conﬁrmed by ANOSIM for both, the 16S rRNA gene and the metaproteomic analysis. There was no signiﬁcant eﬀect of the host animals on the inherent bacterial community structure (Table 2). Bacterial Community Composition and Activity Is Inﬂuenced by Dietary Treatments Dietary impact is the main factor shaping bacterial communities in the rumen (Ley et al., 2008; Henderson et al., 2015). Enzymes of Short-Chain Fatty Acid Production hippie belonging to the family of Succinivibrionaceae reached 31% of the total abundance in the corn silage-based diet and decreased to 19% in the hay- and 18% in the grass silage-based diet. Three formate acetyltransferases of Eubacterium species showed the maximum abundance of 6% in the corn silage- derived samples and were not present in the grass silage- and hay-based diets (Figure 7). Average concentration (mM) and SEM (n = 3) Frontiers in Microbiology \| www.frontiersin.org Metabolites in Rumen Fluid The higher amount of non-ﬁber carbohydrates, mainly starch and sugars, in the corn silage-based diet (Table 1) might be responsible for the increased LFQ and OTU abundances of members of the Proteobacteria phylum and the family of Succinivibrionaceae (Hespell, 1992; Bryant, 2015). Typical members of the Succinivibrionaceae family are Succinimonas amylolytica and Ruminobacter amylophilus. The latter one is restricted to starch and maltose as fermentation substrates and produces succinic, formic and acetic acids (Hamlin and Hungate, 1956). S. amylolytica was shown to increase in abundance when substrates contained starch with succinic, acetic and propionic acid being the main fermentation products (Bryant et al., 1958). Thus, the higher amount of non-ﬁber carbohydrates in the corn silage-based diet (Table 1) and the associated rise in abundance of Proteobacteria and Succinivibrionaceae species (Figure 2) DISCUSSION In this study, the dietary eﬀects of the most common forages in cattle production on the structure and function of the archaeal and bacterial communities inhabiting diﬀerent microenvironments of the rumen ecosystem were analyzed by a combination of shotgun-metaproteomics, Illumina amplicon sequencing and nuclear magnetic resonance to provide deeper insights into the complex microbial adaptation to varying substrates. In general, the bacterial community composition of the rumen samples was dominated by Bacteroidetes and Firmicutes species as reported by other nucleic acid-based studies (Jami and Mizrahi, 2012; Jami et al., 2014). However, the bacterial community arrangement changed signiﬁcantly in response to varying diets and diﬀered between the rumen sample fractions August 2017 \| Volume 8 \| Article 1605 13 Rumen Microbiome Inﬂuenced by Diet and Microenvironments Deusch et al. might have caused an increased production of succinate. The NMR-based metabolomic analysis determined the highest concentration of succinate in the corn silage-based RF fractions (Table 3). Moreover, the rise of Proteobacteria in the fractions of the corn silage-based diet (Figure 2) is in concordance with the increased abundance of monosaccharide transporters in the respective samples and the higher proteobacterial contribution to monosaccharide transporters when compared to the group of oligosaccharide transporters (Figure 5). Additionally, the abundance of proteobacterial glycosyltransferases and particularly carbohydrate esterases increased in the corn silage-derived LP and SP fractions (Figure 4). LFQ-values of proteins of the Selenomonadaceae family assessed in the present study revealed a lower abundance throughout the grass silage-based diet when compared to the corn silage- and the hay-based samples (Figure 2) including proteins involved in propionate formation (Table S7) and with sequence similarity to SLH (Table S3). The Gram-negative staining but phylogenetically Gram-positive rumen anaerobe S. ruminantium exhibits peptidoglycan-associated proteins with SLH that play an important role in the maintenance of the cell surface structure (Kojima et al., 2010). Furthermore, S. ruminantium species are characterized by their ability to use a broad range of substrates including the fermentation products of other bacteria (Bryant, 1956; Cotta, 1990; Rasmussen, 1993). Cross- feeding between S. ruminantium and Butyrivibrio ﬁbrisolvens was reported before (Cotta and Zeltwanger, 1995). Another study showed that the co-cultivation of S. ruminantium and B. ﬁbrisolvens promoted the growth of S. ruminantium (Cotta, 1992). Correspondingly, the LFQ-values of 124 Butyrivibrio proteins assessed by metaproteomics exhibited the lowest abundance in the samples of the grass silage-based diet. DISCUSSION Thus, the lower abundance of the Selenomonadaceae family in the grass silage-based diet might be linked to a similarly decreased abundance of the Butyrivibrio proteins in the respective samples. g g Beyond, the higher abundance of the Proteobacteria phylum and the Succinivibrionaceae family in the corn silage-based samples and the therewith associated increase in succinate formation might explain the rise of OTUs and proteins of the Acidaminococcaceae family in the samples of the corn silage- based diet (Figure 8). All OTUs of the Acidaminococcaceae family were assigned to the genus of the succinate-fermenting Succiniclasticum (van Gylswyk, 1995) and 48 of 70 identiﬁed proteins were produced by the asaccharolytic and succinate- utilizing Phascolarctobacterium species (Watanabe et al., 2012). Succiniclasticum and Phascolarctobacterium species of the Acidaminococcaceae family ferment succinate to produce propionate (Figure 8), the most important carbon source for the ruminant’s gluconeogenesis (Yost et al., 1977). In total six proteins of Phascolarctobacterium species involved in propionate production were identiﬁed (Table S7). There are reports about reduced methane emissions under corn silage-based dietary regimen (Beauchemin and McGinn, 2005; van Gastelen et al., 2015). A metagenomics study linked a decrease in methane emissions and abundance of methanogenic archaea to an increased abundance of the Succinivibrionaceae family (Wallace et al., 2015) whose members utilize hydrogen to produce succinate which is then rapidly converted to propionate and in this compete with the most common, hydrogenotrophic methanogenesis (Liu and Whitman, 2008; McCabe et al., 2015). Concerning the present study, the ratio of acetate to propionate, an indicator for methanogenic activity, was lowest in the corn silage-based diet (Table 3). Furthermore, the increased abundance levels of the Succinivibrionaceae family (Figure 2), the higher amount of succinate (Table 3) and the increased abundance of six propionate-producing Phascolarctobacterium proteins in the fractions of the corn silage-based diet may indicate a consistency with the above-mentioned investigations. The family of Erysipelotrichaceae of the Clostridium subphylum cluster XVII as well belonging to the phylum of Firmicutes exhibited the highest abundance of LFQs and particularly OTUs in the sample fractions of the hay- based diet (Figure 2). Most of the respective OTUs were assigned to the genera of Sharpea and Kandleria while most proteins were produced by species belonging to the genera of Coprobacillus, Catenibacterium, and Eggerthia. Due to phenotypic, chemotaxonomic and phylogenetic data, it was suggested that Lactobacillus catenaformis and L. Frontiers in Microbiology \| www.frontiersin.org DISCUSSION vitulinus should be reclassiﬁed into the genera of Eggerthia and Kandleria, respectively (Salvetti et al., 2011). Similar to Lactobacillus species, most members of Erysipelotrichaceae family probably ferment a wide range of sugars to produce mainly lactic acid (Figure 8) as reported for Sharpea azabuensis (Morita et al., 2008). This assumption may be supported by the higher amounts of lactate identiﬁed in the RF samples of the hay-based diet (Table 3). Moreover, studies of the sheep rumen linked a higher abundance of the Erysipelotrichaceae family in low-methane emitting animals to an increased lactic acid production in which less hydrogen and thus less methane is formed (Kittelmann et al., 2014; Kamke et al., 2016). The higher abundance of the Erysipelotrichaceae family, mainly Sharpea species was further associated with an increased abundance of Megasphaera species that convert formed lactate to butyrate (Kamke et al., 2016). In contrast, the higher abundance of the Erysipelotrichaceae family in samples of the hay-based diet determined in the present study was not accompanied by an increased abundance of Megasphaera species or the Veillonellaceae family that exhibited the highest abundance in the corn silage-based samples (Table S2). Fernando et al. (2010) reported a similar increase of Megasphaera species upon high grain diets. It was hypothesized that the enrichment of the Erysipelotrichaceae family is related No OTUs were assigned to the Firmicutes order of Bacillales or the family of Selenomonadaceae with its typical members of the rumen ecosystem, Selenomonas ruminantium and Anaerovibrio lipolytica. In contrast, 219 proteins were assigned to the family of Selenomonadaceae with the majority belonging to saccharolytic Selenomonas and lipolytic Anaerovibrio species, respectively. This may emphasize the beneﬁts and the necessity of applying multiple, complementary methods to investigate the microbiomes of complex ecosystems like the rumen. A previous study analyzed the identical sample material by quantitative real- time PCR and reported no dietary impact on the abundance of S. ruminantium (Lengowski et al., 2016). Contrarily, the August 2017 \| Volume 8 \| Article 1605 14 Rumen Microbiome Inﬂuenced by Diet and Microenvironments Deusch et al. FIGURE 8 \| Consolidation of metaproteomic, 16S rDNA sequencing and metabolomic analyses showing the most active and abundant rumen bacteria in the respective diets and sample fractions and their main fermentation products. Character size of bacterial family names and metabolites are in accordance to their abundance (LFQ for proteins, concentration for metabolites). Bacterial Community Composition and Activity Is Inﬂuenced by the Microenvironment to the smaller rumen size and higher ruminal turnover rates of low-methane emitting sheep which favors microorganisms with fast, hetero-fermentative growth on sugars (Kittelmann et al., 2014; Kamke et al., 2016). However, in the present study the rise of the Erysipelotrichaceae family in the larger rumen of the Jersey cows was observed exclusively under hay-based dietary regimen and the ratio of acetate to propionate, an indicator for methanogenic activity was highest in the hay-based diet (Table 3). The ﬁndings of the present study suggest further microbiological investigations of the yet sparsely described family of Erysipelotrichaceae to obtain a more valid basis regarding their functional capabilities and their role within the rumen ecosystem. Besides the eﬀect of animal feed composition, large variations in community structure between the diﬀerent microenvironments of the rumen ecosystem were observed emphasizing the importance of sample fractionation in rumen studies to cover the eﬀects of treatments throughout the whole ecosystem and its speciﬁc functional niches. The diﬀerence in bacterial community composition between sample fractions was shown to be greater than the diﬀerence between the same fractions of individual animals (Kong et al., 2010). Variations between the diﬀerent microenvironments have already been discovered before the advent of 16S rRNA gene-based community analysis since the chemical composition of ﬁrmly-attached bacteria was shown to be diﬀerent from the liquid-associated population whereas loosely-attached bacteria were rather similar to the liquid-associated population (Legay-Carmier and Bauchart, 1989) which conﬁrms the ﬁndings of the present study. Within diets, the more ﬁber-rich SP fractions revealed a signiﬁcant increase in abundance of polysaccharide degrading species of the Firmicutes order of Clostridiales including the families of Lachnospiraceae and Ruminococcaceae as found by OTU and protein abundance levels (Figure 2). The order of Clostridiales includes several cellulolytic Clostridium species (Vos et al., 2011) and as well ﬁber-degrading members of the Eubacteriaceae family (Prins et al., 1972; Van Gylswyk and Van der Toorn, 1985). Most of the OTUs within the order of Clostridiales, excepting Lachnospiraceae and Ruminococcaceae, were assigned to unclassiﬁed Clostridiales but no OTUs were assigned to the family of Eubacteriaceae while most respective Biﬁdobacterial genomes comprise features necessary for the metabolism of plant-derived complex carbohydrates like glycoside hydrolases and sugar ABC transporters (Pokusaeva et al., 2011). DISCUSSION Ack, acetate kinase; Fat, formate acetyltransferase; GH, glycosyl hydrolase; Hbd, hydroxybutyryl-CoA dehydrogenase; Lacd, lactate dehydrogenase; MemC methylmalonyl-CoA mutase. FIGURE 8 \| Consolidation of metaproteomic, 16S rDNA sequencing and metabolomic analyses showing the most active and abundant rumen bacteria in the respective diets and sample fractions and their main fermentation products. Character size of bacterial family names and metabolites are in accordance to their abundance (LFQ for proteins, concentration for metabolites). Ack, acetate kinase; Fat, formate acetyltransferase; GH, glycosyl hydrolase; Hbd, hydroxybutyryl-CoA dehydrogenase; Lacd, lactate dehydrogenase; MemC methylmalonyl-CoA mutase. Frontiers in Microbiology \| www.frontiersin.org Bacterial Community Composition and Activity Is Inﬂuenced by the Microenvironment The presence of phenylpropionate identiﬁed by NMR (Table 3) was reported to be essential for adherence to and degradation of cellulose by R. albus (Stack and Hungate, 1984). The phylum of Fibrobacteres is represented by the major cellulose degrader Fibrobacter succinogenes that is restricted to cellulose, hemicellulose or glucose as growth substrates (Hungate, 1950; Puniya et al., 2015). However, there were no signiﬁcant dietary eﬀects on the abundance of Fibrobacteres proteins and OTUs but within diets, the LFQ-values reached their maximum in the SP fractions (Figure 2) similar to the ﬁndings of a previous DNA sequencing study (de Menezes et al., 2011). Large diﬀerences between protein- and the OTU-based abundances might be explained by the fact that metaproteomic investigations depend on the amount and quality of reference sequences available for database searches (Seifert et al., 2013; Tanca et al., 2013). Currently only 15 genomes and thus, comparably small numbers of annotated protein sequences are available for the phylum of Fibrobacteres which limits mass spectrometry-based identiﬁcations. In contrast, 1,863 diﬀerent 16S rRNA gene sequences of the phylum of Fibrobacteres are deposited in the RDP database. The phylum of Spirochaetes was more abundant in the SP fractions as determined by both methods (Figure 2) conﬁrming There is not much information about the members of this phylum that are present in many anaerobic ecosystems including the gastrointestinal tract of animals but usually appear in low abundance within the respective environments (Jumas-Bilak and Marchandin, 2014). Synergistes jonesii was ﬁrst isolated from the rumen of goat and did not ferment carbohydrates but is thought to be involved in the degradation of plant-derived toxins such as mimosine and thus, might be beneﬁcial for the host animal (Allison et al., 1992). The overall low abundant phyla of Elusimicrobia and Verrucomicrobia were not aﬀected by the diﬀerent diets or sample fractions. However, the abundance of Elusimicrobia proteins was lower when compared to the respective OTUs whereas the abundance of Verrucomicrobia proteins was higher when compared to the corresponding OTUs (Table S2). A study employing total RNA sequencing as well as targeted RNA- and DNA amplicon sequencing identiﬁed the phyla of Elusimicrobia and Verrucomicrobia exclusively in the RNA-based datasets and proposed a higher activity of the respective phyla in the rumen (Li et al., 2016). Bacterial Community Composition and Activity Is Inﬂuenced by the Microenvironment Currently only 15 genomes and thus, comparably small numbers of annotated protein sequences are available for the phylum of Fibrobacteres which limits mass spectrometry-based identiﬁcations. In contrast, 1,863 diﬀerent 16S rRNA gene sequences of the phylum of Fibrobacteres are deposited in the RDP database. The Prevotellaceae family comprises common rumen bacteria such as Prevotella ruminicola, P. brevis, P. bryantii, and P. albensis and was shown to increase in abundance upon inclusion of concentrate in diets (Henderson et al., 2015). In the present study, the Prevotellaceae family constituted the most dominant bacterial family within the rumen ecosystem as reported before (Kim et al., 2011; Jami and Mizrahi, 2012). The abundance of proteins and OTUs assigned to the Prevotellaceae family was not aﬀected by the diﬀerent diets but within diets, the LFQ- values were highest in all LP fractions (Figure 2). A higher abundance of the Prevotellaceae family in the liquid rumen fractions has been reported before (Whitford et al., 1998; Kocherginskaya et al., 2001; Pitta et al., 2010). Members of the Prevotellaceae family are characterized by their versatile metabolic capabilities and their ability to utilize a broad range of substrates including peptides, proteins, monosaccharides, and plant polysaccharides (Miyazaki et al., 1997; Matsui et al., 2000; Purushe et al., 2010) and thus, may not be primarily aﬀected by changes in diet composition. The majority of identiﬁed CAZymes were produced by Prevotellaceae species including 46 glycoside hydrolases, 14 glycosyltransferases and six carbohydrate esterases emphasizing their functional prevalence (Table S3). Furthermore, most enzymes involved in acetate and propionate formation were derived from Prevotellaceae species (Figure 8). The high abundance of polysaccharide-degrading bacteria in the SP fractions described above might have caused an increased availability of monosaccharides and a higher abundance of proteins involved in glycolysis by the Embden-Meyerhof-Parnas pathway. Considering the diﬀerent texture and the higher amounts of starch in the corn silage-based diet when compared to the more ﬁbrous grass silage- and hay-based diets most monosaccharides possibly were present in the SP fraction of the corn silage-based diet whereas most monosaccharides from the degradation of structural plant polysaccharides were rather present in the LP fractions of the grass silage- and hay-based diets The phylum of Spirochaetes was more abundant in the SP fractions as determined by both methods (Figure 2) conﬁrming the ﬁndings of de Menezes et al. (2011). The majority of proteins and OTUs belonged to Treponema species. Bacterial Community Composition and Activity Is Inﬂuenced by the Microenvironment This might explain the higher abundance of proteins and OTUs of the family of Biﬁdobacteriaceae in the more ﬁber-rich grass silage- and hay-based samples when compared to the corn silage-based diet (Table S2). Moreover, a Biﬁdobacterium-derived glycosyltransferase of the family GT4 and oligosaccharide, polyol, and lipid transporters from Biﬁdobacteriaceae were more abundant in the grass silage-based diet and the hay-based samples when compared to the relatively low abundance in the corn-silage based diet. Furthermore, the polysaccharide-degrading capabilities of Actinobacteria including the Biﬁdobacteriaceae family might be reﬂected by the higher abundance in the SP fractions of all diets (Figure 2). Contrarily, the study of de Menezes et al. (2011) reported a higher abundance of 16S rRNA gene sequences of Actinobacteria in the LP fractions. August 2017 \| Volume 8 \| Article 1605 Frontiers in Microbiology \| www.frontiersin.org 15 Rumen Microbiome Inﬂuenced by Diet and Microenvironments Deusch et al. proteins were produced by Eubacterium and Clostridium species. The family of Lachnospiraceae comprises the prominent cellulolytic B. ﬁbrisolvens (Bryant and Small, 1956) and the pectinolytic Lachnospira multiparus (Duskova and Marounek, 2001). Members of the Lachnospiraceae family exhibit strong hydrolyzing activities with multiple sets of carbohydrate-active enzymes (Stackebrandt, 2014) which may explain the increased abundance in the SP fractions of all diets (Figure 2). The group of oligosaccharide, polyol and lipid transporters contained 27 Lachnospiraceae proteins that exhibited the highest abundance in the SP fractions of all diets (43%) when compared to the LP (29%) and the RF (28%) fractions. A similar increase in abundance of Lachnospiraceae species in the ﬁber-adherent fractions was reported by Larue et al. (2005). Butyrate formation seemed to be performed by the members of the Lachnospiraceae in the two ﬁber-rich diets (Figure 8) as proteins involved in butyrate formation revealed the highest abundance (Figure 7) and butyrate concentration was higher in the RF fractions of these samples (Table 3). Comparably, protein and OTU abundances of the Ruminococcaceae family with its prominent cellulolytic representatives Ruminococcus ﬂavefaciens and Ruminococcus albus were not signiﬁcantly inﬂuenced by the dietary treatments but the respective proteins were more abundant in the SP fraction of all diets. A comprehensive study of Henderson et al. (2015) reported a relatively even distribution of Ruminococcus species across diﬀerent diets and host animals comparable to the ﬁnding of the present study. Frontiers in Microbiology \| www.frontiersin.org Bacterial Community Composition and Activity Is Inﬂuenced by the Microenvironment The present data may conﬁrm a high metabolic activity of the Verrucomicrobia phylum due to the higher abundance of proteins when compared to the respective OTUs. proteins were produced by Eubacterium and Clostridium species. The family of Lachnospiraceae comprises the prominent cellulolytic B. ﬁbrisolvens (Bryant and Small, 1956) and the pectinolytic Lachnospira multiparus (Duskova and Marounek, 2001). Members of the Lachnospiraceae family exhibit strong hydrolyzing activities with multiple sets of carbohydrate-active enzymes (Stackebrandt, 2014) which may explain the increased abundance in the SP fractions of all diets (Figure 2). The group of oligosaccharide, polyol and lipid transporters contained 27 Lachnospiraceae proteins that exhibited the highest abundance in the SP fractions of all diets (43%) when compared to the LP (29%) and the RF (28%) fractions. A similar increase in abundance of Lachnospiraceae species in the ﬁber-adherent fractions was reported by Larue et al. (2005). Butyrate formation seemed to be performed by the members of the Lachnospiraceae in the two ﬁber-rich diets (Figure 8) as proteins involved in butyrate formation revealed the highest abundance (Figure 7) and butyrate concentration was higher in the RF fractions of these samples (Table 3). Comparably, protein and OTU abundances of the Ruminococcaceae family with its prominent cellulolytic representatives Ruminococcus ﬂavefaciens and Ruminococcus albus were not signiﬁcantly inﬂuenced by the dietary treatments but the respective proteins were more abundant in the SP fraction of all diets. A comprehensive study of Henderson et al. (2015) reported a relatively even distribution of Ruminococcus species across diﬀerent diets and host animals comparable to the ﬁnding of the present study. The presence of phenylpropionate identiﬁed by NMR (Table 3) was reported to be essential for adherence to and degradation of cellulose by R. albus (Stack and Hungate, 1984). The phylum of Fibrobacteres is represented by the major cellulose degrader Fibrobacter succinogenes that is restricted to cellulose, hemicellulose or glucose as growth substrates (Hungate, 1950; Puniya et al., 2015). However, there were no signiﬁcant dietary eﬀects on the abundance of Fibrobacteres proteins and OTUs but within diets, the LFQ-values reached their maximum in the SP fractions (Figure 2) similar to the ﬁndings of a previous DNA sequencing study (de Menezes et al., 2011). Large diﬀerences between protein- and the OTU-based abundances might be explained by the fact that metaproteomic investigations depend on the amount and quality of reference sequences available for database searches (Seifert et al., 2013; Tanca et al., 2013). August 2017 \| Volume 8 \| Article 1605 Bacterial Community Composition and Activity Is Inﬂuenced by the Microenvironment This might explain as well the high abundance of the pentose phosphate pathway in the LP fractions of the grass silage- and hay-based diets that probably contained most sugars within the ﬁber-rich diets (Figure 6D). Most studies of the rumen metabolome identiﬁed higher numbers of diﬀerent metabolites by a combination of NMR- and more sensitive MS-based approaches (Ametaj et al., 2010; Saleem et al., 2012, 2013). However, despite the statistical insigniﬁcancies the metabolite concentrations assessed by NMR in the present study further supported the ﬁndings of the 16S rRNA gene sequencing and the metaproteomic analysis. Insigniﬁcant metabolite patterns might also be related to the 52% identical composition of the experimental diets used in the present study. Next to the hydrogenotrophic methanogenesis, methylotrophic methanogens like the family of Methanomassiliicoccaceae belonging to the class of Thermoplasmata utilize compounds like methylamine, dimethylamine, and trimethylamine as their major energy and carbon sources. Trimethylamines are formed by bacteria from plant-derived glycine betaine and cholin (Neill et al., 1978; Mitchell et al., 1979). A trimethylamine corrinoid protein MttC of the methanogenic archaeon ISO4-H5 was identiﬁed in the metaproteomic analysis and the NMR-based analysis found higher concentrations of methylamine, dimethylamine and trimethylamine in the grass silage- and hay-based diets (Table 3) which is supported by other reports that indicate higher glycine betaine concentrations in grass as when compared to corn (Lerma et al., 1991; Chendrimada et al., 2002). The ratio of acetate to propionate was higher in the two ﬁber-rich diets that further contained more than twice as much methylamines when compared to the corn silage-based diet suggesting a potentially higher activity of methylotrophic methanogenesis in the ﬁber-rich diets. A previous study suggested to focus on methylamines rather than on hydrogen for mitigating methane emission from the rumen (Poulsen et al., 2013). Bacterial Community Composition and Activity Is Inﬂuenced by the Microenvironment Treponema bryantii is known to interact with cellulolytic bacteria (Stanton and Canale-Parola, 1980) and T. succinifaciens is able to ferment carbohydrates (Cwyk and Canale-Parola, 1979) which may explain the increased abundance throughout the SP fractions. Comparably, the low abundant phylum of Synergistetes, characterized in 2009, revealed the highest LFQ-values of proteins and particularly OTUs in the SP fractions (Table S2). August 2017 \| Volume 8 \| Article 1605 16 Rumen Microbiome Inﬂuenced by Diet and Microenvironments Deusch et al. linked to protozoa (Lange et al., 2005). In particular species of Methanobacteriaceae family were found to be associated with protozoa (Janssen and Kirs, 2008). The speciﬁc sample preparation protocols for the shotgun metaproteomic analysis focusing on the enrichment of prokaryotic cells might have caused bias against protozoa-associated archaea, which could further explain the diﬀerences between the DNA- and protein- derived datasets. The metaproteomic approach identiﬁed the alpha, beta and gamma subunits of the methyl-coenzyme M reductase based on 15 proteins (Table S7) that constituted 23% of the total abundance of 55 proteins that belonged to the methane metabolism as deﬁned by the KEGG database. Eleven proteins were produced by species of the Methanobacteriaceae family and four proteins were assigned to the Thermoplasmata phylum. The methyl-coenzyme M reductase catalyzes the ﬁnal step in the formation of methane (Ragsdale, 2014) and the mcrA gene, encoding the alpha subunit of the methyl-coenzyme M reductase, is preferably used as a functional marker since present in all methanogens (Denman et al., 2007). The abundance of the subunits of the methyl-coenzyme M reductase (Table S1) was lowest in the corn silage-based diet corresponding to the ratio of acetate to propionate that was lowest in the corn silage-based diet as well (Table 3). (Figure 6B). Similar to reports of Pitta et al. (2010), more ﬁbrous hay diets included the development of a digesta mat with clearly separated phases whereas wheat-based rumen content was more homogenized without a distinct ﬁbrous mat. Proteins involved in the citric acid cycle showed a remarkably low abundance in the SP fractions of the grass silage- and hay-based diets (Figure 6C) pointing toward a low abundance of substrates like succinate which is in accordance with the abundance values of proteins of the Embden-Meyerhof-Parnas pathway. Archaeal Community Differs in Sequence and Protein Composition p There were no diet, sample fraction or host related shifts in community structure of archaea probably due to the less versatile metabolic capabilities when compared to bacteria (Henderson et al., 2015) and the relatively low numbers of identiﬁed proteins and OTUs. The ﬁndings of the present study support the results of a previous study that analyzed the abundance of total methanogens and the Rumen Cluster C in the same sample material using quantitative real-time PCR (Lengowski et al., 2016). However, in contrast to the bacterial datasets, the archaeal community composition inferred from the quantiﬁed proteins diﬀered clearly from the structure obtained by Illumina amplicon sequencing (Figure 3). OTUs were exclusively assigned to three families of methanogens with members of the family Methanobacteriaceae and Methanomassiliicoccaceae being dominant throughout all diets and sample fractions similar to the results of other 16S rRNA gene-based studies (Janssen and Kirs, 2008; Seedorf et al., 2015). On the other hand, the phylogenetic composition inferred from the quantiﬁed archaeal proteins depicted a higher diversity including the presence of the Crenarchaeota and Thaumarchaeota phyla that were identiﬁed in low abundance in the rumen ecosystem before (Shin et al., 2004; Wang et al., 2016; Jin et al., 2017). The diﬀerent phylogenetic distributions of archaeal proteins and OTUs might be attributed to the prevalence of Methanobrevibacter ruminantium and M. gottschalkii of the Methanobacteriaceae family which dominate the ruminal archaeal community (Janssen and Kirs, 2008; St-Pierre and Wright, 2013; Seedorf et al., 2015; Danielsson et al., 2017). High numbers of the respective 16S rRNA genes might have prevented suﬃcient sequencing reads of low abundant 16S rRNA genes in the present study. Moreover, some archaea are endo- and ectosymbiotically REFERENCES Bryant, M. P., and Small, N. (1956). The anaerobic monotrichous butyric acid-producing curved rod-shaped bacteria of the rumen. J. Bacteriol. 72, 16–21. Allison, M. J., Mayberry, W. R., McSweeney, C. S., and Stahl, D. A. (1992). Synergistes jonesii, gen. nov., sp. nov.: a rumen bacterium that degrades toxic pyridinediols. Syst. Appl. Microbiol. 15, 522–529. doi: 10.1016/S0723-2020(11)80111-6 Bryant, M. P., Small, N., Bouma, C., and Chu, H. (1958). Bacteroides ruminicola. n. sp. and Succinimonas amylolytica; the new genus and species; species of succinic acid-producing anaerobic bacteria of the bovine rumen. J. Bacteriol. 76, 15–23. Ametaj, B. N., Zebeli, Q., Saleem, F., Psychogios, N., Lewis, M. J., Dunn, S. M., et al. (2010). Metabolomics reveals unhealthy alterations in rumen metabolism with increased proportion of cereal grain in the diet of dairy cows. Metabolomics 6, 583–594. doi: 10.1007/s11306-010-0227-6 Burbach, K., Seifert, J., Pieper, D. H., and Camarinha-Silva, A. (2016). Evaluation of DNA extraction kits and phylogenetic diversity of the porcine gastrointestinal tract based on Illumina sequencing of two hypervariable regions. Microbiologyopen 5, 70–82. doi: 10.1002/mbo3.312 Ann Huws, S., Chiariotti, A., Sarubbi, F., Carﬁ, F., and Pace, V. (2012). Eﬀects of feeding Mediterranean buﬀalo sorghum silage versus maize silage on the rumen microbiota and milk fatty acid content. J. Gen. Appl. Microbiol. 58, 107–112. doi: 10.2323/jgam.58.107 Camarinha-Silva, A., Jauregui, R., Chaves-Moreno, D., Oxley, A. P., Schaumburg, F., Becker, K., et al. (2014). Comparing the anterior nare bacterial community of two discrete human populations using Illumina amplicon sequencing. Environ. Microbiol. 16, 2939–2952. doi: 10.1111/1462-2920.12362 jg Beauchemin, K. A., and McGinn, S. M. (2005). Methane emissions from feedlot cattle fed barley or corn diets. J. Anim. Sci. 83, 653–661. doi: 10.2527/2005.833653x Carberry, C. A., Kenny, D. A., Han, S., McCabe, M. S., and Waters, S. M. (2012). Eﬀect of phenotypic residual feed intake and dietary forage content on the rumen microbial community of beef cattle. Appl. Environ. Microbiol. 78, 4949–4958. doi: 10.1128/AEM.07759-11 Belanche, A., de la Fuente, G., Pinloche, E., Newbold, C. J., and Balcells, J. (2012). Eﬀect of diet and absence of protozoa on the rumen microbial community and on the representativeness of bacterial fractions used in the determination of microbial protein synthesis. J. Anim. Sci. 90, 3924–3936. doi: 10.2527/jas.2011-4802 Chendrimada, T. P., Neto, M. G., Pesti, G. M., Davis, A. J., and Bakalli, R. I. (2002). Determination of the betaine content of feed ingredients using high-performance liquid chromatography. J. Sci. Food Agric. 82, 1556–1563. SUPPLEMENTARY MATERIAL The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fmicb. 2017.01605/full#supplementary-material Table S1 \| List of identiﬁed proteins and OTUs including the LFQ-values determined by MaxQuant. Table S2 \| Percentage abundance of archaeal and bacterial proteins and OTUs at phyla, class, order, and family level. Table S3 \| dbCAN assignments. List of identiﬁed CAZymes. AUTHOR CONTRIBUTIONS Table S4 \| WebMGA assignments. List of KEGG Orthology (KO) identiﬁers and Table S4 \| WebMGA assignments. List of KEGG Orthology (KO) identiﬁers and Cluster of Orthologous Groups of proteins (COG) assigned to archaeal and bacterial proteins. Cluster of Orthologous Groups of proteins (COG) assigned to archaeal and bacterial proteins. The authors’ contributions are as follows: MR and JS designed the research. SD, AC, JC, UB, MR, and JS conducted the research. SD, AC, JC, and JS analyzed the data. SD and JS wrote the manuscript. All authors read and approved the ﬁnal version of the manuscript. The authors’ contributions are as follows: MR and JS designed the research. SD, AC, JC, UB, MR, and JS conducted the research. Table S5 \| List of ABC transporters. SD, AC, JC, and JS analyzed the data. SD and JS wrote the manuscript. All authors read and approved the ﬁnal version of the manuscript. SD, AC, JC, and JS analyzed the data. SD and JS wrote the manuscript. All authors read and approved the ﬁnal version of the manuscript. Table S6 \| KEGG REACTIONS in carbon metabolism. Table S7 \| List of enzymes involved in short-chain fatty acid production. CONCLUSION The obtained datasets revealed signiﬁcant alterations of the structure and function of the microbial communities in response to the dietary treatments as determined unanimously by the protein- and the DNA-based analyses. Certain contrasts between the methods employed clearly emphasized the beneﬁts of using combinations of complementary methods to study the microbiome of complex ecosystems like the rumen. Moreover, tremendous variations in community August 2017 \| Volume 8 \| Article 1605 Frontiers in Microbiology \| www.frontiersin.org 17 Deusch et al. Rumen Microbiome Inﬂuenced by Diet and Microenvironments ACKNOWLEDGMENTS composition and functional patterns regarding the diﬀerent microenvironments within the rumen were observed by both methods prompting for the necessity of sample fractionation in rumen studies to cover the eﬀects of applied treatments throughout the whole ecosystem. The role of low abundant phyla such as Elusimicrobia, Synergistetes, Tenericutes, and Verrucomicrobia within the rumen ecosystems remains comparably vague since not much information about the respective species is available. This suggests microbiological investigations of the respective species in pure and mixed cultures to obtain more accurate data regarding their functional capabilities. This study may provide deeper insights into the complicated network of bacterial interactions and adaptions to various substrates. Bioinformatic and technical progress may enhance the metaproteomic coverage of future studies. The authors gratefully acknowledge the ﬁnancial support of the Carl Zeiss Stiftung. We would like to thank the ProteomeCenter in Tübingen for their excellent support. The authors gratefully acknowledge the ﬁnancial support of the Carl Zeiss Stiftung. We would like to thank the ProteomeCenter in Tübingen for their excellent support. REFERENCES Ecol. 62, 313–322. doi: 10.1111/j.1574-6941.2007.00394.x Hungate, R. E. (1966). The Rumen and Its Microbes. New York, NY; London: Academic Press. Hungate, R. E., Bryant, M. P., and Mah, R. A. (1964). The rumen bacteria and protozoa. Annu. Rev. Microbiol. 18, 131–166. doi: 10.1146/annurev.mi.18.100164.001023 Deusch, S., and Seifert, J. (2015). Catching the tip of the iceberg - evaluation of sample preparation protocols for metaproteomic studies of the rumen microbiota. Proteomics 15, 3590–3595. doi: 10.1002/pmic.201400556 Jagtap, P., Goslinga, J., Kooren, J. A., McGowan, T., Wroblewski, M. S., Seymour, S. L., et al. (2013). A two-step database search method improves sensitivity in peptide sequence matches for metaproteomics and proteogenomics studies. Proteomics 13, 1352–1357. doi: 10.1002/pmic.201200352 Deusch, S., Tilocca, B., Camarinha-Silva, A., and Seifert, J. (2015). News in livestock research - use of Omics-technologies to study the microbiota in the gastrointestinal tract of farm animals. Comput. Struct. Biotechnol. J. 13, 55–63. doi: 10.1016/j.csbj.2014.12.005 Jami, E., and Mizrahi, I. (2012). Composition and similarity of bovine rumen microbiota across individual animals. PLoS ONE 7:e33306. doi: 10.1371/journal.pone.0033306 Duskova, D., and Marounek, M. (2001). Fermentation of pectin and glucose, and activity of pectin-degrading enzymes in the rumen bacterium Lachnospira multiparus. Lett. Appl. Microbiol. 33, 159–163. doi: 10.1046/j.1472-765x.2001.00970.x Jami, E., White, B. A., and Mizrahi, I. (2014). Potential role of the bovine rumen microbiome in modulating milk composition and feed eﬃciency. PLoS ONE 9:e85423. doi: 10.1371/journal.pone.0085423 Edgar, R. C., Haas, B. J., Clemente, J. C., Quince, C., and Knight, R. (2011). UCHIME improves sensitivity and speed of chimera detection. Bioinformatics 27, 2194–2200. doi: 10.1093/bioinformatics/btr381 Janssen, P. H., and Kirs, M. (2008). Structure of the archaeal community of the rumen. Appl. Environ. Microbiol. 74, 3619–3625. doi: 10.1128/AEM.02812-07 Edwards, J. R., McEwan, N. J., Travis, A., and John Wallace, R. (2004). 16S rDNA library-based analysis of ruminal bacterial diversity. Anton. Van Leeuw. 86, 263–281. doi: 10.1023/B:ANTO.0000047942.69033.24 Jehmlich, N., Schmidt, F., Hartwich, M., von Bergen, M., Richnow, H. H., and Vogt, C. (2008). Incorporation of carbon and nitrogen atoms into proteins measured by protein-based stable isotope probing (Protein-SIP). Rapid Commun. Mass Spectrom. 22, 2889–2897. doi: 10.1002/rcm.3684 Fernando, S. C., Purvis, H. T. II., Najar, F. Z., Sukharnikov, L. O., Krehbiel, C. R., Nagaraja, T. G., et al. (2010). Rumen microbial population dynamics during adaptation to a high-grain diet. Appl. Environ. Microbiol. 76, 7482–7490. doi: 10.1128/AEM.00388-10 Jin, D., Zhao, S., Zheng, N., Bu, D., Beckers, Y., Denman, S. E., et al. (2017). REFERENCES but a core microbiome is found across a wide geographical range. Sci. Rep. 5:14567. doi: 10.1038/srep14567 Henderson, G., Cox, F., Kittelmann, S., Miri, V. H., Zethof, M., Noel, S. J., et al. (2013). Eﬀect of DNA extraction methods and sampling techniques on the apparent structure of cow and sheep rumen microbial communities. PLoS ONE 8:e74787. doi: 10.1371/journal.pone.0074787 Cox, J., Hein, M. Y., Luber, C. A., Paron, I., Nagaraj, N., and Mann, M. (2014). Accurate proteome-wide label-free quantiﬁcation by delayed normalization and maximal peptide ratio extraction, termed MaxLFQ. Mol. Cell. Proteom. 13, 2513–2526. doi: 10.1074/mcp.M113.031591 8:e74787. doi: 10.1371/journal.pone.0074787 Hespell, R. B. (1992). “The Genera Succinivibrio and Succinimonas,” in The Prokaryotes: A Handbook on the Biology of Bacteria: Ecophysiology, Isolation, Identiﬁcation, Applications, eds. A. Balows, H.G. Trüper, M. Dworkin, W. Harder, and K.-H. Schleifer. (New York, NY: Springer), 3979–3982. Creevey, C. J., Kelly, W. J., Henderson, G., and Leahy, S. C. (2014). Determining the culturability of the rumen bacterial microbiome. Microb. Biotechnol. 7, 467–479. doi: 10.1111/1751-7915.12141 Cwyk, W. M., and Canale-Parola, E. (1979). Treponema succinifaciens sp. nov., an anaerobic spirochete from the swine intestine. Arch. Microbiol. 122, 231–239. doi: 10.1007/BF00411285 Hess, M., Sczyrba, A., Egan, R., Kim, T. W., Chokhawala, H., Schroth, G., et al. (2011). Metagenomic discovery of biomass-degrading genes and genomes from cow rumen. Science 331, 463–467. doi: 10.1126/science.1200387 Danielsson, R., Dicksved, J., Sun, L., Gonda, H., Müller, B., Schnürer, A., et al. (2017). Methane production in dairy cows correlates with rumen methanogenic and bacterial community structure. Front. Microbiol. 8:226. doi: 10.3389/fmicb.2017.00226 Hettich, R. L., Pan, C., Chourey, K., and Giannone, R. J. (2013). Metaproteomics: harnessing the power of high performance mass spectrometry to identify the suite of proteins that control metabolic activities in microbial communities. Anal.Chem. 85, 4203–4214. doi: 10.1021/ac303053e de Menezes, A. B., Lewis, E., O’Donovan, M., O’Neill, B. F., Clipson, N., and Doyle, E. M. (2011). Microbiome analysis of dairy cows fed pasture or total mixed ration diets. FEMS Microbiol. Ecol. 78, 256–265. doi: 10.1111/j.1574-6941.2011.01151.x Heyer, R., Kohrs, F., Benndorf, D., Rapp, E., Kausmann, R., Heiermann, M., et al. (2013). Metaproteome analysis of the microbial communities in agricultural biogas plants. N. Biotechnol. 30, 614–622. doi: 10.1016/j.nbt.2013.01.002 Hungate, R. E. (1950). The anaerobic mesophilic cellulolytic bacteria. Bacteriol. Rev. 14, 1–49. Denman, S. E., Tomkins, N. W., and McSweeney, C. S. (2007). Quantitation and diversity analysis of ruminal methanogenic populations in response to the antimethanogenic compound bromochloromethane. FEMS Microbiol. REFERENCES Diﬀerences in ureolytic bacterial composition between the rumen digesta and rumen wall based on ureC gene classiﬁcation. Front. Microbiol. 8:385. doi: 10.3389/fmicb.2017.00385 Ferrer, M., Ghazi, A., Beloqui, A., Vieites, J. M., Lopez-Cortes, N., Marin- Navarro, J., et al. (2012). Functional metagenomics unveils a multifunctional glycosyl hydrolase from the family 43 catalysing the breakdown of plant polymers in the calf rumen. PLoS ONE 7:e38134. doi: 10.1371/journal.pone.00 38134 Jumas-Bilak, E., and Marchandin, H. (2014). “The phylum Synergistetes,” in The Prokaryotes: Other Major Lineages of Bacteria and the Archaea, eds E. Rosenberg, E. F. DeLong, S. Lory, E. Stackebrandt, and F. Thompson (Berlin; Heidelberg: Springer), 931–954. Gressley, T. F., Hall, M. B., and Armentano, L. E. (2011). Ruminant Nutrition Symposium: productivity, digestion, and health responses to hindgut acidosis in ruminants. J. Anim. Sci. 89, 1120–1130. doi: 10.2527/jas.2010-3460 Kamke, J., Kittelmann, S., Soni, P., Li, Y., Tavendale, M., Ganesh, S., et al. (2016). Rumen metagenome and metatranscriptome analyses of low methane yield sheep reveals a Sharpea-enriched microbiome characterised by lactic acid formation and utilisation. Microbiome 4:56. doi: 10.1186/s40168-016-0201-2 Gupta, N., and Pevzner, P. A. (2009). False discovery rates of protein identiﬁcations: a strike against the two-peptide rule. J. Proteome Res. 8, 4173–4181. doi: 10.1021/pr9004794 Kim, M., Morrison, M., and Yu, Z. (2011). Status of the phylogenetic diversity census of ruminal microbiomes. FEMS Microbiol. Ecol. 76, 49–63. doi: 10.1111/j.1574-6941.2010.01029.x Hackmann, T. J., and Spain, J. N. (2010). Invited review: ruminant ecology and evolution: perspectives useful to ruminant livestock research and production. J. Dairy Sci. 93, 1320–1334. doi: 10.3168/jds.2009-2071 Kittelmann, S., Pinares-Patiño, C. S., Seedorf, H., Kirk, M. R., Ganesh, S., McEwan, J. C., et al. (2014). Two diﬀerent bacterial community types are linked with the low-methane emission trait in sheep. PLoS ONE 9:e103171. doi: 10.1371/journal.pone.0103171 y j Hamlin, L. J., and Hungate, R. E. (1956). Culture and physiology of a starch- digesting bacterium (Bacteroides amylophilus n. sp.) from the bovine rumen. J. Bac. 72, 548–554. Kocherginskaya, S. A., Aminov, R. I., and White, B. A. (2001). Analysis of the rumen bacterial diversity under two diﬀerent diet conditions using denaturing gradient gel electrophoresis, random sequencing, and statistical ecology approaches. Anaerobe 7, 119–134. doi: 10.1006/anae.2001.0378 Hansen, S. H., Stensballe, A., Nielsen, P. H., and Herbst, F. A. (2014). Metaproteomics: evaluation of protein extraction from activated sludge. Proteomics 14, 2535–2539. doi: 10.1002/pmic.201400167 Koike, S., Yoshitani, S., Kobayashi, Y., and Tanaka, K. (2003). REFERENCES doi: 10.1002/jsfa.1214 Bladen, H. A., Bryant, M. P., and Doetsch, R. N. (1961). A study of bacterial species from the rumen which produce ammonia from protein hydrolyzate. Appl. Microbiol. 9, 175–180. Chourey, K., Jansson, J., VerBerkmoes, N., Shah, M., Chavarria, K. L., Tom, L. M., et al. (2010). Direct cellular lysis/protein extraction protocol for soil metaproteomics. J. Proteome Res. 9, 6615–6622. doi: 10.1021/pr100787q Bray, J. R., and Curtis, J. T. (1957). An ordination of the upland forest communities of southern Wisconsin. Ecol. Monogr. 27, 325–349. doi: 10.2307/1942268 Clarke, K. R., and Warwick, R. M. (2001). Change in Marine Communities: An Approach to Statistical Analysis and Interpretation, 2nd Edn. Plymouth: PRIMER-E. Brulc, J. M., Antonopoulos, D. A., Miller, M. E., Wilson, M. K., Yannarell, A. C., Dinsdale, E. A., et al. (2009). Gene-centric metagenomics of the ﬁber-adherent bovine rumen microbiome reveals forage speciﬁc glycoside hydrolases. Proc. Natl. Acad. Sci. U.S.A. 106, 1948–1953. doi: 10.1073/pnas.0806191105 Cole, J. R., Wang, Q., Fish, J. A., Chai, B., McGarrell, D. M., Sun, Y., et al. (2014). Ribosomal Database Project: data and tools for high throughput rRNA analysis. Cole, J. R., Wang, Q., Fish, J. A., Chai, B., McGarrell, D. M., Sun, Y., et al. (2014). Ribosomal Database Project: data and tools for high throughput rRNA analysis. Nucleic Acids Res. 42, D633–D642. doi: 10.1093/nar/gkt1244 Nucleic Acids Res. 42, D633–D642. doi: 10.1093/nar/gkt1244 Bryant, M. P. (1956). The characteristics of strains of Selenomonas isolated from bovine rumen contents. J. Bacteriol. 72, 162–167. Cotta, M. A. (1990). Utilization of nucleic acids by Selenomonas ruminantium and other ruminal bacteria. Appl. Environ. Microbiol. 56, 3867–3870. yant, M. P. (1959). Bacterial species of the rumen. Bacteriol. Rev. 2 Bryant, M. P. (2015). “Succinivibrio,” in Bergey’s Manual of Systematics of Archaea and Bacteria, ed W. B. Whitman (New York, NY: John Wiley & Sons, Ltd). 581–586. Cotta, M. A. (1992). Interaction of ruminal bacteria in the production and utilization of maltooligosaccharides from starch. Appl. Environ. Microbiol. 58, 48–54. August 2017 \| Volume 8 \| Article 1605 Frontiers in Microbiology \| www.frontiersin.org 18 Rumen Microbiome Inﬂuenced by Diet and Microenvironments Deusch et al. but a core microbiome is found across a wide geographical range. Sci. Rep. 5:14567. doi: 10.1038/srep14567 Cotta, M. A., and Zeltwanger, R. L. (1995). Degradation and utilization of xylan by the ruminal bacteria Butyrivibrio ﬁbrisolvens and Selenomonas ruminantium. Appl. Environ. Microbiol. 61, 4396–4402. REFERENCES – Omics’ of the mammalian gut—new insights into function. Curr. Opin. Biotechnol. 23, 491–500. doi: 10.1016/j.copbio.2012.01.016 Mitchell, A. D., Chappell, A., and Knox, K. L. (1979). Metabolism of betaine in the ruminant. J. Anim. Sci. 49, 764–774. doi: 10.2527/jas1979.493764x Lange, M., Westermann, P., and Ahring, B. K. (2005). Archaea in protozoa and metazoa. Appl. Microbiol. Biotechnol. 66, 465–474. doi: 10.1007/s00253-004-1790-4 Miyazaki, K., Martin, J. C., Marinsek-Logar, R., and Flint, H. J. (1997). Degradation and utilization of xylans by the rumen anaerobe Prevotella bryantii (formerly P. ruminicola subsp. brevis) B14. Anaerobe 3, 373–381. doi: 10.1006/anae.1997.0125 Larue, R., Yu, Z., Parisi, V. A., Egan, A. R., and Morrison, M. (2005). Novel microbial diversity adherent to plant biomass in the herbivore gastrointestinal tract, as revealed by ribosomal intergenic spacer analysis and rrs gene sequencing. Environ. Microbiol. 7, 530–543. doi: 10.1111/j.1462-2920.2005.00721.x Mizrahi, I. (2011). “The role of the rumen microbiota in determining the feed eﬃciency of dairy cows,” in Beneﬁcial Microorganisms in Multicellular Life Forms, eds E. Rosenberg and U. Gophna (Berlin; Heidelberg: Springer), 203–210. j Lee, H. J., Jung, J. Y., Oh, Y. K., Lee, S. S., Madsen, E. L., and Jeon, C. O. (2012). Comparative survey of rumen microbial communities and metabolites across one caprine and three bovine groups, using bar-coded pyrosequencing and (1)H nuclear magnetic resonance spectroscopy. Appl. Environ. Microbiol. 78, 5983–5993. doi: 10.1128/AEM.00104-12 Morita, H., Shiratori, C., Murakami, M., Takami, H., Toh, H., Kato, Y., et al. (2008). Sharpea azabuensis gen. nov., sp. nov., a Gram-positive, strictly anaerobic bacterium isolated from the faeces of thoroughbred horses. Int. J. Syst. Evol. Microb. 58, 2682–2686. doi: 10.1099/ijs.0.65543-0 Muth, T., Renard, B. Y., and Martens, L. (2016). Metaproteomic data analysis at a glance: advances in computational microbial community proteomics. Expert. Rev. Proteomics 13, 757–769. doi: 10.1080/14789450.2016.1209418 Legay-Carmier, F., and Bauchart, D. (1989). Distribution of bacteria in the rumen contents of dairy cows given a diet supplemented with soya-bean oil. Br. J. Nutr. 61, 725–740. doi: 10.1079/BJN19890158 Neill, A. R., Grime, D. W., and Dawson, R. M. (1978). Conversion of choline methyl groups through trimethylamine into methane in the rumen. Biochem. J. 170, 529–535. doi: 10.1042/bj1700529 Lengowski, M. B., Witzig, M., Möhring, J., Seyfang, G. M., and Rodehutscord, M. (2016). Eﬀects of corn silage and grass silage in ruminant rations on diurnal changes of microbial populations in the rumen of dairy cows. Anaerobe 42, 6–16. doi: 10.1016/j.anaerobe.2016.07.004 Petri, R. M., Schwaiger, T., Penner, G. REFERENCES B., Beauchemin, K. A., Forster, R. J., McKinnon, J. J., et al. (2013). Characterization of the core rumen microbiome in cattle during transition from forage to concentrate as well as during and after an acidotic challenge. PLoS ONE 8:e83424. doi: 10.1371/journal.pone.0083424 Lerma, C., Rich, P. J., Ju, G. C., Yang, W.-J., Hanson, A. D., and Rhodes, D. (1991). Betaine deﬁciency in maize: complementation tests and metabolic basis. Plant Physiol. 95, 1113–1119. doi: 10.1104/pp.95.4.1113 Physiol. 95, 1113–1119. doi: 10.1104/pp.95.4.1113 Pitta, D. W., Pinchak, E., Dowd, S. E., Osterstock, J., Gontcharova, V., Youn, E., et al. (2010). Rumen bacterial diversity dynamics associated with changing from bermudagrass hay to grazed winter wheat diets. Microb. Ecol. 59, 511–522. doi: 10.1007/s00248-009-9609-6 Ley, R. E., Hamady, M., Lozupone, C., Turnbaugh, P. J., Ramey, R. R., Bircher, J. S., et al. (2008). Evolution of mammals and their gut microbes. Science 320, 1647–1651. doi: 10.1126/science.1155725 Pokusaeva, K., Fitzgerald, G. F., and van Sinderen, D. (2011). Carbohydrate metabolism in Biﬁdobacteria. Genes Nutr. 6, 285–306. doi: 10.1007/s12263-010-0206-6 Li, F., Henderson, G., Sun, X., Cox, F., Janssen, P. H., and Guan, L. L. (2016). Taxonomic assessment of rumen microbiota using total RNA and targeted amplicon sequencing approaches. Front. Microbiol. 7:987. doi: 10.3389/fmicb.2016.00987 Polansky, O., Sekelova, Z., Faldynova, M., Sebkova, A., Sisak, F., and Rychlik, I. (2015). Important metabolic pathways and biological processes expressed by chicken cecal microbiota. Appl. Environ. Microbiol. 82, 1569–1576. doi: 10.1128/AEM.03473-15 Liu, Y., and Whitman, W. B. (2008). Metabolic, phylogenetic, and ecological diversity of the methanogenic archaea. Ann. N. Y. Acad. Sci. 1125, 171–189. doi: 10.1196/annals.1419.019 Poulsen, M., Schwab, C., Jensen, B. B., Engberg, R. M., Spang, A., Canibe, N., et al. (2013). Methylotrophic methanogenic Thermoplasmata implicated in reduced methane emissions from bovine rumen. Nat. Commun. 4, 1428. doi: 10.1038/ncomms2432 Mackie, R. I. (2000). “Molecular ecology and diversity in gut microbial ecosystems,” in Ruminant Physiology: Digestion, Metabolism, Growth and Reproduction, eds C. R. Bell, M. Brylinsky, and P. Johnson-Green (Halifax: Atlantic Canada Society for Microbial Ecology), 61–77. y gy Mackie, R. I. (2002). Mutualistic fermentative digestion in the gastrointestinal tract: diversity and evolution. Integr. Comp. Biol. 42, 319–326. doi: 10.1093/icb/42.2.319 Prins, R. A., Van Vugt, F., Hungate, R. E., and Van Vorstenbosch, C. J. (1972). A comparison of strains of Eubacterium cellulosolvens from the rumen. Anton. Leeuw. J. Microb. 38, 153–161. doi: 10.1007/BF02328087 Mackie, R. I., and Cann, I. K. O. (2005). REFERENCES Phylogenetic analysis of ﬁber-associated rumen bacterial community and PCR Henderson, G., Cox, F., Ganesh, S., Jonker, A., Young, W., and Janssen, P. H. (2015). Rumen microbial community composition varies with diet and host, August 2017 \| Volume 8 \| Article 1605 Frontiers in Microbiology \| www.frontiersin.org 19 Rumen Microbiome Inﬂuenced by Diet and Microenvironments Deusch et al. detection of uncultured bacteria. FEMS Microbiol. Lett. 229, 23–30. doi: 10.1016/S0378-1097(03)00760-2 McAllister, T. A., Bae, H. D., Jones, G. A., and Cheng, K. J. (1994). Microbial attachment and feed digestion in the rumen. J. Anim. Sci. 72, 3004–3018. doi: 10.2527/1994.72113004x Kojima, S., Ko, K. C., Takatsuka, Y., Abe, N., Kaneko, J., Itoh, Y., et al. (2010). Cadaverine covalently linked to peptidoglycan is required for interaction between the peptidoglycan and the periplasm-exposed S-layer- homologous domain of major outer membrane protein Mep45 in Selenomonas ruminantium. J. Bacteriol. 192, 5953–5961. doi: 10.1128/JB.00417-10 McCabe, M. S., Cormican, P., Keogh, K., O’Connor, A., O’Hara, E., Palladino, R. A., et al. (2015). Illumina MiSeq phylogenetic amplicon sequencing shows a large reduction of an uncharacterised Succinivibrionaceae and an increase of the Methanobrevibacter gottschalkii clade in feed restricted cattle. PLoS ONE 10:e0133234. doi: 10.1371/journal.pone.0133234 Kong, Y., Teather, R., and Forster, R. (2010). Composition, spatial distribution, and diversity of the bacterial communities in the rumen of cows fed diﬀerent forages. FEMS Microbiol. Ecol. 74, 612–622. doi: 10.1111/j.1574-6941.2010.00977.x McMichael, A. J., Powles, J. W., Butler, C. D., and Uauy, R. (2007). Food, livestock production, energy, climate change, and health. Lancet 370, 1253–1263. doi: 10.1016/S0140-6736(07)61256-2 Kozich, J. J., Westcott, S. L., Baxter, N. T., Highlander, S. K., and Schloss, P. D. (2013). Development of a dual-index sequencing strategy and curation pipeline for analyzing amplicon sequence data on the MiSeq Illumina sequencing platform. Appl. Environ. Microbiol. 79, 5112–5120. doi: 10.1128/AEM.01 043-13 McSweeney, C. S., Blackall, L. L., Collins, E., Conlan, L. L., Webb, R. I., Denman, S. E., et al. (2005). Enrichment, isolation and characterisation of ruminal bacteria that degrade non-protein amino acids from the tropical legume Acacia angustissima. Anim. Feed Sci. Technol. 121, 191–204. doi: 10.1016/j.anifeedsci.2005.02.018 Krause, D. O., and Russell, J. B. (1996). How many ruminal bacteria are there? J. Dairy Sci. 79, 1467–1475. doi: 10.3168/jds.S0022-0302(96)76506-2 Miron, J., Ben-Ghedalia, D., and Morrison, M. (2001). Invited review: adhesion mechanisms of rumen cellulolytic bacteria. J. Dairy Sci. 84, 1294–1309. doi: 10.3168/jds.S0022-0302(01)70159-2 Lamendella, R., VerBerkmoes, N., and Jansson, J. K. (2012). REFERENCES B., and Rychlik, J. L. (2001). Factors that alter rumen microbial ecology. Science 292, 1119–1122. doi: 10.1126/science.1058830 Van Gylswyk, N. O., and Van der Toorn, J. J. T. K. (1985). Eubacterium uniforme sp. nov. and Eubacterium xylanophilum sp. nov., ﬁber-digesting bacteria from the rumina of sheep fed corn stover. Int. J. Syst. Evol. Microb. 35, 323–326. doi: 10.1099/00207713-35-3-323 Saleem, F., Ametaj, B. N., Bouatra, S., Mandal, R., Zebeli, Q., Dunn, S. M., et al. (2012). A metabolomics approach to uncover the eﬀects of grain diets on rumen health in dairy cows. J. Dairy Sci. 95, 6606–6623. doi: 10.3168/jds.2012-5403 Vizcaino, J. A., Csordas, A., del-Toro, N., Dianes, J. A., Griss, J., Lavidas, I., et al. (2016). 2016 update of the PRIDE database and its related tools. Nucleic Acids Res. 44, D447–D456. doi: 10.1093/nar/gkw880 Saleem, F., Bouatra, S., Guo, A., Psychogios, N., Mandal, R., Dunn, S., et al. (2013). The bovine ruminal ﬂuid metabolome. Metabolomics 9, 360–378. doi: 10.1007/s11306-012-0458-9 Vos, P., Garrity, G., Jones, D., Krieg, N. R., Ludwig, W., Rainey, F. A., et al. (2011). Bergey’s Manual of Systematic Bacteriology: Volume 3: The Firmicutes. New York, NY: Springer. Salvetti, E., Felis, G. E., Dellaglio, F., Castioni, A., Torriani, S., and Lawson, P. A. (2011). Reclassiﬁcation of Lactobacillus catenaformis (Eggerth 1935, Moore and Holdeman 1970) and Lactobacillus vitulinus (Sharpe et al. 1973) as Eggerthia catenaformis gen. nov., comb. nov. and Kandleria vitulina gen. nov., comb. nov., respectively. Int. J. Syst. Evol. Microbiol. 61, 2520–2524. doi: 10.1099/ijs.0.029231-0 Wallace, R. J., Rooke, J. A., McKain, N., Duthie, C. A., Hyslop, J. J., and Ross, D. W. (2015). The rumen microbial metagenome associated with high methane production in cattle. BMC Genomics 16:839. doi: 10.1186/s12864-015-2032-0 Saude, E. J., Slupsky, C. M., and Sykes, B. D. (2006). Optimization of NMR analysis of biological ﬂuids for quantitative accuracy. Metabolomics 2, 113–123. doi: 10.1007/s11306-006-0023-5 Wang, L., Xu, Q., Kong, F., Yang, Y., Wu, D., Mishra, S., et al. (2016). Exploring the goat rumen microbiome from seven days to two years. PLoS ONE 11:e0154354. doi: 10.1371/journal.pone.0154354 Seedorf, H., Kittelmann, S., and Janssen, P. H. (2015). Few highly abundant operational taxonomic units dominate within rumen methanogenic archaeal species in New Zealand sheep and cattle. Appl. Environ. Microbiol. 81, 986–995. doi: 10.1128/AEM.03018-14 Wang, Q., Garrity, G. M., Tiedje, J. M., and Cole, J. R. (2007). Naive Bayesian classiﬁer for rapid assignment of rRNA sequences into the new bacterial taxonomy. Appl. Environ. Microbiol. REFERENCES 73, 5261–5267. doi: 10.1128/AEM.00062-07 Seifert, J., Herbst, F.-A., Nielsen, P. H., Planes, F. J., Jehmlich, N., Ferrer, M., et al. (2013). Bioinformatic progress and applications in metaproteogenomics for bridging the gap between genomic sequences and metabolic functions in microbial communities. Proteomics 13, 2786–2804. doi: 10.1002/pmic.201200566 Watanabe, Y., Nagai, F., and Morotomi, M. (2012). Characterization of Phascolarctobacterium succinatutens sp. nov., an asaccharolytic, succinate- utilizing bacterium isolated from human feces. Appl. Environ. Microbiol. 78, 511–518. doi: 10.1128/AEM.06035-11 Weljie, A. M., Newton, J., Mercier, P., Carlson, E., and Slupsky, C. M. (2006). Targeted proﬁling: quantitative analysis of 1H NMR metabolomics data. Anal. Chem. 78, 4430–4442. doi: 10.1021/ac060209g Shabat, S. K. B., Sasson, G., Doron-Faigenboim, A., Durman, T., Yaacoby, S., Berg Miller, M. E., et al. (2016). Speciﬁc microbiome-dependent mechanisms underlie the energy harvest eﬃciency of ruminants. ISME J. 10, 2958–2972. doi: 10.1038/ismej.2016.62 Whitford, M. F., Forster, R. J., Beard, C. E., Gong, J., and Teather, R. M. (1998). Phylogenetic analysis of rumen bacteria by comparative sequence analysis of cloned 16S rRNA genes. Anaerobe 4, 153–163. doi: 10.1006/anae.1998.0155 Shin, E. C., Choi, B. R., Lim, W. J., Hong, S. Y., An, C. L., Cho, K. M., et al. (2004). Phylogenetic analysis of archaea in three fractions of cow rumen based on the 16S rDNA sequence. Anaerobe 10, 313–319. doi: 10.1016/j.anaerobe.2004.08.002 g Wishart, D. S. (2008). Quantitative metabolomics using NMR. Trends Anal. Chem. Wishart, D. S. (2008). Quantitative metabolomics using NMR. Trends Anal. Chem 27, 228–237. doi: 10.1016/j.trac.2007.12.001 27, 228–237. doi: 10.1016/j.trac.2007.12.001 Wu, S., Zhu, Z., Fu, L., Niu, B., and Li, W. (2011). WebMGA: a customizable web server for fast metagenomic sequence analysis. BMC Genomics 12:444. doi: 10.1186/1471-2164-12-444 j Stack, R. J., and Hungate, R. E. (1984). Eﬀect of 3-phenylpropanoic acid on capsule and cellulases of Ruminococcus albus 8. Appl. Environ. Microbiol. 48, 218–223. Yin, Y., Mao, X., Yang, J., Chen, X., Mao, F., and Xu, Y. (2012). dbCAN: a web resource for automated carbohydrate-active enzyme annotation. Nucleic Acids Res. 40, W445–W451. doi: 10.1093/nar/gks479 Stackebrandt, E. (2014). “The family Lachnospiraceae,” in The Prokaryotes: Firmicutes and Tenericutes, eds E. Rosenberg, E. F. DeLong, S. Lory, E. Stackebrandt, and F. Thompson (Berlin; Heidelberg: Springer), 197–201. Stanton, T. B., and Canale-Parola, E. (1980). Treponema bryantii sp. nov., a rumen spirochete that interacts with cellulolytic bacteria. Arch. Microbiol. 127, 145–156. doi: 10.1007/BF00428018 Yost, W. M., Young, J. W., Schmidt, S. P., and McGilliard, A. D. (1977). REFERENCES A Review of Gastrointestinal Microbiology with Special Emphasis on Molecular Microbial Ecology Approaches. Dordrecht: Springer. Puniya, A. K., Singh, R., and Kamra, D. N. (2015). Rumen Microbiology: From Evolution to Revolution. New Delhi: Springer India. Purushe, J., Fouts, D. E., Morrison, M., White, B. A., Mackie, R. I., Coutinho, P. M., et al. (2010). Comparative genome analysis of Prevotella ruminicola and Prevotella bryantii: insights into their environmental niche. Microb. Ecol. 60, 721–729. doi: 10.1007/s00248-010-9692-8 Matsui, H., Ogata, K., Tajima, K., Nakamura, M., Nagamine, T., Aminov, R. I., et al. (2000). Phenotypic characterization of polysaccharidases produced by four Prevotella type strains. Curr. Microbiol. 41, 45–49. doi: 10.1007/s002840010089 August 2017 \| Volume 8 \| Article 1605 Frontiers in Microbiology \| www.frontiersin.org 20 Rumen Microbiome Inﬂuenced by Diet and Microenvironments Deusch et al. of microbial proﬁles in cow rumen fed with diﬀerent dietary ﬁber by tagged 16S rRNA gene pyrosequencing. Curr. Microbiol. 67, 130–137. doi: 10.1007/s00284-013-0336-3 of microbial proﬁles in cow rumen fed with diﬀerent dietary ﬁber by tagged 16S rRNA gene pyrosequencing. Curr. Microbiol. 67, 130–137. doi: 10.1007/s00284-013-0336-3 Ragsdale, S. W. (2014). “Biochemistry of methyl-coenzyme M reductase: the nickel metalloenzyme that catalyzes the ﬁnal step in synthesis and the ﬁrst step in anaerobic oxidation of the greenhouse gas methane,” in The Metal-Driven Biogeochemistry of Gaseous Compounds in the Environment, eds P. M. H. Kroneck and M. E. S. Torres (Dordrecht: Springer), 125–145. van Gastelen, S., Antunes-Fernandes, E. C., Hettinga, K. A., Klop, G., Alferink, S. J., Hendriks, W. H., et al. (2015). Enteric methane production, rumen volatile fatty acid concentrations, and milk fatty acid composition in lactating Holstein- Friesian cows fed grass silage- or corn silage-based diets. J. Dairy Sci. 98, 1915–1927. doi: 10.3168/jds.2014-8552 Rappsilber, J., Mann, M., and Ishihama, Y. (2007). Protocol for micro-puriﬁcation, enrichment, pre-fractionation and storage of peptides for proteomics using StageTips. Nat. Protoc. 2, 1896–1906. doi: 10.1038/nprot.2007.261 Rasmussen, M. A. (1993). Isolation and characterization of Selenomonas ruminantium strains capable of 2-deoxyribose utilization. Appl. Environ. Microbiol. 59, 2077–2081. van Gylswyk, N. O. (1995). Succiniclasticum ruminis gen. nov., sp. nov., a ruminal bacterium converting succinate to propionate as the sole energy- van Gylswyk, N. O. (1995). Succiniclasticum ruminis gen. nov., sp. nov., a ruminal bacterium converting succinate to propionate as the sole energy- yielding mechanism. Int. J. Syst. Bacteriol. 45, 297–300. doi: 10.1099/00207713- 45-2-297 yielding mechanism. Int. J. Syst. Bacteriol. 45, 297–300. doi: 10.1099/00207713- 45-2-297 Russell, J. REFERENCES Gluconeogenesis in ruminants: propionic acid production from a high-grain diet fed to cattle. J. Nutr. 107, 2036–2043. St-Pierre, B., and Wright, A. D. (2013). Diversity of gut methanogens in herbivorous animals. Animal 7, 49–56. doi: 10.1017/S1751731112000912 Zhang, R., Zhu, W., Zhu, W., Liu, J., and Mao, S. (2013). Eﬀect of dietary forage sources on rumen microbiota, rumen fermentation and biogenic amines in dairy cows. J. Sci. Food Agric. 94, 1886–1895. doi: 10.1002/jsfa.6508 Tajima, K., Aminov, R. I., Nagamine, T., Matsui, H., Nakamura, M., and Benno, Y. (2001). Diet-dependent shifts in the bacterial population of the rumen revealed with Real-Time PCR. Appl. Environ. Microbiol. 67, 2766–2774. doi: 10.1128/AEM.67.6.2766-2774.2001 Conﬂict of Interest Statement: The authors declare that the research was conducted in the absence of any commercial or ﬁnancial relationships that could be construed as a potential conﬂict of interest. Conﬂict of Interest Statement: The authors declare that the research was conducted in the absence of any commercial or ﬁnancial relationships that could be construed as a potential conﬂict of interest. Tanca, A., Palomba, A., Deligios, M., Cubeddu, T., Fraumene, C., Biosa, G., et al. (2013). Evaluating the impact of diﬀerent sequence databases on metaproteome analysis: insights from a lab-assembled microbial mixture. PLoS ONE 8:e82981. doi: 10.1371/journal.pone.0082981 Copyright © 2017 Deusch, Camarinha-Silva, Conrad, Beifuss, Rodehutscord and Seifert. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. Tanca, A., Palomba, A., Fraumene, C., Pagnozzi, D., Manghina, V., Deligios, M., et al. (2016). The impact of sequence database choice on metaproteomic results in gut microbiota studies. Microbiome 4:51. doi: 10.1186/s40168-016-0196-8 Thoetkiattikul, H., Mhuantong, W., Laothanachareon, T., Tangphatsornruang, S., Pattarajinda, V., Eurwilaichitr, L., et al. (2013). Comparative analysis August 2017 \| Volume 8 \| Article 1605 Frontiers in Microbiology \| www.frontiersin.org 21
https://openalex.org/W2059206459	https://bmcpregnancychildbirth.biomedcentral.com/counter/pdf/10.1186/s12884-015-0495-2	English	null	A prospective cohort study comparing the reactogenicity of trivalent influenza vaccine in pregnant and non-pregnant women	BMC pregnancy and childbirth	2,015	cc-by	5,341	* Correspondence: Annette.Regan@health.wa.gov.au 1School or Pathology and Laboratory Medicine, University of Western Australia, 227 Stubbs Terrace Shenton Park, Western Australia, Australia 2Communicable Disease Control Directorate, Western Australia Department of Health, Shenton Park, Western Australia, Australia Full list of author information is available at the end of the article Abstract Background: Influenza vaccination during pregnancy can prevent serious illness in expectant mothers and provide protection to newborns; however, historically uptake has been limited due to a number of factors, including safety concerns. Symptomatic complaints are common during pregnancy and may be mistakenly associated with reactions to trivalent influenza vaccine (TIV). To investigate this, we compared post-vaccination events self-reported by pregnant women to events reported by non-pregnant women receiving TIV. Methods: A prospective cohort of 1,086 pregnant women and 314 non-pregnant female healthcare workers (HCWs) who received TIV between March-May 2014 were followed-up seven days post-vaccination to assess local and systemic adverse events following immunisation (AEFIs). Women were surveyed by text message regarding perceived reactions to TIV. Those reporting an AEFI completed an interview by telephone or mobile phone to ascertain details. Logistic regression models adjusting for age and residence were used to compare reactions reported by pregnant women and non-pregnant HCWs. Results: Similar proportions of pregnant women and non-pregnant, female HCWs reported ≥1 reaction following vaccination with TIV (13.0% and 17.3%, respectively; OR = 1.2 [95% CI: 0.8-1.8]). Non-pregnant, female HCWs were more likely to report fever or headache compared to pregnant women (OR: 4.6 [95% CI 2.1-10.3] and OR: 2.2 [95% CI 1.0-4.6], respectively). No other significant differences in reported symptoms were observed. No serious vaccine-associated adverse events were reported, and less than 2% of each group sought medical advice for a reaction. Conclusions: We found no evidence suggesting pregnant women are more likely to report adverse events following influenza vaccination when compared to non-pregnant female HCWs of similar age, and in some cases, pregnant women reported significantly fewer adverse events. These results further support the safety of TIV administered in pregnant women. Keywords: Trivalent influenza vaccine, Pregnancy, Vaccine safety, Antenatal immunisation pregnant women in Australia are immunised against sea- sonal influenza [2,3]. A number of studies have confirmed influenza antenatal vaccination is safe for mother and baby [4-10]. However, continued monitoring is warranted, con- sidering the antigenic composition can vary from year to year and ongoing concerns about side-effects remain a common factor contributing to non-vaccination among antenatal patients [2,11-13]. Even the expectation of minor post-vaccination reactions can negatively affect the decision to be immunised against influenza [14,15]. Pregnancy can be associated with a variety of symptomatic Regan et al. BMC Pregnancy and Childbirth (2015) 15:61 DOI 10.1186/s12884-015-0495-2 Regan et al. BMC Pregnancy and Childbirth (2015) 15:61 DOI 10.1186/s12884-015-0495-2 A prospective cohort study comparing the reactogenicity of trivalent influenza vaccine in pregnant and non-pregnant women Annette K Regan1,2, Lauren Tracey2, Christopher C Blyth3,4, Donna B Mak2, Peter C Richmond3,4, Geoffrey Shellam1, Caroline Talbot4 and Paul V Effler1,2 Annette K Regan1,2, Lauren Tracey2, Christopher C Blyth3,4, Donna B Mak2, Peter C Richmond3,4, Geoffrey Shellam1, Caroline Talbot4 and Paul V Effler1,2 RESEARCH ARTICLE Open Access © 2015 Regan et al.; licensee BioMed Central. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Outcome measurement Th f Persons who replied “Y”, “yes” or some other affirma- tive response by SMS were sent a follow-up message soliciting details regarding the possible AEFI they re- ported experiencing. The second message read: The occurrence of any AEFI was defined as a “yes” re- sponse to the initial SMS message. A systemic reaction was defined as a “yes” response to fever, headache, fatigue, vomiting, rigors, or self-reported cold and flu-like symp- toms, myalgia, nausea, or malaise. A local reaction was de- fined as replying “yes” to pain or swelling at the injection, or self-reported redness at the site of injection. A reaction requiring telephone advice was defined as any AEFI where the woman reported calling a GP, a nurse helpline, or other healthcare service for advice regarding their reaction. A re- action requiring medical attention was defined as any AEFI where the woman reported visiting a GP or other health service for the reaction. An AEFI requiring treatment in- cluded any AEFI which was self-treated with an anti- pyretic/analgesic following vaccination and those receiving treatment by a medical professional. “Thank you, your ongoing health is important to us. Please click here to answer a five minute survey about your reaction. Alternatively, please respond CALL if you would prefer to be telephoned about your reaction”. The second SMS included an embedded link to a survey which could be completed on a mobile phone. Research nurses subsequently attempted to telephone and interview anyone who had not responded to either the first or second SMS, or had not completed the mobile phone survey, as well as those who had replied “Call” by SMS. For this analysis, all non-pregnant, female healthcare workers (HCWs) in the follow-up program, who were of reproductive age and were vaccinated with the same brand of TIV, i.e. Vaxigrip® (Sanofi Pasteur) were selected for Background The World Health Organisation has identified pregnant women as the highest priority for influenza vaccination [1]. Despite national recommendations in Australia and the availability of free vaccine under the National Immunisa- tion Program, surveys have found that less than 30% of Regan et al. BMC Pregnancy and Childbirth (2015) 15:61 Page 2 of 7 comparison with pregnant women. Female HCWs were eligible for the analysis if they were between the ages of 18 and 45 years and had indicated on their consent form that they were not pregnant at the time of vaccination. The ma- jority (82%) of pregnant women included in the analysis were in their second or third trimester of pregnancy; 93% of reported vaccinations in pregnant women and non- pregnant female HCWs were included in the follow-up. Participants who provided no telephone number (5%), pro- vided only a home telephone number (2%) or an incorrect mobile telephone number (<1%) on their consent form were excluded. Ethics approval for this assessment was obtained by the University of Western Australia Human Research Ethics Committee (RA/4/1/6095). complaints and whether these impact the side effects re- ported by antenatal influenza vaccine recipients is currently unknown. To assess this, we compared post-vaccination reactions among pregnant women to those reported by non-pregnant females of similar age in Western Australia. Methods In 2012, the Western Australia Department of Health (WA-DOH) initiated a program for active surveillance of adverse events following immunisation (AEFI) in pregnant women. The Follow-up and Active Surveillance of Triva- lent influenza vaccine in Mums (FASTMum) program fol- lows up a subset of pregnant women who receive trivalent influenza vaccine (TIV), beginning in March each year. Antenatal women receiving government-procured TIV are asked by their provider at the time of immunisation if they are willing to be contacted by the WA-DOH for quality as- surance purposes. In 2014, the opportunity for post- vaccination follow-up was extended to healthcare workers (HCWs) immunised against influenza at government hos- pitals and health centres. Survey instrument h b l h The mobile phone and telephone questionnaires asked if the vaccinee had experienced fever, headache, fatigue, rigors, convulsions, vomiting, or pain or swelling at the in- jection site - indicated by a “yes” or “no” response to each. The presence or absence of other symptoms was solicited and, if present, recorded verbatim. Respondents were also asked to recall the time between vaccination and first symptom onset as well as the duration of any symptoms reported. Consumption of over-the-counter antipyretic or pain relievers following the vaccination was queried, as well as whether the vaccinee had called a general practitioner (GP) or other health service for telephone medical advice regarding the reaction, or had visited a GP, after-hours clinic, or emergency department (ED) to receive treatment for a reported reaction. Using an automated system, pregnant women and HCWs consenting to follow up were sent a short mes- sage service (SMS) seven days after they had been vacci- nated with TIV. The SMS read: “This is a message from the WA Department of Health. Our records show that you recently had a flu vaccine and we are conducting routine follow up. Please respond Y if you experienced any kind of reaction, fever, or illness in the week following your vaccination, or N if there was no reaction”. Results Between 19 March and 15 May 2014, a total of 1,400 women (1,086 pregnant and 314 non-pregnant HCW) were sent the SMS asking about possible AEFI (Figure 1); 1,205 (86%) women replied by SMS (918 [85%] pregnant and 287 [91%] non-pregnant), and another 71 (64 [6%] pregnant and 7 [2%] non-pregnant) did not reply but were surveyed later by telephone. A total of 52 women (35 [4%] pregnant and 17 [6%] non-pregnant) who re- plied to the initial SMS and indicated they had experi- enced a reaction did not provide AEFI details and were excluded from analysis. The final analysis included 947 pregnant women and 275 non-pregnant women. The overall response rates in pregnant women (87.2% [95% CI 85.2-89.2%]) and non-pregnant, female HCWs (87.6% [95% CI 83.9-91.2%]) were similar (p > .05). Almost twice as many non-pregnant, female HCWs re- ported a reaction for which they obtained some form of treatment, such as self-treatment with an antipyretic or pain reliever or treatment by a doctor, medical centre or hospital emergency department (10.7% [95% CI 6.6-14.8%]) com- pared with pregnant women (5.5% [95% CI 4.0-7.0%]). However, this difference was not statistically significant (p = .06) (Table 2). This difference in proportion of reac- tions treated between pregnant and non-pregnant, fe- male HCWs can be largely attributed to the increased rates of fever and headache reported by non-pregnant, female HCWs. Among women reporting any reaction, headache and fever were the only symptoms signifi- cantly associated with seeking some form of treatment (p = .03 and p < .01, respectively). Reactions requiring tele- phone advice or medical attention were uncommon in both pregnant women and non-pregnant, female HCWs (1.3% [95% CI 0.5-2.0%] vs 0.4% [95% CI 0.0-1.2%], p = .25). Two significant differences in demographic character- istics were observed between the cohorts of pregnant women and non-pregnant, female HCWs included in our study. Non-pregnant, female HCWs were on average 2.6 years older than our cohort of pregnant women (33.7 years vs. 31.1 years, respectively, p < .01), and were also more likely to reside in a non-urban area (47.8% [95% CI 42.2-53.3%]) compared to pregnant women (15.6% [95% CI 13.3-17.9%]; p < .01). The greater propor- tion of non-pregnant female HCWs residing in non- urban areas is likely because many metropolitan health care facilities offered HCWs an intra-dermal influenza vaccine in preference to Vaxigrip® (Sanofi Pasteur). Statistical analysis Statistical analysis was performed using SAS version 9.3 (SAS Institute, North Carolina, United States). Overall re- sponse rate was calculated based on the proportion of Regan et al. BMC Pregnancy and Childbirth (2015) 15:61 Page 3 of 7 women who replied by either SMS or telephone and pro- vided complete details regarding any adverse events out of all women contacted. Initial comparisons between AEFI re- ported by pregnant women and non-pregnant, female HCWs were made using Fisher’s exact test. Adjusted ana- lyses controlling for demographic differences observed be- tween groups were performed using multivariate logistic regression models. Differences in the mean symptom onset and symptom duration were compared with independent sample t-tests using the Satterthwaite approximation for degrees of freedom. A power analysis indicated the ac- quired sample size was sufficient to determine differences between groups at a power level of 0.98. reported. Systemic reactions were reported by similar pro- portions of pregnant women and non-pregnant, female HCWs, overall (9.0% and 10.2% among pregnant women and non-pregnant HCWs, respectively). However, fever (OR 4.6 [95% CI 2.1-10.3]) and headache (OR 2.2 [95% CI 1.0-4.6]) were both reported more frequently by non- pregnant HCWs than pregnant women. Four of the 16 non-pregnant HCWs and five of the 46 pregnant women who reported a fever reported measuring their temperature. On average, reported fever began within 24 hours of vac- cination (median: 24 hours; IQR: 6–48 hours) and lasted between 8–120 hours (median: 27 hours; IQR: 12–48 hours). The time to onset and duration of fever were simi- lar in pregnant women and non-pregnant, female HCWs (p = .52 and p = .14, respectively). Other reported systemic reactions usually occurred within 24 hours of vaccination (median: 24 hours; IQR: 6–48 hours) and lasted for a me- dian of 48 hours (IQR: 24–72 hours). The onset and dur- ation of these reactions did not differ between pregnant women and non-pregnant, female HCWs (p = .26 and p = .21, respectively). Local reactions typically began on the day of vaccination (median: 8 hours; IQR: 3–24 hours) and had a median duration of 48 hours (IQR: 24–72 hours). The onset and duration of local reactions did not differ be- tween pregnant women and non-pregnant, female HCWs (p = .18 and p = .24, respectively). Results A total of 192 (15.7%) women reported a suspected reac- tion, with similar proportions of pregnant and non- pregnant, female HCWs reporting at least one AEFI (13.0% [95% CI 11.0-15.0%] and 17.3% [95% CI 13.0- 21.6%], respectively; p = .34) (Table 1). The rate of reaction was constant for both pregnant women and non-pregnant, female HCWs throughout the study period (Figure 2). The most common reaction reported by both pregnant and non-pregnant HCWs was a local reaction (4.5% [95% CI 3.4-6.1%] and 7.3% [95% CI 4.1-10.5%], respectively, p = .13). No serious vaccine-associated reactions were Four pregnant women reported attending a hospital emer- gency department in the week following influenza vaccin- ation. One woman reported fever and rigors, a second reported gastroenteritis, the third reported an upper respira- tory tract infection, and the fourth woman reported nausea, dizziness, malaise and a miscarriage. Follow-up assessment by the physician caring for the woman who reported a mis- carriage indicated the woman had a history of obstetric complications, including polycystic ovarian syndrome and multiple previous miscarriages. The physician reported the Regan et al. BMC Pregnancy and Childbirth (2015) 15:61 Page 4 of 7 Figure 1 Follow-up of adverse events following trivalent influenza vaccine in pregnant women and non-pregnant female healthcare workers – FASTMum, Western Australia, Australia, 19 March- 15 May 2014. Figure 1 Follow-up of adverse events following trivalent influenza vaccine in pregnant women and non-pregnant female healthcare workers – FASTMum, Western Australia, Australia, 19 March- 15 May 2014. event was consistent with a spontaneous abortion and un- likely to be related to vaccination. group was a local reaction at the injection site, occurring in about one of every 15–20 women vaccinated. This in- formation is useful in reassuring pregnant women and antenatal immunisation providers regarding the reacto- gencity of seasonal influenza vaccination during preg- nancy. However, because the antigenic characteristics of the influenza vaccine can change from year to year, on- going assessments of safety and reactogencity are war- ranted. Secondarily, these results indicate that SMS is a feasible method of rapidly collecting data for monitoring vaccine safety in both pregnant and non-pregnant women. Discussion We used SMS to collect information on post-influenza vaccination events in a sample of pregnant and non- pregnant women, and found no evidence that pregnant women are more likely to experience a reaction follow- ing administration of the 2014 influenza vaccination when compared to non-pregnant, female HCWs of simi- lar age. Using active surveillance, we found that 1-in-10 pregnant women experienced some sort of reaction, but fewer than 2% developed a fever. These results were similar for non-pregnant, female HCWs, although this group reported slightly higher rates of fever and head- ache. The most common side-effect reported by either Previous active surveillance initiatives in Western Australia in 2012 [9] and 2013 [10] found AEFI rates similar to those reported here for the 2014 influenza vaccine. Comparable rates of AEFI among pregnant Regan et al. BMC Pregnancy and Childbirth (2015) 15:61 Page 5 of 7 Table 1 Adverse events following influenza immunisation reported by pregnant and non-pregnant women – FASTMum, Western Australia, Australia, 19 March-15 May 2014 Pregnant (n = 947) Non-pregnant (n = 275) Fisher’s exact test p-value AOR p-value n Percent (95% CI) n Percent (95% CI) Any reaction** 141 13.0 (11.0-15.0) 51 17.5 (13.1-21.8) .19 .33 Systemic reaction 85 9.0 (7.1-10.8) 28 10.2 (6.6-13.8) .55 .36 Fever 15 1.6 (0.8-2.4) 16 5.8 (3.0-8.6) <.01* <.01* Headache 27 2.9 (1.8-3.9) 13 4.7 (2.2-7.3) <.01* .04* Fatigue 40 4.2 (2.9-5.5) 13 4.7 (2.2-7.3) .74 .68 Vomiting 7 0.7 (0.2-1.3) 0 (0.0-0.7) .36 .95 Rigors 5 0.5 (0.1-1.0) 2 0.7 (0.0-1.7) .66 .89 Cold/flu-like 37 3.9 (2.7-5.1) 10 3.6 (1.4-5.9) .50 .69 Myalgia 11 1.2 (0.5-1.8) 5 1.8 (0.2-3.4) .37 .71 Nausea 8 0.8 (0.3-1.4) 1 0.4 (0.0-1.1) .69 .59 Malaise 4 0.4 (0.0-0.8) 1 0.4 (0.0-1.1) .69 .51 Local reaction 45 4.8 (3.4-6.1) 20 7.3 (4.2-10.4) .13 .13 Other reaction 6 0.6 (0.1-1.1) 2 0.7 (0.0-1.7) .57 .89 *Any reaction was defined as replying “yes” to the question “did you experience any fever, illness, or reaction following your vaccination?”. Significant at α = .05. AOR, adjusted odds ratio – adjusted for age and residence (metropolitan/non-metropolitan). CI, confidence interval. Table 1 Adverse events following influenza immunisation reported by pregnant and non-pregnant women – FASTMum, Western Australia, Australia, 19 March-15 May 2014 including spontaneous abortion, in pregnant women who receive TIV [17]. However, the majority of vaccine safety studies in pregnant women, including our own, have been observational in nature [18]. Discussion women have been reported from other settings [10,16]. In the United States, Nordin et al. [16] investigated the incidence of medically-attended events in pregnant women 42 days following TIV vaccination, finding a low frequency of such events and no increased risk of medically-attended events in pregnant women. Screen- ing of the Vaccine Adverse Event Reporting System in the United States has also indicated there are no differ- ences in pregnancy complications or fetal outcomes, women have been reported from other settings [10,16]. In the United States, Nordin et al. [16] investigated the incidence of medically-attended events in pregnant women 42 days following TIV vaccination, finding a low frequency of such events and no increased risk of medically-attended events in pregnant women. Screen- ing of the Vaccine Adverse Event Reporting System in the United States has also indicated there are no differ- ences in pregnancy complications or fetal outcomes, To our knowledge, this is the first study to directly compare the reactogenicity of influenza vaccine in preg- nant women to a sample of non-pregnant women. It is interesting to note the higher incidence of fever ob- served in non-pregnant female HCWs, which may Figure 2 Proportion of pregnant and non-pregnant women reporting an adverse event following trivalent influenza vaccination – FASTMum, Western Australia, Australia, 19 March- 15 May 2014. Figure 2 Proportion of pregnant and non-pregnant women reporting an adverse event following trivalent influenza vaccination – FASTMum, Western Australia, Australia, 19 March- 15 May 2014. Figure 2 Proportion of pregnant and non-pregnant women reporting an adverse event following trivalent influenza vaccination – FASTMum, Western Australia, Australia, 19 March- 15 May 2014. Regan et al. Discussion BMC Pregnancy and Childbirth (2015) 15:61 Page 6 of 7 Table 2 Medical attendance of adverse events following influenza immunisation among pregnant and non-pregnant women – FASTMum, Western Australia, Australia, 19 March-15 May 2014 Table 2 Medical attendance of adverse events following influenza immunisation among pregnant and non-pregnant women – FASTMum, Western Australia, Australia, 19 March-15 May 2014 Pregnant (n = 947) Non-pregnant (n = 275) Fisher’s exact test p-value AOR p-value n Percent (95% CI) n Percent (95% CI) Reaction requiring any treatment 52 5.5 (4.0-6.9) 26 9.5 (6.0-12.9) .02 .06 Reaction requiring telephone advice 11 1.2 (0.5-1.8) 2 0.8 (0.0-2.0) .74 .95 Telephoned a doctor 7 0.7 (0.2-1.3) 0 0.0 (0.0-0.1) .36 .95 Telephoned other 4 0.4 (0.0-0.8) 2 0.8 (0.0-2.0) .62 .78 Reaction requiring medical attention 12 1.3 (0.5-2.0) 1 0.4 (0.0-1.2) .32 .18 Visited a doctor 8 0.8 (0.3-1.4) 1 0.4 (0.0-1.2) .69 .45 Visited a hospital emergency department 4 0.4 (0.0-0.8) 0 (0.0-0.1) .58 .96 A reaction requiring treatment was defined as any reaction where the woman reported self-treating with an antipyretic or pain reliever or visiting a doctor, medical centre or hospital emergency department to seek treatment. AOR, adjusted odds ratio – adjusted for age and residence (metropolitan/non-metropolitan). CI, confidence interval. tendance of adverse events following influenza immunisation among pregnant and non-pregnant m, Western Australia, Australia, 19 March-15 May 2014 **A reaction requiring treatment was defined as any reaction where the woman reported self-treating with an antipyretic or pain reliever or visiting a doctor, medical centre or hospital emergency department to seek treatment. AOR, adjusted odds ratio – adjusted for age and residence (metropolitan/non-metropolitan). CI, confidence interval. controlling for differences in age and residence in the analysis. suggest a protective effect of pregnancy against febrile events. Such an occurrence is not implausible consider- ing previous research has shown pregnancy can have a protective effect against medical conditions, such as breast cancer [19,20] and rheumatoid arthritis, due to hormonal and immunological changes induced by preg- nancy [21]. Alternatively, it is possible these differences were observed due to reporting differences in the groups of women. Non-pregnant female HCWs are likely not a perfect comparison group. Because of their profession and potential knowledge of simple remedies, HCWs are a unique subset of vaccinees with distinctive health- seeking behaviours and perceptions of health. Discussion As a re- sult, it is possible that the incidence of reported fever is more a reflection on the perception of fever, and the threshold for subjective fever may differ between HCWs and other cohorts. Another possible explan- ation for the observation that more HCWs reported fever is that pregnant women may expect or be accus- tomed to fluctuations in symptomatology related to their pregnancy and therefore do not attribute such symptoms to vaccination. However, these explanations are speculative, and additional studies would be re- quired to explore further. In any event, our study found nothing to suggest pregnant women are more likely to report experiencing a reaction to inactivated influenza vaccine, compared to non-pregnant women of similar age. Authors’ contributions AR i i d i h d AR participated in the design of the study, coordinated the recruitment of participants and data collection, and performed all analyses included in the manuscript. LT participated in the development of technologies for data collection, assisted in data collection, and contributed to the study design. CB and PR coordinated the involvement of research nurses and participated in the study design. DB, PE, and GS participated in the study design and advised on the data analysis. CT assisted in data collection and performed all telephone interviews. All authors were directly involved in the drafting of the manuscript and have read and approved the final version. There are several other limitations which should be considered when evaluating these results. First, the reac- tions were self-reported and generally not medically attended, thus they are subject to reporting biases. Sec- ondly, we also identified demographic differences be- tween these groups of women, most likely due to the younger age distribution of pregnant women compared to non-pregnant women and the preferential distribution of intradermal TIV in metropolitan HCW vaccination programs. However, we addressed these differences by Competing interests AR PE LT GS d DM p g AR, PE, LT, GS, and DM have no conflicts of interest to disclose. CB, PR, and CT work for vaccine trials group which has previously received funding from vaccine manufacturers, including GlaxoSmithKline, Novartis, and Sanofi Pasteur. Conclusions Our results indicate that pregnant women experience similar rates of vaccine-associated side effects as non- pregnant women, and these findings can be used to reassure pregnant women who are wary of influenza vac- cination due to concerns about side effects. Continued monitoring of vaccine safety and reactogenicity is an in- tegral component of vaccination campaigns [22]. Rapid, timely and relevant vaccine safety information can be collected using systems such as FASTMum. Integration of such data collection into vaccination programs would facilitate communication of vaccine safety information in a timely manner to pregnant women and antenatal care providers, promoting better informed decision- making regarding antenatal vaccination. 20. Peschken CA, Robinson DB, Hitchon CA, Smolik I, Hart D, Bernstein CN, et al. Pregnancy and the risk of rheumatoid arthritis in a highly predisposed North American native population. J Rheumatol. 2012;39(12):2253–60. 21. Russo IH, Koszalka M, Russo J. Human chorionic gonadotropin and rat mammary cancer prevention. J Natl Cancer Inst. 1990;82:1286–9. 22. Duclos P. A global perspective on vaccine safety. Vaccine. 2004;22(15–16):2059–63. Received: 3 October 2014 Accepted: 4 March 2015 Received: 3 October 2014 Accepted: 4 March 2015 Regan et al. BMC Pregnancy and Childbirth (2015) 15:61 Regan et al. BMC Pregnancy and Childbirth (2015) 15:61 Page 7 of 7 References 1. World Health Organisation. Vaccines against influenza WHO position paper - November 2012. Wkly Epidemiol Rec. 2012;87(47):461–6. 2. Taksdal SE, Mak DB, Joyce S, Tomlin S, Carcione D, Armstrong PK, et al. Predictors of uptake of influenza vaccination – a survey of pregnant women in Western Australia. Aust Fam Physician. 2013;42(8):585–6. 3. Wiley KE, Massey PD, Cooper SC, Wood NJ, Ho J, Quinn HE, et al. Uptake of influenza vaccine by pregnant women: a cross-sectional survey. Med J Aust. 2013;198(7):373–5. 3. Wiley KE, Massey PD, Cooper SC, Wood NJ, Ho J, Quinn HE, et al. Uptake of influenza vaccine by pregnant women: a cross-sectional survey. Med J Aust. 2013;198(7):373–5. 4. Legge A, Dodds L, MacDonald NE, Scott J, McNeil S. Rates and determinants of seasonal influenza vaccination in pregnancy and association with neonatal outcomes. Can Med Assoc J. 2014;186(4):E157–64. 5. Nordin JD, Kharbanda EO, Benitez GV, Lipkind H, Vellozi C, Destefano F. Maternal influenza vaccine and risks for preterm or small for gestational age birth. J Paediatr. 2014;164(5):1051–7. 6. Kharbanda EO, Vazquez-Benitez G, Lipkind H, Naleway A, Lee G, Nordin JD. Inactivated influenza vaccine during pregnancy and risks for adverse obstetric events. Obstet Gynecol. 2013;122(3):659–67. 7. Haberg SE, Trogstad L, Gunnes N, Wilcox AJ, Gjessing HK, Samuelsen SO, et al. Risk of fetal death after pandemic influenza virus infection or vaccination. New Engl J Med. 2013;368(4):333–40. 8. Naleway AL, Irving SA, Henninger ML, Li D-K, Shifflet P, Ball S, et al. Safety of influenza vaccination during pregnancy: a review of subsequent maternal obstetric events and findings from two recent cohort studies. Vaccine. 2014;32:3122–7. 8. Naleway AL, Irving SA, Henninger ML, Li D-K, Shifflet P, Ball S, et al. Safety of influenza vaccination during pregnancy: a review of subsequent maternal obstetric events and findings from two recent cohort studies. Vaccine. 2014;32:3122–7. 9. Carcione D, Blyth CC, Richmond PC, Mak DB, Effler PV. Safety surveillance of influenza vaccine in pregnant women. Aust N Z J Obstet Gynaecol. 2013;53:98–103. 10. Regan AK, Blyth CC, Effler PV. Using SMS technology to verifty the safety of seasonal trivalent influenza vaccine for pregnant women in real time. Med J Aust. 2013;199(11):744–6. 11. Henninger M, Naleway A, Crane B, Donahue J, Irving S. Predictors of seasonal influenza vaccination during pregnancy. Obstet Gynecol. 2013;121(4):741–9. 11. Henninger M, Naleway A, Crane B, Donahue J, Irving S. Predictors of seasonal influenza vaccination during pregnancy. Obstet Gynecol. 2013;121(4):741–9. 12. Author details 1 h l h 1School or Pathology and Laboratory Medicine, University of Western Australia, 227 Stubbs Terrace Shenton Park, Western Australia, Australia. 2Communicable Disease Control Directorate, Western Australia Department of Health, Shenton Park, Western Australia, Australia. 3School of Paediatrics and Child Health, University of Western Australia, Crawley, Western Australia, Australia. 4Vaccine Trials Group, Telethon Kids Institute, University of Western Australia, Subiaco, Western Australia, Australia. Acknowledgements ld l k k We would like to acknowledge the Improvement Foundation for their development of a centralised system for text messaging. We’d also like to thank Sharon Gough for her contribution to the data management of the study. Page 7 of 7 References Yuet Sheung Yuen C, Yee Tak Fong D, Lai Yin Lee I, Chu S, Sau-mei Siu E, Tarrant M. Prevalence and predictors of maternal seasonal influenza vaccination in Hong Kong. Vaccine. 2013;31(45):5281–8. 12. Yuet Sheung Yuen C, Yee Tak Fong D, Lai Yin Lee I, Chu S, Sau-mei Siu E, Tarrant M. Prevalence and predictors of maternal seasonal influenza vaccination in Hong Kong. Vaccine. 2013;31(45):5281–8. 13. Bednarczyk RA, Adjaye-Gbewonyo D, Omer SB. Safety of influenza immunization during pregnancy for the fetus and the neonate. Am J Obstet Gynecol. 2012;207(3):S38–46. 13. Bednarczyk RA, Adjaye-Gbewonyo D, Omer SB. Safety of influenza immunization during pregnancy for the fetus and the neonate. Am J Obstet Gynecol. 2012;207(3):S38–46. 14. Effler PV, Tomlin S, Joyce S, Mak DB. Adult preference for influenza vaccines with lower likelihood of side effects. Hum Vaccin Immunother. 2013;102:456–60. 14. Effler PV, Tomlin S, Joyce S, Mak DB. Adult preference for influenza vaccines with lower likelihood of side effects. Hum Vaccin Immunother. 2013;102:456–60. Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color ﬁgure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit 15. Bartlett MJ, Burgess MA, McIntyre PB, Heath TC. Parent and general practitioner preferences for infant immunisation. Reactogencity or multiple injections? Aust Fam Physician. 1999;28 Suppl 1:S22–7. 15. Bartlett MJ, Burgess MA, McIntyre PB, Heath TC. Parent and general practitioner preferences for infant immunisation. Reactogencity or multiple injections? Aust Fam Physician. 1999;28 Suppl 1:S22–7. References Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color ﬁgure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color ﬁgure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit Submit your next manuscript to BioMed Central and take full advantage of: Submit your next manuscript to BioMed Central and take full advantage of: 16. Nordin JD, Kharbanda EO, Benitez GV, Nichol K, Lipkind H, Naleway A, et al. Maternal safety of trivalent inactivated influenza vaccine in pregnant women. Obstet Gynecol. 2013;121(3):519–25. 16. Nordin JD, Kharbanda EO, Benitez GV, Nichol K, Lipkind H, Naleway A, et al. Maternal safety of trivalent inactivated influenza vaccine in pregnant women. Obstet Gynecol. 2013;121(3):519–25. • Convenient online submission • Thorough peer review 17. Moro PL, Broder K, Zheteyeva Y, Walton K, Rohan P, Sutherland A, et al. Adverse events in pregnant women following administration of trivalent inactivated influenza vaccine and live attenuated influenza vaccine in the Vaccine Adverse Event Reporting system, 1990–2009. Am J Obstet Gynecol. 2011;204(2):146. e141-147. 17. Moro PL, Broder K, Zheteyeva Y, Walton K, Rohan P, Sutherland A, et al. Adverse events in pregnant women following administration of trivalent inactivated influenza vaccine and live attenuated influenza vaccine in the Vaccine Adverse Event Reporting system, 1990–2009. Am J Obstet Gynecol. 2011;204(2):146. e141-147. 18. Fell DB, Dodds L, MacDonald NE, Allen VM, McNeil S. Influenza vaccination and fetal and neonatal outcomes. Expert Rev Vaccines. 2013;12(12):1417–30. 18. Fell DB, Dodds L, MacDonald NE, Allen VM, McNeil S. Influenza vaccination and fetal and neonatal outcomes. Expert Rev Vaccines. 2013;12(12):1417–30. 19. Russo JR, Moral R, Balogh GA, Mailo D, Russo IH. The protective role of pregnancy in breast cancer. Breast Cancer Res. 2005;7:131–42. 19. Russo JR, Moral R, Balogh GA, Mailo D, Russo IH. The protective role of pregnancy in breast cancer. Breast Cancer Res. 2005;7:131–42.
https://openalex.org/W2269545336	https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0148891&type=printable	English	null	Association of PON1, P2Y12 and COX1 with Recurrent Ischemic Events in Patients with Extracranial or Intracranial Stenting	PloS one	2,016	cc-by	8,026	RESEARCH ARTICLE Methods Patients with symptomatic extracranial or intracranial stenosis scheduled for stenting and receiving dual antiplatelets (clopidogrel 75 mg and aspirin 100 mg daily) for at least 5 days before intervention were enrolled. Ischemic events including recurrent transient ischemic attack, stroke, myocardial infarction, and vascular-related mortality within 12 months follow- up were recorded. We examined the influence of genetic polymorphisms on treatment out- come in our patients. Data Availability Statement: All relevant data are within the paper. Abstract Citation: Li X-Q, Ma N, Li X-G, Wang B, Sun S-S, Gao F, et al. (2016) Association of PON1, P2Y12 and COX1 with Recurrent Ischemic Events in Patients with Extracranial or Intracranial Stenting. PLoS ONE 11(2): e0148891. doi:10.1371/journal.pone.0148891 Background and Purpose Short-term combined use of clopidogrel and aspirin improves cerebrovascular outcomes in patients with symptomatic extracranial or intracranial stenosis. Antiplatelet non-responsive- ness is related to recurrent ischemic events, but the culprit genetic variants responsible for the non-responsiveness have not been well studied. We aimed to identify the genetic vari- ants associated with poor clinical outcomes. Editor: Jens Minnerup, University of Münster, GERMANY Editor: Jens Minnerup, University of Münster, GERMANY Received: August 2, 2015 Accepted: January 4, 2016 Published: February 12, 2016 Received: August 2, 2015 Accepted: January 4, 2016 Published: February 12, 2016 Copyright: © 2016 Li et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Association of PON1, P2Y12 and COX1 with Recurrent Ischemic Events in Patients with Extracranial or Intracranial Stenting Xiao-Qing Li1,2☯, Ning Ma1☯, Xin-Gang Li3☯, Bo Wang1, Shu-Sen Sun4, Feng Gao1, Da- Peng Mo1, Li-Gang Song1, Xuan Sun1, Lian Liu1, Xing-Quan Zhao5, Yi-Long Wang5, Yong- Jun Wang5, Zhi-Gang Zhao3, Zhong-Rong Miao1 1 Department of Interventional Neuroradiology, Beijing Tiantan Hospital, Capital Medical University, National Clinical Research Center for Neurological Diseases, Center of Stroke, Beijing Institute for Brain Disorders, Beijing, China, 2 Department of Neurology, Shaanxi Provincial People’s Hospital, Xi’an, China, 3 Department of Pharmacy, Beijing Tiantan Hospital, Capital Medical University, Beijing, China, 4 College of Pharmacy, Western New England University, Springfield, Massachusetts, United States of America, 5 Department of Neurology, Beijing Tiantan Hospital, Capital Medical University, Beijing, China ☯These authors contributed equally to this work. ☯These authors contributed equally to this work. * 1022zzg@sina.com (ZGZ); zhongrongm@163.com (ZRM) * 1022zzg@sina.com (ZGZ); zhongrongm@163.com (ZRM) Data Availability Statement: All relevant data are within the paper. Funding: This study was funded by the National Natural Science Foundation of China (81371290, 81503157), Beijing High-level Personnel Funds (2013-2-19), National Science and Technology Support Program of ‘The 12th Five-Year Plan’ of the Ministry of Science and Technology, Beijing Key Laboratory of Translational Medicine for Cerebrovascular Diseases (2014NXGZ01), Organization Department of Beijing Municipal Committee (2014000021469G258) and Social Development Key Projects of Shannxi Province Introduction Clopidogrel is a prodrug, which requires conversion to an active metabolite by multiple enzymes. The active metabolite acts by inhibiting the ADP receptor P2Y12 on platelet cell membranes [1]. Aspirin is a COX-1 inhibitor, preventing the production of TBXA2, which plays an important role in platelet aggregation. Short- term dual antiplatelet therapy appears to be safe and effective in reducing stroke recurrence and combined vascular events in patients with acute ischemic stroke or TIA as compared with monotherapy [2,3]. Although the clinical benefit of a course of dual antiplatelet therapy for patients undergoing stenting of extracranial or intracranial stenosis is undisputed, inter-individual variability in clo- pidogrel and aspirin response may account for the poor outcome in some patients even after a successful procedure [4–6]. Patients treated with clopidogrel who demonstrate higher in vitro platelet reactivity are at an increased risk of ischemic events [6]. Genetic polymorphism is one of the reasons for clopidogrel and aspirin treatment failure [7,8]. The loss-of-function CYP2C192 variant had been shown to be associated with decreased active metabolites and increased adverse clinical outcomes in patients treated with clopidogrel [9,10]. Despite its robust association with poor outcomes of clopidogrel therapy, CYP2C19 loss-of-function alleles do not account for all of the variability. The clinical use of the CYP2C19 genotype as a prediction tool for personalized antiplatelet therapy remains debatable. Aside from CYP2C19, there are many other related genes in the pathway of aspirin and clo- pidogrel metabolism, such as ABCB1, PON1, CYP2C9, CYP2C18, CES1, P2Y12, COX1 and UCP2. The ABCB1 gene encode transporter plays an important role in the first-pass elimina- tion of orally administered drugs to limit their bioavailability by effluxing them [11,12]. The enzyme encoded by PON1 is an arylesterase that is also involved in the transformation of clopi- dogrel into its active state [13]. A genome-wide association analysis identified 13 SNPs on chromosome 10q24 within the CYP2C18-CYP2C19-CYP2C9-CYP2C8 cluster, showing strong evidence for association with clopidogrel response in an Amish population [14]. In a compet- ing metabolic reaction, about 85% of the clopidogrel is hydrolyzed by CES1 to an inactive metabolite [15]. P2Y12 belongs to the G-protein coupled receptor family. It plays a key role in the ADP-dependent amplification of platelet aggregation induced by other agonists such as TBXA2 and thrombin [16,17], and it is the target of action of clopidogrel. Activation of PLA2 releases AA, which is a precursor for TBXA2 synthesis. Conclusions PON1, P2Y12 and COX1 polymorphisms were associated with poorer vascular outcomes. Testing for these polymorphisms may be valuable in the identification of patients at risk for recurrent ischemic events. Abbreviations: AA, arachidonic acid; ABCB1, ATP- binding cassette sub-family B member 1; ACS, acute coronary syndromes; ADP, adenosine diphosphate; AR, aspirin resistance; BMI, body mass index; CES1, carboxylesterase 1; CI, confidence interval; COX-1, cyclooxygenase-1; COX-2, cyclooxygenase-2; CYP1A2, cytochrome P450 family 1 subfamily A polypeptide 2; CYP2B6, cytochrome P450 family 2 subfamily B polypeptide 6; CYP2C8, cytochrome P450 family 2 subfamily C polypeptide 8; CYP2C9, cytochrome P450 family 2 subfamily C polypeptide 9; CYP2C18, cytochrome P450 family 2 subfamily C polypeptide 18; CYP2C19, cytochrome P450 family 2 subfamily C polypeptide 19; CYP3A4, cytochrome P450 family 3 subfamily A polypeptide 4; CYP3A5, cytochrome P450 family 3 subfamily A polypeptide 5; DSA, digital subtraction angiography; DWI, diffusion weighted imaging; HWE, Hardy-Weinberg equilibrium; LD, linkage disequilibrium; LTA, light transmission aggregometry; MAF, minor allele frequency; MALDI-TOF, matrix-assisted laser desorption ionization time-of-flight; OR, odds ratios; P2Y12, purinergic receptor P2Y, G-protein coupled, 12; PCI, percutaneous coronary intervention; PLA2, phospholipase A2; PON1, paraoxonase 1; SD, standard deviation; SE, standard error; SNPs, single nucleotide polymorphisms; TBXA2, thromboxane A2; TIA, transient ischemic attack; UCP, uncoupling protein 2; UTR, untranslated region. Competing Interests: The authors have declared that no competing interests exist. Competing Interests: The authors have declared that no competing interests exist. Results A total of 268 patients were enrolled into our study and ischemic events were observed in 39 patients. For rs662 of paraoxonase 1 (PON1), allele C was associated with an increased risk of ischemic events (OR = 1.64, 95%CI = 1.03–2.62, P = 0.029). The A-allele carriers of rs2046934 of P2Y12 had a significant association with adverse events (OR = 2.01, 95%CI = 1.10–3.67, P = 0.041). The variant T-allele of cyclooxygenase-1 (COX1) rs1330344 significantly increased the risk of recurrent clinical events (OR = 1.85, 95%CI = 1.12–3.03, 1 / 12 PLOS ONE \| DOI:10.1371/journal.pone.0148891 February 12, 2016 PLOS ONE \| DOI:10.1371/journal.pone.0148891 February 12, 2016 Genetic Variants and Events after Stenting P = 0.017). The other single nucleotide polymorphism (SNP) had no association with ische- mic events. P = 0.017). The other single nucleotide polymorphism (SNP) had no association with ische- mic events. (12015SF023). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. Patient selection This case-control study recruited consecutive ischemic stroke patients who underwent stenting for extracranial or intracranial arterial stenosis in Beijing Tiantan Hospital between May 2013 and September 2013. Patients were selected according to the following criteria: Inclusion crite- ria: 1) diagnosis of ischemic cerebrovascular disease with 70% to 99% stenosis of a major intra- cranial artery (internal carotid artery, M1 segment of middle cerebral artery, vertebral artery or basilar artery) or an extracranial artery (common carotid artery, internal carotid artery, subcla- vian artery, innominate artery or vertebral artery), confirmed by DSA, 2) clopidogrel (75 mg/ day) plus aspirin (100 mg/day) were started at least 5 days before enrollment, 3) informed con- sent available. Exclusion criteria: 1) contraindications to extracranial or intracranial stenting, 2) known allergy or contraindication to aspirin, clopidogrel, heparin, local or general anesthes- tics, 3) active peptic ulcer disease, bleeding tendency, severe liver or kidney impairment, 4) comorbid conditions that may limit survival to less than one year, 5) enrollment in another study that would conflict with the current study. This study was approved by the Institutional Review Board of Beijing Tiantan Hospital, Capital Medical University (Ethics approval num- ber: qx2012-012-01 and KY2014-051-01), and written informed consents were obtained from patients or their close relatives. Introduction COX1 catalyzes the first step in the for- mation of TBXA2 from AA. This reaction is irreversibly blocked by aspirin, which also leads to the blockage of platelet aggregation. The effects of UCP2 polymorphism on platelet reactivity and prognosis in Chinese patients with type 2 diabetes and ischemic stroke was investigated in a study, which showed that the -866G>A polymorphism was associated with clopidogrel resis- tance and platelet reactivity [18]. 2 / 12 PLOS ONE \| DOI:10.1371/journal.pone.0148891 February 12, 2016 Genetic Variants and Events after Stenting Non-responsiveness is related to recurrent ischemic events in patients with extracranial or intracranial occlusive disease on dual antiplatelet therapy, but it is not well studied. We per- formed this study to investigate the relationship between genetic polymorphisms and poor clinical outcomes. PLOS ONE \| DOI:10.1371/journal.pone.0148891 February 12, 2016 Genotyping The most frequent genetic polymorphisms related to aspirin and clopidogrel resistance were selected. The final SNP alleles are listed in Table 1. Genomic DNA was extracted from leuko- cytes in the blood using the EZNA™Blood DNA Midi Kit (Omega Bio-Tek, Norcross, GA, USA). Genotyping was performed by Boao Biotechnology Co., Ltd (Beijing, China) using the MassARRAY system (Sequenom, San Diego, CA, USA) by means of MALDI-TOF mass spec- trometry method according to the manufacturer’s instructions. Single-base extension and PCR primers were designed using the Sequenom Assay Design 3.1 software (Sequenom, San Diego, CA, USA). Genotype calling was performed in real-time with the MassARRAY RT software version 3.0.0.4 and analyzed using the MassARRAY Typer software version 3.4 (Sequenom, San Diego, CA, USA). A repeat analysis of a randomly chosen subgroup of 10% of the cases and controls was conducted for quality control; the reproducibility was 100%. Genetic Variants and Events after Stenting Table 1. Selected variants for aspirin and clopidogrel. Gene Variant Allele SNP position MAF in control HWE P-value ABCB1 rs3213619 A>G 5' UTR 0.044 1.000 ABCB1 rs1128503 A>G Exonic, (Gly412Gly) 0.317 0.264 ABCB1 rs1045642 A>G Exonic, (lle1145lle) 0.383 0.367 CYP2C19 rs12248560 C>T 5' UTR 0.007 1.000 CYP2C19 rs4244285 G>A Exonic, (Pro227Pro) 0.319 0.313 CYP2C19 rs4986893 G>A Exonic, (Trp212null) 0.046 1.000 CYP2C19 rs3758580 C>T Exonic, (Val330Val) 0.103 1.000 CYP2C9 rs4086116 C>T Intronic 0.107 0.141 CYP2C18 rs2104543 C>T Other 0.402 1.000 CYP2C18 rs12772169 C>T Other 0.395 1.000 CYP2C18 rs1998591 G>A 3' UTG 0.432 0.805 CYP2C18 rs1042194 G>T 3' UTG 0.321 0.574 PON1 rs662 T>C Exonic, (Gln192Arg) 0.332 0.105 CES1 rs1968753 A>G Intronic 0.398 0.365 CES1 rs8192950 T>G Intronic 0.191 0.207 P2Y12 rs2046934 G>A Intronic 0.181 0.121 P2Y12 rs6798347 A>G Intronic 0.281 0.298 P2Y12 rs6801273 C>T Intronic 0.411 0.521 P2Y12 rs6787801 A>G Intronic 0.445 0.535 COX1 rs1330344 C>T Promoter 0.367 0.897 COX1 rs10306114 C>G 5' Flanking 0.066 0.610 UCP2 rs659366 C>T Promoter 0.507 0.902 All allele are given on the positive chromosomal strand. doi:10.1371/journal.pone.0148891.t001 Table 1. Selected variants for aspirin and clopidogrel. primary endpoints during a 1-year follow up. The different genotype between the two groups was compared. primary endpoints during a 1-year follow up. The different genotype between the two groups was compared. Study design Demographic and clinical characteristics of the patients, including gender, age, BMI, intracra- nial stent, extracranial stent, smoking, drinking, presence of hypertension, diabetes mellitus or hyperlipidemia were retrieved from the medical records. Clopidogrel 75 mg and aspirin 100 mg were given for at least 5 days before stenting. On the day before the procedure, 5 mL of venous blood from each of the patients was collected in heparin-coated tubes and stored at -70°C for genotyping. The procedure was performed by experienced neurointerventionists, who had each done at least 100 endovascular procedures for intracranial atherosclerotic stenosis. After the procedure, dual antiplatelet therapy consisting of clopidogrel 75 mg and aspirin 100 mg were given for 90 days, followed by single antiplatelet therapy. Other medical interventions were the manage- ment of the atherosclerotic risk factors including elevated systolic blood pressure and low-den- sity lipoprotein levels, diabetes, smoking, obesity, and insufficient exercise. The primary endpoints were TIA, ischemic stroke, myocardial infarction and vascular- related mortality. Ischemic stroke was defined as a new focal neurologic deficit of sudden onset, lasting at least 24 hours, with no hemorrhage on CT or MRI. TIA was defined as a tran- sient episode of neurological dysfunction caused by focal brain or retinal ischemia that lasts for at least 10 minutes but resolves within 24 hours regardless of DWI changes. The occurrence of adverse events was identified on followup visits at 1, 2, 3, 6 and 12 months, or by phone inter- view if the patients could not attend followups. Inpatient hospital readmission records and out- patient clinic records were reviewed. All clinical events were classified and adjudicated by two independent physicians who were blind to the follow-up process. Patients were stratified into the “case group” and the “control group”. The "control group" patients were those with no 3 / 12 Characteristics of enrolled population This study enrolled a total of 268 patients including 39 cases (patients with clinical adverse events) and 229 controls (events-free patients). Their clinical characteristics are listed in Table 2. Compared with controls, patients with events had similar baseline characteristics. Variable Case (n = 39) Control (n = 229) P-value Age, mean±SD 62.90±7.78 62.97±9.14 0.963 Male, n (%) 36 (92.3) 192 (83.8) 0.170 Intracranial stent, n (%) 11 (28.2) 53 (23.1) 0.493 Extracranial stent, n (%) 28 (71.8) 176 (76.9) 0.493 Risk factors BMI, mean±SD 25.30±3.10 24.96±3.60 0.574 Hypertension, n (%) 24 (61.5) 163 (71.2) 0.226 Diabetes, n (%) 12 (30.8) 72 (31.4) 0.933 Hyperlipidemia, n (%) 12 (30.8) 89 (38.9) 0.335 Family history of stroke, n (%) 5 (12.8) 25 (10.9) 0.727 Prior cerebral infarction, n (%) 8 (20.5) 47 (20.5) 0.999 Smoker, n(%) 0.476 Never 12 (30.8) 94 (41.0) - Current 18 (46.2) 88 (38.4) - Ex-smoker 9 (23.1) 47 (20.5) - Drinker, n(%) 0.918 Never 20 (51.3) 125 (54.6) - Social drinker 13 (33.3) 73 (31.9) - Regular drinker 6 (15.4) 31 (13.5) - Outcome, n (%) - Death 5 (12.8) 0 - Ischemic stroke 13 (33.3) 0 - Coronary ischemic event 6 (15.4) 0 - Transient ischemic attack 15 (38.5) 0 - doi:10.1371/journal.pone.0148891.t002 Table 2. Clinical characteristics of patients with events and without events. P-value exact test. The HWE was tested by a χ2 test. The Fisher’s exact test (χ2 test) was used to com- pare the distributions of genotypes between cases and controls. Cochran-Armitage trend test was also used in our data for testing genetic association. OR and 95%CI were applied to evalu- ate the association of genetic variants with the primary outcome events during the follow-up period using unconditional logistic regression. Power analysis was performed using the PS Power and Sample Size Calculation program version 3.1.2. The LD pattern and haplotype structure were measured by the Haploview software 4.2 (Daly Lab, USA). The significance of any haplotypic association was evaluated using χ2 test. A P-value of less than 0.05 was consid- ered statistically significant. Clinical Trial Registration-URL. Unique identifier: NCT01925872. Statistical analysis Data were analyzed using SPSS statistical package version 17.0 (SPSS Inc., Chicago, Illinois, USA) and PLINK v1.07 software. Continuous variables were analyzed using the Student's t- test, and expressed as mean±SD. Categorical data were analyzed using Pearson's χ2 test, and presented as number and percentages or in the case of small expected cell frequencies, Fisher’s 4 / 12 PLOS ONE \| DOI:10.1371/journal.pone.0148891 February 12, 2016 Genetic Variants and Events after Stenting exact test. The HWE was tested by a χ2 test. The Fisher’s exact test (χ2 test) was used to com- pare the distributions of genotypes between cases and controls. Cochran-Armitage trend test was also used in our data for testing genetic association. OR and 95%CI were applied to evalu- ate the association of genetic variants with the primary outcome events during the follow-up period using unconditional logistic regression. Power analysis was performed using the PS Power and Sample Size Calculation program version 3.1.2. The LD pattern and haplotype structure were measured by the Haploview software 4.2 (Daly Lab, USA). The significance of any haplotypic association was evaluated using χ2 test. A P-value of less than 0.05 was consid- ered statistically significant. Clinical Trial Registration-URL. Unique identifier: NCT01925872. Table 2. Clinical characteristics of patients with events and without events. Variable Case (n = 39) Control (n = 229) P-value Age, mean±SD 62.90±7.78 62.97±9.14 0.963 Male, n (%) 36 (92.3) 192 (83.8) 0.170 Intracranial stent, n (%) 11 (28.2) 53 (23.1) 0.493 Extracranial stent, n (%) 28 (71.8) 176 (76.9) 0.493 Risk factors BMI, mean±SD 25.30±3.10 24.96±3.60 0.574 Hypertension, n (%) 24 (61.5) 163 (71.2) 0.226 Diabetes, n (%) 12 (30.8) 72 (31.4) 0.933 Hyperlipidemia, n (%) 12 (30.8) 89 (38.9) 0.335 Family history of stroke, n (%) 5 (12.8) 25 (10.9) 0.727 Prior cerebral infarction, n (%) 8 (20.5) 47 (20.5) 0.999 Smoker, n(%) 0.476 Never 12 (30.8) 94 (41.0) - Current 18 (46.2) 88 (38.4) - Ex-smoker 9 (23.1) 47 (20.5) - Drinker, n(%) 0.918 Never 20 (51.3) 125 (54.6) - Social drinker 13 (33.3) 73 (31.9) - Regular drinker 6 (15.4) 31 (13.5) - Outcome, n (%) - Death 5 (12.8) 0 - Ischemic stroke 13 (33.3) 0 - Coronary ischemic event 6 (15.4) 0 - Transient ischemic attack 15 (38.5) 0 - doi:10.1371/journal.pone.0148891.t002 Table 2. Clinical characteristics of patients with events and without events. There was no significant deviation from HWE for any SNP in all the patients (P > 0.05). Table 3 lists the results of Fisher’s exact test and logistic regression analysis between cases and controls. Compared with controls, the cases had a significant higher mutant frequency of the allele C (PON1 rs662) (OR = 1.64, 95%CI = 1.03–2.62, P = 0.029), allele A (P2Y12 rs2046934) (OR = 2.01, 95%CI = 1.10–3.67, P = 0.041) and allele T (COX1 rs1330344) (OR = 1.85, 95% CI = 1.12–3.03, P = 0.017). The Cochran-Armitage trend test also suported the results of Fish- er’s exact test and logistic regression analysis. The other genotypes were not significantly differ- ent between the two groups (P > 0.05). Statistical power was calculated to verify whether the non-significant results were really due to no relation in the sample or due to a lack of statistical power. The P-value of the power analysis for CYP2C19 rs4986893 was lower than 0.05, and it may be due to the the low MAF (0.046) and small sample size. Gene Variant Fisher test, P-value Cochran-Armitage trend test, P-value Logistic regression, OR, 95%CI Power analysis ABCB1 rs3213619 0.340 0.184 0.27, 0.04–2.10 0.170 ABCB1 rs1128503 0.191 0.177 1.39, 0.86–2.25 0.264 ABCB1 rs1045642 0.803 0.817 1.06, 0.66–1.71 0.057 CYP2C19 rs12248560 0.472 0.559 1.96, 0.20–19.30 0.162 CYP2C19 rs4244285 0.787 0.813 1.06, 0.64–1.78 0.057 CYP2C19 rs4986893 1.000 0.760 0.82, 0.23–2.90 0.047 CYP2C19 rs3758580 0.175 0.180 1.60, 0.80–3.20 0.299 CYP2C9 rs4086116 0.130 0.149 1.59, 0.84–3.01 0.298 CYP2C18 rs2104543 0.172 0.159 1.42, 0.87–2.31 0.301 CYP2C18 rs12772169 0.136 0.124 1.46, 0.90–2.38 0.341 CYP2C18 rs1998591 0.270 0.261 1.32, 0.81–2.15 0.207 CYP2C18 rs1042194 0.694 0.719 1.10, 0.66–1.82 0.068 PON1 rs662 0.029 0.035* 1.64, 1.03–2.62 0.517 CES1 rs1968753 0.900 0.877 1.04, 0.64–1.68 0.053 CES1 rs8192950 0.076 0.059 0.47, 0.21–1.04 0.493 P2Y12 rs2046934 0.041* 0.021* 2.01, 1.10–3.67 0.710 P2Y12 rs6798347 0.219 0.202 1.38, 0.84–2.27 0.248 P2Y12 rs6801273 0.212 0.188 0.71, 0.42–1.19 0.257 P2Y12 rs6787801 0.903 0.885 0.97, 0.60–1.55 0.051 COX1 rs1330344 0.017* 0.014* 1.85, 1.12–3.03 0.709 COX1 rs10306114 0.804 0.660 0.78, 0.26–2.36 0.057 UCP2 rs659366 0.083 0.073 0.64, 0.39–1.05 0.434 P-value < 0.05 doi:10.1371/journal.pone.0148891.t003 Genetic Variants and Events after Stenting PLOS ONE \| DOI:10.1371/journal.pone.0148891 February 12, 2016 6 / 12 Genetic Variants and Events after Stenting Table 3. Summary of SNPs detection and the results of statistical analysis. Gene Variant Fisher test, P-value Cochran-Armitage trend test, P-value Logistic regression, OR, 95%CI Power analysi ABCB1 rs3213619 0.340 0.184 0.27, 0.04–2.10 0.170 ABCB1 rs1128503 0.191 0.177 1.39, 0.86–2.25 0.264 ABCB1 rs1045642 0.803 0.817 1.06, 0.66–1.71 0.057 CYP2C19 rs12248560 0.472 0.559 1.96, 0.20–19.30 0.162 CYP2C19 rs4244285 0.787 0.813 1.06, 0.64–1.78 0.057 CYP2C19 rs4986893 1.000 0.760 0.82, 0.23–2.90 0.047 CYP2C19 rs3758580 0.175 0.180 1.60, 0.80–3.20 0.299 CYP2C9 rs4086116 0.130 0.149 1.59, 0.84–3.01 0.298 CYP2C18 rs2104543 0.172 0.159 1.42, 0.87–2.31 0.301 CYP2C18 rs12772169 0.136 0.124 1.46, 0.90–2.38 0.341 CYP2C18 rs1998591 0.270 0.261 1.32, 0.81–2.15 0.207 CYP2C18 rs1042194 0.694 0.719 1.10, 0.66–1.82 0.068 PON1 rs662 0.029 0.035* 1.64, 1.03–2.62 0.517 CES1 rs1968753 0.900 0.877 1.04, 0.64–1.68 0.053 CES1 rs8192950 0.076 0.059 0.47, 0.21–1.04 0.493 P2Y12 rs2046934 0.041* 0.021* 2.01, 1.10–3.67 0.710 P2Y12 rs6798347 0.219 0.202 1.38, 0.84–2.27 0.248 P2Y12 rs6801273 0.212 0.188 0.71, 0.42–1.19 0.257 P2Y12 rs6787801 0.903 0.885 0.97, 0.60–1.55 0.051 COX1 rs1330344 0.017* 0.014* 1.85, 1.12–3.03 0.709 COX1 rs10306114 0.804 0.660 0.78, 0.26–2.36 0.057 UCP2 rs659366 0.083 0.073 0.64, 0.39–1.05 0.434 P-value < 0.05 doi:10 1371/journal pone 0148891 t003 Table 3. Summary of SNPs detection and the results of statistical analysis. 0.05. Association of end-point and genotype The allelic frequencies and HWE test results are shown in Table 1. Two SNPs of rs4986893 (CYP2C193) and rs12248560 (CYP2C1917) had minor allele frequency (MAF) lower than 5 / 12 PLOS ONE \| DOI:10.1371/journal.pone.0148891 February 12, 2016 0.05. There was no significant deviation from HWE for any SNP in all the patients (P > 0.05). Table 3 lists the results of Fisher’s exact test and logistic regression analysis between cases and controls. Compared with controls, the cases had a significant higher mutant frequency of the allele C (PON1 rs662) (OR = 1.64, 95%CI = 1.03–2.62, P = 0.029), allele A (P2Y12 rs2046934) (OR = 2.01, 95%CI = 1.10–3.67, P = 0.041) and allele T (COX1 rs1330344) (OR = 1.85, 95% CI = 1.12–3.03, P = 0.017). The Cochran-Armitage trend test also suported the results of Fish- er’s exact test and logistic regression analysis. The other genotypes were not significantly differ- ent between the two groups (P > 0.05). Statistical power was calculated to verify whether the non-significant results were really due to no relation in the sample or due to a lack of statistical power. The P-value of the power analysis for CYP2C19 rs4986893 was lower than 0.05, and it may be due to the the low MAF (0.046) and small sample size. Haplotype analysis and association The LD block and haplotype structure were measured by D' among the selected SNPs. The final LD analysis revealed three haplotypes in our patients, and the three blocks were located in CYP2C18, CES1 and P2Y12 respectively (Fig 1. Linkage disequilibrium plot of selected SNP). The significance of any haplotypic association was shown in Table 4. Although the haplotype frequency was low (0.082), the TTTG haplotype was found to increase the risk for recurrent clinical events (P = 0.032). The other haplotypes may not be risk factors for recurrent clinical events for Chinese patients with extracranial or intracranial stenting. Table 3. Summary of SNPs detection and the results of statistical analysis. Haplotype analysis and association The LD block and haplotype structure were measured by D' among the selected SNPs. The final LD analysis revealed three haplotypes in our patients, and the three blocks were located in CYP2C18, CES1 and P2Y12 respectively (Fig 1. Linkage disequilibrium plot of selected SNP). The significance of any haplotypic association was shown in Table 4. Although the haplotype frequency was low (0.082), the TTTG haplotype was found to increase the risk for recurrent clinical events (P = 0.032). The other haplotypes may not be risk factors for recurrent clinical events for Chinese patients with extracranial or intracranial stenting. 6 / 12 PLOS ONE \| DOI:10.1371/journal.pone.0148891 February 12, 2016 Genetic Variants and Events after Stenting Discussion This study evaluated the impact of gene polymorphisms on adverse clinical events in pat with extracranial or intracranial occlusive disease on dual antiplatelet therapy. The result Fig 1. Linkage disequilibrium plot of selected SNP. SNPs in CYP2C19, ABCB1, COX1, P2Y12, CYP2C18 and CES1 genes, respectively. The dep red color presented the computed pair-wise D'. The values in the squares are D'. doi:10.1371/journal.pone.0148891.g001 Genetic Variants and Events after S Linkage disequilibrium plot of selected SNP. SNPs in CYP2C19, ABCB1, COX1, P2Y12, CYP2C18 and CES1 genes, respectively. The depth of or presented the computed pair-wise D'. The values in the squares are D'. Fig 1. Linkage disequilibrium plot of selected SNP. SNPs in CYP2C19, ABCB1, COX1, P2Y12, CYP2C18 and CES1 genes, respectively. The depth of red color presented the computed pair-wise D'. The values in the squares are D'. doi:10.1371/journal.pone.0148891.g001 doi:10.1371/journal.pone.0148891.g001 The COX-1 gene contain 11 exons and most of the polymorphisms in human COX-1 gene are present at low frequency In this study, we selected two variants with higher frequencies and analyzed their impacts on clinical outcomes Our results showed that allele T (rs1330344) had a significant impact on Table 4. The frequency of each haplotype within a block and the association of haplotype and clinical endpoint. Gene Block Haplotype Frequency Case,Control Frequencies P-val CYP2C18 CCCG 0.556 0.500, 0.566 0.280 CYP2C18 TTTT 0.326 0.344, 0.323 0.723 CYP2C18 TTTG 0.082 0.144, 0.071 0.032 CYP2C18 CTCG 0.034 0.013, 0.037 0.269 CES1 AT 0.603 0.596, 0.604 0.895 CES1 AC 0.217 0.290, 0.205 0.092 CES1 CC 0.180 0.114, 0.191 0.101 P2Y12 GG 0.436 0.395, 0.443 0.429 P2Y12 AA 0.283 0.312, 0.278 0.538 P2Y12 AG 0.274 0.252, 0.277 0.648 P-value < 0.05 doi:10.1371/journal.pone.0148891.t004 Table 4. The frequency of each haplotype within a block and the association of haplotype and clinical endpoint. Gene Block Haplotype Frequency Case,Control Frequencies P-value CYP2C18 CCCG 0.556 0.500, 0.566 0.280 CYP2C18 TTTT 0.326 0.344, 0.323 0.723 CYP2C18 TTTG 0.082 0.144, 0.071 0.032* CYP2C18 CTCG 0.034 0.013, 0.037 0.269 CES1 AT 0.603 0.596, 0.604 0.895 CES1 AC 0.217 0.290, 0.205 0.092 CES1 CC 0.180 0.114, 0.191 0.101 P2Y12 GG 0.436 0.395, 0.443 0.429 P2Y12 AA 0.283 0.312, 0.278 0.538 P2Y12 AG 0.274 0.252, 0.277 0.648 P-value < 0.05 . The frequency of each haplotype within a block and the association of haplotype and clinical endpoint. Table 4. The frequency of each haplotype within a block and the association of haplotype and Table 4. The frequency of each haplotype within a blo showed that PON1 rs662 (Q192R), P2Y12 rs2046934 and COX1 rs1330344 polymorphisms were significantly associated with the development of adverse clinical events. Clopidogrel is a prodrug and is activated in the liver. Multiple enzymes, such as CYP1A2, CYP2B6, CYP2C9, CYP2C19 and CYP3A4/5 are involved in the metabolism of clopidogrel. However, the relative importance of the individual enzyme is controversial [19–21]. PON1 gene is located on the long-arm of chromosome 7, and genetic polymorphism has the biggest effect on the PON1 activity level [22]. The coding region PON1 Q192R polymorphism deter- mines a substrate dependent effect on activity. Bouman et al. found that PON1 is also involved in the transformation of clopidogrel into its active state [13]. Discussion This study evaluated the impact of gene polymorphisms on adverse clinical events in patients with extracranial or intracranial occlusive disease on dual antiplatelet therapy. The results 7 / 12 PLOS ONE \| DOI:10.1371/journal.pone.0148891 February 12, 2016 Genetic Variants and Events after Stenting showed that PON1 rs662 (Q192R), P2Y12 rs2046934 and COX1 rs1330344 polymorphisms were significantly associated with the development of adverse clinical events. Clopidogrel is a prodrug and is activated in the liver. Multiple enzymes, such as CYP1A2, CYP2B6, CYP2C9, CYP2C19 and CYP3A4/5 are involved in the metabolism of clopidogrel. However, the relative importance of the individual enzyme is controversial [19–21]. PON1 gene is located on the long-arm of chromosome 7, and genetic polymorphism has the biggest effect on the PON1 activity level [22]. The coding region PON1 Q192R polymorphism deter- mines a substrate dependent effect on activity. Bouman et al. found that PON1 is also involve in the transformation of clopidogrel into its active state [13]. PON1 QQ192 homozygous indi viduals showed a considerably higher risk of stent thrombosis than RR192 homozygous indi- viduals, lower PON1 plasma activity, lower plasma concentrations of active metabolite and lower platelet inhibition. Thus, they identified PON1 as a major determinant of clopidogrel efficacy. Several studies supported these results [23–25]. We found, in addition, that allele C o the PON1 gene was more associated with risk of clinical adverse events than allele T. Howeve this association was not supported by many other studies, some reporting the converse [26– 29]. Other genetic and clinical factors may also influence a patient's response to clopidogrel. P2Y12 is selectively expressed in human platelets and is a target of clopidogrel [30]. Two functional haplotypes (H1 and H2) of P2Y12 have been identified, and these haplotypes can b differentiated by any of the four tagging SNPs in absolute linkage disequilibrium. One of thes SNP (rs2046934) is located in the intron [31]. This SNP showed an association with periphera arterial disease and coronary artery disease [32,33]. However, this result was controversial, an Cuisset et al. found that this variant showed no influence on clopidogrel response in patients with non-ST elevation acute coronary syndrome [34]. In our patients, we found that allele A was more associated with an increased risk of clinical outcome events (TIA, ischemic stroke, myocardial infarction, and death) than allele G. Aspirin directly and irreversibly inhibits the activity of COX-1 and COX-2 to decrease the formation of prostaglandins and thromboxane precursors from AA. PLOS ONE \| DOI:10.1371/journal.pone.0148891 February 12, 2016 Genetic Variants and Events after Stenting was significantly associated with AR determined by LTA and thromboelastography platelet mapping assay using AA as a stimulus [36]. Similar results were also reported using AA- induced LTA combined with ADP-induced LTA to distinguish AR and non-AR [37]. However, there is no sufficient evidence on how the rs1330344 loci leads to AR. The rs1330344 polymor- phism is located in the 5' UTR and may be essential for transcription. We speculate that the TT genotype may up-regulate COX-1 RNA and protein expression. Another possible reason is that rs1330344 is in linkage disequilibrium with other relevant polymorphisms which cause AR [35]. The polymorphism of the CYP2C193 allele (rs4986893) is a G>A transition in exon 4 that results in a premature termination codon at amino acid 212 [38]. CYP2C1917 (rs12248560) is a C>T transition in the promoter region that creates a consensus binding site for the GATA transcription factor family, resulting in increased CYP2C19 expression and activity [39–41]. However, the CYP2C193 allele frequency in our patients is 4.5% and the MAF of rs12248560 is below 1%. Considering the low frequency, the two variants are unlikely significant contribu- tors to drug non-responsiveness in our patients. CYP2C192 (c.681G>A, rs4244285) is a com- mon polymorphism that results in a splicing defect and nonfunctional CYP2C19 protein [42]. CYP2C192 is associated with adverse cardiovascular outcomes in ACS or PCI in patients treated with clopidogrel [43–45]. However, our study did not show the same findings. The rea- son for these conflicting results remains unclear but may be due to several factors including a difference in study design and patient population (patients with symptomatic extracranial or intracranial occlusive disease). In our dataset, males accounted for the majority of both cases and controls, and a separate analysis of males was performed. We got similar results: the polymorphisms of PON1 rs662 (OR = 1.86, 95%CI = 1.12–3.11, P = 0.016), P2Y12 rs2046934 (OR = 1.95, 95%CI = 1.10–3.47, P = 0.021) and COX1 rs1330344 (OR = 1.76, 95%CI = 1.06–2.92, P = 0.028) had significant association with recurrent clinical events. In our enrolled female patients, the data was very limited (only three cases and 37 controls), and the analytical results showed that no genetic ploymorphism was associated with the recurrent ischemic events. Due to the limited data, it was difficult to demonstrate gender differences. PON1 QQ192 homozygous indi- viduals showed a considerably higher risk of stent thrombosis than RR192 homozygous indi- viduals, lower PON1 plasma activity, lower plasma concentrations of active metabolite and lower platelet inhibition. Thus, they identified PON1 as a major determinant of clopidogrel efficacy. Several studies supported these results [23–25]. We found, in addition, that allele C of the PON1 gene was more associated with risk of clinical adverse events than allele T. However, this association was not supported by many other studies, some reporting the converse [26– 29]. Other genetic and clinical factors may also influence a patient's response to clopidogrel. g y p p p g P2Y12 is selectively expressed in human platelets and is a target of clopidogrel [30]. Two functional haplotypes (H1 and H2) of P2Y12 have been identified, and these haplotypes can be differentiated by any of the four tagging SNPs in absolute linkage disequilibrium. One of these SNP (rs2046934) is located in the intron [31]. This SNP showed an association with peripheral arterial disease and coronary artery disease [32,33]. However, this result was controversial, and Cuisset et al. found that this variant showed no influence on clopidogrel response in patients with non-ST elevation acute coronary syndrome [34]. In our patients, we found that allele A was more associated with an increased risk of clinical outcome events (TIA, ischemic stroke, myocardial infarction, and death) than allele G. Aspirin directly and irreversibly inhibits the activity of COX-1 and COX-2 to decrease the formation of prostaglandins and thromboxane precursors from AA. The COX-1 gene contains 11 exons and most of the polymorphisms in human COX-1 gene are present at low frequency. In this study, we selected two variants with higher frequencies and analyzed their impacts on clinical outcomes. Our results showed that allele T (rs1330344) had a significant impact on adverse clinical outcome. Cao et al. reported that in Chinese patients with ischemic stroke treated with aspirin, TT genotype of rs1330344 might increase the risk of subsequent vascular events [35]. Another study by Fan et al. found that the variant T-allele of COX-1 rs1330344 8 / 12 PLOS ONE \| DOI:10.1371/journal.pone.0148891 February 12, 2016 Conclusions In summary, in Chinese patients with extracranial or intracranial occlusive disease, PON1 rs662, P2Y12 rs2046934 and COX1 rs1330344 genetic polymorphisms may increase the risk of subsequent vascular events. None of the other previously reported SNPs with suggested phar- macogenetic influence on antiplatelet therapy efficacy were associated with outcome events during 1 year follow-up in our study. However, this is a relatively small-scale study and the results still need to be verified by a larger study. To investigate the impacts of other drugs (anti- hypertensive, anti-diabetic and anti-hyperlipidemic drugs) taken by patients on clinical events, logistic regression analysis was performed using SPSS software. The other drugs may not have effects on the clinical events (Table 5, P-value > 0.05). Based on the pathway of aspirin and clopidogrel metabolism, PON1 and P2Y12 are the clo- pidogrel-metabolizing enzyme and target, and COX1 is the pharmacological target of aspirin. According to genetic testing, several therapeutic options may be selected. Firstly, an increase in the dose of aspirin or clopidogrel might reduce the rate of poor response. Secondly, aspirin or clopidgrel may be switched to other antiplatelet drugs. Table 5. Multivariate logistic regression analysis of other drugs taken by the patients related to recurrent clinical events. Variables β SE# Wald χ2 P-value OR (95% CI) Anti-hypertensive drugs 0.395 0.374 1.116 0.291 1.485 (0.713–3.091) Anti-diabetic drugs -0.092 0.387 0.057 0.812 0.912 (0.427–1.946) Anti-hyperlipidemic drugs 0.294 0.382 0.594 0.441 1.342 (0.635–2.838) Constant -2.031 0.386 27.716 0.000 0.131 β: regression coefﬁcient #SE: standard error of regression coefﬁcient. doi:10.1371/journal.pone.0148891.t005 OS O \| O / / stic regression analysis of other drugs taken by the patients related to recurrent clinical events. Table 5. Multivariate logistic regression analysis of other drugs taken by the patients related to recurrent cl PLOS ONE \| DOI:10.1371/journal.pone.0148891 February 12, 2016 9 / 12 Genetic Variants and Events after Stenting Several limitations of this study need to be mentioned: 1) the sample size is small with only 268 patients (39 cases) meeting the inclusion and exclusion criteria, 2) not all platelet function tests were performed in the current study, and we will analyze the association of genetic vari- ants and platelet function in future study, and 3) P2Y12 and COX1 gene expression had not been analyzed. Acknowledgments We would like to thank Dr. Lin-Feng Zhang, MD, PhD of the Department of Epidemiology, the Cardiovascular Institute, Fuwai Hospital of the Chinese Academy of Medical Sciences and Peking Union Medical College, and the National Center for Cardiovascular Disease Control and Research, Beijing, China, for his assistance in statistical analysis, and Dr. Wai Ting Joyce Lo, MD, of the Department of Medicine, Queen Elizabeth Hospital, Hong Kong, China, for editing of the article. Author Contributions Conceived and designed the experiments: ZRM ZGZ NM XQZ YJW. Performed the experi- ments: XQL NM XGL BW FG DPM LGS XS LL YLW. Analyzed the data: XGL NM BW. Con- tributed reagents/materials/analysis tools: ZRM ZGZ. Wrote the paper: XGL NM SSS. PLOS ONE \| DOI:10.1371/journal.pone.0148891 February 12, 2016 References 1. Brown DG, Wilkerson EC, Love WE. A review of traditional and novel oral anticoagulant and antiplatelet therapy for dermatologists and dermatologic surgeons. J Am Acad Dermatol. 2015; 72(3):524–534. doi: 10.1016/j.jaad.2014.10.027 PMID: 25486915 2. Wang X, Lin WH, Zhao YD, Chen XY, Leung TW, Chen C, et al. The effectiveness of dual antiplatelet treatment in acute ischemic stroke patients with intracranial arterial stenosis: a subgroup analysis of CLAIR study. Int J Stroke. 2013; 8(8):663–668. doi: 10.1111/j.1747-4949.2012.00828.x PMID: 22883712 3. Geeganage CM, Diener HC, Algra A, Chen C, Topol EJ, Dengler R, et al. Dual or mono antiplatelet ther- apy for patients with acute ischemic stroke or transient ischemic attack: systematic review and meta- analysis of randomized controlled trials. Stroke. 2012; 43(4):1058–1066. doi: 10.1161/STROKEAHA. 111.637686 PMID: 22282894 4. Angiolillo DJ, Fernandez-Ortiz A, Bernardo E, Alfonso F, Macaya C, Bass TA, et al. Variability in individ- ual responsiveness to clopidogrel: clinical implications, management, and future perspectives. J Am Coll Cardiol. 2007; 49(14):1505–1516. doi: 10.1016/j.jacc.2006.11.044 PMID: 17418288 5. Gurbel PA, Bliden KP, Guyer K, Cho PW, Zaman KA, Kreutz RP, et al. Platelet reactivity in patients and recurrent events post-stenting: results of the PREPARE POST-STENTING Study. J Am Coll Cardiol. 2005; 46(10):1820–1826. doi: 10.1016/j.jacc.2005.07.041 PMID: 16286165 6. Bliden KP, DiChiara J, Tantry US, Bassi AK, Chaganti SK, Gurbel PA. Increased risk in patients with high platelet aggregation receiving chronic clopidogrel therapy undergoing percutaneous coronary intervention: is the current antiplatelet therapy adequate? J Am Coll Cardiol. 2007; 49(6):657–666. doi: 10.1016/j.jacc.2006.10.050 PMID: 17291930 7. Frelinger AL 3rd, Bhatt DL, Lee RD, Mulford DJ, Wu J, Nudurupati S, et al. Clopidogrel pharmacokinet- ics and pharmacodynamics vary widely despite exclusion or control of polymorphisms (CYP2C19, ABCB1, PON1), noncompliance, diet, smoking, co-medications (including proton pump inhibitors), and PLOS ONE \| DOI:10.1371/journal.pone.0148891 February 12, 2016 10 / 12 Genetic Variants and Events after Stenting pre-existent variability in platelet function. J Am Coll Cardiol. 2013; 61(8):872–879. doi: 10.1016/j.jac 2012.11.040 PMID: 23333143 pre-existent variability in platelet function. J Am Coll Cardiol. 2013; 61(8):872–879. doi: 10.1016/j.jacc. 2012.11.040 PMID: 23333143 8. Angiolillo DJ. Variability in responsiveness to oral antiplatelet therapy. Am J Cardiol. 2009; 103(3 Sup- pl):27A–34A. doi: 10.1016/j.amjcard.2008.11.020 PMID: 19166710 9. Mega JL, Close SL, Wiviott SD, Shen L, Hockett RD, Brandt JT, et al. Cytochrome p-450 polymor- phisms and response to clopidogrel. N Engl J Med. 2009; 360(4):354–362. doi: 10.1056/ NEJMoa0809171 PMID: 19106084 10. References Ingelman-Sundberg M, Sim SC, Gomez A, Rodriguez-Antona C. Influence of cytochrome P450 poly- morphisms on drug therapies: pharmacogenetic, pharmacoepigenetic and clinical aspects. Pharmacol Ther. 2007; 116(3):496–526. doi: 10.1016/j.pharmthera.2007.09.004 PMID: 18001838 11. Mooij MG, Nies AT, Knibbe CA, Schaeffeler E, Tibboel D, Schwab M, et al. Development of Human Membrane Transporters: Drug Disposition and Pharmacogenetics. Clin Pharmacokinet. 2015. doi: 10. 1007/s40262-015-0328-5 PMID: 26410689 12. Zhu P, Zhu Q, Zhang Y, Ma X, Li Z, Li J, et al. ABCB1 variation and treatment response in AIDS patients: initial results of the Henan cohort. PLoS One. 2013; 8(1):e55197. doi: 10.1371/journal.pone. 0055197 PMID: 23372834 13. Bouman HJ, Schomig E, van Werkum JW, Velder J, Hackeng CM, Hirschhauser C, et al. Paraoxo- nase-1 is a major determinant of clopidogrel efficacy. Nat Med. 2011; 17(1):110–116. doi: 10.1038/nm. 2281 PMID: 21170047 14. Shuldiner AR, O'Connell JR, Bliden KP, Gandhi A, Ryan K, Horenstein RB, et al. Association of cyto- chrome P450 2C19 genotype with the antiplatelet effect and clinical efficacy of clopidogrel therapy. JAMA. 2009; 302(8):849–857. doi: 10.1001/jama.2009.1232 PMID: 19706858 15. Lins R, Broekhuysen J, Necciari J, Deroubaix X. Pharmacokinetic profile of 14C-labeled clopidogrel. Semin Thromb Hemost. 1999; 25 Suppl 2:29–33. PMID: 10440420 16. Storey RF. Biology and pharmacology of the platelet P2Y12 receptor. Curr Pharm Des. 2006; 12 (10):1255–1259. PMID: 16611109 17. Haynes SE, Hollopeter G, Yang G, Kurpius D, Dailey ME, Gan WB, et al. The P2Y12 receptor regulates microglial activation by extracellular nucleotides. Nat Neurosci. 2006; 9(12):1512–1519. doi: 10.1038/ nn1805 PMID: 17115040 18. Chai Y, Gu B, Qiu JR, Yi HG, Zhu Q, Zhang L, et al. Effects of uncoupling protein 2 -866G/A polymor- phism on platelet reactivity and prognosis in Chinese patients with type 2 diabetes and ischemic stroke. Int J Neurosci. 2013; 123(11):752–758. doi: 10.3109/00207454.2013.798733 PMID: 23621569 19. Savi P, Pereillo JM, Uzabiaga MF, Combalbert J, Picard C, Maffrand JP, et al. Identification and biologi- cal activity of the active metabolite of clopidogrel. Thromb Haemost. 2000; 84(5):891–896. PMID: 11127873 20. Clarke TA, Waskell LA. The metabolism of clopidogrel is catalyzed by human cytochrome P450 3A and is inhibited by atorvastatin. Drug Metab Dispos. 2003; 31(1):53–59. PMID: 12485953 21. Farid NA, Payne CD, Small DS, Winters KJ, Ernest CS 2nd, Brandt JT, et al. Cytochrome P450 3A inhi- bition by ketoconazole affects prasugrel and clopidogrel pharmacokinetics and pharmacodynamics dif- ferently. Clin Pharmacol Ther. 2007; 81(5):735–741. doi: 10.1038/sj.clpt.6100139 PMID: 17361128 22. PLOS ONE \| DOI:10.1371/journal.pone.0148891 February 12, 2016 References Adenosine diphosphate- induced platelet aggregation is associated with P2Y12 gene sequence variations in healthy subjects. Circulation. 2003; 108(8):989–995. doi: 10.1161/01.CIR.0000085073.69189.88 PMID: 12912815 32. Fontana P, Gaussem P, Aiach M, Fiessinger JN, Emmerich J, Reny JL. P2Y12 H2 haplotype is associ- ated with peripheral arterial disease: a case-control study. Circulation. 2003; 108(24):2971–2973. doi: 10.1161/01.CIR.0000106904.80795.35 PMID: 14662702 33. Cavallari U, Trabetti E, Malerba G, Biscuola M, Girelli D, Olivieri O, et al. Gene sequence variations of the platelet P2Y12 receptor are associated with coronary artery disease. BMC Med Genet. 2007; 8:59. doi: 10.1186/1471-2350-8-59 PMID: 17803810 34. Cuisset T, Frere C, Quilici J, Morange PE, Saut N, Lambert M, et al. Role of the T744C polymorphism of the P2Y12 gene on platelet response to a 600-mg loading dose of clopidogrel in 597 patients with non-ST-segment elevation acute coronary syndrome. Thromb Res. 2007; 120(6):893–899. doi: 10. 1016/j.thromres.2007.01.012 PMID: 17337040 35. Cao L, Zhang Z, Sun W, Bai W, Sun W, Zhang Y, et al. Impacts of COX-1 gene polymorphisms on vas- cular outcomes in patients with ischemic stroke and treated with aspirin. Gene. 2014; 546(2):172–176. doi: 10.1016/j.gene.2014.06.023 PMID: 24930730 36. Fan L, Cao J, Liu L, Li X, Hu G, Hu Y, et al. Frequency, risk factors, prognosis, and genetic polymor- phism of the cyclooxygenase-1 gene for aspirin resistance in elderly Chinese patients with cardiovas- cular disease. Gerontology. 2013; 59(2):122–131. doi: 10.1159/000342489 PMID: 23038044 37. Li XL, Cao J, Fan L, Wang Q, Ye L, Cui CP, et al. Genetic polymorphisms of HO-1 and COX-1 are asso- ciated with aspirin resistance defined by light transmittance aggregation in Chinese Han patients. Clin Appl Thromb Hemost. 2013; 19(5):513–521. doi: 10.1177/1076029612444002 PMID: 22609818 38. De Morais SM, Wilkinson GR, Blaisdell J, Meyer UA, Nakamura K, Goldstein JA. Identification of a new genetic defect responsible for the polymorphism of (S)-mephenytoin metabolism in Japanese. Mol Pharmacol. 1994; 46(4):594–598. PMID: 7969038 39. Sim SC, Risinger C, Dahl ML, Aklillu E, Christensen M, Bertilsson L, et al. A common novel CYP2C19 gene variant causes ultrarapid drug metabolism relevant for the drug response to proton pump inhibi- tors and antidepressants. Clin Pharmacol Ther. 2006; 79(1):103–113. doi: 10.1016/j.clpt.2005.10.002 PMID: 16413245 40. Rudberg I, Mohebi B, Hermann M, Refsum H, Molden E. Impact of the ultrarapid CYP2C1917 allele on serum concentration of escitalopram in psychiatric patients. Clin Pharmacol Ther. 2008; 83(2):322– 327. doi: 10.1038/sj.clpt.6100291 PMID: 17625515 41. References Mackness B, Durrington PN, Mackness MI. Human serum paraoxonase. Gen Pharmacol. 1998; 31 (3):329–336. PMID: 9703197 23. Campo G, Ferraresi P, Marchesini J, Bernardi F, Valgimigli M. Relationship between paraoxonase Q192R gene polymorphism and on-clopidogrel platelet reactivity over time in patients treated with per- cutaneous coronary intervention. J Thromb Haemost. 2011; 9(10):2106–2108. doi: 10.1111/j.1538- 7836.2011.04457.x PMID: 21819538 24. Wu H, Qian J, Xu J, Sun A, Sun W, Wang Q, et al. Besides CYP2C19, PON1 genetic variant influences post-clopidogrel platelet reactivity in Chinese patients. Int J Cardiol. 2013; 165(1):204–206. doi: 10. 1016/j.ijcard.2012.08.017 PMID: 22974728 25. Park KW, Park JJ, Kang J, Jeon KH, Kang SH, Han JK, et al. Paraoxonase 1 gene polymorphism does not affect clopidogrel response variability but is associated with clinical outcome after PCI. PLoS One. 2013; 8(2):e52779. doi: 10.1371/journal.pone.0052779 PMID: 23418418 26. Xie C, Ding X, Gao J, Wang H, Hang Y, Zhang H, et al. The effects of CES1A2 A(-816)C and CYP2C19 loss-of-function polymorphisms on clopidogrel response variability among Chinese patients with coro- nary heart disease. Pharmacogenet Genomics. 2014; 24(4):204–210. doi: 10.1097/FPC. 0000000000000035 PMID: 24535487 27. Tresukosol D, Suktitipat B, Hunnangkul S, Kamkaew R, Poldee S, Tassaneetrithep B, et al. Effects of cytochrome P450 2C19 and paraoxonase 1 polymorphisms on antiplatelet response to clopidogrel 11 / 12 PLOS ONE \| DOI:10.1371/journal.pone.0148891 February 12, 2016 Genetic Variants and Events after Stenting therapy in patients with coronary artery disease. PLoS One. 2014; 9(10):e110188. doi: 10.1371/journal. pone.0110188 PMID: 25329996 therapy in patients with coronary artery disease. PLoS One. 2014; 9(10):e110188. doi: 10.1371/journal. pone.0110188 PMID: 25329996 28. 28. Viviani Anselmi C, Briguori C, Roncarati R, Papa L, Visconti G, Focaccio A, et al. Routine assessment of on-clopidogrel platelet reactivity and gene polymorphisms in predicting clinical outcome following drug-eluting stent implantation in patients with stable coronary artery disease. JACC Cardiovasc Interv. 2013; 6(11):1166–1175. doi: 10.1016/j.jcin.2013.06.010 PMID: 24262617 29. Zhang L, Chen Y, Jin Y, Qu F, Li J, Ma C, et al. Genetic determinants of high on-treatment platelet reac- tivity in clopidogrel treated Chinese patients. Thromb Res. 2013; 132(1):81–87. doi: 10.1016/j. thromres.2013.05.006 PMID: 23726091 30. Hollopeter G, Jantzen HM, Vincent D, Li G, England L, Ramakrishnan V, et al. Identification of the plate- let ADP receptor targeted by antithrombotic drugs. Nature. 2001; 409(6817):202–207. doi: 10.1038/ 35051599 PMID: 11196645 31. Fontana P, Dupont A, Gandrille S, Bachelot-Loza C, Reny JL, Aiach M, et al. PLOS ONE \| DOI:10.1371/journal.pone.0148891 February 12, 2016 References Sibbing D, Koch W, Gebhard D, Schuster T, Braun S, Stegherr J, et al. Cytochrome 2C1917 allelic variant, platelet aggregation, bleeding events, and stent thrombosis in clopidogrel-treated patients with coronary stent placement. Circulation. 2010; 121(4):512–518. doi: 10.1161/CIRCULATIONAHA.109. 885194 PMID: 20083681 42. de Morais SM, Wilkinson GR, Blaisdell J, Nakamura K, Meyer UA, Goldstein JA. The major genetic defect responsible for the polymorphism of S-mephenytoin metabolism in humans. J Biol Chem. 1994; 269(22):15419–15422. PMID: 8195181 43. Oh IY, Park KW, Kang SH, Park JJ, Na SH, Kang HJ, et al. Association of cytochrome P450 2C192 polymorphism with clopidogrel response variability and cardiovascular events in Koreans treated with drug-eluting stents. Heart. 2012; 98(2):139–144. doi: 10.1136/hrt.2011.227272 PMID: 21700758 44. Delaney JT, Ramirez AH, Bowton E, Pulley JM, Basford MA, Schildcrout JS, et al. Predicting clopido- grel response using DNA samples linked to an electronic health record. Clin Pharmacol Ther. 2012; 91 (2):257–263. doi: 10.1038/clpt.2011.221 PMID: 22190063 45. Wu H, Qian J, Xu J, Sun A, Sun W, Wang Q, et al. Effects of CYP2C19 variant alleles on postclopido- grel platelet reactivity and clinical outcomes in an actual clinical setting in China. Pharmacogenet Geno- mics. 2012; 22(12):887–890. doi: 10.1097/FPC.0b013e328359253a PMID: 22990067 12 / 12 PLOS ONE \| DOI:10.1371/journal.pone.0148891 February 12, 2016
W2313133738.txt	https://zenodo.org/records/2508005/files/article.pdf	de		Dermatologisches Zentralblatt	Archives of dermatological research	1,916	public-domain	423	,668 Fachzeitschrif~;en. Dermatologisches Zentralblatt. Bd. X[X. Nr. 3--4. des Arztes zur Verweig e r u n g d e s Z e u g n i s s e s u n d d e s G u t a c h t e n s . p. 50 und p. 68. Wenn das Strafgesetzbuch einerseits dea unbefugten Geheimnisbruch des hrztes mit Strafe bedroht, andererseits die Strafprozel~ordnung die Arzte zur Verweigerung des Zeughisses hinsichtlich dessen, was ihnen bei Ausiibung ihres Berufes anvertraut ist, berechtigt, so ist die Ablegung oder Weigerung des Zeugnisses vor Gericht in des Arztes freies Ermessen gestellt. Aus Griinden einer effolgreichen Gesundheitspfiege hat der Staat das hSchste Iateresse daran, dal~ der Kranke sieh in vollstem Vertrauen auf die eventuell notwendige Geheimhaltung seiner Mitteilungen an seinen Arzt wenden kann. Das Schutzbediirfnis des Kranken tritt besonders dort hervor, wo es sich um sog. geheime Krankheiten handelt, die in den hugen der Mitwelt dem Kranken zum Nachteil oder zur Unehre gereichen. F i n g e r schl~igt vor, um eine allzu h~ufige Ver weigerung zum Schaden der Rechtspfiege zu verhindern, dutch Gesetzesnormeu daffir zu sorgen, da[~ das Geheimnis nur in einem engeu Kreis bekannt werde. Angesiehts der Verweigerung des Gutachtens seitens eines Arztes, der slch aut seine Pfiicht zur Verschwiegenheit berufen hatte, erkl~rte das Reichsgericht die Weigerung fiir berechtigt. Zur Abgabe mediziniseher Gutachten sind die ~4.rzte wohl verpfliehtet: doch berechtigen dieselben Griinde, welche einen Zeugen berechtigen, das Zeugnis zu verweigern, den zum Sachverst~ndigen Ernannten zur Verweigerung des Gutachtens. Eine Wegnahme des Krankenjournals und Einsicht in dasselbe ist ungesetzlich, da die Strafprozeflordnung gegeniiber den zur Verweigerung des Zeugnisses Berechtigten diese Zwangsmittel ausdriicklich ausschlieBt. Brohl. E i n R h i n o p h y m a y o n s e l t e n e r h r t . p. 66. Bei einem 71j~ihrigen Kapit~n bildete sich eine kartoffelhhnliche Geschwulst der Nase, die his fiber den Mund herabhing. Der histologische Befund des anstandslos operierten Tumors ergab gef~l~fiihrendes Bindegewebe mit hyperplastisehen Talgdrtisen, sowie vereinzelte, mit Plattenepithel ausgekleidete Zysten, die mit atheromghnlichen Massen ausgefiillt waren. Hugo F a s a 1 (Wien). Lieske, H. V o m R e c h t e Zeitschrift fiir Sexualwissenschaft. 7.--9. Heft. Wilhelm, E. D i e I m p o t e n t i a coeundi und gener a n d i d e s W e i b e s in i h r e n B e z i e h u n g e n zur Eheanfechtung und Eheseheidung. p. 225.
https://openalex.org/W4206492695	https://europepmc.org/articles/pmc2778169?pdf=render	English	null	Aspirin Resistance	Advances in hematology	2,009	cc-by	8,056	Khaled Mansour,1 Ali T. Taher,2 Khaled M. Musallam,2 and Samir Alam1 1Division of Cardiology, Department of Internal Medicine, American University of Beirut Medical Center, Beirut 1107 2020, Lebanon 2Division of Hematology-Oncology, Department of Internal Medicine, American University of Beirut Medical Center, Beirut 1107 2020, Lebanon Correspondence should be addressed to Samir Alam, salam@aub.edu.lb Received 24 October 2008; Revised 9 February 2009; Accepted 15 February 2009 Recommended by Charles S. Greenberg Recommended by Charles S. Greenberg The development of adverse cardiovascular events despite aspirin use has established an interest in a possible resistance to the drug. Several deﬁnitions have been set and various laboratory testing modalities are available. This has led to a wide range of prevalence reports in diﬀerent clinical entities. The etiologic mechanism has been related to clinical, genetic, and other miscellaneous factors. The clinical implications of this phenomenon are signiﬁcant and warrant concern. Management strategies are currently limited to dosing alteration and introduction of other anitplatelet agents. However, these measures have not met the expected eﬃcacy or safety. Copyright © 2009 Khaled Mansour et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Hindawi Publishing Corporation Advances in Hematology Volume 2009, Article ID 937352, 10 pages doi:10.1155/2009/937352 Hindawi Publishing Corporation Advances in Hematology Volume 2009, Article ID 937352, 10 pages doi:10.1155/2009/937352 1. Introduction In this paper, the prevalence, mechanism, and clini- cal implications of aspirin resistance will be highlighted. Moreover, the available laboratory tests used to assess this phenomenon and the possible ways to overcome it will be described. Despite the development of newer antiplatelet drugs in the last decade, aspirin is still the most widely used antiplatelet agent across the world to prevent cardiovascular diseases [1– 4]. In the 19th century, willow leaves became an attraction after their extract, salicylic acid, was found eﬀective as an analgesic for arthralgias and an antirheumatic for a variety of rheumatic disease [5–7]. At the end of the 19th century the acetylated form of salicylic acid was manufactured [8], with less gastrointestinal side eﬀects, and consequently became more widespread and commonly used. Long-term aspirin administration in patients at high risk of occlusive vascular events reduced up to 34% of nonfatal myocardial infarction (MI), 25% of nonfatal stroke, and 18% of all-cause mortality [4]. Ever since, several patients have reported developing adverse vascular events despite aspirin intake, an observation that was later coined the term “aspirin resistance” (AR) [9]. Nowadays, the term has been employed to express the occurrence of cardiovascular events in spite of regular intake of aspirin at recommended doses [10–13]. Recent advances in evaluating platelet function and the introduction of the point-of-care platelet function machinery made assessing the degree of platelet response to a certain antiplatelet drug more reasonable, accessible, and easier to perform [14]. 2. Terminology The lack of agreement on a standardized deﬁnition for “aspirin resistance” has contributed to the disparity in reports of its incidence among diﬀerent studies. Whereas some use the term “aspirin treatment failure,” while others like to call it “aspirin nonresponsiveness.” The term “resis- tance” was assigned based on biochemical and laboratory ﬁndings in which aspirin was unable to inhibit one of the many available in vitro tests of platelet function [15, 16]. Hence, from a pharmacologic point of view, resistance to aspirin may be deﬁned as lack of ability to attain the expected inhibition of platelet cyclooxygenase-(COX-)1 with avoidance of platelet thromboxane (TX) A2 formation [17]. “Aspirin treatment failure” is deﬁned based on clinical outcomes, when aspirin fails to prevent recurrent vascular ischemic events. However, reinfarction after aspirin use in the setting of an acute coronary event may be due to thrombus spread mediated by adenosine diphosphate (ADP) rather 2 2 Advances in Hematology Table 1: Prevalence of aspirin resistance. Reference Patients Test used Prevalence of AR Comments N(%) Christiaens et al. [32] N = 97 PFA-100 analyzer 29 (29.9) ♀> ♂(38 versus 15%) Stable CAD patients already on aspirin No clinical correlation with laboratory parameters after 2.5 years follow-up Pamukcu et al. [33] N = 234 PFA-100 analyzer 52 (22.2) Similar risk in resistant and nonresistant patients after 20.6 ± 6.9 months follow-up. Risk in aspirin resistant patients increased after cessation of clopidogrel Stable CAD Pamukcu et al. [34] N = 105 PFA-100 analyzer 20 (19) Greater risk of MACE in patients resistant to aspirin ACS Akay et al. [35] N = 280 Optical platelet aggregometry 77 (27.5) Large trial evaluating the frequency of AR in healthy subjects Healthy Turkish volunteers (ADP, AA) Lee et al. [36] N = 468 VerifyNow-Aspirin 128 (27.4) 100 mg or less daily dose were associated with a higher incidence of AR in patients with CAD Stable CAD Harrison et al. [30] N = 100 PFA-100 22 (22) Poor agreement between the diﬀerent tests leads to the conclusion that aspirin resistance is highly test-speciﬁc Patients after TIA or Stroke VerifyNow-Aspirin 17 (17) Optical platelet aggregometry 5 (5) Gum et al. 2. Terminology [28] N = 325 Optical platelet aggregometry 18 (5.5) Trend toward increased age in patients with AR Stable CAD PFA-100 analyzer 31 (9.5) AR: aspirin resistance; CAD: coronary artery disease; ACS: acute coronary syndrome; MACE: major adverse cardicac events; ADP: adenosine diphosphate; AA: arachidonic acid; TIA: transient ischemic attack. one laboratory test (5 by PFA-100, 1 by aggregometry, and urinary 11dhTxB2), which leads to the conclusion that, as with adults, the incidence of AR is also assay-dependent in the pediatric population [31]. Table 1 summarizes the main studies investigating AR prevalence in diﬀerent clinical entities [28, 30, 32–36]. than continuing TX-induced platelet aggregation; which may render the terminology “failure” improper [18]. Some suggested that until the various possible reasons of treatment failure with aspirin have been recognized, the more suitable term may be “aspirin nonresponsiveness” [15]. 4. Proposed Etiology To stress the importance of this factor, a recent meta-analysis including 50 000 patients at high risk of ischemic coronary disease found that noncompliance or early discontinuation of the drug carried a 3 times higher risk of cardiac events (odds ratio [OR] 3.14, 95% conﬁdence interval [CI] 1.75– 5.61; P = .0001) [43]. Thus, explaining the beneﬁts of antiplatelet therapy to the patient may help improving compliance [44]. Poor glucose control and body weight are also proposed to contribute to AR, where in a recent study assessing 48 patients with type 2 diabetes mellitus using the PFA-100 assay, AR was signiﬁcantly associated with HbA1c ≥8% (P = .002) and obesity (BMI ≥30 kg/m2; P = .01) [54]. Although this might implicate that better glucose control leads to less incidence of AR, the clinical signiﬁcance of such ﬁndings should be carefully inspected, since in two of the largest trials [55, 56] assessing the role of aspirin on primary prevention of cardiovascular events in patients with type2 diabetes, low- dose aspirin did not decrease the risk of cardiovascular events when compared to placebo (13.6 per 1000 person-year in the aspirin group versus 17.0 per 1000 person-year in the placebo group, hazard ratio [HR] 0.80; 95% CI 0.58–1.10 in the JPAD trial [55]; 116 of 638 primary events in the aspirin group compared with 117 of 638 in the no aspirin group, HR 0.98; 95% CI 0.76–1.26 in the POPADAD trial [56]). In the JPAD trial both groups (aspirin versus no aspirin) had similar baseline characteristics in terms of glycosylated hemoglobin (7.0 versus 7.1, resp.). On the other hand, the aspirin group in the POPADAD trial had a mean HbA1c of 8.0 compared to 7.9 in the no aspirin group. The lack of beneﬁcial eﬀect of aspirin in the latter study may be partly explained by the fact that AR is signiﬁcantly associated with HbA1c levels ≥8% [54]. Some drugs may compete with aspirin at the COX-1 receptor site; of those, the most commonly encountered are nonsteroidal anti-inﬂammatory drugs such as ibuprofen which can oﬀset the clinical beneﬁt of aspirin in a variety of vascular diseases [45]. The bioavailability of a drug is dependent on numerous factors, absorption being the most relevant. 4. Proposed Etiology Due to the lack of standardized testing for AR, prevalence rates of nonresponders to aspirin among adults diﬀer according to the platelet function test used and the threshold of response, with a wide range reported (5.5 to 60%) [19– 22]. For instance, there are up to seven diﬀerent thresholds for deﬁning aspirin response using the PFA-100 [23–29]. When using the combination of diﬀerent laboratory tests to deﬁne resistance (VerifyNow-Aspirin, optical aggregometry, and PFA-100), a lower resistance rate (2%) was reported as compared to using each test alone [30]. 4.1. Pharmacology of Aspirin. Low-dose aspirin (as low as 81 mg) irreversibly inhibits the COX-1 enzyme, by acetylat- ing the serine residue at position 529, consequently impair- ing the transformation of arachidonic acid to prostaglandin (G2/H2), and TX A2, which is a potent mediator of platelet aggregation and activation. This eﬀect explains the clinical beneﬁt of aspirin in patients with high risk vascular disease [37–39]. Aspirin’s eﬀect on COX-2 is minimal in doses <1200 mg [40, 41]. The majority of studies on AR were conducted on adult patients, but recently the prevalence of AR was studied in 44 children aged 1 to 17 years taking aspirin for diﬀerent indications, by using diﬀerent laboratory tests. Six out of 44 were considered aspirin resistant according to at least 4.2. Mechanism of Resistance. The diﬀerent mechanisms by which AR might take place are not yet well understood [10, 12, 13, 19]. Medication compliance is one preventable and important contributor to the phenomenon of resistance 3 Advances in Hematology 3 Clinical factors: 1. Diabetes 2. Heart failure 3. Acute coronary syndrome 4. Infection/inflammation 5. Obesity 6. CABG Aspirin resistance Genetic factors: 1. GPIIIa: PIA1/A2 polymorphism 2. COX-1 gene mutation 3. Overexpression of COX-2 Miscellaneous: 1. Drug-drug interaction: NSAIDS/PPI . 2. Poor compliance 3. Tachyphylaxis 4. Alternative pathways other than COX-1 Figure 1: Proposed factors contributing to aspirin resistance (CABG: coronary artery bypass grafting; COX: cyclooxygenase; NSAIDS: nonsteroidal anti-inﬂamatory drugs; PPI: proton pump inhibitors). Figure 1: Proposed factors contributing to aspirin resistance (CABG: coronary artery bypass grafting; COX: cyclooxygenase; NSAIDS: nonsteroidal anti-inﬂamatory drugs; PPI: proton pump inhibitors). P = .009), but combining data from both healthy individuals and those with cardiovascular disease reduced the size of the observed eﬀect (OR 1.14, 95% CI 0.84–1.54; P = .40) [50]. and to the overreporting of aspirin nonresponsiveness [42]. 6. Clinical Implication Another signiﬁcant question to consider is whether this phenomenon is conﬁned to laboratory ﬁndings or aﬀects the expected clinical outcome. A recent meta-analysis on 2930 patients with cardiovascular disease, who were on aspirin (75–325 mg daily) alone or in combination with other antiplatelet therapy, found resistance to be more prevalent in females as well as in patients with renal impairment. These populations were found to carry a fourfold higher risk of death due to vascular events and higher risk of nonfatal cere- brovascular and cardiovascular events compared to aspirin sensitive patients (39% of AR patients versus 16% of aspirin sensitive patients had a cardiovascular event, OR 3.85, 95% CI 3.08–4.80; P < .001), regardless of the assay used to assess resistance [71]. The authors conclusions are reinforced by a previous meta-analysis evaluating clinical outcome in AR, where again, resistant patients had considerably higher risk of recurrent vascular events compared to aspirin sensitive patients [72]. In a recent trial evaluating the relationship between AR (assessed by thrombelastography) and stroke in 45 patients with ischemic stroke, it was found that AR was more frequent in the stroke than the control arm (67% versus 40%; P = .028). Within the stroke group, the AR arm had more severe stroke (assessed by Rankin score). In addition AR was greater in lacunar than embolic strokes [73]. In a follow-up of 468 patients with stable coronary artery disease and/or a high risk for vascular events (diabetic and/or hypertensive) for a mean period of 379 ± 200 days, cardiovascular death and/or nonfatal events were more frequent among patients with AR [74]. Moreover, after measuring urinary 11-dehydroTX B2 levels in 976 high-risk patients (half of them had sustained previous vascular events) who were initially enrolled in the heart outcomes prevention evaluation (HOPE) trial (on aspirin for 5 years), it was demonstrated that patients with levels in the highest quartile sustained more MI and cardiovascular death versus those in the lowest quartile [49], the drawbacks of this nested case control study were the issue of compliance which was not veriﬁed adequately by objective laboratory methods. Another point-of-care newly introduced assay is the VerifyNow-Aspirin (the Ultegra Rapid Platelet Function Assay, Accumetrics Inc., San Diego, Calif, USA) which corre- lates well with light transmittance aggregometry [66]. Results from the VerifyNow-Aspirin were highly reproducible in 21 healthy volunteers and 40 patients with stable coronary artery disease [67]. 5. Platelet-Function Testing pathway of arachidonic acid, which is blocked by aspirin at very high doses (1200 mg) only [40, 41]. Urinary 11- dehydroTX B2 concentration is aﬀected by kidneys pro- duction of this substance; however, measurement of this metabolite is still commonly used in trials assessing AR due to its low cost and ease to carry out [49, 69]. Multiple assays for platelet function and response to aspirin have emerged in the past decade. The test that is considered gold standard for assessment of the degree of aspirin response is light transmittance aggregometry (LTA) [61]. This assay measures the increase in light transmittance across platelet-rich plasma as a consequence of aggregation of platelets and development of clumps in response to diﬀerent agonists (ADP, collagen, arachidonic acid). Several drawbacks limit the use of this assay. However, it has been described as time consuming, operator dependent, and of high cost. Inability to reproduce results, even in the same laboratory, has also been reported [62]. According to this assay, the most accepted deﬁnition of AR is ≥20% platelet aggregation with 1 mg/mL arachidonic acid and ≥70% aggregation with 10 μmol/l ADP despite regular intake of aspirin [63]. A relevant question is the extent to which these lab- oratory methods correlate with one another. The various laboratory assays used to identify AR are compared weakly with each other. This was demonstrated in a study using six diﬀerent platelet function tests in 201 patients with stable coronary artery disease who were on daily aspirin use. The encounter of AR varied according to the assay used, being uppermost for the PFA-100 (60%) and lowermost using LTA (4%) [70]. Workup of a patient with suspected aspirin resistance, eventually leading to appropriate platelet- function testing, is highlighted in Figure 2. The point-of-care platelet function analyzer PFA-100 device (Dade Behring, Leiderbach, Germany) acts like an injured artery, where high shear stress conditions are present, and works in the presence of erythrocytes; unlike LTA where there is no interaction between platelet and other blood components. Hence, platelet function is evaluated by the time needed to form a platelet plug to occlude the gap present in this device. Using this test, AR is generally deﬁned as a closure time of <164 s despite regular aspirin intake. The device is easy to use and requires only a small amount of blood. The test is also fast and reproducible. 4. Proposed Etiology Lower doses of acetylsalicylic acid may be hydrolyzed to a higher extent into an inactive form, by gastrointestinal mucosal esterases when given with proton pump inhibitors due to acid suppression, thus reducing the absorption of the active drug. However, current evidence failed to conﬁrm this argument [46, 47]. Another hypothesis considers resistance at the target site of the drug’s action. This may highlight the role of genetic studies to determine the potential contribution of some genetic polymorphisms in AR. Polymorphisms in the COX- 1 gene have been implicated in the partial nonresponse to low-dose aspirin [48, 49]. However, a recent large systematic review addressing the role of diﬀerent genetic polymor- phisms did not ﬁnd a clear association between COX-1 gene polymorphisms (speciﬁcally C50T/A842G polymorphism) and AR [50]. Another major genetic contributor to biological AR is thought to be the PlA1/A2 polymorphism in the GPIIIa platelet receptor [51–53], which is according to the same systematic review, the most frequently investigated parameter (19 studies involving 1389 subjects) [50]. This variant was noticeably associated with AR when measured in the healthy population (OR 2.36, 95% CI 1.24–4.49; Finally in all conditions associated with a high platelet turnover (coronary artery bypass grafting, acute coronary syndrome (ACS), acute or chronic infection, or inﬂamma- tion), low-dose aspirin is associated with a short half-life (15 to 20 minutes) and might not be able to suppress COX-1 in fresh platelets that are continuously and quickly released into the circulation in such stressful circumstances, leading to higher platelet reactivity (Figure 1) [57–60]. 4 Advances in Hematology 5. Platelet-Function Testing Unfortunately, correlation between clinical outcomes with the PFA-100 is poor [64]. The lack of correlation with clinical outcome was also demonstrated in a recent prospective study assessing the prevalence of resistance in 97 patients with stable coronary artery disease on 160 mg aspirin for at least one month using the PFA-100 and a follow-up of 2.5 years for the composite of death, MI, and ischemic cerebral infarction or acute limb ischemia. It was found that aspirin resistant patients (29.9%) did not have a higher risk of death, MI, or ischemic vascular event compared with aspirin-sensitive patients [65]. Moreover, the reliance upon hematocrit and plasma von Willebrand factor, along with high cost, limits the use of PFA-100. 6. Clinical Implication It also showed poor sensitivity and good speciﬁcity with a cut-oﬀvalue at 550 aspirin reaction units (ARU), compared to LTA, which makes the signiﬁcance of the cut-oﬀlevel at 550 ARU for detecting AR controversial [67]. Other test measures consider the end products of the TX A2 pathway such as serum TX B2 [68], or urine 11- dehydroTX B2 [49], for assessing aspirin activity [49]. In fact, these two tests may better reﬂect the amount of TX A2 derived from sources other than platelets such as macrophages and monocytes, and on the COX-2 linked 5 Advances in Hematology Yes Yes Yes Yes Yes No No No No No No No Is the optimal dose of aspirin prescribed? (e.g. 300 mg for ACS) Is the patient compliant to aspirin intake? Is the proper form of aspirin used? (avoid use of enteric coated formulations) Are there any potential drug-drug interactions? (e.g. current NSAIDs or PPI intake) Are there any concomitant infections? Family history of premature CAD History of previous MI, stroke, PVD Stenting of unprotected left main coronary artery or last patient coronary vessel Dose the patient have a history of one of the following? Proceed with platelet function testing (e.g. VerifyNow-Aspirin, LTA, PFA-100) Follow-up for adverse vascular events Treat infection Use proper form Ensure compliance Use optimal dose Modify concurrent drugs . . . Figure 2: Algorithm highlighting approach to a patient with suspected aspirin resistance. ACS: acute coronary syndrome; NSAIDs: nonsteroidal antiinﬂamatory drugs; PPI: proton pump inhibitors; CAD: coronary artery disease; MI: myocardial infarction; PVD: peripheral vascular disease; LTA: light transmittance aggregometry; PFA: platelet function analyzer. Is the patient compliant to aspirin intake? Is the proper form of aspirin used? (avoid use of enteric coated formulations) Are there any concomitant infections? Proceed with platelet function testing (e.g. VerifyNow-Aspirin, LTA, PFA-100) Figure 2: Algorithm highlighting approach to a patient with suspected aspirin resistance. ACS: acute coronary syndrome; NSAIDs: nonsteroidal antiinﬂamatory drugs; PPI: proton pump inhibitors; CAD: coronary artery disease; MI: myocardial infarction; PVD: peripheral vascular disease; LTA: light transmittance aggregometry; PFA: platelet function analyzer. when the PFA-100 was used and HR 5.2, 95% CI 1.1–23; P = .03 using the LTA) [77]. 6. Clinical Implication The clinical correlation of AR was also elucidated in patients undergoing percutaneous coronary intervention (PCI) in a recent study that measured the cardiac biomarkers (CK-MB and Troponin I) in 151 patients who underwent nonurgent PCI and who were already on various doses of aspirin and a 300 mg loading dose of clopidogrel. The study demonstrated that patients with AR, assessed using the VerifyNow-Aspirin test, sustained a considerably higher incidence of post-PCI elevation in cardiac biomarkers [75]. Furthermore, in 216 patients with ST elevation myocardial infarction (STEMI), enhanced platelet activity assessed with PFA-100, was an independent predictor of markers of cardiac necrosis [76]. The laboratory response to glycoprotein (GP) IIB/IIIA inhibitors and the clinical outcome in 70 patients with STEMI undergoing PCI correlated well with baseline platelet reactivity (PR) (assessed with the PFA-100) before intervention. High-baseline PR was associated with a 5–11 times increase in the risk of death, reinfarction, and target vessel revascularization (HR 11, 95% CI 1.5–78; P = .02 Finally, aspirin resistance has been implicated recently as a potential cause of the rare but very serious complication of drug eluting stents thrombosis. This was demonstrated in a prospective systematic analysis of platelet aggregation in four subsequent cases of late thrombosis, where all four cases showed resistance to either aspirin (evaluated with the PFA- 100) or clopidogrel, and two cases showed dual resistance to both of these platelet antiaggregants [78]. 7. Management After correlating AR with incidence of more adverse car- diovascular events many investigators are trying to ﬁnd solutions to overcome this poor response. The idea of increasing aspirin dose in order to overcome resistance has been assayed in many studies, since there is some evidence that response to aspirin may be dose-dependent 6 Advances in Hematology Advances in Hematology [27]. Laboratory and genetic inconsistency as well as dose dependence is seen when agonists other than arachidonic acid (the most speciﬁc in assessing AR), such as (ADP, collagen, epinephrine), are used for in vitro assessment of platelet inhibition by aspirin. This inconsistency and dose- dependence may be explained by the fact that those agonists maybe only in part reliant upon, or even independent of, the COX-1 pathway which is the main target of aspirin [79–82]. More solid evidence that reﬂects on the appropriate dosage of aspirin was portrayed by the antithrombotic trialists collaboration. It showed that the most eﬀective aspirin dose with the fewest adverse consequences is 75 to 150 mg once daily with similar eﬃcacy as doses up to 1500 mg daily [8, 83]. Even though low-aspirin doses might relate to resistance by reducing absorption, administration of higher doses looks unwarranted and is outweighed by a higher risk of gastrointestinal bleeding [84]. Although the subgroup analysis of the CAPRIE trial failed to show a signiﬁcant beneﬁcial eﬀect of clopidogrel over aspirin in patients with history of recent stroke or MI [93], the recommendations issued by the American Heart Association/American Stroke Association (AHA/ASA) in 2006 on stroke prevention [94] supported initial antiplatelet therapy with the combination of aspirin and extended- release dipyridamole (ER-DP), 25 mg/200 mg twice a day (aggrenox), rather than aspirin (Grade 2A evidence) or the use of clopidogrel for those not treated with aggrenox (Grade 2B evidence) in patients with a history of noncardioembolic stroke or TIA of atherothrombotic, lacunar (small vessel occlusive type), or cryptogenic type. Hence, adding an addi- tional antiplatelet such as (ER-DP) or substituting aspirin by another potent antiplatelet (namely, clopidogrel) would be a logical alternative to patients who develop CVA despite using aspirin on a daily basis. The addition of clopidogrel to aspirin for secondary stroke prevention has fallen out of favor since the combination of the two drugs does not oﬀer better beneﬁt for stroke prevention than either drug alone but does considerably amplify the risk of bleeding complications [95]. 7. Management Unlike patients who develop stroke, those being on aspirin and develop ACS (unstable angina and non-STEMI) would beneﬁt from adding clopidogrel to aspirin as per the famous CURE trial [96]. [27]. Laboratory and genetic inconsistency as well as dose dependence is seen when agonists other than arachidonic acid (the most speciﬁc in assessing AR), such as (ADP, collagen, epinephrine), are used for in vitro assessment of platelet inhibition by aspirin. This inconsistency and dose- dependence may be explained by the fact that those agonists maybe only in part reliant upon, or even independent of, the COX-1 pathway which is the main target of aspirin [79–82]. More solid evidence that reﬂects on the appropriate dosage of aspirin was portrayed by the antithrombotic trialists collaboration. It showed that the most eﬀective aspirin dose with the fewest adverse consequences is 75 to 150 mg once daily with similar eﬃcacy as doses up to 1500 mg daily [8, 83]. Even though low-aspirin doses might relate to resistance by reducing absorption, administration of higher doses looks unwarranted and is outweighed by a higher risk of gastrointestinal bleeding [84]. In a number of studies, sensitivity of platelets to ADP and levels of this agonist in patients with AR revealed to be considerably ampliﬁed [25, 85]. Moreover, in a randomized cross-over trial, patients with AR turned out to be highly responsive to platelet ADP receptor antagonist [86]. One could argue, based on the above that replacing or adding a diﬀerent antiplatelet could cancel out the incidence of adverse events resulting from resistance to aspirin. This theory was refuted by a recent meta-analysis of 20 studies (6 of them had an additional antiplatelet used) which found that concomitant treatment with additional antiplatelet (namely, clopidogrel or tiroﬁban) provided no clinical beneﬁt to those patients identiﬁed as aspirin resistant (OR 2.52, 95% CI 1.79–3.56 for aspirin alone versus OR 3.06, 95% CI 1.99 to 4.70 for the dual antiplatelet group) [71]. To elaborate on the issue of dual antiplatelet therapy, an analysis of asymptomatic low-risk patients on both aspirin and clopidogrel in the CHARISMA trial revealed a higher incidence of death due to cardiovascular disease [87], and that dual antiplatelet therapy should be reserved for patients with high risk for vascular events [88]. The addition of dipyridamole to aspirin has also been subject to much debate [4, 89, 90]. 7. Management In a recent review involving 23 019 patients with vascular disease [91], dipyridamole alone or in combination with aspirin did not lessen the risk of vascular death (relative risk [RR] 0.99, 95% CI 0.87–1.12), though the risk of vascular events was lowered (RR 0.88, 95% CI 0.8– 0.95). 8. Conclusion Aspirin resistance is a true phenomenon that needs to be fur- ther elucidated. A single deﬁnition should be provided when describing resistance or nonresponse to aspirin. Moreover, consensus should be made about the optimal laboratory method that allows objective assessment of response to aspirin. These points, if achieved, may normalize the wide range of prevalence reported among diﬀerent studies. The correlation between AR and higher incidence of adverse vascular events is established by a considerable number of well-designed trials. On the other hand, despite the absence of optimal methods to overcome this phenomenon of resistance, physicians must emphasize on proper patients compliance, as well as avoidance of potential drug-drug interactions. Finally one should always keep in mind that no single platelet activation pathway is responsible for all thrombotic complications, and a single treatment strategy directed against a speciﬁc receptor/target cannot overcome all thrombotic complications. In particular, dealing with some patients who develop adverse vascular events (e.g., cerebrovascular accidents (CVA) and ACS) despite being on aspirin remains chal- lenging, and many questions persist without clear cut answers. There is no clear evidence and no consensus yet on whether using diﬀerent laboratory assays to detect aspirin resistance would guide therapy in those patients, especially after a recent report assessing AR status with PFA-100 in 129 patients with transient ischemic attack (TIA), stroke, or vascular cognitive impairment failed to predict new thrombotic events in patients found to be AR during mean follow-up of 56 months, as new thrombotic events occurred in 15.4% of AR patients and in 14.6% of those without resistance (P = 1.00) [92]. [1] The Second International Study of Infarct Survival Collabo- rative Group, “Randomized trial of intravenous streptokinase, oral aspirin, both, or neither among 17,187 cases of suspected acute myocardial infarction: ISIS-2,” The Lancet, vol. 2, no. 8607, pp. 349–360, 1988. References [1] The Second International Study of Infarct Survival Collabo- rative Group, “Randomized trial of intravenous streptokinase, oral aspirin, both, or neither among 17,187 cases of suspected acute myocardial infarction: ISIS-2,” The Lancet, vol. 2, no. 8607, pp. 349–360, 1988. [2] Antithrombotic Trialists’ Collaboration, “Collaborative over- view of randomised trials of antiplatelet therapy—II: main- tenance of vascular graft or arterial patency by antiplatelet therapy,” British Medical Journal, vol. 308, no. 6922, pp. 159– 168, 1994. Advances in Hematology 7 [22] T. S. Poulsen, B. Jørgensen, L. Korsholm, P. B. Licht, T. Haghfelt, and H. Mickley, “Prevalence of aspirin resistance in patients with an evolving acute myocardial infarction,” Thrombosis Research, vol. 119, no. 5, pp. 555–562, 2007. [3] Antithrombotic Trialists’ Collaboration, “Collaborative over- view of randomised trials of antiplatelet therapy—III: reduc- tion in venous thrombosis and pulmonary embolism by antiplatelet prophylaxis among surgical and medical patients,” British Medical Journal, vol. 308, no. 6923, pp. 235–246, 1994. [23] A. J. Grau, S. Reiners, C. Lichy, F. Buggle, and A. Ruf, “Platelet function under aspirin, clopidogrel, and both after ischemic stroke: a case-crossover study,” Stroke, vol. 34, no. 4, pp. 849– 854, 2003. [4] C. Baigent, C. Sudlow, R. Collins, and R. Peto, “Collaborative meta-analysis of randomised trials of antiplatelet therapy for prevention of death, myocardial infarction, and stroke in high risk patients,” British Medical Journal, vol. 324, no. 7329, pp. 71–86, 2002. [24] S. Ziegler, T. Maca, E. Alt, W. Speiser, B. Schneider, and E. Minar, “Monitoring of antiplatelet therapy with the PFA-100® in peripheral angioplasty patients,” Platelets, vol. 13, no. 8, pp. 493–497, 2002. [5] P. Marson and G. Pasero, “The Italian contributions to the history of salicylates,” Reumatismo, vol. 58, no. 1, pp. 66–75, 2006. [25] K. Grundmann, K. Jaschonek, B. Kleine, J. Dichgans, and H. Topka, “Aspirin non-responder status in patients with recurrent cerebral ischemic attacks,” Journal of Neurology, vol. 250, no. 1, pp. 63–66, 2003. [6] H. L´evesque and O. Lafont, “Aspirin throughout the ages: an historical review,” La Revue de M´edecine Interne, vol. 21, supplement 1, pp. S8–S17, 2000. [26] L. Macchi, L. Christiaens, S. Brabant, et al., “Resistance to aspirin in vitro is associated with increased platelet sensitivity to adenosine diphosphate,” Thrombosis Research, vol. 107, no. 1-2, pp. 45–49, 2002. [7] J. R. Vane, “The ﬁght against rheumatism: from willow bark to COX-1 sparing drugs,” Journal of Physiology and Pharmacology, vol. 51, no. 4, part 1, pp. References 573–586, 2000. [8] H. Vainio and G. Morgan, “Aspirin for the second hundred years: new uses for an old drug,” Pharmacology and Toxicology, vol. 81, no. 4, pp. 151–152, 1997. [27] J. M. ten Berg, W. B. M. Gerritsen, F. J. L. M. Haas, H. C. Kelder, F. W. A. Verheugt, and H. W. Thijs Plokker, “High-dose aspirin in addition to daily low-dose aspirin decreases platelet activation in patients before and after percutaneous coronary intervention,” Thrombosis Research, vol. 105, no. 5, pp. 385– 390, 2002. [9] C. Patrono, L. A. Garc´ıa Rodr´ıguez, R. Landolﬁ, and C. Baigent, “Low-dose aspirin for the prevention of atherothrom- bosis,” The New England Journal of Medicine, vol. 353, no. 22, pp. 2373–2383, 2005. [28] P. A. Gum, K. Kottke-Marchant, E. D. Poggio, et al., “Proﬁle and prevalence of aspirin resistance in patients with cardio- vascular disease,” American Journal of Cardiology, vol. 88, no. 3, pp. 230–235, 2001. [10] A. D. Michelson, M. Cattaneo, J. W. Eikelboom, et al., “Aspirin resistance: position paper of the Working Group on Aspirin Resistance,” Journal of Thrombosis and Haemostasis, vol. 3, no. 6, pp. 1309–1311, 2005. [29] K. Andersen, M. Hurlen, H. Arnesen, and I. Seljeﬂot, “Aspirin non-responsiveness as measured by PFA-100 in patients with coronary artery disease,” Thrombosis Research, vol. 108, no. 1, pp. 37–42, 2002. [11] G. J. Hankey and J. W. Eikelboom, “Aspirin resistance,” The Lancet, vol. 367, no. 9510, pp. 606–617, 2006. [12] T. H. Wang, D. L. Bhatt, and E. J. Topol, “Aspirin and clopidogrel resistance: an emerging clinical entity,” European Heart Journal, vol. 27, no. 6, pp. 647–654, 2006. [30] P. Harrison, H. Segal, K. Blasbery, C. Furtado, L. Silver, and P. M. Rothwell, “Screening for aspirin responsiveness after transient ischemic attack and stroke: comparison of 2 point- of-care platelet function tests with optical aggregometry,” Stroke, vol. 36, no. 5, pp. 1001–1005, 2005. [13] E. D. Michos, R. Ardehali, R. S. Blumenthal, R. A. Lange, and H. Ardehali, “Aspirin and clopidogrel resistance,” Mayo Clinic Proceedings, vol. 81, no. 4, pp. 518–526, 2006. [31] D. L. Yee, B. R. Dinu, C. W. Sun, et al., “Low prevalence and assay discordance of “aspirin resistance” in children,” Pediatric Blood & Cancer, vol. 51, no. 1, pp. 86–92, 2008. [14] A. D. Michelson, “Platelet function testing in cardiovascular diseases,” Circulation, vol. 110, no. 19, pp. e489–e493, 2004. [15] C. H. Hennekens, K. Schror, S. Weisman, and G. A. References FitzGerald, “Terms and conditions: semantic complexity and aspirin resistance,” Circulation, vol. 110, no. 12, pp. 1706–1708, 2004. [32] L. Christiaens, S. Ragot, J. Mergy, J. Allal, and L. Macchi, “Major clinical vascular events and aspirin-resistance status as determined by the PFA-100 method among patients with stable coronary artery disease: a prospective study,” Blood Coagulation & Fibrinolysis, vol. 19, no. 3, pp. 235–239, 2008. [16] J. Mehta, P. Mehta, C. Burger, and C. J. Pepine, “Platelet aggregation studies in coronary artery disease: past 4. Eﬀect of aspirin,” Atherosclerosis, vol. 31, no. 2, pp. 169–175, 1978. g y [33] B. Pamukcu, H. Oﬂaz, I. Onur, et al., “Clinical relevance of aspirin resistance in patients with stable coronary artery disease: a prospective follow-up study (PROSPECTAR),” Blood Coagulation & Fibrinolysis, vol. 18, no. 2, pp. 187–192, 2007. [17] P. A. Gurbel and U. S. Tantry, “Clopidogrel resistance?” Thrombosis Research, vol. 120, no. 3, pp. 311–321, 2007. [18] P. Andr´e, S. M. Delaney, T. LaRocca, et al., “P2Y12 regulates platelet adhesion/activation, thrombus growth, and thrombus stability in injured arteries,” The Journal of Clinical Investiga- tion, vol. 112, no. 3, pp. 398–406, 2003. g y pp [34] B. Pamukcu, H. Oﬂaz, A. Oncul, et al., “The role of aspirin resistance on outcome in patients with acute coronary syn- drome and the eﬀect of clopidogrel therapy in the prevention of major cardiovascular events,” Journal of Thrombosis and Thrombolysis, vol. 22, no. 2, pp. 103–110, 2006. [19] S. Sanderson, J. Emery, T. Baglin, and A.-L. Kinmonth, “Nar- rative review: aspirin resistance and its clinical implications,” Annals of Internal Medicine, vol. 142, no. 5, pp. 370–380, 2005. [35] O. M. Akay, Z. Canturk, E. Akin, C. Bal, and Z. Gulbas, “Aspirin-resistance frequency: a prospective study in 280 healthy Turkish volunteers,” Clinical and Applied Thrombo- sis/Hemostasis, vol. 15, no. 1, pp. 98–102, 2009. [20] L. Macchi, N. Sorel, and L. Christiaens, “Aspirin resistance: deﬁnitions, mechanisms, prevalence, and clinical signiﬁ- cance,” Current Pharmaceutical Design, vol. 12, no. 2, pp. 251– 258, 2006. [36] P.-Y. Lee, W.-H. Chen, W. Ng, et al., “Low-dose aspirin increases aspirin resistance in patients with coronary artery disease,” The American Journal of Medicine, vol. 118, no. 7, pp. 723–727, 2005. [21] E. Shantsila, T. Watson, and G. Y. H. Lip, “Aspirin resistance: what, why and when?” Thrombosis Research, vol. 119, no. 5, pp. 551–554, 2007. Advances in Hematology 8 [53] J. Dropinski, J. Musial, M. Sanak, W. References Agostoni, et al., “A systematic review and meta-analysis on the hazards of discontinuing or not adhering to aspirin among 50 279 patients at risk for coronary artery disease,” European Heart Journal, vol. 27, no. 22, pp. 2667–2674, 2006. [58] N. Zimmermann, M. Kurt, A. Wenk, J. Winter, E. Gams, and T. Hohlfeld, “Is cardiopulmonary bypass a reason for aspirin resistance after coronary artery bypass grafting?” European Journal of Cardio-Thoracic Surgery, vol. 27, no. 4, pp. 606–610, 2005. [44] D. Rottlaender, M. Scherner, T. Schneider, and E. Erdmann, “Polypharmacy, compliance and non-prescription medication in patients with cardiovascular disease in Germany,” Deutsche Medizinische Wochenschrift, vol. 132, no. 4, pp. 139–144, 2007. [59] S. Guthikonda, E. I. Lev, R. Patel, et al., “Reticulated platelets and uninhibited COX-1 and COX-2 decrease the antiplatelet eﬀects of aspirin,” Journal of Thrombosis and Haemostasis, vol. 5, no. 3, pp. 490–496, 2007. [45] F. Catella-Lawson, M. P. Reilly, S. C. Kapoor, et al., “Cyclooxy- genase inhibitors and the antiplatelet eﬀects of aspirin,” The New England Journal of Medicine, vol. 345, no. 25, pp. 1809– 1817, 2001. [60] A. Modica, F. Karlsson, and T. Mooe, “Platelet aggregation and aspirin non-responsiveness increase when an acute coronary syndrome is complicated by an infection,” Journal of Throm- bosis and Haemostasis, vol. 5, no. 3, pp. 507–511, 2007. [46] B. S. Anand, S. K. Sanduja, and L. M. Lichtenberger, “Eﬀect of omeprazole on the bioavailability of aspirin: a randomized controlled study on healthy volunteers,” Gastroenterology, vol. 116, p. A371, 1999. [61] P. Harrison, “Platelet function analysis,” Blood Reviews, vol. 19, no. 2, pp. 111–123, 2005. [47] P. I˜narrea, F. Esteva, R. Cornudella, and A. Lanas, “Omepra- zole does not interfere with the antiplatelet eﬀect of low-dose aspirin in man,” Scandinavian Journal of Gastroenterology, vol. 35, no. 3, pp. 242–246, 2000. [62] N. S. Nicholson, S. G. Panzer-Knodle, N. F. Haas, et al., “Assessment of platelet function assays,” American Heart Journal, vol. 135, no. 5, part 2, pp. S170–S178, 1998. [48] M. K. Halushka and P. V. Halushka, “Why are some individuals resistant to the cardioprotective eﬀects of aspirin? Could it be thromboxane A2?” Circulation, vol. 105, no. 14, pp. 1620– 1622, 2002. [63] P. A. Gum, K. Kottke-Marchant, E. D. Poggio, et al., “Proﬁle and prevalence of aspirin resistance in patients with cardio- vascular disease,” American Journal of Cardiology, vol. 88, no. 3, pp. 230–235, 2001. [49] J. W. Eikelboom, J. Hirsh, J. References We¸grzyn, R. Nizankowski, and A. Szczeklik, “Antithrombotic eﬀects of aspirin based on PLA1/A2 glycoprotein IIIa polymorphism in patients with coronary artery disease,” Thrombosis Research, vol. 119, no. 3, pp. 301–303, 2007. [37] G. H. R. Rao, “Inﬂuence of anti-platelet drugs on platelet- vessel wall interactions,” Prostaglandins, Leukotrienes and Medicine, vol. 30, no. 2-3, pp. 133–145, 1987. [38] J. Koutts, “Aspirin and coronary heart disease. Clinical appli- cations,” Australian Family Physician, vol. 19, no. 2, pp. 217, 221–223, 1990. [54] H. W. Cohen, J. P. Crandall, S. M. Hailpern, and H. H. Billett, “Aspirin resistance associated with HbA1c and obesity in diabetic patients,” Journal of Diabetes and Its Complications, vol. 22, no. 3, pp. 224–228, 2008. [39] K. A. Louden, F. B. Pipkin, S. Heptinstall, et al., “Neonatal platelet reactivity and serum thromboxane B2 production in whole blood: the eﬀect of maternal low dose aspirin,” British Journal of Obstetrics and Gynaecology, vol. 101, no. 3, pp. 203– 208, 1994. [55] H. Ogawa, M. Nakayama, T. Morimoto, et al., “Low-dose aspirin for primary prevention of atherosclerotic events in patients with type 2 diabetes: a randomized controlled trial,” Journal of the American Medical Association, vol. 300, no. 18, pp. 2134–2141, 2008. [40] F. Bucchi, A. Bodzento, G. de Gaetano, and C. Cerletti, “Eﬀects of 1 gram oral or intravenous aspirin on urinary excretion of thromboxane B2 and 6-keto-PGF1α in healthy subjects,” Prostaglandins, vol. 32, no. 5, pp. 691–701, 1986. [56] J. Belch, A. MacCuish, I. Campbell, et al., “The prevention of progression of arterial disease and diabetes (POPADAD) trial: factorial randomised placebo controlled trial of aspirin and antioxidants in patients with diabetes and asymptomatic peripheral arterial disease,” British Medical Journal, vol. 337, p. a1840, 2008. [41] J. C. Fr¨olich, “A classiﬁcation of NSAIDs according to the relative inhibition of cyclooxygenase isoenzymes,” Trends in Pharmacological Sciences, vol. 18, no. 1, pp. 30–34, 1997. [42] G. Cotter, E. Shemesh, M. Zehavi, et al., “Lack of aspirin eﬀect: aspirin resistance or resistance to taking aspirin?” American Heart Journal, vol. 147, no. 2, pp. 293–300, 2004. [57] N. Zimmermann, A. Wenk, U. Kim, et al., “Functional and biochemical evaluation of platelet aspirin resistance after coronary artery bypass surgery,” Circulation, vol. 108, no. 5, pp. 542–547, 2003. [43] G. G. L. Biondi-Zoccai, M. Lotrionte, P. References Loke, “Risk of gastrointestinal haemorrhage with long term use of aspirin: meta-analysis,” British Medical Journal, vol. 321, no. 7270, pp. 1183–1187, 2000. [70] M. Lordkipanidz´e, C. Pharand, E. Schampaert, J. Turgeon, D. A. Palisaitis, and J. G. Diodati, “A comparison of six major platelet function tests to determine the prevalence of aspirin resistance in patients with stable coronary artery disease,” European Heart Journal, vol. 28, no. 14, pp. 1702–1708, 2007. [84] C. L. Campbell, S. Smyth, G. Montalescot, and S. R. Steinhubl, “Aspirin dose for the prevention of cardiovascular disease: a systematic review,” Journal of the American Medical Associa- tion, vol. 297, no. 18, pp. 2018–2024, 2007. [85] C. Borna, E. Lazarowski, C. van Heusden, H. ¨Ohlin, and D. Erlinge, “Resistance to aspirin in increased by ST-elevation myocardial infarction and correlates with adenosine diphos- phate levels,” Thrombosis Journal, vol. 3, article 10, pp. 1–9, 2005. [71] G. Krasopoulos, S. J. Brister, W. S. Beattie, and M. R. Buchanan, “Aspirin “resistance” and risk of cardiovascular morbidity: systematic review and meta-analysis,” British Med- ical Journal, vol. 336, no. 7637, pp. 195–198, 2008. [72] J. D. Snoep, M. M. C. Hovens, J. C. J. Eikenboom, J. G. van der Bom, and M. V. Huisman, “Association of laboratory- deﬁned aspirin resistance with a higher risk of recurrent cardiovascular events: a systematic review and meta-analysis,” Archives of Internal Medicine, vol. 167, no. 15, pp. 1593–1599, 2007. [86] J. W. Eikelboom, G. J. Hankey, J. Thom, et al., “Enhanced antiplatelet eﬀect of clopidogrel in patients whose platelets are least inhibited by aspirin: a randomized crossover trial,” Journal of Thrombosis and Haemostasis, vol. 3, no. 12, pp. 2649–2655, 2005. [87] T. H. Wang, D. L. Bhatt, K. A. A. Fox, et al., “An analysis of mortality rates with dual-antiplatelet therapy in the primary prevention population of the CHARISMA trial,” European Heart Journal, vol. 28, no. 18, pp. 2200–2207, 2007. [73] N. A. Englyst, G. Horsﬁeld, J. Kwan, and C. D. Byrne, “Aspirin resistance is more common in lacunar strokes than embolic strokes and is related to stroke severity,” Journal of Cerebral Blood Flow and Metabolism, vol. 28, no. 6, pp. 1196–1203, 2008. [88] D. L. Bhatt, M. D. Flather, W. Hacke, et al., “Patients with prior myocardial infarction, stroke, or symptomatic peripheral arterial disease in the CHARISMA trial,” Journal of the American College of Cardiology, vol. 49, no. 19, pp. 1982– 1988, 2007. [74] W.-H. Chen, X. References Cheng, P.-Y. Lee, et al., “Aspirin resistance and adverse clinical events in patients with coronary artery disease,” The American Journal of Medicine, vol. 120, no. 7, pp. 631–635, 2007. [89] The ESPS-2 Group, “European stroke prevention study 2: eﬃcacy and safety data,” Journal of the Neurological Sciences, vol. 151, supplement 1, pp. S1–S77, 1997. [75] W.-H. Chen, P.-Y. Lee, W. Ng, H.-F. Tse, and C.-P. Lau, “Aspirin resistance is associated with a high incidence of myonecrosis after non-urgent percutaneous coronary inter- vention despite clopidogrel pretreatment,” Journal of the American College of Cardiology, vol. 43, no. 6, pp. 1122–1126, 2004. [90] P. H. Halkes, J. van Gijn, L. J. Kappelle, P. J. Koudstaal, and A. Algra, “Aspirin plus dipyridamole versus aspirin alone after cerebral ischaemia of arterial origin (ESPRIT): randomised controlled trial,” The Lancet, vol. 367, no. 9523, pp. 1665–1673, 2006. [76] M. Frossard, I. Fuchs, J. M. Leitner, et al., “Platelet function predicts myocardial damage in patients with acute myocardial infarction,” Circulation, vol. 110, no. 11, pp. 1392–1397, 2004. [91] E. De Schryver, A. Algra, and J. van Gijn, “Dipyridamole for preventing stroke and other vascular events in patients with vascular disease,” Cochrane Database of Systematic Reviews, vol. 18, no. 3, Article ID CD001820, 2007. [77] G. Campo, M. Valgimigli, D. Gemmati, et al., “Value of platelet reactivity in predicting response to treatment and clinical outcome in patients undergoing primary coronary intervention. Insights into the STRATEGY study,” Journal of the American College of Cardiology, vol. 48, no. 11, pp. 2178– 2185, 2006. [92] G. B. Boncoraglio, A. Bodini, C. Brambilla, E. Corsini, M. R. Carriero, and E. A. Parati, “Aspirin resistance determined with PFA-100 does not predict new thrombotic events in patients with stable ischemic cerebrovascular disease,” Clinical Neurology and Neurosurgery, vol. 111, no. 3, pp. 270–273, 2009. [78] G. Barone-Rochette, O. Ormezzano, B. Polack, G. Vanzetto, B. Bertrand, and J. Machecourt, “Resistance to platelet antiaggregants: an important cause of very late thrombosis of drug eluting stents? Observations from ﬁve cases,” Archives of Cardiovascular Diseases, vol. 101, no. 2, pp. 100–107, 2008. [93] M. Gent, “A randomised, blinded, trial of clopidogrel versus aspirin in patients at risk of ischaemic events (CAPRIE),” The Lancet, vol. 348, no. 9038, pp. 1329–1339, 1996. [79] N. Faraday, L. R. Yanek, R. Mathias, et al., “Heritability of platelet responsiveness to aspirin in activation pathways directly and indirectly related to cyclooxygenase-1,” Circula- tion, vol. 115, no. References I. Weitz, M. Johnston, Q. Yi, and S. Yusuf, “Aspirin-resistant thromboxane biosynthesis and the risk of myocardial infarction, stroke, or cardiovascular death in patients at high risk for cardiovascular events,” Circulation, vol. 105, no. 14, pp. 1650–1655, 2002. [64] E. F. Mammen, P. C. Comp, R. Gosselin, et al., “PFA-100TM system: a new method for assessment of platelet dysfunction,” Seminars in Thrombosis and Hemostasis, vol. 24, no. 2, pp. 195– 202, 1998. [65] L. Christiaens, S. Ragot, J. Mergy, J. Allal, and L. Macchi, “Major clinical vascular events and aspirin-resistance status as determined by the PFA-100 method among patients with stable coronary artery disease: a prospective study,” Blood Coagulation & Fibrinolysis, vol. 19, no. 3, pp. 235–239, 2008. [50] T. Goodman, A. Ferro, and P. Sharma, “Pharmacogenetics of aspirin resistance: a comprehensive systematic review,” British Journal of Clinical Pharmacology, vol. 66, no. 2, pp. 222–232, 2008. [66] J. L. Coleman, J. C. Wang, and D. I. Simon, “Determination of individual response to aspirin therapy using the Accumetrics Ultegra RFPA-ASA system,” Point of Care, vol. 3, no. 2, pp. 77– 82, 2004. [51] A. Szczeklik, A. Undas, M. Sanak, M. Frolow, and W. We¸grzyn, “Relationship between bleeding time, aspirin and the PlA1/A2 polymorphism of platelet glycoprotein IIIa,” British Journal of Haematology, vol. 110, no. 4, pp. 965–967, 2000. [67] H. L. Nielsen, S. D. Kristensen, S. S. Thygesen, et al., “Aspirin response evaluated by the VerifyNowTM Aspirin System and Light Transmission Aggregometry,” Thrombosis Research, vol. 123, no. 2, pp. 267–273, 2008. [52] E. Papp, V. Havasi, J. Bene, et al., “Glycoprotein IIIA gene (PlA) polymorphism and aspirin resistance: is there any correlation?” Annals of Pharmacotherapy, vol. 39, no. 6, pp. 1013–1018, 2005. Advances in Hematology 9 [82] P.-Y. Lee, W.-H. Chen, W. Ng, et al., “Low-dose aspirin increases aspirin resistance in patients with coronary artery disease,” The American Journal of Medicine, vol. 118, no. 7, pp. 723–727, 2005. [68] R. G. Hart, A. D. Leonard, R. L. Talbert, et al., “Aspirin dosage and thromboxane synthesis in patients with vascular disease,” Pharmacotherapy, vol. 23, no. 5, pp. 579–584, 2003. [69] A. Bruno, J. P. McConnell, S. N. Cohen, et al., “Serial urinary 11-dehydrothromboxane B2, aspirin dose, and vascular events in blacks after recent cerebral infarction,” Stroke, vol. 35, no. 3, pp. 727–730, 2004. [83] S. Derry and Y. K. References 19, pp. 2490–2496, 2007. [94] R. L. Sacco, R. Adams, G. Albers, et al., “Guidelines for prevention of stroke in patients with ischemic stroke or transient ischemic attack: a statement for healthcare pro- fessionals from the American Heart Association/American Stroke Association council on stroke: co-sponsored by the council on cardiovascular radiology and intervention. The American Academy of Neurology aﬃrms the value of this guideline,” Stroke, vol. 37, no. 2, pp. 577–617, 2006. [80] A. Assadian, J. Lax, U. Meixner-Loicht, G. W. Hagm¨uller, P. M. Bayer, and W. H¨ubl, “Aspirin resistance among long- term aspirin users after carotid endarterectomy and controls: ﬂow cytometric measurement of aspirin-induced platelet inhibition,” Journal of Vascular Surgery, vol. 45, no. 6, pp. 1142–1147, 2007. [95] H.-C. Diener, J. Bogousslavsky, L. M. Brass, et al., “Aspirin and clopidogrel compared with clopidogrel alone after recent ischaemic stroke or transient ischaemic attack in high-risk [81] D. J. Fitzgerald and A. Maree, “Aspirin and clopidogrel resistance,” Hematology, vol. 2007, no. 1, pp. 114–120, 2007. 10 Advances in Hematology patients (MATCH): randomised, double-blind, placebo- controlled trial,” The Lancet, vol. 364, no. 9431, pp. 331–337, 2004. [96] S. Yusuf, F. Zhao, S. R. Mehta, S. Chrolavicius, G. Tognoni, and K. K. Fox, “Eﬀects of clopidogrel in addition to aspirin in patients with acute coronary syndromes without ST-segment elevation,” The New England Journal of Medicine, vol. 345, no. 7, pp. 494–502, 2001.
W2994054988.txt	https://link.springer.com/content/pdf/10.1007/s11614-019-00367-y.pdf	de		Forumtheater als Werkzeug szenisch-partizipativer Forschung in der Sozialen Arbeit. Verfahrensweisen, Projektbeispiele und methodische Reflexionen	Österreichische Zeitschrift für Soziologie/ÖZS. Österreichische Zeitschrift für Soziologie	2,019	cc-by	8,665	Österreich Z Soziol (2019) (Suppl 3) 44:83–102 https://doi.org/10.1007/s11614-019-00367-y HAUPTBEITRÄGE Forumtheater als Werkzeug szenisch-partizipativer Forschung in der Sozialen Arbeit. Verfahrensweisen, Projektbeispiele und methodische Reflexionen Michael Wrentschur © Der/die Autor(en) 2019 Zusammenfassung Forumtheater ist eine interaktive Theaterform des „Theaters der Unterdrückten“, die in vielerlei Hinsicht zu einem Werkzeug partizipativer Forschung in der Sozialen Arbeit werden kann: Dies betrifft die Einbeziehung von Adressat_innen in die Gestaltungs- und Forschungsprozesse ebenso wie die aktive Beteiligung des Publikums bei Aufführungen und schließlich den Transfer der Ergebnisse in soziale und politische Räume. Ausgehend von grundlegenden methodologischen Überlegungen zu szenischen Forschungsprozessen werden Methodik und Verfahrensweisen des Forumtheaters vorgestellt, mit Projektbeispielen aus Handlungs- und Problemfeldern Sozialer Arbeit illustriert sowie methodisch reflektiert. Diskutiert wird die Frage, welche Ansprüche partizipativer Forschung mit dem Forumtheater realisiert und welche Bezüge zu einer performativen Ethnografie und kritisch-lebensweltorientierten Adressat_innenforschung in der Sozialen Arbeit hergestellt werden können. Schlüsselwörter Szenisches und partizipatives Forschen · Forumtheater und Soziale Arbeit Forum Theatre as a tool of drama based, participatory research in Social Work. Methods, examples and methodical reflections Abstract Forum Theatre is an interactive Theatre technique of the “Theatre of the Oppressed”, which can be used as a tool of participatory research in Social Work involving the clients of Social Work into the creation and research process, the active participation of the audience during the performances and the transfer of M. Wrentschur () Institut für Erziehungs- und Bildungswissenschaft, Universität Graz, Merangasse 70/II, 8010 Graz, Österreich E-Mail: michael.wrentschur@uni-graz.at K 84 M. Wrentschur results into social and political spaces. Starting with methodological reflections on drama-based research, methods of the Forum Theatre are presented and illustrated by examples of projects in fields of Social Work. This will be combined with methodical reflections. The question will be discussed, which claims of participatory research can be realized and which references to performative ethnography and critical, living environment orientated user research in Social Work can be made. Keywords Drama-based and participatory research · Forum Theatre and social work 1 Einleitung Auf welche Weise kann Forumtheater als methodisches Verfahren im Sinne einer szenischen und partizipativen Forschungskonzeption in der Sozialen Arbeit angewendet und in der methodologischen Landschaft verortet werden? Diese Frage hat in meiner eigenen Biografie als Forscher seit meiner Dissertation (Wrentschur 2004) Relevanz, sie wurde in Projekten und Publikationen zu szenisch-partizipativen Forschungsansätzen fortgeführt (Wrentschur 2007, 2008a, 2012) und schließlich in meiner Habilitationsschrift (Wrentschur 2019) stärker methodologisch reflektiert und fundiert. Dabei ist Forumtheater zunächst eine interaktive Aufführungsform des sogenannten Theaters der Unterdrückten (= TdU), bei der das Publikum eingeladen wird, sich am Spielgeschehen zu beteiligen, um vielfältige Handlungs- und Veränderungsideen für einen in szenischen Verläufen dargestellten sozialen Konflikt oder ein Problem zu erproben (Staffler 2009; Boal 2013; Fritz 2013). Die Folgen des Handelns, das besonders auf die Veränderung diskriminierender, ausgrenzender und unterdrückender Interaktionen und Strukturen abzielt, werden in diesem dramatischen Labor unmittelbar sichtbar und erlebbar. Gemeinsam werden Handlungsspielräume untersucht, verbunden mit der Frage, welche strukturellen Veränderungen notwendig sind, um vor allem sozial ausgegrenzten und benachteiligten Gruppen umfassende Möglichkeiten zur gesellschaftlichen (und politischen) Partizipation zu eröffnen. Mit Forumtheater ist aber auch ein partizipativer und szenischer Forschungsprozess in Gruppen gemeint, bei dem sich die Teilnehmenden unter Anwendung szenisch-theatralischer Methoden mit einschränkenden, diskriminierenden oder unterdrückenden Situationen und/oder Strukturen in ihrem Alltagsleben und ihren Lebenswelten auseinandersetzen. Die Theatralisierung und Reflexion von individuellen, subjektiven Erfahrungen führen zu Verdichtungen in Form von Szenen und Bildern. Erkenntnisund Bewusstwerdungsprozesse werden mit der Suche nach Veränderung der persönlichen, sozialen und politischen Wirklichkeit verbunden (vgl. Wrentschur 2012, 2019). Kommt Forumtheater in Handlungs- und Problemfeldern Sozialer Arbeit zur Anwendung, wird die Einbeziehung von Adressat_innen Sozialer Arbeit in die Gestaltungs- und Forschungsprozesse ebenso wie die aktive Beteiligung des Publikums bei Aufführungen und schließlich der Transfer der Ergebnisse in soziale und politische Räume angestrebt. Damit werden nicht nur Ansprüche partizipativer Forschung realisiert, sondern es zeigen sich Bezüge zu einer performativen Ethnografie und zu einer K Forumtheater als Werkzeug szenisch-partizipativer Forschung in der Sozialen Arbeit.... 85 kritischen Adressat_innenforschung. Um das zu verdeutlichen, stelle ich zunächst grundlegende Überlegungen zu szenischen und performativen Forschungsprozessen an. In weiterer Folge werden Methodik und Verfahrensweisen vorgestellt und mit Projektbeispielen illustriert, bei denen Forumtheater als Werkzeug szenischen Forschens in Problem- und Handlungsfeldern der Sozialen Arbeit zur Anwendung gekommen ist. Dies wird mit methodischen und methodologischen Reflexionen zu Möglichkeiten, Herausforderungen und Grenzen von Forumtheater als Forschungsansatz verbunden. 2 Vom Theaterspiel(en) als Forschung zur performativen und szenischen Sozialforschung Aus der sozialwissenschaftlichen Forschung sind Beispiele bekannt, bei denen Methoden des Theaterspiels zumindest implizit angewendet wurden. Bei den Breaching Experiments der „Ethnomethodologen“ (Garfinkel 1985) wurden alltägliche Regeln und Normen auf irritierende Weise gebrochen, um aufgrund der Reaktionen der Menschen deren alltagssoziologischen Sinn und Zustandekommen besser zu verstehen. Und viele Experimente in der Psychologie wie das Milgram-Experiment (Milgram 1974) können als verdecktes oder unsichtbares Theater bezeichnet werden, bei dem die Versuchspersonen nicht wussten, welche Rollen ihnen im wahrsten Sinn des Wortes vorgespielt wurden, sie agierten in der Als-ob-Realität des Labors, als ob es Realität wäre. Mittlerweile sind vielfältige szenische, performative und theaterpädagogische Verfahren dokumentiert (vgl. Wrentschur 2019, S. 383 ff.), die das Medium „Theaterspiel“ explizit als Grundlage oder Vertiefung für sozialwissenschaftliche Forschungsprozesse nutzbar machen. So wird festgestellt, dass Theaterschaffende zumeist – wenn auch nicht bewusst – in einem qualitativen Sinn forschen, da es ihnen um ein Verstehen und Interpretieren der sozialen Welt sowie um ein Präsentieren der Ergebnisse der Theaterarbeit in Form von Aufführungen (vgl. Kramer 2007, S. 174) geht. Für Theaterschaffende ist es oft selbstverständliche Praxis, Daten zu sammeln, zu recherchieren und ein Stück zu produzieren, das aus persönlichen Geschichten entwickelt wird. Dabei sieht Kramer in Anlehnung an Denzin den qualitativen, ethnografischen Forscher/die qualitative, ethnografische Forscherin selbst als Performer: „the world is understood to be performance, culture is a verb, not a noun, fieldwork is a collaborative process – a performance – and knowledge is performative, not informative“ (Denzin 1977, S. 104, zit. nach Kramer 2007, S. 176). Die Rede ist von der performance ethnography, bei der „leiblich verankerte Interaktionen und deren Bedeutung“ im Zentrum stehen (Winter 2010, Abs. 34). Dabei führen die Ethnograf_innen und zum Teil auch die von ihnen Beforschten die Forschungsergebnisse öffentlich auf, wodurch es zum „gegenseitigen Austausch und Abgleichen von gelebten Erfahrungen, Emotionen, Perspektiven und Formen des Verstehens“ (ebd.) kommt. Persönliche Geschichten, die Menschen einander erzählen, werden kulturell und gesellschaftlich kontextualisiert, indem sie zu anderen geschriebenen Texten und Diskursen in Bezug gesetzt werden (vgl. Winter 2012, S. 458). Dazu werden in erster Linie vielfältige szenische Verfahren angewendet, um verschiedene Stimmen und Perspektiven der jeweils Betroffenen zu K 86 M. Wrentschur präsentieren. Diese Berücksichtigung verschiedener Perspektiven bildet die Voraussetzung für einen Text, „der die differenzierten Verschränkungen persönlicher, interpersonaler und politischer Ebenen erfassen kann“. In der Interaktion zwischen den Welten der Forschenden und „denen der Erforschten spielt sich der auf Kooperation aufbauende Forschungsprozess ab“ (Winter 2010, Abs. 18). Auf diese Weise können Zuschauer_innen Verbindungen zu ihren eigenen Erfahrungen herstellen und die Forschungsergebnisse können so für mehr Menschen zugänglich werden als allein für jene, die akademische Textproduktion gewohnt sind und diese studiert haben: „Theatre, I would argue, has the potential of reaching a much wider audience, and also offers a possibility of giving back to a community that one has conducted research in. It further offers multiple ways of communication, through different genres and setting up of spaces, which allow for audience accessibility and even participation, if that is a desired feature“ (Kramer 2007, S. 178). Auch für Liamputtong (2007, S. 172 ff.) sind Theaterstücke und Performances Ort und Mittel, um sozialen, politischen und kulturellen Dialog zu verhandeln – Performances können gerade über die Darstellung von Gefühlen die Herzen von Menschen und nicht nur deren Verstand erreichen: „Performance can be a powerful instrument for social reform, for righting some wrongs, and for helping those in need“ (Fontana und Frey 2005, S. 714, zit. nach Liamputtong 2007, S. 173). Performative Ansätze in der Ethnografie wie die Autoethnografie oder die performative Ethnografie werden von Winter (2010, Abs. 4) in einem kritischen Verständnis von qualitativer Forschung verortet, gerade dann, wenn sie sozial und politisch im Sinne einer emanzipatorischen Agenda transformativ „zur Verwirklichung sozialer Gerechtigkeit und einer radikalen Demokratie beitragen“ (ebd.). Durch die szenisch-performative Darstellung von Phänomenen und Problemen können sich alternative Sicht- und Denkweisen entwickeln, was mit dem Ziel verbunden wird, Veränderungen im Alltagsleben von Menschen zu bewirken. Dies geht einher mit kooperativen Formen der Forschung, der Ermächtigung der Beforschten und der Demokratisierung von Wissen und Macht (vgl. Winter 2010, Abs. 16), wie es in der partizipativen Forschung angestrebt wird. In sozialwissenschaftlichen Forschungskontexten wird zudem von performativer Sozialforschung gesprochen, wie sie etwa in einer Sondernummer vom Forum Qualitative Sozialforschung (9/2008) anhand zahlreicher performativer Verfahren und Konzeptionen vorgestellt und diskutiert wird. Performative Sozialforschung folgt dem wachsenden Interesse am Einbezug von Kunst und dabei vor allem von Drama in den Sozialwissenschaften. Seitz zufolge entstand performative research vor dem Hintergrund einer Kunstpraxis, „die sich als artistic research gegenüber den Wissenschaften zu behaupten sucht und als practice-led research innerhalb der empirischen Sozialforschung inzwischen sogar einen eigenständigen Status beansprucht“ (Seitz 2012, S. 81). Damit wird „ein eigenes, an künstlerischen Verfahren orientiertes Forschungsparadigma“ befördert (ebd., S. 84). Bei performative research stehen das Handeln und die Alltagspraktiken im Vordergrund, weniger die inneren und äußeren Gründe dafür. Dabei geht es „um ein spezifisches, praktisches Einlassen, um ein umsichtiges Behandeln von sozialer Praxis, von Material, Körper, Raum, ja selbst von Ideen“ (ebd., S. 87). K Forumtheater als Werkzeug szenisch-partizipativer Forschung in der Sozialen Arbeit.... 87 Der im vorliegenden Beitrag verwendete Begriff des szenischen Forschens wird in Bezug zu performativen Ansätzen der Sozialforschung gestellt, ist aber stärker an szenisch-theatralischen Vorgangsweisen und Vorgängen im engeren Sinn orientiert (vgl. Wrentschur 2019, S. 381 ff.) Dies bedeutet, dass alle Elemente des Theaterspiels dabei eine Rolle spielen (können), wie Raum, Bewegung, Rhythmus, Gestik, Mimik, Körperhaltungen, Emotionen, Sprache, Stimme, Interaktionen, Handlungsabläufe, Rollen und Figuren. Der ästhetische Raum des Theaterspiels ist nach Boal (vgl. 1999, S. 31 ff.) plastisch und gestaltbar, Vergangenes kann gegenwärtig, Fiktives real und Reales fiktiv werden. Darsteller_innen wie Zuschauer_innen sind dabei in einem bestimmten Raum bzw. an einem bestimmten Ort und gleichzeitig in der Welt der jeweiligen Szene(n) anwesend. Und auch die telemikroskopische Dimension des ästhetischen Raumes bekommt für Forschungsprozesse insofern Bedeutung, als Vorgänge und Prozesse vor- und zurückgespult oder auf andere Weise verfremdet werden, damit sich Fremdes im Alltäglichen offenbaren kann (vgl. Boal 1999, S. 27–39; Koch 1997, S. 81–85). Zentral sind jedenfalls die in einem szenischen Raum mitwirkenden Akteur_innen, deren Handlungen, Haltungen und Verhaltensweisen sowie deren Zusammenspiel und Zusammenwirken im theatralischen Raum: „Auf diese Weise näher herangeholt und vergrößert, können menschliche Handlungen besser beobachtet werden“ (Boal 1999, S. 38). Zumeist stehen Inhalte und Themen szenischer Verfahren in Bezug zu Alltags- und Lebenswelterfahrungen, die in szenisch-theatralische Bilder und Handlungsabläufe transformiert und zu einer eigenen ästhetischen wie sozialen Wirklichkeit werden. Dabei gestaltet sich szenisches Forschen „während der Herstellung und der assoziativen Interpretation notwendigerweise als partizipatorischer Prozess“ (Weitzel 2012b, S. 269). Weitzel bindet szenisches Forschen an die Praxeologie (vgl. Weitzel 2012a, S. 111 ff.), deren zentrale Frage es ist, auf welche Weise Menschen handeln (können). Insofern steht eine erkenntnisleitende Handlungsperspektive im Vordergrund und damit eine performative Perspektive auf Soziale Arbeit. Dabei gestaltet sich szenisches Forschen aus Weitzels Sicht „während der Herstellung und der assoziativen Interpretation notwendigerweise als partizipatorischer Prozess“ (Weitzel 2012b, S. 269). Und dies gilt in starkem Maße für Forschungsprozesse mit dem Forumtheater, die ich im folgenden Abschnitt in Bezug zu partizipativen Forschungsansätzen stelle. 3 Forumtheater im Diskurs partizipativer Forschungsansätze In fachlichen Beiträgen und Diskursen zur Aktionsforschung und zur partizipativen Forschung finden sich immer wieder Bezüge zu partizipativer Theaterarbeit wie dem Forumtheater: So verweist Liamputtong (2007, S. 133) darauf, dass Dramamethoden in Projekten der Participatory Action Research (= PAR) häufig eine Rolle spielen. Im aktuellen Handbuch für Action Research berichten Wakefort et al. (2015) über Projekte der PAR, bei denen interaktive und partizipative Theaterformen im Sinne des Forumtheaters zur Anwendung kamen. Guhathakurta (2015) diskutiert „Theatre in Participatory Action Research“ in Bangladesch sowie in Indien und bezieht sich dabei vor allem auf interaktives Theater im Sinne des TdU. Im Gegensatz zu vielen Entwicklungstheater-Projekten werden der Bevölkerung nicht Lösungen vor- K 88 M. Wrentschur gespielt, vielmehr identifizieren die Menschen selbst ihre Probleme, Themen und Anliegen, verbunden mit der Möglichkeit der Transformation (vgl. ebd., S. 101). Streck und Holliday (2015) gehen in Zusammenhang mit der PAR auf lateinamerikanische Wurzeln und Perspektiven ein (vgl. ebd., S. 477 ff.). Sie verweisen u. a. auf das Forumtheater und die Transformation der Zuschauer_innen in die Protagonist_innen der Aktion. Partizipative Aktion und Forschung in ihren verschiedenen Ausprägungen stehen in Lateinamerika in enger Verbindung zu sozialen Bewegungen für soziale Gerechtigkeit. Dies wird auch von Fritz (2013, S. 205 ff.) bestätigt, die sich mit der PAR und der Creación Colectiva in Südamerika auseinandersetzt und Parallelen zum TdU herstellt (vgl. ebd., S. 219): Die PAR wird von impulsgebenden Akteur_innen und der Gemeinschaft durchgeführt, die ihre Realität verändern wollen, wobei die zu bearbeitenden Probleme gemeinsam definiert werden. Praktisch bedeutet dies etwa, dass eine Community eine Einladung an eine/n Forscher_in ausspricht, der/die sich in der Folge mit dem geschichtlichen Kontext des Ortes und der Menschen, mit denen er/sie arbeitet, auseinandersetzt. Die Auswahl der Anliegen und Forschungsmethoden wird gemeinsam entschieden und im Laufe der Forschung gibt es einen Methodentransfer in die einladende Community. Zudem werden alle Ergebnisse in mehreren Schleifen in den größeren Kontext der Gemeinschaft zurückgeführt, damit auf dieser Grundlage neue Schritte zur Verbesserung der Situation geplant werden können (vgl. ebd., S. 260). Aus der Sicht von Fritz (ebd., S. 260 f.) wird dieser Anspruch von der indischen Forumtheatergruppe Jana Sanskriti realisiert: Ausgangspunkte sind Forderungen und Notwendigkeiten der Betroffenen, mit denen gemeinsam über die Vorgangsweise entschieden wird. Gemeinsam wird die Situation analysiert und die Ergebnisse werden in die größere Community zurückgeführt, woraus weitere Vorgehensweisen entwickelt werden. „Der Prozess besteht aus Aktion und Reflexion. Die Beteiligten ,erobern‘ sich quasi die Situation, sie eignen sich durch den Prozess die Welt als ,Ding für uns‘ anstelle eines ,Dings an sich‘ an“ (ebd., S. 261). Für Ganguly, den Leiter von Jana Sanskriti, ist Forumtheater in diesem Sinn ein Ort, an dem „Sozialwissenschaft betrieben wird. Das Resultat ist, dass sich der Prozess der Stückerarbeitung zu einem Prozess entwickelt, in dem Macht in die SchauspielerInnen eingeschrieben wird. Sie fühlen sich der Veränderung ihrer Realität verpflichtet“ (Interview mit Ganguly in Fritz 2011, S. 171). Im deutschsprachigen Raum widmete sich das universitäre Forschungsprojekt zum „Brüchigen Habitus“ über mehrere Jahre der empirischen Erforschung kooperativer Handlungsmöglichkeiten von Studierenden und Hochschullehrer_innen (vgl. Bülow-Schramm und Gipser 1997, 2007; Gipser 1996, 2007). Dabei wurde in erster Linie das Forumtheater als Methode einer handlungsbezogenen und emanzipatorischen Forschung angewendet, um Machtbeziehungen und Herrschaftsstrukturen im Hochschulalltag zu analysieren und über das dramatische Probehandeln Handlungsalternativen zu finden, die in der Folge in den universitären Alltag übertragen wurden. Ausgehend vom Projekt zum „Brüchigen Habitus“ gibt es von Gipser forschungsmethodologische Überlegungen zum Forumtheater (vgl. Gipser 1996, S. 28 ff., 2007). Sie sieht in der „biografischen Selbstreflexion“ – gemeint als Aneignung der eigenen Biografie vor dem Hintergrund gesellschaftlicher Bedingungen durch Aktualisierung und Reflexion eigener vergangener Erfahrungen – einen wichtigen Anknüpfungspunkt, vor allem dort, wo es um die Aktivierung des K Forumtheater als Werkzeug szenisch-partizipativer Forschung in der Sozialen Arbeit.... 89 Körpergedächtnisses im szenischen Spiel geht. Bezüge werden auch zum sogenannten „soziologischen Experimentieren“ nach Brecht hergestellt (Gipser 1996, S. 29). Lehr- und Lernprozesse werden in diesem Sinn als „handlungs- und subjektorientierte Sozialforschungsprozesse“ (ebd.) verstanden, biografische Selbstreflexion und soziologisches Experimentieren sind in Hinblick auf Wirklichkeitsveränderung verknüpft. Forumtheater als emanzipatorischer Forschungs- und Bewusstwerdungsprozess basiert auf der Verknüpfung von Forschung und Aktion, Erkennen ist von praktischen und politischen Interessen geleitet und mit der Lebenspraxis verknüpft. Forschungsprozesse mit Mitteln des Forumtheaters speisen sich weniger aus rein akademischen Fragestellungen, sondern aus den Erkenntnisinteressen, Anliegen und Fragen der Mitwirkenden. Gipser (1996, S. 30) sieht darin Möglichkeiten, „Postulate einer emanzipatorischen Handlungsforschung praktisch umzusetzen“, zumal sich deren paradigmatische Elemente wie der dialogische Zugang zur Realität, die Wahrheitsfindung über die kritische Argumentation im Diskurs sowie das zyklische Modell von Datensammlung, Diskussion und Handeln im Feld mit szenischer Forschung gut vereinbaren ließen (vgl. Gipser 1996, S. 29 ff.). Ausgehend von diesen Bezügen des Forumtheaters zu partizipativen und emanzipatorischen Forschungskonzeptionen gehe ich im Folgenden auf Verfahrensweisen und Methoden ein, die bei Forumtheaterprojekten in Handlungs- und Problemfeldern der Sozialen Arbeit Anwendung finden. 4 Forumtheater in der Sozialen Arbeit: Verfahrensweisen, Methoden und Projektbeispiele Grundlage für die Beschreibung von methodischen Verfahrensweisen bilden Forumtheaterprojekte, die in Handlungs- und Problemfeldern Sozialer Arbeit realisiert wurden und in denen sich vielfältige Merkmale und Dimensionen von szenischen und partizipativen Forschungsprozessen zeigen. Die Projekte folgten einem methodischen Grundkonzept, das ich im Folgenden im Überblick vorstelle.1 In die Darstellung der Verfahrensweisen werden Beispiele aus Forumtheaterprojekten miteinbezogen (und kursiv hervorgehoben) und methodologische Bezüge hergestellt. Wesentlich für szenische Forschungsprozesse mit Forumtheater in Handlungsund Problemfeldern der Sozialen Arbeit sind die subjektiven, individuellen Wahrnehmungen und Erfahrungen der Mitwirkenden zu den grundlegenden Themen und Fragen eines Projekts. Zudem können theoretische und fachliche Bezüge (etwa in empirischen Studien zu einer bestimmten Problemlage) in der grundlegenden Frage, die sich ein Projekt stellt, und in der damit verbundenen Gesamtkonzeption eine Rolle spielen und in weiteren Phasen der Stückentwicklung einbezogen werden. In der Regel richten sich Zielsetzungen – wie Unger (2014, S. 53) in Bezug auf partizipative Forschung schreibt – „in erster Linie nach den Relevanzsetzungen der Betroffenen und dem praktischen Handlungsbedarf“, wobei sich das Thema und dessen 1 Eingehend beschrieben und diskutiert werden diese Verfahrensweisen in Abschnitt I, II und V meiner bei Beltz Juventa publizierten Habilitationsschrift (Wrentschur 2019). Dabei kommen besonders in Abschnitt V, Fallstudie 3, die Projektmitwirkenden selbst zu Wort. K 90 M. Wrentschur Fokussierung je nach Relevanz auch ändern können. So bilden thematische Interessen und Anliegen von bestimmten Communitys oder Gruppen, die mit dem Wunsch nach Veränderung verbunden sind, den Rahmen für ein Projekt. Oder es werden Menschen, die bestimmte Erfahrungen (z. B. mit Diskriminierung, Ausgrenzung, finanziellen Notlagen) teilen, eingeladen, an einem szenischen Forschungsprojekt mitzuwirken. Jedenfalls ist wesentlich, dass ein thematisches Interesse der Mitwirkenden besteht, das für die Mitwirkenden in ihren Lebenswelten und in ihrer Praxis bedeutsam ist. Darüber hinaus werden Kooperationen mit sozialen Bewegungen und Organisationen gesucht, die thematische Interessen und Anliegen nach Veränderung teilen. Und aus projektstrategischen Gründen wird überlegt, welche Gruppen und Einrichtungen (Behörden, Politik) eingebunden werden, die für die Umsetzung von Veränderungsanliegen und -vorschlägen relevant sind. 4.1 Phase 1: Stückentwicklung als emanzipatorischer und partizipativer Gestaltungs-, Recherche- und Erkenntnisprozess Szenisches Forschen mit dem Forumtheater wird in Gruppen praktiziert, was in der Regel mit einer hohen Präsenzzeit aller Mitwirkenden verbunden ist. Damit die Teilnahme an den sich oft auch über mehrere Tage, Monate, manchmal sogar Jahre währenden Forumtheaterprojekten möglich ist, müssen die zeitlichen Rahmenbedingungen für die Mitwirkenden lebensweltlich angepasst und mit finanziellen Abgeltungen verbunden werden. So steht am Beginn eines Projekts oft ein mehrtägiger Workshop, bei dem zunächst versucht wird, eine vertrauens- und respektvolle, wertschätzende und kooperativen Atmosphäre in der jeweiligen Projektgruppe herzustellen. Dabei kommen angeleitete Körper-, Bewegungs- und Improvisationsübungen zur Anwendung, über die auch Spiellust und Kreativität angeregt werden. Dies ist eine wesentliche Voraussetzung dafür, dass die Teilnehmer_innen in der Folge Mut fassen, sich den für sie oft belastenden Themen und Problemfeldern zu stellen und diese zu artikulieren. Ähnlich den Prämissen einer kritischen Adressat_innenforschung (Anhorn und Stehr 2012; Hanses 2005; Graßhoff 2013) und der partizipativen Forschung (Unger 2014; Bergold und Thomas 2012) ist bedeutsam, dass sich die Mitwirkenden als Subjekte mit ihren Erfahrungen ernst nehmen und sich individuell und kollektiv Stimme und Ausdruck verschaffen. Dabei wird auch der Körper als wesentliche Quelle von Wissen und Erkenntnis betrachtet, d. h. körperliche Wahrnehmungen, Bedeutungen, Haltungen und Wertungen eines Themas werden in den Forschungsprozess miteinbezogen. Das kann sich auf das Anregen der Körpererinnerung, auf das Bewusstmachen von routinisierten, einverleibten Körperhaltungen oder auf das Aufspüren von gesellschaftlich mitproduzierten Bewegungsmustern sowie auf die nonverbalen und symbolhaften Ausdrucksmöglichkeiten und Dimensionen eines Themas beziehen. So werden die subjektiven Erfahrungen und Perspektiven der Mitwirkenden besonders über nonverbale, körperbezogene Ausdrucksformen wie den vielfältigen Bildertheater-Techniken artikuliert, bei denen soziale Situationen, Erfahrungen, Geschichten und Probleme der Teilnehmer_innen mithilfe von (eingefrorenen) Körperhaltungen, Gestik und Mimik dargestellt werden (vgl. Wrentschur 2008a, S. 103). „Image and imagination thus become an interplay of structure and de-structure, the image providing a form K Forumtheater als Werkzeug szenisch-partizipativer Forschung in der Sozialen Arbeit.... 91 of closure to play with; the imagination providing a way of opening up the form to possibility“ (Linds und Vettraino 2008, Abs. 7). Diese methodische Vorgangsweise unterstützt die Teilnehmenden dabei, Momente und Erfahrungen aus ihrem Alltag, ihrer Lebenswelt zum Ausdruck zu bringen, ästhetisch zu verdichten, zu verfremden, zu betrachten und kreativ zu bearbeiten. Oft erkennen Projektmitwirkende Aspekte ihrer eigenen lebensweltlichen Erfahrungen in theatralischen Bildern, Szenen und Geschichten, die von anderen Teilnehmer_innen herrühren, wieder. Zudem werden die symbolisch verdichteten Bilder und deren gemeinsame Analyse zu einem bedeutsamen Impuls für Austausch und Diskussionen unter den Mitwirkenden über die gezeigten und erlebten Situationen und Problemlagen. Im Zuge des Projekts „Legislatives Theater mit wohnungslosen Menschen“ (Wrentschur 2008a) wurde ausgehend von bewusst provokant gestellten Fragen wie „Was empört euch in Bezug auf Wohnungslosigkeit? Was findet ihr ungerecht? Was müsste am dringendsten verändert werden?“ die Gruppe dazu angeregt, im Sinne des Bildertheaters „lebende Skulpturen“ zu entwickeln und darzustellen. Der Wechsel zwischen dem szenischen Gestalten, dem Beobachten, dem Reflektieren und Diskutieren war eine entscheidende Triebkraft dafür, dass die oftmals mit Gefühlen von Ohnmacht, Scham und Hilflosigkeit verbundenen Erfahrungen von Wohnungslosigkeit artikuliert werden konnten. Dabei wurden besonders Situationen zum Ausdruck gebracht, in denen sich wohnungslose Menschen mit Ausgrenzung, Diskriminierung oder Respektlosigkeit konfrontiert sahen, oder in denen es als Herausforderung erlebt wurde, sich im System der angebotenen (und verwehrten) sozialen Dienste und Unterstützungsleistungen zurechtzufinden. Die Mitwirkenden bringen ihre persönliche Ansicht in die gemeinsame Auseinandersetzung ein und geben Impulse für Verständigungs-, Interpretations- und Assoziationsprozesse. Dokumentiert wird dies schriftlich in Form von Protokollen mit Beschreibungen und Assoziationen zu den einzelnen Bildern, oft auch in Stichworten auf Plakaten in Kombination mit Fotos. Dies gilt vor allem als Erinnerungsstütze und in weiterer Folge als Grundlage für weitere Entscheidungen der thematischen Vertiefung und szenischen Gestaltung. Im Sinne von „Daten“ sind jedoch die entwickelten Bilder und Szenen vorrangig. Davon ausgehend wird verstärkt nach den verbindenden, gemeinsamen Problem- und Konfliktlagen gesucht, es wird auf intersubjektive Themen und auf die damit verbundenen (gemeinsamen) Anliegen nach Veränderung fokussiert. Dafür bieten sich unterschiedliche methodische Möglichkeiten an, wie etwa das sogenannte Bild der Bilder, in dem Aspekte und Dimensionen bereits zum Ausdruck gebrachter theatralischer Bilder verbunden und analysiert werden (vgl. Boal 1999, S. 74 ff.) oder auch das Magnetische Bild (vgl. Diamond 2013, S. 152 ff.), bei dem Projektmitwirkende übergreifende Themen ihrer gemeinsamen Erfahrungen zum Ausdruck bringen. Zur Relevanzsetzung und Priorisierung übergreifender Themen können – über die theatralische Arbeit und den Erfahrungsraum der Gruppe hinaus – Recherchen in Form von (aktivierenden) Befragungen, aber auch andere Recherche- und Forschungsmethoden wie Sozialraumbegehungen oder Medienanalysen Anwendung finden. So wurden beim Projekt „Legislatives Theater mit wohnungslosen Menschen“ Rückkopplungsschleifen zur Community wohnungsloser Menschen in Graz gebildet (vgl. Wrentschur 2008a, S. 102 f.). Die Gruppenmitglieder betätigten sich als Sozi- K 92 M. Wrentschur alforscher_innen und führten kurze Fragebogeninterviews mit wohnungslosen Menschen durch, um zu erfahren, welche Probleme und Anliegen von den wohnungslosen Menschen als besonders dringend angesehen werden. Als selbst Betroffene konnten die Gruppenmitglieder das Vertrauen ihrer ,Kolleg_innen‘ gewinnen, die Vorarbeiten innerhalb der Projektgruppe kamen ihnen dabei zugute. Die Ergebnisse dieser Befragung bestärkten die Gruppe in ihrer Themenwahl, sie wurden zu einer wertvollen Stütze und Rückbindung für die weitere Entwicklung der Szenen und Bilder. Die Themen, Problemlagen und Anliegen wurden schließlich nach ihrer Dringlichkeit gereiht, wobei Schwierigkeiten und Diskriminierungen bei der Arbeits- und Wohnungssuche, bürokratische Barrieren und mangelnder Zugang zu relevanten Informationen (das „Herumgeschickt-Werden“) sowie die Verschuldensdynamik zu den wichtigsten zählten. Schließlich werden aus dem gemeinsam entwickelten, szenischen Rohmaterial eine oder mehrere ästhetisch-theatralische Szenen oder Bilder gestaltet. Dabei soll deutlich werden, welches Problem im Fokus steht und welche Fragen in den jeweiligen Szenen bzw. mit dem Stück aufgeworfen werden. Parallel dazu spielen ästhetische und szenische (Verdichtungs-)Prozesse eine große Rolle. Methodisch kommen dabei zumeist Formen der Improvisation mit anschließenden Reflexionen sowie spezielle Probetechniken zur Fokussierung und Gestaltung der szenischen Abläufe zur Anwendung. Folgende Fragen stehen im Zentrum: Was braucht es an Handlungen, an Gefühlen und räumlichen Arrangements, an Gesten und Haltungen, damit die Szene stimmiger und realer wird, damit die in den Situationen wirkenden Macht- und Statusbeziehungen deutlich und den von der Gruppe intendierten Problemen und Anliegen gerecht werden? Sind die sozialen Rollen und deren Habitus bzw. Haltungen entsprechend dargestellt? Sind die Szenen entsprechend vielschichtig und komplex? In welcher Weise berühren die Szenen die gemeinsamen Erfahrungen und/ oder Anliegen der Mitwirkenden? Neben dem ästhetischen Entwicklungs- und Gestaltungsprozess ist in dieser Phase vor allem die Rollenarbeit sehr bedeutsam, um die Darsteller_innen gut auf die interaktive Forumphase vorzubereiten, bei der diese aus dem Wissen und Habitus einer Rolle heraus auf Veränderungsideen der Zuschauer_innen reagieren. Unterstützt wird dies mit Recherchen zu Fakten, Gesetzen, Abläufen und Handlungsspielräumen der beteiligten Rollen. Beim Forumtheaterprojekt „Kein Kies zum Kurven Kratzen“ (Wrentschur 2019, S. 787) führte ein Darsteller, der selbst mit finanziellen Notlagen zu kämpfen hatte, Interviews mit Gerichtsvollzieher_innen durch, um für die Darstellung seiner Rolle im Stück Informationen über gesetzliche Vorgaben und Handlungsspielräume zu bekommen. Und beim gleichen Projekt setzte sich eine Akteurin, die das AMS viele Jahre lang aus der Perspektive einer arbeitslosen Person erlebt hatte und im Forumtheaterstück eine AMS-Beraterin spielte, mit den Perspektiven und Handlungsräumen von diesen auseinander. Hier fand schon bei der Recherche ein Rollen- und Statuswechsel statt: Die Rolle als Darstellerin und Forscherin eröffnete nicht nur Zugänge zu Informationen, sondern erlaubte Gespräche auf Augenhöhe. Ein weiteres methodisches Element in dieser Phase bildet eine Form der dialogischen Rückkopplung und Validierung, indem Menschen, die aufgrund ihrer Betroffenheit eine persönliche oder eine professionelle Expertise zu den erarbeiteten K Forumtheater als Werkzeug szenisch-partizipativer Forschung in der Sozialen Arbeit.... 93 Themen aufweisen, zu offenen Proben eingeladen werden. Dabei werden szenische Entwürfe präsentiert, um gemeinsam zu reflektieren und zu diskutieren, inwieweit sie den Erfahrungen und Realitäten der Eingeladenen entsprechen. Im Sinne eines Wissens- und Informationstransfers werden zudem szenische Improvisationen zum Kontext und Inhalt der Szene durchgeführt, an denen die eingeladenen Expert_innen unmittelbar mitwirken. Diese offenen Proben dienen dem Austausch relevanter Informationen zum Thema, die einen wichtigen Hintergrund für die Szenen bilden. Dies kann zu weiteren Recherchen zu fachlichen, gesetzlichen oder politischen Hintergründen anregen, die für die jeweiligen Szenen relevant sind. Dieser szenische Forschungsprozess führt in ein von Text und Ablauf her fixiertes Forumtheaterstück, das ästhetisch gestaltet und verdichtet Erkenntnisse und Phänomene des bisherigen Gestaltungs- und Reflexionsprozesses enthält. Die Mitwirkenden eines Forumtheaterprojekts sind in allen bisherigen Phasen des szenischen Forschens mit dem Forumtheater aktive Subjekte und Akteur_innen der Forschung, die ihre individuellen Erfahrungen und Sichtweisen zu gemeinsam gestalteten, kollektiven theatralischen Bildern und Szenen verbinden. In den szenischen Verläufen werden individuelle, alltägliche biografische Verläufe und Handlungsverläufe in ihrem dialektischen Wechsel- und Spannungsverhältnis zu gesellschaftlichen und strukturellen Rahmenbedingungen und Kontexten dargestellt. Diese szenisch verdichteten Erkenntnisse werden in der Folge einem Publikum vermittelt, verbunden mit einer oder mehreren Frage(n), die bei den interaktiven Aufführungen bearbeitet werden. Dies setzt gemeinsame Klärungen darüber voraus, was die Projektmitwirkenden öffentlich zeigen wollen – Text und Handlungsablauf bedürfen der Zustimmung aller. 4.2 Phase 2: Interaktive Aufführungen als kollektive und partizipative Forschungsforen Mit den interaktiven Aufführungen wird die in offenen Proben realisierte dialogische Rückkopplung intensiviert, indem vor allem betroffene Communitys, einschlägige Organisationen und Initiativen sowie gesellschaftliche Öffentlichkeiten angesprochen und einbezogen werden. Über eine Aufführungsserie hinweg kann sich auf diesem Weg in umfassenderer Weise herauskristallisieren, inwieweit die dargestellten Szenen als Ergebnisse der szenischen Forschungsprozesse kontextualisierbar und über die Erfahrungen und Erkenntnisse der Projektmitwirkenden hinaus generalisierbar sind, gerade auch, weil es dabei nicht um eine Reproduktion sozialer Wirklichkeit geht, sondern um deren szenische und ästhetische Verdichtung. Forumtheateraufführungen erweisen sich als eine Form einer intervenierenden und kommunikativen Feldforschung, bei der regionale Eindrücke, Befunde und Erkenntnisse zur Wahrnehmung und zum Umgang mit den gezeigten sozialen Problemen aufgenommen und analysiert werden können. Dieser Erkenntnisprozess kann in der Folge unter Anwendung anderer Methoden (regional) vertieft werden (vgl. Wrentschur 2008b, S. 717 f.). Szenisches Forschen mit dem Forumtheater ist in der interaktiven Forumphase von einem spielerisch-experimentellen Charakter geprägt. In dieser Phase wird gemeinsam mit dem Publikum nach Ideen zur Veränderung bzw. zu Lösungsansätzen für die gezeigten Probleme und Konflikte gesucht, die auch hinsichtlich ihrer Über- K 94 M. Wrentschur tragbarkeit reflektiert werden: Szenen werden unterbrochen und in neuen Varianten gespielt; es wird zwischen der unmittelbaren Beteiligung im Spiel, der Beobachtung und der Reflexion gewechselt. Die Forumphase gleicht dem angesprochenen soziologischen Experimentieren mit Haltungen, Verhaltensweisen und Handlungen (vgl. Koch 1988, S. 45), indem versucht wird, dem Geschehen auf der Bühne eine Wendung zu geben, während die Darsteller_innen aus der habituellen Perspektive der jeweiligen Rolle darauf reagieren. Dadurch werden mögliche Folgen des Handelns unmittelbar sichtbar und erlebbar gemacht, genauso wie jene habituellen, soziokulturellen und strukturellen Barrieren bzw. Grenzen, auf die individuelles Handeln stoßen kann (Wrentschur 2019, S. 416 f.). In zahlreichen interaktiven Forumtheateraufführungen wurde beim bereits erwähnten Projekt „Legislatives Theater mit wohnungslosen Menschen“ (Wrentschur 2008a) das Publikum, das sich aus wohnungslosen Menschen, in sozialen Einrichtungen Tätigen und Interessierten zusammensetzte, dazu eingeladen, sich im Sinne des Forumtheaters am Spielgeschehen zu beteiligen, die Rolle des oder der Wohnungslosen zu ersetzen und Veränderungsideen auszuprobieren. Das Erkenntnisinteresse galt vor allem der Frage, wie es möglich ist, aus dem nach unten führenden Kreislauf auszusteigen und auf diskriminierende und ausgrenzende Praktiken zu reagieren. Die „Einstiege“ in die Forumtheaterszenen loteten Handlungsspielräume der Rollen in den jeweiligen Szenen aus und ließen erkennen, welche institutionellen oder gesetzlichen Veränderungen notwendig wären, um die Reintegration und Partizipation von wohnungslosen Menschen zu erleichtern. Als große Hürde stellten sich dabei die eklatanten Macht- und Statusunterschiede heraus, mit denen wohnungslose Menschen konfrontiert sind. Vor dem Hintergrund des zumeist geringen sozialen, ökonomischen, kulturellen und symbolischen Kapitals, auf das wohnungslose Menschen zurückgreifen können, wurde die Suche nach Möglichkeiten angeregt, den dargestellten Machtungleichgewichten zumindest insofern zu begegnen, als wohnungslose Menschen in den für sie existenziell bedeutsamen Situationen als Dialogpartner_innen ernst genommen werden. Dieser Prozess machte beschränkende oder beherrschende habituelle Praktiken bewusst, denen wohnungslose Menschen selbst unterliegen bzw. mit denen sie konfrontiert sind. Es zeigte sich das für die individuelle Handlungsebene interessante und zugleich widersprüchliche Ergebnis, dass es durchaus Wege gab, um aus der Sicht wohnungsloser Menschen schwierige Situationen zu bewältigen oder zu wichtigen Auskünften und Unterstützungsleistungen zu gelangen. Dafür schienen etwa selbstbewusstes oder hartnäckiges Auftreten, Freundlichkeit und Humor, das Deutlichmachen von Notsituationen, das Aushandeln von Lösungen sowie das Umdeuten von Situationen notwendig. Die meisten dieser Eigenschaften und Verhaltensweisen stehen aber im Widerspruch dazu, dass akut wohnungslose Menschen in der Regel zumeist über Scham- und Versagensgefühle und über wenig Selbstwert und Selbstbewusstsein verfügen, was oftmals destruktive oder aggressive Verhaltensweisen befördert. Insofern bestand eine wesentliche Erkenntnis darin, dass die Forderung nach Veränderung der individuellen Haltungen der Betroffenen zu kurz greift, wenn soziale Unterstützungssysteme und Behörden ihrerseits Barrieren aufbauen und damit Ausgrenzungspraktiken vollziehen, welche die rechtlichen Ansprüche wohnungsloser Menschen auf Information und Hilfestellungen konterkarieren (vgl. Wrentschur 2008a, S. 106). K Forumtheater als Werkzeug szenisch-partizipativer Forschung in der Sozialen Arbeit.... 95 Der Erprobung von Handlungsalternativen bei den Einstiegen folgt eine gemeinsame Reflexion, indem Einsteiger_innen, Zuschauer_innen und manchmal auch Darsteller_innen ihre Wahrnehmungen und Ansichten artikulieren. Die spielerischexperimentelle Auseinandersetzung mit theatralisch verdichteten Lebens- und Alltagswelten trägt zu einem besseren Verstehen der Handlungen der einzelnen Akteur_innen, der gezeigten sozialen Probleme und ihrer Auswirkungen bei: Durch Einstiege können bislang verborgene, auch tabuisierte oder zunächst unbewusste Facetten eines Themas zutage treten und dessen Komplexität und Vielschichtigkeit deutlich werden. Entlang der zentralen Frage, ob eher individuelle, handlungsbezogene und/oder eben eher strukturbezogene, politische Herangehensweisen notwendig sind, um Veränderungen zu initiieren und zu realisieren, verstehen sich Forumtheateraufführungen als handlungs- und dialogorientierte Forschungsprozesse. Wie bei Ansätzen der performativen Ethnografie schaffen die Forumtheateraufführungen einen gemeinsamen Erfahrungsraum, in dem zwischen Darsteller_innen und Publikum Erlebnisse, Emotionen und Verständnisse geweckt, entfaltet und geteilt werden können. Das Publikum wird zum Ko-Produzenten der Forschung und der Geschichten, die weitergeschrieben, umgedeutet, neu geschrieben werden. Damit wird ein Raum für Reflexion, Diskussion und Dialog eröffnet, der zu einem Wandel von Einstellungen und Perspektiven führen kann, indem übernommene Bedeutungen und normative Traditionen über Performance aufgebrochen, in Bewegung gebracht, hegemoniale Strukturen infrage gestellt und demokratische, partizipatorische sowie emanzipatorische Perspektiven unterstützt werden. „Performance wird auf diese Weise zu einem „umfassenden und integrierenden Konzept, um kulturelle und soziale Prozesse verstehen und verändern zu können“ (Winter 2010, Abs. 36). Bei Projekten, die stärker auf die Veränderung politischer Rahmenbedingungen abzielen, wird das Publikum auch dazu eingeladen, aus den Erfahrungen der interaktiven Aufführungsprozesse heraus strukturell-politische Anliegen und Vorschläge zu formulieren, die sich aus der Auseinandersetzung mit dem Stück ergeben und die Suche nach eher individuellen Handlungsmöglichkeiten ergänzen. Alle Einstiege, Diskussionsbeiträge und Vorschläge, die im Verlauf einer Aufführung entstehen, werden jedenfalls schriftlich, manches Mal auch audiovisuell über Filmmitschnitte dokumentiert. 4.3 Phase 3: Auswertung und Transfer Schon während der Forumtheateraufführungen werden in Zusammenhang mit den Einstiegen des Publikums gemeinsam Handlungsmöglichkeiten reflektiert. Dadurch kann es zu vielfältigen Erkenntnissen kommen, die mit dem unmittelbaren und oft auch körperlichen Erleben der Aufführungen in Zusammenhang stehen. Diese Erkenntnisarbeit setzt sich im Projektteam aber über die Aufführungen hinaus fort und regt zur Frage an, wie die gewonnenen Erkenntnisse in die jeweiligen Lebensrealitäten übertragen werden können bzw. welche Lösungsvorschläge und Handlungsansätze auf persönlicher, gesellschaftlicher und politischer Ebene damit verbunden sind. Grundlage dafür sind die inhaltsanalytische Auswertung und Strukturierung der in den Protokollen dokumentierten Einstiege, Lösungsvorschläge und Diskussi- K 96 M. Wrentschur onsbeiträge des Publikums, die in der Gruppe der Projektmitwirkenden hinsichtlich zweier grundlegender Perspektiven reflektiert und diskutiert werden: Zum einen wird untersucht, ob sich in den über die Einstiege des Publikums zum Ausdruck gebrachten Lösungsideen und Handlungsalternativen Ansätze zeigen, die für die Mitwirkenden am Projekt als sinnvoll, bedeutsam und in ihren Lebensrealitäten als realisierbar erscheinen. Dazu werden in der Gruppe der Projektmitwirkenden die Einstiege und ihre Wirkungen gemeinsam besprochen und reflektiert, und es wird überlegt, wie Haltungen und Handlungsweisen, die im szenischen Spiel als wirksam und stimmig erlebt werden, in der Folge im Alltag bzw. in der gesellschaftlichen Praxis auf die Probe gestellt werden. Handlungsmöglichkeiten beziehen sich dabei zumeist auf Situationen und Dynamiken von Ausgrenzung, Unterdrückung und Ohnmacht, sie können persönliche, soziale, berufliche und politische Kontexte betreffen. Über individuelle Handlungsansätze hinaus können zudem gemeinsame, kollektive Handlungsformen angedacht und realisiert werden. Zum anderen geht es um die Frage, ob in den Einstiegen und formulierten Lösungsvorschlägen generelle, verbindende Vorschläge und Lösungsansätze gefunden werden können und daraus ein gesellschaftlicher, politischer oder behördlicher Handlungs- bzw. Veränderungsbedarf artikuliert werden kann. Auf dieser Grundlage werden Vorschläge, Lösungsansätze, Anliegen und Forderungen von der Gruppe der Projektmitwirkenden formuliert, die zur Verbesserung bzw. Veränderung einer bestimmten sozialen oder politischen Problematik beitragen können. Diese Vorschläge, Lösungsideen und Forderungen können anschließend mit weiteren betroffenen und professionellen Expert_innen hinsichtlich ihrer Sinnhaftigkeit, ihres Beitrags zur Lösung der thematisierten Probleme reflektiert und diskutiert werden. Dabei ist die Frage wesentlich, ob die Vorschläge und Lösungsansätze Betroffene in einer ähnlichen Lage unterstützen können, ihre Situation besser zu bewältigen. In einem weiteren Schritt können die Vorschläge und Ideen gemeinsam mit professionellen Expert_innen dahingehend überprüft werden, welche politischen, behördlichen oder rechtlichen Kontexte davon berührt werden. Nach einer nochmaligen Überarbeitung der Vorschläge und Texte – wieder unter Miteinbeziehung von betroffenen und fachlichen Expert_innen – werden diese für den Transfer an unterschiedliche Entscheidungsträger_innen aufbereitet. Diese partizipative Produktion eines Berichts mit Vorschlägen und Forderungen ist Ergebnis eines kollektiven Reflexions- und Klärungsprozesses, der die Erfahrungen und Sichtweisen von Menschen ernst nimmt, die eine Problematik aus der eigenen Lebenswelt kennen und die sich in einem länger währenden Forschungs- und Bewusstseinsprozess über ihre Anliegen und Veränderungsvorschläge klargeworden sind. Unter Zuhilfenahme der jeweiligen Forumtheaterstücke kann in der Folge in unterschiedliche behördliche und politische Kontexte interveniert werden, um die Vorschläge, Anliegen und Lösungsideen mit Entscheidungs- und Verantwortungsträger_innen in Bezirken, Kommunen, Ländern oder im Parlament zu diskutieren (vgl. etwa Wrentschur 2008a, 2008b, 2010, 2011, 2012; Wrentschur und Moser 2014). Die Forumtheaterstücke haben dabei nicht nur die Funktion, ästhetisch verdichtete Ergebnisse des szenischen Forschungsprozesses an Entscheidungs- und Verantwortungsträger_innen zu vermitteln, sondern dadurch die bei den bisherigen Aufführungs- und Reflexionsprozessen entwickelten Lösungs- und Veränderungsideen ver- K Forumtheater als Werkzeug szenisch-partizipativer Forschung in der Sozialen Arbeit.... 97 ständlich und nachvollziehbar zu machen. Dabei spielt besonders die Frage eine Rolle, wie die Vorschläge und Anliegen von den jeweiligen Verantwortungsträger_innen umgesetzt werden können. Das Projekt „ZusammenSpiel“ konzentrierte sich auf das Zusammenleben in einem Grazer Park und versuchte, Konflikte zwischen Benutzer_innengruppen in Szene zu setzen und partizipativ zu verhandeln (vgl. Wrentschur 2019, S. 615 ff.). Bei Interviews mit Expert_innen, Sozialraumanalysen und bei offenen Proben im Park zeigte sich, dass der Erholungswert im Fröbelpark im Konflikt mit dem Fröbelpark als Spielplatz steht. Da es in diesem Stadtteil sonst keine öffentlichen Spielplätze gibt, wird der Raum für die spielenden Kinder enger, die aufgrund der Wohnsituation oft viel Zeit im Freien verbringen (müssen). Diese räumliche Enge sorgt für Konflikte zwischen Fußball spielenden Burschen und Mädchen (die Burschen sind mit ihrem Fußballspiel sehr platz- und raumgreifend, die Mädchen fühlen sich davon sehr eingeschränkt und gestört), für Konflikte mit Anrainer_innen (Lärm, Bälle, die gegen Autos, Zäune etc. fliegen), aber auch mit Eltern, die mit ihren Kleinkindern im Park sind (die Bälle fliegen oft gefährlich knapp an Kinderköpfen vorbei, manchmal auch darauf ...). Die Darsteller_innen des Forumtheaters waren ortskundige Jugendliche, gemeinsam mit dem anwesenden Publikum (Parkbenutzer_innen, Eltern, Geschwister, angrenzende Bewohner_innen) wurde im interaktiven Forumtheaterprozess nach Lösungsideen und neuen Handlungsmöglichkeiten gesucht. Unmittelbar im Anschluss an die zweite Aufführung fand eine Diskussion statt, an der neben den jugendlichen Projektmitwirkenden, Vertreter_innen der kommunalen Politik sowie von umliegenden Schulen und Jugendeinrichtungen teilnahmen. Allen Beteiligten war klar, dass mehr freie, bespielbare Flächen im öffentlichen Raum benötigt werden, was bei der Stadtplanung und den Bebauungsplänen aber nicht berücksichtigt wird. Als ein Ergebnis der Diskussion nach der Aufführung wurde eine schnelle, kurzfristige Lösung für die Sommerferien ermöglicht: Im Sinne der Mehrfachnutzung vorhandener Sportplätze wurde der Sportplatz einer angrenzenden Schule wochentags von 13–18 Uhr geöffnet und entsprechend betreut. Ein Konzept, das schon lange in den Schubladen gelegen ist, wurde als Folge des partizipativen Theaterprozesses und des Erlebens der Forumtheateraufführung Wirklichkeit und in der Folge zu einem Modellfall für die gesamte Stadt Graz. 5 Methodische Reflexionen und Ausblicke Forumtheater lässt sich in unterschiedlicher Akzentuierung als Forschungsprozess gestalten, wobei emanzipatorische und partizipative Gestaltungs-, Recherche- und Erkenntnisprozesse bei der Stückentwicklung, die interaktiven Aufführungen als kollektive und partizipative Forschungsforen und schließlich die Auswertung und der Transfer ins Soziale und Politische als wesentliche Phasen beschrieben werden können. Neben erfahrungs- und handlungsbezogenen Erkenntnissen im szenischen Forschungsprozess sind Forumtheaterszenen und -stücke, die in ästhetisch verdichteter Weise Erkenntnisse über das Zusammenspiel von sozialen Handlungen, Dynamiken und Strukturen beinhalten, ein zentrales Ergebnis dieses Prozesses. Die Theatralisierung von Erfahrungen aus dem sozialen Alltag und von sozialer Wirklichkeit K 98 M. Wrentschur generiert Wissen und Erkenntnisse, ebenso wie die gemeinsame Stückentwicklung, die damit verbundene Rollenarbeit und die interaktiven Aufführungen. Zudem werden Berichte zum Projektprozess und zu inhaltlichen Ergebnissen auf Basis der dokumentierten, ausgewerteten und rückgekoppelten Lösungs- und Veränderungsvorschläge verfasst. Ähnlich der partizipativen Forschung und der ethnografischen Forschung im Kontext einer kritischen qualitativen Forschung versteht sich szenisches Forschen mit dem Forumtheater als Auslöser für sozialen Wandel sowie als eine in soziale und politische Räume eingreifende Forschungsstrategie. Die Reflexions- und Erkenntnisprozesse beim Forumtheater sind vom Wunsch und von der Suche nach Veränderungsmöglichkeiten hinsichtlich der Überwindung von Unterdrückung, Ausgrenzung und Diskriminierung in Richtung von mehr sozialer Gerechtigkeit, Teilhabe, Partizipation und Empowerment geleitet. Insofern ist es ein Ziel, in starkem Maße Veränderungswissen zu produzieren, das zu den lebensweltlichen Bedingungen und Lebenslagen der Projektmitwirkenden rückgebunden ist. So stehen die Forschungsprozesse in starkem Zusammenhang mit alltäglichen Erfahrungen in der Lebenswelt der Teilnehmer_innen, die mit theatralen Werkzeugen artikuliert, verdichtet, verfremdet und transformiert werden, um damit Problemlagen und generative Themen spezifischer Gruppen und Communitys bewusst zu machen. Diese gemeinsame Untersuchung lebensweltlicher Verhältnisse macht es möglich, Konfliktverhältnisse im Alltag zu rekonstruieren und diese „in ihrer Dialektik als Kritik und Handeln, Zwang und Freiheit wie auch Unterdrückung und Emanzipation zu beschreiben“ (Schimpf 2012, S. 258), was sich nicht nur mit den Zielen der kritischen Adressat_innenforschung, sondern auch mit der performativen Ethnografie und Ansätzen der partizipativen Forschung verbinden lässt. Dabei ist wesentlich, welchen Nutzen und welchen Gebrauchswert die Mitwirkenden aus den Erkenntnissen für ihr Handeln und ihre Handlungsspielräume schöpfen – im Sinne der Erweiterung des Wissens und der gemeinsamen Handlungsfähigkeit aller Beteiligten. Zentral ist das Kennenlernen neuer Perspektiven und Handlungsmöglichkeiten auf persönlicher, sozialer und politischer Ebene, gerade auch dort, wo es um soziale Probleme und Konflikte, um Prozesse von Ausgrenzung, Macht und Unterdrückung geht. Der szenische Forschungsprozess ist maßgeblich von Menschen getragen, deren Stimmen und Perspektiven nur selten einbezogen werden „und die selbst kaum Möglichkeiten haben, ihre Interessen begründet einzubringen und durchzusetzen“ (Bergold und Thomas 2012, S. 19), die aber von den Umständen, unter denen sie leben bzw. denen sie ausgesetzt sind, am meisten betroffen sind. Forumtheater eröffnet gerade gesellschaftlich ausgegrenzten oder marginalisierten Gruppen die Möglichkeiten zur Artikulation ihrer Stimmen und Sichtweisen. In der jeweiligen (sozialen) Standpunktperspektive kann ein Wissen elaboriert und artikuliert werden, das die dominanten Diskurse und Praktiken umgeht, indem die Welt aus der Perspektive der Betroffenen gezeigt wird und Strukturen von Unterdrückung dargestellt werden. Über vielschichtige und vielstimmige Aufführungstexte werden soziale Probleme und deren Komplexität in der Verbindung von Alltagshandeln und gesellschaftlichpolitischer Struktur in einer Weise zum Ausdruck gebracht, die viele Menschen weit über akademische Kontexte hinaus erreichen und beteiligen kann. Die Emotionalität und Vielschichtigkeit in den dargestellten Forumtheaterszenen regen dazu an, K Forumtheater als Werkzeug szenisch-partizipativer Forschung in der Sozialen Arbeit.... 99 Verbindungen zu eigenen lebensweltlichen Erfahrungen herzustellen, die Wissensund Erkenntnisproduktion realisiert sich nicht nur über Texte, sondern auch über Handlungen, Haltungen, Gestik und emotionales Erleben. Eine derart verstandene aufführungsorientierte Sozialwissenschaft wird durch eine kritische soziologische Imagination geprägt, die eine radikale Demokratisierung der Lebensverhältnisse anstrebt. „Dabei ist es Aufgabe einer kritisch orientierten Sozialwissenschaft, Erfahrung, Politik, Performativität und Ermächtigung miteinander zu verknüpfen“ (Winter 2010, Abs. 40). Dies alles erweist sich insgesamt als voraussetzungsvoll und ist mit speziellen Herausforderungen und Spannungsfeldern verbunden. So liegt in der Stärke dieses Ansatzes zugleich eine Schwierigkeit: Erst die notwendige hohe Präsenzzeit der Projektmitwirkenden eröffnet vielfältige Reflexions- und Erkenntnismöglichkeiten und macht die Entwicklung von qualitätsvollen Forumtheaterstücken möglich. Die mit der Teilnahme einhergehende Involviertheit und intensive Auseinandersetzung mit belastenden, oft auch schmerzhaften Themen kann als sehr herausfordernd und anstrengend erlebt werden. Umso wichtiger sind transparente und unterstützende Rahmenbedingungen, eine auf Wertschätzung und Respekt basierende Zusammenarbeit und die mit der Theaterarbeit verbundenen Räume für Kreativität, Ausdruck und gemeinsames Handeln. Herausfordernd ist zudem, dass inhaltlich und ästhetisch überzeugende Forumtheaterstücke notwendige, aber nicht hinreichende Voraussetzung für deren soziale und politische Wirkung sind. Vielmehr ist bei den jeweiligen Projekten viel an Vernetzungs- und Öffentlichkeitsarbeit zu leisten, die von Projektbeginn an mitgedacht werden muss. Nur so können über die Aufführungen viele Menschen aus gesellschaftlichen, professionellen, behördlichen oder politischen Kontexten angesprochen, erreicht und am Dialog über Lösungsansätze beteiligt werden. So waren beim Projekt „ZusammenSpiel“ die Recherchen und Gespräche mit relevanten Akteur_innen im Stadtteil nicht nur hinsichtlich deren Sichtweisen und Einschätzungen bedeutsam, sondern auch, um diese in ihrer Kooperations- und Veränderungsbereitschaft anzusprechen. Abgesehen davon müssen in Bezug auf die Veränderungsanliegen und -vorschläge wiederholt Rückkopplungen zum Projektteam, zu relevanten Communitys, Initiativen und Menschen mit einschlägiger Expertise erfolgen. Dies alles ist eine Grundlage für politische Folgehandlungen, die aber außerhalb der Gestaltungsmöglichkeiten und des Machtbereichs eines Projekts liegen. So hat sich im Zuge von Forumtheaterprojekten immer wieder herausgestellt, dass die „Stimmen der Adressat_innen“ zwar beim unmittelbaren Erleben der Aufführungen Gehör finden und bei politischen Entscheidungsträger_innen Verständnis und Umsetzungsbereitschaft für die Veränderungsvorschläge geweckt wird, dass diese aber aufgrund der Dynamiken innerhalb der jeweiligen politischen Gremien nicht realisiert werden (vgl. Wrentschur 2014). Zur weiterführenden und vertiefenden methodischen und methodologischen Entwicklung wird die Realisierung weiterer Projekte, deren Reflexion in Publikationen und innerhalb und außerhalb der Community der szenisch Forschenden und darauf basierend die Re- bzw. Neuformulierung von Kriterien und Ansprüchen notwendig sein. Fragen nach dem Grad, der Art und Weise der Partizipation der Mitwirkenden auch beim Verfassen von fachlichen Publikationen im Sinne vielstimmiger Dar- K 100 M. Wrentschur stellungsformen sowie die Weiterentwicklung und Anwendung ethischer Kriterien und Regeln werden dabei eine besondere Rolle spielen, wenn sich Forumtheater als szenischer und partizipativer Forschungsansatz in der sozialwissenschaftlichen und damit auch sozialpädagogischen Forschungslandschaft weiter etablieren will. Die spezifischen Beiträge und Leistungen von Forumtheater als szenischem Forschungsansatz in Handlungs- und Forschungsfeldern der Sozialen Arbeit bestehen meines Erachtens in der unmittelbaren Beteiligung der Adressat_innen, der Untersuchung von Handlungsspielräumen bei gleichzeitiger Bewusstheit der strukturellen Rahmungen sowie in der vielschichtigen, interaktiven Vermittlung von Erkenntnissen an gesellschaftliche und politische Öffentlichkeit(en). Dies entspricht einem von Schneider (2014) formulierten Verständnis von Forschung in der Sozialen Arbeit, das von „der Sozialen Arbeit als Handlungswissenschaft in der Sorge um die soziale Existenz des Menschen ausgeh[t]“ (Schneider 2014, S. 396). Dabei ist sie dem Handeln auf einer individuellen, institutionellen und politischen Ebene verpflichtet. Ebenso wichtig ist aber auch, dass sich Forschung stärker als bisher um Dissemination aktiv bemühen sollte. Denn eine Soziale Arbeit, so Schneider (2012, S. 165 f.), die gesellschaftlichen Wandel betreiben will, muss proaktiv und professionell agieren. Und dazu kann szenisches Forschen mit dem Forumtheater jedenfalls beitragen. Funding Open access funding provided by University of Graz. Open Access Dieser Artikel wird unter der Creative Commons Namensnennung 4.0 International Lizenz (http://creativecommons.org/licenses/by/4.0/deed.de) veröffentlicht, welche die Nutzung, Vervielfältigung, Bearbeitung, Verbreitung und Wiedergabe in jeglichem Medium und Format erlaubt, sofern Sie den/die ursprünglichen Autor(en) und die Quelle ordnungsgemäß nennen, einen Link zur Creative Commons Lizenz beifügen und angeben, ob Änderungen vorgenommen wurden. Literatur Anhorn, Roland, und Johannes Stehr. 2012. Grundmodelle von Gesellschaft und soziale Ausschließung: Zum Gegenstand einer kritischen Forschungsperspektive in der Sozialen Arbeit. In Kritisches Forschen in der Sozialen Arbeit. Gegenstandbereiche – Kontextbedingungen – Positionierungen – Perspektiven, Hrsg. Elke Schimpf, Johannes Stehr, 57–76. Wiesbaden: VS. Bergold, Jarg, und Stefan Thomas. 2012. Partizipative Forschungsmethoden: Ein methodischer Ansatz in Bewegung. Forum Qualitative Sozialforschung https://doi.org/10.17169/fqs-13.1.1801. Boal, Augusto. 1999. Der Regenbogen der Wünsche: Methoden aus Theater und Therapie. Seelze (Velber): Kallmeyer. Boal, Augusto. 2013. Übungen und Spiele für Schauspieler und Nicht-Schauspieler. Aktualisierte und erweiterte Ausgabe. Frankfurt a. Main: Suhrkamp. Herausgegeben und aus dem brasilianischen Portugiesisch übersetzt von Till Baumann. Bülow-Schramm, Margret, und Dietlinde Gipser (Hrsg.). 1997. Spielort Universität. 10 Jahre Lehr-/ Lernprojekt „Der brüchige Habitus“. Hochschuldidaktische Arbeitspapiere, Bd. 30. Hamburg: Interdisziplinäres Zentrum für Hochschuldidaktik der Universität Hamburg. Bülow-Schramm, Margret, und Dietlinde Gipser. 2007. Theater mit dem Habitus. Zur forschungsbasierten szenischen Arbeit am brüchigen Habitus von HochschullehrerInnen. In Bühne frei für Forschungstheater. Theatrale Inszenierungen als wissenschaftlicher Erkenntnisprozess, Hrsg. Margret BülowSchramm, Dietlinde Gipser, und Doris Krohn, 87–104. Oldenburg: Paolo Freire. Diamond, David. 2013. Theater zum Leben. Über die Kunst und die Wissenschaft des Dialogs in Gemeinwesen. Stuttgart: Ibidem. Fritz, Birgit. 2011. InExActArt. Ein Handbuch zur Praxis des Theaters der Unterdrückten. Stuttgart: Ibidem. K Forumtheater als Werkzeug szenisch-partizipativer Forschung in der Sozialen Arbeit.... 101 Fritz, Birgit. 2013. Von Revolution zu Autopoiese. Auf den Spuren Augusto Boals ins 21. Jahrhundert. Das Theater der Unterdrückten im Kontext von Friedensarbeit und einer Ästhetik der Wahrnehmung. Stuttgart: Ibidem. Garfinkel, Harold. 1985. Studies in Ethnomethodology. Oxford: Polity Press. Gipser, Dietlinde. 1996. Grenzüberschreitungen: Theater der Unterdrückten an Hochschulen in Nah-Ost und West – Emanzipatorische Forschungsprozesse. zeitschrift für befreiende pädagogik 10:26–31. Gipser, Dietlinde. 2007. Hochschultheater. Wege zu einer emanzipatorischen Forschungspraxis. In Bühne frei für Forschungstheater. Theatrale Inszenierungen als wissenschaftlicher Erkenntnisprozess, Hrsg. Margret Bülow-Schramm, Dietlinde Gipser, und Doris Krohn, 127–160. Oldenburg: Paolo Freire. Graßhoff, Gunther (Hrsg.). 2013. Adressaten, Nutzer, Agency. Akteursbezogene Forschungsperspektiven in der Sozialen Arbeit. Wiesbaden: VS. Guhathakurta, Meghna. 2015. Theatre in participatory action research: Experiences from Bangladesh. In The SAGE Handbook of Action Research, Hrsg. Hilary Bradbury, 100–108. Los Angeles: SAGE. Hanses, Andreas. 2005. AdressatInnenforschung in der Sozialen Arbeit – Zwischen disziplinärer Grundlegung und Provokation. In Sozialpädagogik als forschende Disziplin, Hrsg. Cornelia Schweppe, Werner Thole, 185–199. Weinheim, München: Juventa. Koch, Gerd. 1988. Lernen mit Bert Brecht. Bertolt Brechts politisch-kulturelle Pädagogik. Frankfurt a. Main: Brandes & Apsel. Koch, Gerd. 1997. Theater-Spiel als szenische Sozialforschung. In TheaterSpiel. Ästhetik des Schul- und Amateurtheaters, Hrsg. Jürgen Belgrad, 81–96. Hohengehren: Schneider. Kramer, Paula. 2007. Performing Research—Researching Performance in South Africa. Methodological Musings on Research Praxis and Theatre as Research. In Bühne frei für Forschungstheater. Theatrale Inszenierungen als wissenschaftlicher Erkenntnisprozess, Hrsg. Margret Bülow-Schramm, Dietlinde Gipser, und Doris Krohn, 171–182. Oldenburg: Paolo Freire. Liamputtong, Pranee. 2007. Researching the vulnerable. A guide to sensitive research methods. London, Thousand Oaks, New Dehli: SAGE. Linds, Warren, und Elinor Vettraino. 2008. Collective imagining: Collaborative story telling through image theater [54 paragraphs]. Forum Qualitative Sozialforschung/Forum: Qualitative Social Research 9(2):56. Milgram, Stanley. 1974. Obedience to authority. An experimental view. New York: Harper. Schimpf, Elke. 2012. Widersprüchliche Deutungsmuster und Praktiken lebensweltorientierten Forschens. In Kritisches Forschen in der Sozialen Arbeit. Gegenstandbereiche – Kontextbedingungen – Positionierungen – Perspektiven, Hrsg. Elke Schimpf, Johannes Stehr, 233–261. Wiesbaden: VS. Schneider, Armin. 2012. Forschung und Soziale Arbeit – Chancen und Grenzen zur Politikbeeinflussung. In Forschung, Politik und Soziale Arbeit, Hrsg. Armin Schneider, Rebekka Streck, und Eppler Natalie, 163–177. Opladen, Berlin, Toronto: Budrich. Schneider, Armin. 2014. Forschung. In Handlungsräume Sozialer Arbeit, Hrsg. Günter J. Friesenhain, Daniela Braun, und Rainer Ningel, 392–400. Opladen, Toronto: Barbara Budrich. Seitz, Hanne. 2012. Performative Research. In Die Kunst, über kulturelle Bildung zu forschen. Theorie- und Forschungsansätze, Hrsg. Tobias Fink, Burkhard Hill, Vanessa-Isabelle Reinwand, und Alexander Wenzlik, 81–95. München: kopaed. Staffler, Armin. 2009. Augusto Boal. Einführung. Essen: Oldib. Streck, Danilo R., und Oscar Jara Holliday. 2015. Research, participation and social transformation: Grounding systematization of experiences in Latin American perspectives. In The SAGE Handbook of Action Research, Hrsg. Hilary Bradbury, 472–480. Los Angeles: SAGE. von Unger, Hella. 2014. Partizipative Forschung. Einführung in die Forschungspraxis. Wiesbaden: VS. Wakefort, Tom, Michael Pimbert, und Eric Walcon. 2015. Re-fashioning citizens’ juries: Participatory democracy in action. In The SAGE Handbook of Action Research, Hrsg. Hilary Bradbury, 230–246. Los Angeles: SAGE. Weitzel, Julia. 2012a. Existenzielle Bildung. Zur ästhetischen und szenologischen Aktualisierung einer bildungstheoretischen Leitidee. Bielefeld: transcript. Weitzel, Julia. 2012b. Forschungslabor: Szenisches Forschen. In Die Kunst, über kulturelle Bildung zu forschen. Theorie- und Forschungsansätze, Hrsg. Tobias Fink, Burkhard Hill, Vanessa-Isabelle Reinwand, und Alexander Wenzlik, 262–271. München: kopaed. Winter, Rainer. 2010. Ein Plädoyer für kritische Perspektiven in der qualitativen Forschung [41 Absätze]. Forum Qualitative Sozialforschung/Forum: Qualitative Social Research 12(1):7 Winter, Rainer. 2012. Macht, Kultur und soziale Intervention. Cultural Studies als kritische Theorieperspektive in der Sozialen Arbeit. In Kritik der Sozialen Arbeit – Kritische Soziale Arbeit, Hrsg. Roland Anhorn, Frank Bettinger, Cornelis Horlacher, und Kerstin Rathgeb, 449–462. Wiesbaden: VS. K 102 M. Wrentschur Wrentschur, Michael. 2004. Theaterpädagogische Wege in den öffentlichen Raum. Zwischen struktureller Gewalt und lebendiger Beteiligung. Stuttgart: Ibidem. Wrentschur, Michael. 2007. Forschen mit Methoden des Theaters und des szenischen Spiels. Ein Zwischenbericht. In Bühne frei für Forschungstheater. Theatrale Inszenierungen als wissenschaftlicher Erkenntnisprozess, Hrsg. Margret Bülow-Schramm, Dietlinde Gipser, und Doris Krohn, 127–160. Oldenburg: Paolo Freire. Wrentschur, Michael. 2008a. Forum theatre as a participatory tool for social research and development: a reflection on nobody is perfect: a project with homeless people. In Qualitative research and social change in European contexts, Hrsg. Pat Cox, Thomas Geisen, und Roger Green, 94–111. New York: palgrave macmillan. Wrentschur, Michael. 2008b. Kein Kies zum Kurven Kratzen – Neuer Armut entgegenwirken. Ein Projekt zur kreativen Beteiligung von Armutsbetroffenen. In Armut, Gesellschaft und Soziale Arbeit. Perspektiven gegen Armut und soziale Ausgrenzung in Österreich, Hrsg. Gerald Knapp, Heinz Pichler, 692–723. Klagenfurt, Ljubljana, Wien: Hermagoras. Wrentschur, Michael. 2010. Neuer Armut entgegenwirken: Politisch-partizipative Theaterarbeit als kreativer Impulse für soziale und politische Partizipationsprozesse. In Macht – Eigensinn – Engagement. Lernprozesse gesellschaftlicher Teilhabe, Hrsg. Angela Pilch Ortega, Andrea Felbinger, Regina Mikula, und Rudolf Egger, 211–232. Wiesbaden: VS. Wrentschur, Michael. 2011. Theater an die Macht: Neuer Armut entgegenwirken – bis ins Parlament! Zeitschrift für Theaterpädagogik, Korrespondenzen 27(58):55–58. Wrentschur, Michael. 2012. Szenisch-partizipatives Forschen als Beitrag zu sozialer Teilhabe und politischer Beteiligung. In Empirische Forschung in der Sozialen Arbeit. Methoden und methodologische Herausforderungen, Hrsg. Arno Heimgartner, Ulrike Loch, und Stephan Sting, 137–153. Wien, Berlin, Münster: LIT. Wrentschur, Michael. 2014. Die Stimmen der AdressatInnen – werden sie gehört? Situative Blitzlichter auf das Zusammenspiel von Anerkennung, Respekt und Partizipation. In Perspektiven der AkteurInnen in der Sozialen Arbeit, Hrsg. Arno Heimgartner, Karin Lauermann, und Stephan Sting, 69–82. Wien, Berlin, Münster: LIT. Wrentschur, Michael. 2019. Forumtheater, szenisches Forschen und Soziale Arbeit. Diskurse – Verfahren – Fallstudien. Weinheim, Basel: Beltz Juventa. Wrentschur, Michael, und Michaela Moser. 2014. ,Stop: Now we are speaking!‘ A creative and dissident approach of empowering disadvantaged young people. International Social Work 57(4):398–410. Michael Wrentschur PD Mag. Dr., lehrt und forscht am Institut für Erziehungs- und Bildungswissenschaft der Karl-Franzens-Universität Graz im Arbeitsbereich Sozialpädagogik. Schwerpunkte: Theaterarbeit in sozialen Feldern, Soziokultur und soziale Kulturarbeit, Armut und soziale Ausgrenzung, Partizipation; szenische und partizipative Forschungsmethoden; Künstlerischer Leiter von InterACT, der Werkstatt für Theater und Soziokultur. michael.wrentschur@uni-graz.at www.interact-online.org K
https://openalex.org/W2096112302	https://kar.kent.ac.uk/37656/1/The%20structure%20of%20phosphate%20glass%20biomaterials%20from%20neutron%20diffraction%20and%20P-31%20nuclear%20magnetic%20resonance%20data%20%28Oct%202007%29.pdf	English	null	The structure of phosphate glass biomaterials from neutron diffraction and<sup>31</sup>P nuclear magnetic resonance data	Journal of physics. Condensed matter	2,007	cc-by	5,666	Licence for this version CC BY (Attribution) Additional information Versions of research works Versions of Record If this version is the version of record, it is the same as the published version available on the publisher's web site. Cite as the published version. Downloaded from Downloaded from https://kar.kent.ac.uk/37656/ The University of Kent's Academic Repository KAR Author Accepted Manuscripts p p If this document is identified as the Author Accepted Manuscript it is the version after peer review but before type setting, copy editing or publisher branding. Cite as Surname, Initial. (Year) 'Title of article'. To be published in Title of Journal , Volume and issue numbers [peer-reviewed accepted version]. Available at: DOI or URL (Accessed: date). Enquiries If this document is identified as the Author Accepted Manuscript it is the version after peer review but before type setting, copy editing or publisher branding. Cite as Surname, Initial. (Year) 'Title of article'. To be published in Title of Journal , Volume and issue numbers [peer-reviewed accepted version]. Available at: DOI or URL (Accessed: date). The version of record is available from https://doi.org/10.1088/0953-8984/19/41/415116 The version of record is available from https://doi.org/10.1088/0953-8984/19/41/415116 Kent Academic Repository Pickup, David M., Ahmed, Ifty, Guerry, Paul, Knowles, Jonathan C., Smith, Mark E. and Newport, Robert J. (2007) The structure of phosphate glass biomaterials from neutron diffraction and 31P nuclear magnetic resonance data. Journal of Physics: Condensed Matter, 19 (41). ISSN 0953-8984. Downloaded from https://kar.kent.ac.uk/37656/ The University of Kent's Academic Repository KAR Home Search Collections Journals About Contact us My IOPscience Enquiries q If you have questions about this document contact ResearchSupport@kent.ac.uk. Please include the URL of the record in KAR. If you believe that your, or a third party's rights have been compromised through this document please see our Take Down policy (available from https://www.kent.ac.uk/guides/kar-the-kent-academic-repository#policies). The structure of phosphate glass biomaterials from neutron diffraction and 31P nuclear magnetic resonance data This article has been downloaded from IOPscience. Please scroll down to see the full text article. 2007 J. Phys.: Condens. Matter 19 415116 (http://iopscience.iop.org/0953-8984/19/41/415116) View the table of contents for this issue, or go to the journal homepage for more Please note that terms and conditions apply. IOP PUBLISHING JOURNAL OF PHYSICS: CONDENSED MATTER doi:10.1088/0953-8984/19/41/415116 J. Phys.: Condens. Matter 19 (2007) 415116 (8pp) Abstract Neutron diffraction and 31P nuclear magnetic resonance spectroscopy were used to probe the structure of phosphate glass biomaterials of general composition (CaO)0.5−x(Na2O)x(P2O5)0.5 (x = 0, 0.1 and 0.5). The results suggest that all three glasses have structures based on chains of Q2 phosphate groups. Clear structural differences are observed between the glasses containing Na2O and CaO. The P–O bonds to bridging and non-bridging oxygens are less well resolved in the neutron data from the samples containing CaO, suggesting a change in the nature of the bonding as the ﬁeld strength of the cation increases Na+ →Ca2+. In the (CaO)0.5(P2O5)0.5 glass most of the Ca2+ ions are present in isolated CaOx polyhedra whereas in the (Na2O)0.5(P2O5)0.5 glass the NaOx polyhedra share edges leading to a Na–Na correlation. The results of the structural study are related to the properties of the (CaO)0.4(Na2O)0.1(P2O5)0.5 biomaterial. The structure of phosphate glass biomaterials from neutron diffraction and 31P nuclear magnetic resonance data D M Pickup1, I Ahmed2, P Guerry3, J C Knowles2, M E Smith3 and R J Newport1 1 School of Physical Sciences, University of Kent, Canterbury CT2 7NH, UK 2 Division of Biomaterials and Tissue Engineering, UCL Eastman Dental Institute, 256 Gray’s Inn Road, London WC1X 8LD, UK 3 3 Department of Physics, University of Warwick, Coventry CV4 7AL, UK Received 6 March 2007, in ﬁnal form 11 April 2007 Published 27 September 2007 Received 6 March 2007, in ﬁnal form 11 April 2007 Published 27 September 2007 Received 6 March 2007, in ﬁnal form 11 April 2007 Published 27 September 2007 Online at stacks.iop.org/JPhysCM/19/415116 0953-8984/07/415116+08$30.00 © 2007 IOP Publishing Ltd Printed in the UK 1. Introduction The general trend in biomaterials research is a move from passive, inert implant materials to those that degrade and play an active part in the regeneration of tissue [1]. A greater understanding of the interaction of materials with cells has allowed implant materials to be designed with the aim of promoting a speciﬁc biological response. This new class of materials is often referred to as the ‘Third Generation’ of biomaterials, and included in this class are biocompatible phosphate glasses. Phosphate based glasses have many unique properties, the most interesting of which, from a biomedical standpoint, is their ability to dissolve completely in aqueous media. Furthermore, this dissolution behaviour may be easily controlled via the chemistry of the glasses [2]. Of 0953-8984/07/415116+08$30.00 © 2007 IOP Publishing Ltd Printed in the UK 1 1 J. Phys.: Condens. Matter 19 (2007) 415116 D M Pickup et al beneﬁt to biomedical applications is the fact that these glasses can be prepared to include ions routinely found in the human body. Because of these properties, phosphate based glasses can ﬁnd application as degradable temporary implants in the human body: reacting and dissolving in the physiological environment, and eventually being replaced by hard or soft tissue. Further to the properties described above, phosphate based glasses can also have a beneﬁcial effect on cell proliferation. Bitar et al [3] demonstrated that ternary phosphate glasses with high calcium content can support the attachment, growth and differentiation of human osteoblasts and ﬁbroblasts. Consequently, there has been signiﬁcant interest in these glasses when manufactured as ﬁbres for potential use in tissue engineering, and in particular, for any tissue with a medium to high anisotropy, such as muscle and ligament [4]. Other potential clinical applications of phosphate based glasses include bacterial control devices when doped with antibacterial ions, as neural repair devices when manufactured in a tubular form and as a component in oral healthcare products when doped with ﬂuoride ions [1]. Whilst signiﬁcant work has been carried out to elucidate the structure of phosphate glasses [5], there is a scarcity of data to correlate this information with the physical properties such as dissolution rate. In this paper, we have studied the structure of phosphate glass biomaterials of general composition (CaO)0.5−x(Na2O)x(P2O5)0.5 with neutron diffraction and 31P NMR spectroscopy in order to better understand the structure/properties relationship. 2. Experimental: preparation of the phosphate glasses The phosphate glasses were prepared using sodium dihydrogen orthophosphate (NaH2PO4, 99%, BDH), calcium carbonate (CaCO3, 99 + %, BDH) and phosphorus pentoxide (P2O5, 97%, BDH). The precursors were weighed out, mixed, placed in a Pt/10%Rh crucible (Type 71040, Johnson Matthey) and loaded into a preheated furnace (Carbolite, RHF 1500, UK). The mixture was left at 300 ◦C for half an hour, raised to 600 ◦C for a further half an hour and ﬁnally melted at 1100 ◦C for 1 h. The molten glass was then poured into a preheated (370 ◦C) graphite mould, and left to cool to room temperature. Characterization necessary for analysis of the neutron diffraction data was performed: elemental analysis (ICP-AES and gravimetric) was carried out by a commercial company (Medac Ltd) and macroscopic densities were determined by helium pycnometry using a Quantachrome Multipycnometer. The experimentally determined compositions were found to agree with the expected values within experimental error. 3. Experimental: structural studies The Q-space simulation is generated using the following equation: p(Q)i j = Ni jwi j c j sin QRi j QRi j exp " −Q2σ 2 i j 2 # (2) (2) where p(Q)i j is the pair function in reciprocal space, Ni j, Ri j and σi j are the coordination number, atomic separation and disorder parameter, respectively, of atom i with respect to j, c j is the concentration of atom j and wi j is the weighting factor, given by: (3) wi j = 2cic jbib j if i ̸= j wi j = 2cic jbib j if i ̸= j (3) or, wi j = c2 i b2 i if i = j wi j = c2 i b2 i if i = j (4) (4) where b represents the coherent scattering length. where b represents the coherent scattering length. where b represents the coherent scattering length. The 31P NMR experiments were carried out on a CMX Inﬁnity spectrometer attached to an 8.5 T magnet giving a 31P Larmor frequency of 145.85 MHz. Samples were placed in the magnet using a Doty 4 mm MAS (magic angle spinning) probe and spun at ∼12 kHz. Spectra were recorded using the Spinsight software running a simple one (90◦) pulse experiment. The 90◦pulse length was 2 µs. A repetition time of 30 s was chosen to avoid saturation. Typically, 70 scans were accumulated to obtain an adequate signal/noise ratio. Spectra were referenced to the resonance of NH4H2PO4 at 0.9 ppm. 3. Experimental: structural studies The neutron diffraction data presented here were collected on the GEM diffractometer on the ISIS spallation neutron source at the Rutherford Appleton Laboratory, UK. The samples were in the form of 8 mm diameter rods, which negated the need for a container (thus simplifying the necessary corrections to the data), and time-of-ﬂight data collected over a wide range of Q (up to 60 ˚A −1, where Q = 4π sin θ/λ). The program GUDRUN was used to reduce and correct the data [6]. The initial stage of analysis of diffraction data from an amorphous material involves the removal of background scattering, normalization, correction for absorption, inelastic and multiple scattering and subtraction of the self-scattering term [6]. The resultant scattered intensity, i(Q), can reveal structural information by Fourier transformation to obtain the pair distribution function: T (r) = T 0(r) + 2 π ! ∞ 0 Qi(Q)M(Q) sin(Qr) dQ ( (1) 2 J. Phys.: Condens. Matter 19 (2007) 415116 D M Pickup et al where T 0(r) = 4πrρ0 ¯b2 is the average density contribution (r is the atomic separation between atoms, ρ0 the macroscopic number density and ¯b is the average coherent scattering length) and M(Q) is a window function necessitated by the ﬁnite maximum experimentally attainable value of Q. where T 0(r) = 4πrρ0 ¯b2 is the average density contribution (r is the atomic separation between atoms, ρ0 the macroscopic number density and ¯b is the average coherent scattering length) and M(Q) is a window function necessitated by the ﬁnite maximum experimentally attainable value of Q. Structural information can be obtained from the diffraction data by simulating the Q-space data and converting the results to r-space by Fourier transformation to allow comparison with the experimentally determined correlation function [7]. 4. Results Each PO3− 4 can be connected to a maximum of three other such units to form a three- dimensional network, as in v-P2O5. Additions of metal oxides cause the connectivity of the PO3− 4 groups to be reduced and chain-like structures or structures containing isolated phosphate anions are formed. Two P–O distances may be observed in phosphate glasses: a shorter 4 Figure 1. Neutron diffraction data from the (CaO)0.5(P2O5)0.5 glass: (a) Q-space interference function, i(Q), and (b) pair distribution function, T (r), (solid line) together with ﬁt (dashed line). Figure 1. Neutron diffraction data from the (CaO)0.5(P2O5)0.5 glass: (a) Q-space interference function, i(Q), and (b) pair distribution function, T (r), (solid line) together with ﬁt (dashed line). Figure 1. Neutron diffraction data from the (CaO)0.5(P2O5)0.5 glass: (a) Q-space interference function, i(Q), and (b) pair distribution function, T (r), (solid line) together with ﬁt (dashed line). Figure 2. Neutron diffraction data from the (Na2O)0.5(P2O5)0.5 glass: (a) Q-space interference function, i(Q), and (b) pair distribution function, T (r), (solid line) together with ﬁt (dashed line). Figure 2. Neutron diffraction data from the (Na2O)0.5(P2O5)0.5 glass: (a) Q-space interference function, i(Q), and (b) pair distribution function, T (r), (solid line) together with ﬁt (dashed line). 4. Results Figures 1, 2 and 3 show the neutron diffraction data from the (CaO)0.5(P2O5)0.5, (Na2O)0.5(P2O5)0.5 and (CaO)0.4(Na2O)0.1(P2O5)0.5 glasses, respectively. Both the r-space and Q-space data are shown, together with the ﬁts to the pair distributions functions obtained using the method described above. The structural parameters obtained from the ﬁtting of the neutron diffraction data are given in table 1. The peak assignments used in the ﬁtting process were taken from the literature reports of previous work on phosphate glass containing CaO and Na2O [5]. The strongest peaks in the pair distributions functions at ∼1.5 ˚A and ∼2.5 ˚A are due to the P–O and O–O nearest-neighbour distances, respectively. The shoulder to the low r side of the O–O peak is the metal–oxygen correlation and the shoulder at ∼3 ˚A is due to the shortest P–P distance in the structure. The second nearest-neighbour oxygen–oxygen distance (O–O2nd) was taken from an x-ray diffraction study of vitreous P2O5(v-P2O5) [8]. Figure 4 shows the 31P MAS NMR spectra from the (Na2O)0.5(P2O5)0.5, (CaO)0.5(P2O5)0.5 and (CaO)0.4(Na2O)0.1(P2O5)0.5 samples. For the (CaO)0.5(P2O5)0.5 sam- ple, two peaks are present at −10.4 ppm and −27.4 ppm. These are ascribed to Q1 and Q2 structural units, where n in Qn refers to the number of bridging oxygen atoms in the PO3− 4 structural unit (and hence the number of other PO3− 4 groups connected to it). Fitting of the spectrum revealed the presence of 96% Q2 and 4% Q1. The chemical shifts (−10.4 ppm for Q1 and −27.4 ppm for Q2) agree with previously published results [9]. The spectrum from the (CaO)0.4(Na2O)0.1(P2O5)0.5 glass is almost identical to that from the (CaO)0.5(P2O5)0.5 sam- ple. The spectrum from the (Na2O)0.5(P2O5)0.5 sample shows only the presence of Q2 with a peak at −19.8 ppm, in agreement with previously published work [5]. 3 3 J. Phys.: Condens. Matter 19 (2007) 415116 D M Pickup et al Figure 1. Neutron diffraction data from the (CaO)0.5(P2O5)0.5 glass: (a) Q-space interference function, i(Q), and (b) pair distribution function, T (r), (solid line) together with ﬁt (dashed line). Figure 2. Neutron diffraction data from the (Na2O)0.5(P2O5)0.5 glass: (a) Q-space interference function, i(Q), and (b) pair distribution function, T (r), (solid line) together with ﬁt (dashed line). 5. Discussion It is well known that the building blocks of phosphate based glasses are PO3− 4 tetrahedra [5]. distance of ∼1.49 ˚A ascribed to bonds to non-bridging oxygens (NBOs) and a longer distance of ∼1.60 ˚A due to bonds to bridging oxygens (BOs). The results in table 1 indicate that, in the glasses studied here, the number of NBOs and BOs connected to each phosphorus atom are equal. This is to be expected from the composition 5 distance of ∼1.49 ˚A ascribed to bonds to non-bridging oxygens (NBOs) and a longer distance of ∼1.60 ˚A due to bonds to bridging oxygens (BOs). 5. Discussion If it is assumed that all the phosphorus atoms are bonded to four oxygens, and that all the oxygens are bonded to either one or two P atoms then the P–NBO coordination number is: nP−NBO = 2 NO NP −4 nP−NBO = 2 NO NP −4 (5) (5) and similarly the P–BO coordination number is: and similarly the P–BO coordination number is: nP−BO = 8 −2 NO NP (6) nP−BO = 8 −2 NO NP (6) where NO and NP are the numbers of oxygen and phosphorus atoms in the sample. The samples studied here are predicted to have coordination numbers of two for both the P–O ﬁrst-shell correlations, in agreement (within the errors associated with the measurements) with the experimental results from the neutron diffraction study. This result is supported by the analysis of the 31P MAS NMR spectra. PO3− 4 groups with two NBOs and two BOs correspond to Q2 units in the NMR terminology. The NMR results suggest structures based on Q2 units. The small amount of Q1 (<5%) observed in the Ca containing samples is likely to be due to partial hydrolysis of the structure which is known to occur in phosphate glasses [5]. The P– P coordination numbers of close to two for all three glass samples also provide evidence for structures based on chains of Q2 groups. where NO and NP are the numbers of oxygen and phosphorus atoms in the sample. The samples studied here are predicted to have coordination numbers of two for both the P–O ﬁrst-shell correlations, in agreement (within the errors associated with the measurements) with the experimental results from the neutron diffraction study. This result is supported by the analysis of the 31P MAS NMR spectra. PO3− 4 groups with two NBOs and two BOs correspond to Q2 units in the NMR terminology. The NMR results suggest structures based on Q2 units. The small amount of Q1 (<5%) observed in the Ca containing samples is likely to be due to partial hydrolysis of the structure which is known to occur in phosphate glasses [5]. The P– P coordination numbers of close to two for all three glass samples also provide evidence for structures based on chains of Q2 groups. Of particular interest in relation to understanding the properties of (CaO)0.5−x(Na2O)x (P2O5)0.5 glasses is the effect the cations have on the phosphate skeleton that forms the basis of the structure. 5. Discussion It is well known that the building blocks of phosphate based glasses are PO3− 4 tetrahedra [5]. Each PO3− 4 can be connected to a maximum of three other such units to form a three- dimensional network, as in v-P2O5. Additions of metal oxides cause the connectivity of the PO3− 4 groups to be reduced and chain-like structures or structures containing isolated phosphate anions are formed. Two P–O distances may be observed in phosphate glasses: a shorter 4 4 J. Phys.: Condens. Matter 19 (2007) 415116 D M Pickup et al Figure 3. Neutron diffraction data from the (CaO)0.4(Na2O)0.1(P2O5)0.5 glass: (a) Q-space interference function, i(Q), and (b) pair distribution function, T (r), (solid line) together with ﬁt (dashed line). Figure 3. Neutron diffraction data from the (CaO)0.4(Na2O)0.1(P2O5)0.5 glass: (a) Q-space interference function, i(Q), and (b) pair distribution function, T (r), (solid line) together with ﬁt (dashed line). Table 1. Structural parameters obtained from the simulations of the neutron diffraction data. Note that reasonable estimates of the errors are ±0.02 ˚A in R, ±15% in N and ±0.010 ˚A in σ. Table 1. Structural parameters obtained from the simulations of the neutron diffraction data. Note that reasonable estimates of the errors are ±0.02 ˚A in R, ±15% in N and ±0.010 ˚A in σ. Sample Density (g cm−3) Correlation R ( ˚A) N σ ( ˚A) (CaO)0.5(P2O5)0.5 2.61 P–NBO 1.49 1.9 0.034 P–BO 1.60 2.0 0.046 Ca–O 2.34 4.9 0.120 O–O 2.51 4.0 0.080 P–P 2.94 1.9 0.078 O–O2nd 2.82 0.9 0.115 (Na2O)0.5(P2O5)0.5 2.47 P–NBO 1.48 1.8 0.034 P–BO 1.61 2.0 0.048 Na–O 2.33 4.0 0.130 O–O 2.52 4.0 0.078 P–P 2.93 2.0 0.070 Na–Na 3.07 1.4 0.080 O–O2nd 2.82 0.6 0.100 (CaO)0.4(Na2O)0.1(P2O5)0.5 2.59 P–NBO 1.49 1.9 0.036 P–BO 1.60 2.0 0.048 Na–O 2.33 3.5 0.130 Ca–O 2.34 4.9 0.120 O–O 2.52 4.2 0.080 P–P 2.93 2.1 0.070 O–O2nd 2.82 1.0 0.115 J. Phys.: Condens. Matter 19 (2007) 415116 D M Pickup et al Figure 4. 31P MAS NMR spectra: (a) (Na2O)0.5(P2O5)0.5, (b) (CaO)0.5(P2O5)0.5 and (c) (CaO)0.4(Na2O)0.1(P2O5)0.5 glasses. Figure 4. 31P MAS NMR spectra: (a) (Na2O)0.5(P2O5)0.5, (b) (CaO)0.5(P2O5)0.5 and (c) (CaO)0.4(Na2O)0.1(P2O5)0.5 glasses. of the glasses. If it is assumed that all the phosphorus atoms are bonded to four oxygens, and that all the oxygens are bonded to either one or two P atoms then the P–NBO coordination number is: of the glasses. 5. Discussion With regard to this, it is useful to consider the ratio of NBOs per Mev+ cation, MNBO, which is given by [10]: MNBO = NNBO/NMe = v(y + 1)/y (7) (7) MNBO = NNBO/NMe = v(y + 1)/y where NNBO and NMe are the numbers of NBOs and cations, respectively, and y is the molar ratio of metal oxide to phosphorus oxide (y = n(Me2/vO)/n(P2O5)). In the case of the (CaO)0.5(P2O5)0.5 sample MNBO = 4, whereas for the (Na2O)0.5(P2O5)0.5 sample MNBO = 2. Given that the cations need to be associated with the NBOs for charge balancing purposes, we can use the MNBO ratios in conjunction with the Me–O coordination number derived 6 6 J. Phys.: Condens. Matter 19 (2007) 415116 D M Pickup et al from the neutron data to predict how the cations coordinate with the NBOs. In the case of the (CaO)0.5(P2O5)0.5 glass, the Ca–O coordination number is ∼5 and there are four NBOs available for each Ca2+. This means that only a quarter of the NBOs need to be coordinated to two Ca2+ ions and that most of the Ca2+ ions can be present in isolated CaOx polyhedra. This is not the case for the (Na2O)0.5(P2O5)0.5 glass where there are only two NBOs per Na+ ion and the Na–O coordination number is ∼4. The result of this is that each NBO must be shared by two Na+ ions and the NaOx polyhedra must share edges leading to a Na–Na correlation across these edges. Experimentally it was found the data from the (Na2O)0.5(P2O5)0.5 glass could not be satisfactorily ﬁtted without the inclusion of a Me–Me correlation whereas this was not the case for the other datasets. Thus, it seems that the experimental results conﬁrm the arrangement of NaOx polyhedra predicted on the basis of the glass composition. Previous Raman spectroscopy studies on phosphate glasses have suggested that the covalency of the Me–NBO bond increases as the ﬁeld strength of the modiﬁer cation increases [11, 12]. The results in table 1 show that the splitting of the P–O peak is greater in the data from the (Na2O)0.5(P2O5)0.5 sample than in that from the two calcium containing glasses. Although, the change in bond lengths is only small, −0.01 ˚A for the P–NBO bond and +0.01 ˚A for the P–BO bond, the effect can clearly be observed in the pair distribution functions shown in ﬁgures 1–3. 5. Discussion This provides evidence of increasing covalency in the Me– NBO bond with increasing ﬁeld strength of the cation (Ca2+ > Na+) as electron density moves away from the non-bridging oxygen to the Me–NBO bond causing the P–NBO bond to lengthen and the P–BO bond to shorten. This change in the nature of the P–NBO–Me and P–BO bonding interactions is further evidenced by the NMR spectra which exhibit a change in the Q2 chemical shift from −19.8 to −27.4 ppm as the modiﬁer cation changes from Na+ to Ca2+. As the Na+ ions are replaced by Ca2+, and the P–NBO and P–BO bond lengths become more similar with less localized electron density on the NBO, the environment of phosphorus becomes more similar to that in a Q3 unit (i.e. surrounded by three equivalent oxygens). Since the chemical shift of phosphorus in a Q3 unit is −40 ppm [13], it is expected that the Q2 chemical shift will become more negative as the oxygens in the PO3− 4 group become more equivalent. We can apply the structural information discussed above to understand better the properties of the high Ca content (CaO)0.4(Na2O)0.1(P2O5)0.5 biocompatible glass. We have shown that glasses of this general composition have structures comprising of Q2 phosphate chains linked together by modifying cations. Such a structure is compatible with ﬁbre drawing: previous work on the CaO–Na2O–P2O5 system has shown that ﬁbres cannot be drawn from glasses containing less than 50 mol% P2O5 due to the presence of a signiﬁcant amount of Q1 species [14]. In fact, a 2D rotor-synchronized 31P NMR MAS study of an extruded calcium metaphosphate glass by J¨ager et al showed clear ordering of the Q2 chains [15]. Studies on the dissolution properties of CaO–Na2O–P2O5 glasses have shown that the solubility decreases as the calcium content increases [2]. This can be explained in terms of the reduction in ionicity in the Me–NBO bond as the ﬁeld strength of the cation is increased (Ca2+ > Na+). 6. Conclusions High quality neutron diffraction data collected over a wide range of Q (up to 60 ˚A −1) have provided high enough real-space resolution to separate and ﬁt the P–BO and P–NBO correlations. The results show that as the ﬁeld strength of the cation increases Na+ →Ca2+, the P–BO and P–NBO distances move closer together, suggesting an increase in covalency in the Me–NBO bonding. This conclusion is supported by the 31P NMR data which exhibit a change in the chemical shift of the Q2 phosphorus atoms as the Na+ ions are replaced by Ca2+. 7 J. Phys.: Condens. Matter 19 (2007) 415116 D M Pickup et al There is another clear structural difference between the glasses containing Na2O and CaO which relates to the connectivity of the MeOx polyhedra. In the (CaO)0.5(P2O5)0.5 glass most of the Ca2+ ions are present in isolated CaOx polyhedra whereas in the (Na2O)0.5(P2O5)0.5 glass the NaOx polyhedra share edges leading to a Na–Na correlation. The high Ca containing (CaO)0.4(Na2O)0.1(P2O5)0.5 glass, which has potential for use in biomedical applications, is structurally similar to the (CaO)0.5(P2O5)0.5 glass. It is proposed that the increase in covalency in the Me–NBO bond as the ﬁeld strength of the cation is increased (Ca2+ > Na+) is responsible for the reduction in solubility of CaO–Na2O–P2O5 glasses with increasing calcium content. Acknowledgments The authors wish to acknowledge funding from the Engineering and Physical Sciences Research Council (EP/C000714, EP/C000633 and GR/T21080). We thank A C Hannon of the Council for the Central Laboratory of the Research Council's Rutherford Appleton Laboratory for his help in running GEM. for his help in running GEM. References [1] Knowles J C 2003 J. Mater. Chem. 13 2395 [1] Knowles J C 2003 J. Mater. Chem. 13 2395 [2] Ahmed I, Lewis M, Olsen I and Knowles J C 2004 Biomaterials 25 491 [3] Bitar M, Salih V, Mudera V, Knowles J C and Lewis M P 2004 Biomaterials 25 2283 [5] Brow R K 2000 J. Non-Cryst. Solids 263/264 1 [5] Brow R K 2000 J. Non-Cryst. Solids 263/264 1 [6] Hannon A C 2005 Nucl. Instrum. Methods A 551 88 [6] Hannon A C 2005 Nucl. Instrum. Methods A 551 88 [7] Gaskell P H 1991 Glasses and Amorphous Materials (Materials Science and Technology vol 9) ed Anomalous dispersion neutron diffraction and its potential role in the elucidation of structure in glasses Isaac Abrahams Centre for Materials Research, School of Biological and Chemical Sciences, Queen Mary, University of London, Mile End Road, London E1 4NS, UK E-mail: i.abrahams@qmul.ac.uk E-mail: i.abrahams@qmul.ac.uk Received 11 December 2006 Published 15 January 2007 Online at stacks.iop.org/JPhysCM/19/051001 Received 11 December 2006 Published 15 January 2007 Online at stacks.iop.org/JPhysCM/19/051001 Received 11 December 2006 Published 15 January 2007 Online at stacks iop org/JPhys 0953-8984/07/051001+02$30.00 © 2007 IOP Publishing Ltd Printed in the UK J Zarzycki (Weinheim: VCH) p 175 [8] Hoppe U, Walter G, Kranold R and Stachel D 1998 Z. Naturf. a 53 93 [8] Hoppe U, Walter G, Kranold R and Stachel D 1998 Z. Naturf. a 53 93 [9] Witter R, Hartmann P, Vogel J and J¨ager C 1998 Solid State Nucl. Magn. Reson. 13 189 [9] Witter R, Hartmann P, Vogel J and J¨ager C 1998 Solid State Nucl. Magn. Reson. 13 [10] Hoppe U, Walter G, Kranold R and Stachel D 2000 J. Non-Cryst. Solids 263/2 [10] Hoppe U, Walter G, Kranold R and Stachel D 2000 J. Non-Cryst. Solids 26 [10] Hoppe U, Walter G, Kranold R and Stachel D 2000 J. N [11] Matic A and B¨orjesson L 1998 Phil. Mag. B 77 357 [12] Swenson J, Matic A, Brodin A, B¨orjesson L and Howells W S 1998 Phys. Rev. B 58 11331 [12] Swenson J, Matic A, Brodin A, B¨orjesson L and Howells W S 1998 Phys. Rev. B 58 11331 [12] Swenson J, Matic A, Brodin A, B¨orjesson L and Howe j y [13] Sato R K, Kirkpatrick R J and Brow R K 1992 J. Non-Cryst. Solids 143 2 3] Sato R K, Kirkpatrick R J and Brow R K 1992 J. Non-Cryst. Sol [14] Ahmed I, Lewis M, Olsen I and Knowles J C 2004 Biomaterials 25 501 [14] Ahmed I, Lewis M, Olsen I and Knowles J C 2004 Biomaterials 25 501 [15] J¨ager C, Hartmann P, Witter R and Braun M 2000 J. Non-Cryst. Solids 263/264 61 [15] J¨ager C, Hartmann P, Witter R and Braun M 2000 J. Non-Cryst. Solids 263/264 61 JOURNAL OF PHYSICS: CONDENSED MATTER doi:10.1088/0953-8984/19/5/051001 JOURNAL OF PHYSICS: CONDENSED MATTER INSTITUTE OF PHYSICS PUBLISHING doi:10.1088/0953-8984/19/5/051001 J. Phys.: Condens. Matter 19 (2007) 051001 (2pp) Centre for Materials Research, School of Biological and Chemical Sciences, Queen Mary, University of London, Mile End Road, London E1 4NS, UK Abstract The anomalous dispersion neutron diffraction method is brieﬂy discussed in the context of structure elucidation in glasses. The method represents a novel contrast technique that might be applied more generally provided suitable isotopes with accessible and isolated absorption resonances are available. Determining the structure of glasses represents a challenging problem in structural science. Unlike crystalline solids, where long-range order of the periodic lattice can be probed by standard diffraction methods, in amorphous solids the lack of periodicity necessitates a different approach to elucidation of the structure. In both types of solids powder diffraction represents a powerful technique, but data analysis differs signiﬁcantly between the two. The powder diffraction patterns of crystalline solids are characterized by sharp Bragg peaks that can usually be modelled by Rietveld analysis yielding the asymmetric unit of the crystal structure. In contrast, the diffraction patterns of glasses exhibit broad diffuse features, and lack the detail and intensity of crystalline patterns. While the lack of periodicity in amorphous solids means that directional information on interatomic vectors is lost in their diffraction patterns, their magnitude is not and this information can be accessed in the form of a total correlation function via Fourier transformation. Analysis of these data is complicated by the fact that the total correlation function is a summation over all atoms and as such contains overlapping pair correlations, the number of which increases with radial distance. This means that analysis is often limited to relatively short-range correlations typically below 3–4 ˚A. Analysis is often aided by results from element-speciﬁc techniques such as EXAFS, XANES and solid state NMR, while vibrational spectroscopy can allow for interpretation of bonding interactions in the glass framework. Deconvolution of medium- to long-range pair correlations in total correlation functions is key to an improved understanding of structure/property relationships in glasses and is particularly important in optoelectronic glasses, where metal–metal distances inﬂuence the optical and electronic properties. Since the total correlation function is a sum of individual 0953-8984/07/051001+02$30.00 © 2007 IOP Publishing Ltd Printed in the UK 1 J. Phys.: Condens. Matter 19 (2007) 051001 Viewpoint partial correlations it is possible, using suitable contrast experiments and subsequent ﬁrst- order difference methods, to extract individual partial correlation functions. A number of interesting approaches have been used in this context. [1] Hoppe U, Kranold R, Barz A, Stachel D, Neuefeind J and Keen D A 2001 J. Non-Cryst. Solids 293–295 158 [2] Clark E B, Mead R N and Mountjoy G 2006 J. Phys.: Condens. Matter 18 6815 [3] Newport R J, Skipper L J, Carta D, Pickup D M, Sowrey F E, Smith M E, Saravanapavan P and Hench L L 2006 J. Mater. Sci., Mater. Med. 17 1003 [4] Cole J M, Hannon A C, Martin R A and Newport R J 2006 Phys. Rev. B 73 104210 [5] Cole J M, Wright A C, Newport R J, Fisher C E, Clarke S J, Sinclair R N, Fischer H E and Cuello G J 2007 J. Phys.: Condens. Matter 19 056002 [6] Shikerkar A G, Desa J A E, Krishna P S R and Chitra R 2000 J Non-Cryst. Solids 270 234 [7] Martin R A, Salmon P S, Fischer H E and Cuello G J 2003 Phys. Rev. Lett. 90 185501 Abstract For example, a combination of x- ray and neutron diffraction methods involving reverse Monte Carlo simulation has been used successfully to resolve short- and medium-range correlations in divalent metal phosphate glasses [1]. Similarly, molecular dynamics simulations have also been employed to this end in Tb phosphate glasses [2]. Where appropriate, isotopic substitution can also be used to increase contrast [3]. A recent novel approach is that of magnetic difference neutron diffraction, where differences in the diffraction patterns with and without an external applied ﬁeld yield a correlation pattern exclusively associated with metal–metal correlations [4]. p y Hot on the heels of the magnetic difference method is another novel contrast method, this time applied to Sm phosphates [5]. The anomalous neutron dispersion method exploits wavelength-dependent variations in the neutron scattering length of particular isotopes at an absorption resonance to provide contrast. The beauty of this method is that through generation of difference correlation functions not only can the metal–metal pair correlations be separated, but in addition all other correlations involving the metal can be obtained separately. While this work focuses on Sm phosphate glasses, its impact lies in the ﬁeld of amorphous solids as a whole. The method described takes advantage of the 14% natural abundance of 149Sm, its large neutron absorption cross-section, and in particular the accessibility of a suitable isolated absorption resonance within the wavelength range of the diffractometer. However, in principle, similar experiments could be performed on isotopes of other elements, provided their absorption resonances are within the operational range of the diffractometer. Indeed, the range of neutron wavelengths available is already quite large on reactor source instruments such as D4 at ILL, as well as time-of-ﬂight instruments such as GEM at ISIS. This situation should be further improved with proposed instruments such as NIMROD on TS2 at ISIS. In addition, the similarity of structural chemistry in rare earth phosphate glasses [6] means that distance correlations extracted for partial structure factor calculations on Sm could serve as initial values in the analysis of total correlation data from analogous systems involving other lanthanides. These could proceed using the isomorphic substitution method, which is based on the assumption that diffraction patterns in compositional analogues differ only in the coherent neutron scattering length of the modifying cations and have been used successfully in analysing Dy and Ho phosphates as well as La and Ce phosphates [7]. References 2
https://openalex.org/W2164493094	https://biomedical-engineering-online.biomedcentral.com/counter/pdf/10.1186/1475-925X-11-66	English	null	Simultaneous monitoring of static and dynamic intracranial pressure parameters from two separate sensors in patients with cerebral bleeds: comparison of findings	BioMedical engineering online	2,012	cc-by	11,094	RESEARCH Open Access Simultaneous monitoring o pressure parameters from with cerebral bleeds: comp Eide et al. Simultaneous monitoring of static and dynamic intracranial pressure parameters from two separate sensors in patients with cerebral bleeds: comparison of findings Eide et al. Eide et al. BioMedical Engineering OnLine 2012, 11:66 Simultaneous monitoring of static and dynamic intracranial pressure parameters from two separate sensors in patients with cerebral bleeds: comparison of findings Eide et al. Eide et al. BioMedical Engineering OnLine 2012, 11:66 http://www.biomedical-engineering-online.com/content/11/1/66 Eide et al. BioMedical Engineering OnLine 2012, 11:66 http://www.biomedical-engineering-online.com/content/11/1/66 Eide et al. BioMedical Engineering OnLine 2012, 11:66 Eide et al. BioMedical Engineering OnLine 2012, 11:66 http://www.biomedical-engineering-online.com/content/11/1/66 Simultaneous monitoring of static and dynamic intracranial pressure parameters from two separate sensors in patients with cerebral bleeds: comparison of findings Per Kristian Eide1,2, Sverre Holm3 and Wilhelm Sorteberg1 Per Kristian Eide1,2, Sverre Holm3 and Wilhelm Sorteberg1 * Correspondence: peide@ous-hf.no 1Department of Neurosurgery, Oslo University Hospital, Rikshospitalet, Oslo, Norway 2Faculty of Medicine, University of Oslo, Oslo, Norway Full list of author information is available at the end of the article © 2012 Eide et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. * Correspondence: peide@ous-hf.no 1Department of Neurosurgery, Oslo University Hospital, Rikshospitalet, Oslo, Norway 2Faculty of Medicine, University of Oslo, Oslo, Norway Full list of author information is available at the end of the article © 2012 Eide et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Common Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Patient recordings The ICP recordings were retrieved from patients managed for aneurysmal subarach- noid haemorrhage (SAH) and/or intra-cerebral haemorrhage (ICH) at the Department of Neurosurgery, Oslo University Hospital – Rikshospitalet during the time period 2002–2011. Only patients wherein management included simultaneous monitoring from two separate ICP sensors were included. The Regional Committee for Medical and Health Research Ethics (REK) of Health Region South-East, Norway was informed in writing, and had no objections to the study. The study was approved by the Oslo University Hospital – Rikshospitalet as a quality study. Abstract Background: We recently reported that in an experimental setting the zero pressure level of solid intracranial pressure (ICP) sensors can be altered by electrostatics discharges. Changes in the zero pressure level would alter the ICP level (mean ICP); whether spontaneous changes in mean ICP happen in clinical settings is not known. This can be addressed by comparing the ICP parameters level and waveform of simultaneous ICP signals. To this end, we retrieved our recordings in patients with cerebral bleeds wherein the ICP had been recorded simultaneously from two different sensors. Materials and Methods: During a time period of 10 years, 17 patients with cerebral bleeds were monitored with two ICP sensors simultaneously; sensor 1 was always a solid sensor while Sensor 2 was a solid -, a fluid - or an air- pouch sensor. The simultaneous signals were analyzed with automatic identification of the cardiac induced ICP waves. The output was determined in consecutive 6-s time windows, both with regard to the static parameter mean ICP and the dynamic parameters (mean wave amplitude, MWA, and mean wave rise time, MWRT). Differences in mean ICP, MWA and MWRT between the two sensors were determined. Transfer functions between the sensors were determined to evaluate how sensors reproduce the ICP waveform. Results: Comparing findings in two solid sensors disclosed major differences in mean ICP in 2 of 5 patients (40%), despite marginal differences in MWA, MWRT, and linear phase magnitude and phase. Qualitative assessment of trend plots of mean ICP and MWA revealed shifts and drifts of mean ICP in the clinical setting. The transfer function analysis comparing the solid sensor with either the fluid or air- pouch sensors revealed more variable transfer function magnitude and greater differences in the ICP waveform derived indices. Conclusions: Simultaneous monitoring of ICP using two solid sensors may show marked differences in static ICP but close to identity in dynamic ICP waveforms. This indicates that shifts in ICP baseline pressure (sensor zero level) occur clinically; trend plots of the ICP parameters also confirm this. Solid sensors are superior to fluid – and air pouch sensors when evaluating the dynamic ICP parameters. Eide et al. BioMedical Engineering OnLine 2012, 11:66 http://www.biomedical-engineering-online.com/content/11/1/66 Eide et al. BioMedical Engineering OnLine 2012, 11:66 http://www.biomedical-engineering-online.com/content/11/1/66 Page 2 of 20 Page 2 of 20 Background Monitoring of the intracranial pressure (ICP) is crucial in the management of neurosurgical patients [1–3]. The goal is then usually to keep the static pressure parameter mean ICP or the dynamic pressure parameter mean ICP wave ampli- tude (MWA) below certain threshold levels; i.e. the mean ICP <20-25 mmHg [1,2], or the MWA <5 mmHg [3]. The ICP is most often measured using a solid sensor or through a fluid catheter placed within a cerebral ventricle [4]. Though today’s practice of ICP monitoring spans decades [5], newer data have shown that the level of the ICP (i.e. the static ICP or the mean ICP) can be sensitive to inherent weaknesses in the sensors. We hence recently reported that in an experimental setting spontaneous shifts in the baseline pressure (sensor zero pressure) of a solid sensor can be triggered by elec- trostatic discharges (ESDs) [6]. From this, one may ask if this also happens in clinical settings, and whether the ICP scores that are presented on the monitoring screen represent reality or not. The questions raised above may be addressed by comparing simultaneous signals obtained from two separate sensors placed intracranially in the same patient. To this end, we retrieved our ICP recordings in all 17 patients with cerebral bleeds wherein the ICP had been monitored simultaneously from two separate sensors. We then compared the mean ICP (static pressure parameter) as well as the mean wave amplitude (MWA) and mean wave rise time (MWRT) (dynamic pressure parameters) of the two sensors. With special emphasis on the quality of the dy- namic ICP signals, the transfer function of the sensor types used in the 17 patients were also assessed. ICP monitoring and analysis The setup for the simultaneous ICP monitoring was as follows: Sensor 1 was al- ways a solid (strain-gauge) sensor (Codman Microsensor, Codman MicroSensor, Johnson and Johnson, Raynham, Massachusetts, USA), while Sensor 2 was either (a) another solid sensor (Codman Microsensor, Codman MicroSensor, Johnson and Johnson, Raynham, Massachusetts, USA; Category A), (b) a fluid sensor (Edward’s fluid sensor) connected to an external ventricular drain (Truwave PX-600 F Pres- sure Monitoring Set, Edwards Life sciences LLC, Irvine, CA, USA; Category B), or (c) Eide et al. BioMedical Engineering OnLine 2012, 11:66 http://www.biomedical-engineering-online.com/content/11/1/66 Eide et al. BioMedical Engineering OnLine 2012, 11:66 http://www.biomedical-engineering-online.com/content/11/1/66 Page 3 of 20 Page 3 of 20 an air-pouch sensor (Spiegelberg intraparenchymal probe 3PN, Spiegelberg KG, Hamburg, GE; Category C). Both ICP sensors were implanted at the same time. The ICP sensors were introduced to the intracranial compartment either via a small burr hole and a minimal opening in the dura or via the craniotomy used for aneurysm clipping/hematoma evacuation. The solid sensor was placed within the brain parenchyma, and connected via cable to the ICP Express (Codman ICP Ex- press, Johnson and Johnson, Raynham, Massachusetts, USA). The fluid sensor was connected outside the patient to an external ventricular drain (EVD) that had been placed in the ventricular fluid, while the air-pouch sensor was placed inside the brain parenchyma, and connected to a Spiegelberg ICP Monitor (Spiegelberg KG, Hamburg, Ge). The ICP signals from all sensors were passed to a vital signs Sie- mens 9000 XL Series Monitor (Siemens Medical Systems Inc., Danvers, MA, USA). By means of the Siemens Infinity Gateway Software (Siemens Medical Systems Inc., Danvers, MA, USA), the continuous ICP signals were transferred online via the hospital network to a computer server and stored as raw data files (sampling rate 100 Hz). The analysis of the continuous ICP waveforms was done using a previously pub- lished method for automatic cardiac induced ICP waves [7] that has been imple- mented in the software (Sensometrics Software, dPCom As, Oslo). In short, the process is as follows: (1) From the continuous pressure signal, each pressure wave is identified by its beginning and ending diastolic minimum pressure, and its sys- tolic maximum pressure. (2) For each pressure wave a set of single wave para- meters is determined such as rise time (dT), amplitude (dP) and rise time coefficient (RT). ICP monitoring and analysis (3) Each pressure wave is differentiated as either a cardiac-beat- induced ICP wave, or as an artefact-induced wave, depending on whether the sin- gle wave parameters meet defined requirements. Thus, cardiac induced waves have single wave parameters (e.g. dT, dP, RT) within defined threshold values 3) Then, the identified cardiac-induced waves are applied for further analysis; analysis is done during time windows of 6-s duration containing. Only 6-s time windows con- taining minimum 4 cardiac beat induced waves were considered to be of good quality, and were used for the present analysis. Thus, for each 6-s time window the mean ICP and the ICP waveform indices mean wave amplitude (MWA) and mean wave rise time (MWRT) were determined. For a 6-s time window to be accepted according to the automatic method, it contained minimum four cardiac induced waves. Accordingly, this method automatically differentiates between pres- sure waves induced by the cardiac contractions and artefact waves due to noise in the pressure signal (e.g. due to patient movement, or sensor movement or dysfunc- tion); artefact waves were hence omitted from the analysis. For this particular study, intracranial pressure recordings from the corresponding 6-s time windows containing simultaneous ICP signals and with identical time refer- ences were then compared. A 6-s time window with two ICP signals (PatID 2) is shown in Figure 1. For each signal in every recording we determined the average values of mean ICP, MWA and MWRT during the whole observation period as well as the per- centage of 6-s time windows, with differences between signals of mean ICP Eide et al. BioMedical Engineering OnLine 2012, 11:66 Page 4 of 20 http://www.biomedical-engineering-online.com/content/11/1/66 Eide et al. BioMedical Engineering OnLine 2012, 11:66 Page 4 of 20 Page 4 of 20 Eide et al. BioMedical Engineering OnLine 2012, 11:66 http://www.biomedical-engineering-online.com/content/11/1/66 g g http://www.biomedical-engineering-online.com/content/11/1/66 Figure 1 Simultaneous 6-s time windows of Signals 1 and 2 (PatID 2). A 6-s time window of the raw signal of (a) Signal 1 and (b) Signal 2 of PatID 2, showing the cardiac-induced ICP waves. For every cardiac- induced wave the amplitude (dP) and rise time (dT) were automatically determined (c). Visual inspection of trend plots of mean ICP and MWA For the 17 patient recordings, the trend plots of mean ICP and MWA for simultaneous recordings were inspected for the occurrence of shifts or drifts of mean ICP relative to MWA. This was done to qualitatively assess whether shifts in mean ICP occurred in the clinical setting. ICP monitoring and analysis Mean ICP, mean ICP wave amplitude (MWA) and rise time (MWRT) were determined for each 6-s time window; for the 6-s time window shown in a-b, mean ICP was 21.3 mmHg (Signal 1) and 10.9 mmHg (Signal 2), MWA was 4.1 mmHg (Signal 1) and 4.1 mmHg (Signal 2), and MWRT was 0.21 sec (Signal 1) and 0.20 sec (Signal 2). Figure 1 Simultaneous 6-s time windows of Signals 1 and 2 (PatID 2). A 6-s time window of the raw signal of (a) Signal 1 and (b) Signal 2 of PatID 2, showing the cardiac-induced ICP waves. For every cardiac- induced wave the amplitude (dP) and rise time (dT) were automatically determined (c). Mean ICP, mean ICP wave amplitude (MWA) and rise time (MWRT) were determined for each 6-s time window; for the 6-s time window shown in a-b, mean ICP was 21.3 mmHg (Signal 1) and 10.9 mmHg (Signal 2), MWA was 4.1 mmHg (Signal 1) and 4.1 mmHg (Signal 2), and MWRT was 0.21 sec (Signal 1) and 0.20 sec (Signal 2). >5 mmHg or >10 mmHg, MWA >1 mmHg or >2 mmHg, and MWRT >0.1 sec or >0.2 sec, respectively. >5 mmHg or >10 mmHg, MWA >1 mmHg or >2 mmHg, and MWRT >0.1 sec or >0.2 sec, respectively. Assessment of transfer function for the different types of sensors The transfer function was estimated using the Matlab function tfestimate.m (MathworksW, version R2011a with Signal Processing Toolbox version 6.15). It esti- mates the linear filter required to transform the second transducer’s data (Signal 2) into that of the reference transducer (Codman; Signal 1). The transfer function is Eide et al. BioMedical Engineering OnLine 2012, 11:66 Page 5 of 20 http://www.biomedical-engineering-online.com/content/11/1/66 Eide et al. BioMedical Engineering OnLine 2012, 11:66 Page 5 of 20 Page 5 of 20 Eide et al. BioMedical Engineering OnLine 2012, 11:66 http://www.biomedical-engineering-online.com/content/11/1/66 http://www.biomedical-engineering-online.com/content/11/1/66 the quotient of the cross power spectral density (P21) of signals 1 and 2 and the Figure 2 Transfer function of a 6-s time window of PatID 3. Analysis of one 6-s time window of PatID 3 showing (a) signals 1 (Codman) and 2 (dash-dot line, Codman), (b) transfer function of magnitude, and (c) transfer function of phase with dash-dot line indicating estimate based on a delay of 0 sec. In this case, 15.6% of 6-s time windows showed a difference in MWA >1 mmHg (largest MWA in Signal 1 in 11.3% and largest MWA in Signal 2 in 4.3%). In 0.5% of 6-s time windows, the difference in MWRT was >0.1 sec (largest MWRT in Signal 1 in 0.2% and largest MWRT in Signal 2 in 0.3%). Figure 2 Transfer function of a 6-s time window of PatID 3. Analysis of one 6-s time window of PatID 3 showing (a) signals 1 (Codman) and 2 (dash-dot line, Codman), (b) transfer function of magnitude, and (c) transfer function of phase with dash-dot line indicating estimate based on a delay of 0 sec. In this case, 15.6% of 6-s time windows showed a difference in MWA >1 mmHg (largest MWA in Signal 1 in 11.3% and largest MWA in Signal 2 in 4.3%). In 0.5% of 6-s time windows, the difference in MWRT was >0.1 sec (largest MWRT in Signal 1 in 0.2% and largest MWRT in Signal 2 in 0.3%). the quotient of the cross power spectral density (P21) of signals 1 and 2 and the power spectral density (P11) of signal 1, T21(f) = P21(f)/P11(f). The transfer function has a magnitude and a phase as indicated in Figures 2, 3,4 b and c. It has been Eide et al. BioMedical Engineering OnLine 2012, 11:66 Page 6 of 20 http://www.biomedical-engineering-online.com/content/11/1/66 Eide et al. Assessment of transfer function for the different types of sensors BioMedical Engineering OnLine 2012, 11:66 Page 6 of 20 Page 6 of 20 http://www.biomedical-engineering-online.com/content/11/1/66 estimated by averaging over Fourier transformed 6 s time windows spanning from Figure 3 Transfer function of a 6-s time window of PatID 6. Analysis of one 6-s time window of PatID 6 showing (a) the signals 1(Codman) and 2 (dash-dot line, Edwards), (b) transfer function of magnitude, and (c) transfer function of phase with dash-dot line indicating estimate based on a delay of 0.065 sec. In this case, 47.1% of 6-s time windows showed a difference in MWA >1 mmHg (largest MWA in Signal 1 in 18.9% and largest MWA in Signal 2 in 28.2%). In 12.8% of 6-s time windows, the difference in MWRT was >0.1 sec (largest MWRT in Signal 1 in 12.7% and largest MWRT in Signal 2 in 0.1%). Figure 3 Transfer function of a 6-s time window of PatID 6. Analysis of one 6-s time window of PatID 6 showing (a) the signals 1(Codman) and 2 (dash-dot line, Edwards), (b) transfer function of magnitude, and (c) transfer function of phase with dash-dot line indicating estimate based on a delay of 0.065 sec. In this case, 47.1% of 6-s time windows showed a difference in MWA >1 mmHg (largest MWA in Signal 1 in 18.9% and largest MWA in Signal 2 in 28.2%). In 12.8% of 6-s time windows, the difference in MWRT was >0.1 sec (largest MWRT in Signal 1 in 12.7% and largest MWRT in Signal 2 in 0.1%). estimated by averaging over Fourier transformed 6-s time windows, spanning from a minimum of 36 seconds (6 segments). Because the mean value was subtracted for each time window, the transfer function analysis is not influenced by variations Eide et al. BioMedical Engineering OnLine 2012, 11:66 Page 7 of 20 http://www.biomedical-engineering-online.com/content/11/1/66 Eide et al. BioMedical Engineering OnLine 2012, 11:66 Page 7 of 20 Page 7 of 20 Eide et al. BioMedical Engineering OnLine 2012, 11:66 http://www.biomedical-engineering-online.com/content/11/1/66 http://www.biomedical-engineering-online.com/content/11/1/66 in mean ICP. The phase corresponds to the delay, τ, between the two signals di t φ 2 f It h b h k d b ti ti th d l i th ti Figure 4 Transfer function of a 6-s time window of PatID 15. Assessment of transfer function for the different types of sensors Analysis of one 6-s time window of PatID 15 showing (a) signal 1 (Codman) and 2 (dash-dot line, Spiegelberg), (b) transfer function of magnitude, and (c) transfer function of phase with dash-dot line indicating estimate based on a delay of 0.12 sec. In this case, 80.4% of 6-s time windows showed a difference in MWA >1 mmHg (largest MWA in Signal 1 in 51.6% and largest MWA in Signal 2 in 28.8%). In 7.6% of 6-s time windows, the difference in MWRT was >0.1 sec (largest MWRT in Signal 1 in 3.2% and largest MWRT in Signal 2 in 4.4%). Figure 4 Transfer function of a 6-s time window of PatID 15. Analysis of one 6-s time window of PatID 15 showing (a) signal 1 (Codman) and 2 (dash-dot line, Spiegelberg), (b) transfer function of magnitude, and (c) transfer function of phase with dash-dot line indicating estimate based on a delay of 0.12 sec. In this case, 80.4% of 6-s time windows showed a difference in MWA >1 mmHg (largest MWA in Signal 1 in 51.6% and largest MWA in Signal 2 in 28.8%). In 7.6% of 6-s time windows, the difference in MWRT was >0.1 sec (largest MWRT in Signal 1 in 3.2% and largest MWRT in Signal 2 in 4.4%). in mean ICP. The phase corresponds to the delay, τ, between the two signals according to φ = −2πfτ. It has been checked by estimating the delay in the time domain. This can be done by inspecting the time plots (Figures 2,3 and 4a), but a Eide et al. BioMedical Engineering OnLine 2012, 11:66 http://www.biomedical-engineering-online.com/content/11/1/66 Eide et al. BioMedical Engineering OnLine 2012, 11:66 http://www.biomedical-engineering-online.com/content/11/1/66 Eide et al. BioMedical Engineering OnLine 2012, 11:66 http://www.biomedical-engineering-online.com/content/11/1/66 Eide et al. BioMedical Engineering OnLine 2012, 11:66 http://www.biomedical-engineering-online.com/content/11/1/66 Page 8 of 20 Page 8 of 20 more robust measure can be found by estimating the cross-correlation function (Matlab xcorr.m). The phase is a linear function which has been plotted with red dotted lines on top of the transfer function phase. In all of the transfer-function plots it matches very well with the estimated phase. Patient recordings Table 1 gives age, gender and type of bleed in the 17 patients included in the study, while Table 2 shows type and location of the ICP sensors and number of accepted (good quality) 6-s time windows containing two signals. For all patients combined, a total of 441,654 6-s time windows were analyzed (Table 2). ACOM: anterior communicating artery; BA: basilar artery; ICA: internal carotid artery; artery; PCOM: posterior communicating artery; SAH: Subarachnoid haemorrhage; VA b l ACOM: anterior communicating artery; BA: basilar artery; ICA: internal carotid artery; ICH: Intra-cerebral haemorrhage; IVH: intra-ventricular haematoma; MCA: middle cerebra g ; ; artery; PCOM: posterior communicating artery; SAH: Subarachnoid haemorrhage; VA: vertebral artery Mean ICP Table 3, left presents the mean ICP recorded by the two sensors. Differences in mean ICP >5 mmHg in >20% of observations were seen in 2 of 5 patients (40%) when using two solid sensors (Additional file 1: Category A; Table 3, right), in 4 of 5 patients (80%) when using one solid and one fluid sensor (Additional file 2: Category B), and in 1of 7 patients (14%) when using one solid and one air-pouch sensor (Additional file 3: Category C). Table 1 Demographic data of 17 patients with cerebral bleeds PatID Age Gender Type of bleed Category A 1 66 M SAH (ACOM) 2 76 M ICH (right parieto-occipital) 3 39 F SAH (ACOM) 4 72 F SAH (left MCA) 5 59 F SAH (left MCA) Category B 6 56 M SAH (BA) 7 48 M SAH (left MCA) 8 60 M SAH (ACOM) 9 50 F SAH (right VA) 10 55 F SAH (ACOM) Category C 11 66 M ICH (right frontal)/IVH 12 56 F SAH (right MCA) 13 60 F SAH (BA/left ICA) 14 54 M SAH (left PCOM) 15 67 M SAH (right PCOM) 16 71 M ICH (cerebellum) 17 82 F ICH (cerebellum) ACOM: anterior communicating artery; BA: basilar artery; ICA: internal carotid artery; ICH Intra cerebral haemorrhage IVH intra entric lar haematoma MCA middle cerebral Table 1 Demographic data of 17 patients with cerebral bleeds PatID Age Gender Type of bleed Table 1 Demographic data of 17 patients with cerebral bleeds Page 9 of 20 Page 9 of 20 Eide et al. Mean ICP BioMedical Engineering OnLine 2012, 11:66 http://www.biomedical-engineering-online.com/content/11/1/66 Table 2 Sensor type, sensor location and number of accepted simultaneous 6-s time windows PatID Sensor type Sensor location Number of simultaneous 6-s time windows Signal 1 Signal 2 Signal 1 Signal 2 Category A 1 Solid Solid Left frontal lobe Right frontal lobe 6,149 2 “ “ Left frontallobe Left occipital lobe 31,488 3 “ “ Right frontal lobe Right occipital lobe 8,208 4 “ “ Left frontal lobe Left frontal ventricular horn 21,435 5 “ “ Right frontal lobe Right frontal ventricular horn 19,703 Category B 6 Solid Fluid Right frontal lobe Right frontal ventricular horn 6,295 7 “ “ Right frontal lobe Right frontal ventricular horn 26,677 8 “ “ Right frontal lobe Right frontal ventricular horn 4,489 9 “ “ Right frontal lobe Right frontal ventricular horn 2,555 10 “ “ Left frontal lobe Left frontal ventricular horn 6,262 Category C 11 Solid Air-pouch Right frontal lobe Right frontal ventricular horn 44,445 12 “ “ Right frontal lobe Right frontal ventricular horn 93,506 13 “ “ Right frontal lobe Right frontal ventricular horn 52,137 14 “ “ Right frontal lobe Right frontal ventricular horn 37,192 15 “ “ Left frontal lobe Left frontal ventricular horn 38,274 16 “ “ Right frontal lobe Right frontal ventricular horn 588 17 “ “ Right frontal lobe Right frontal ventricular horn 42,251 Table 2 Sensor type, sensor location and number of accepted simultaneous 6-s time i d Mean ICP wave amplitude (MWA) Table 4, gives the MWA recorded by the two sensors and percentage of simultaneous 6-s time windows where the MWA of the two sensors differed >1 mm Hg and >2 mm Hg, respectively. Differences were marginal when using two solid sensors. (Additional file 1: Category A). The patient with two solid sensors that had differences in >1 mmHg in 16% of the 6-s time windows (PatID 3), were one out of two patients where one sen- sor had been placed frontal and the other occipital. The other patient with such sensor placements (PatID 2), presented with considerable differences in mean ICP but with similar MWAs. A difference in MWA >1 mmHg in more than 20% of the observations was seen in 4/5 (80%) of patients when using one solid and one fluid sensor (Additional file 2: Category B), and in all 7 patients (100%) when using one solid and one air-pouch sen- sor (Additional file 3: Category C; Table 4). Examples of trend plots of mean ICPs and MWAs Figures 5,6,7,8,9,10,11,12 and 13 show trend plots of mean ICPs and MWAs in 9 different patients. Figures 5-8 are from Category A patients (PatID´s 1–4) and compared findings in two solid sensors. The figures demonstrate shifts and drifts in mean ICP of both sensors, but with different profiles of shifts between the two simultaneous signals. Figures 9-10 are from Category B patients (PatID´s 9–10), comparing one solid and one fluid sensor. The figures reveal shifts in mean ICP of the solid sensor, but with no accompanying shifts in the fluid sensor. Finally, Figures 11-13 present findings in Category C Patients (PatID´s 11, 13 and 15), comparing one solid and one air pouch sensor. Shifts in mean ICP of the air- pouch sensor were found in both PatID’s 12 and 13. In PatID 15 the shifts of mean ICP oc- curred in opposite directions for the solid and the air-pouch sensor. Mean ICP wave rise time (MWRT) Table 5, presents MWRT as well as percentage of simultaneous 6-s time windows where the MWRT of the two sensors differed >0.1 s and >0.2 s, respectively. Mar- ginal differences in MWRT were seen when comparing two solid sensors were seen (Category A). Also, when comparing one solid and one fluid sensor (Category B), Eide et al. BioMedical Engineering OnLine 2012, 11:66 http://www.biomedical-engineering-online.com/content/11/1/66 Page 10 of 20 Page 10 of 20 Table 3 Differences in mean ICP between simultaneous ICP signals PatID Mean ICP (mmHg; average + std) Percentage of 6-s time windows with difference in mean ICP: Signal 1 Signal 2 >5 mmHg >10 mmHg Category A 1 5.3 + 2.7 8.1 + 2.3 2 - 2 9.2 + 2.9 18.1 + 3.6 100 50 3 9.3 + 3.7 9.5 + 3.0 11 2 4 8.1 + 2.9 2.7 + 2.3 68 - 5 9.1 + 2.3 5.8 + 2.5 - - Category B 6 15.3 + 3.4 20.3 + 2.3 37 2 7 9.3 + 4.6 7.6 + 5.5 35 4 8 6.9 + 4.1 9.5 + 5.3 37 - 9 −0.48 + 2.4 −29.6 + 4.6 100 100 10 11.9 + 5.4 12.5 + 5.2 5 - Category C 11 11.4 + 2.6 14.8 + 2.3 4 3 12 9.3 + 3.5 9.5 + 3.4 1 1 13 10.0 + 3.6 9.0 + 5.1 19 6 14 10.3 + 2.8 5.9 + 6.5 14 9 15 7.8 + 14.3 8.5 + 4.1 27 6 16 14.1 + 1.9 10.6 + 3.7 14 1 17 15.8 + 5.5 15.5 + 3.8 8 5 Table 3 Differences in mean ICP between simultaneous ICP signals differences were small. In contrast, when comparing one solid and one air-pouch sensor, differences were >0.1 s in more than 20% of observations in 3 of 7 patient recordings (43%; Category C). Impact of sensor characteristics on ICP waveform reproduction When comparing one solid and one air-pouch sensor, the transfer function falls off for frequencies above 60–120 heart beats per minute (1–2 Hz) (Figure 4b). In this subject (PatID 15), the delay was linear in the same lim- ited area. The heart rate was about 40 beats per minute (0.63 Hz), and the air pouch sensor’s gain was about 2 dB below that of the solid sensor. This is consistent with the statistical data for this subject since for the 80% of time windows where the MWA dif- fered more than 1 mmHg, the majority (51.6% of all time windows) the MWA was lar- ger for the Codman sensor. In this case the transfer function showed a clear low pass character, but in other cases this was not the case and varied much more over the interesting frequency interval up to 5–6 Hz. Impact of sensor characteristics on ICP waveform reproduction While comparison of two solid sensors disclosed flat transfer function magnitude (Figure 2b) and no phase delay (lower panel), Figure 2 (Figure 2c), comparing one solid and one fluid sensor indicated some delay in the fluid sensor (Figure 3c). The delay was even larger when comparing one solid and one air-pouch sensor (Figure 4c). In the Page 11 of 20 Eide et al. BioMedical Engineering OnLine 2012, 11:66 http://www.biomedical-engineering-online.com/content/11/1/66 Page 11 of 20 Table 4 Differences in mean wave amplitude (MWA) between simultaneous ICP signals PatID MWA (mmHg; average + std) Percentage of 6-s time windows with difference in MWA: Signal 1 Signal 2 >1 mmHg >2 mmHg Category A 1 4.5 + 0.7 4.4 + 0.7 2 - 2 4.5 + 0.7 4.5 + 0.7 1 - 3 6.8 + 1.2 7.0 + 1.2 16 - 4 3.2 + 0.5 3.1 + 0.5 - - 5 1.9 + 0.8 1.6 + 0.6 - - Category B 6 6.4 + 0.9 6.5 + 1.2 47 10 7 4.5 + 1.2 4.3 + 1.3 23 7 8 8.9 + 3.2 8.9 + 4.1 59 21 9 3.0 + 0.3 2.4 + 0.6 19 - 10 7.1 + 3.1 6.6 + 2.9 27 9 Category C 11 7.2 + 1.0 6.2 + 0.9 52 4 12 5.6 + 1.8 4.9 + 1.3 41 18 13 4.4 + 1.4 3.9 + 1.7 29 2 14 6.7 + 1.7 5.2 + 3.2 90 65 15 7.6 + 14.3 7.1 + 14.3 80 64 16 9.3 + 0.5 8.3 + 0.6 30 1 17 8.0 + 2.1 7.3 + 2.3 38 6 MWA: mean wave amplitude. Table 4 Differences in mean wave amplitude (MWA) between simultaneous ICP signals example retrieved from PatID 6 (Figure 3), the heart rate was about 75 beats per mi- nute (1.3 Hz), at which frequency the transfer function was positive (Figure 3b). The phase was linear up to 5–6 Hz (Figure 3c). The positive gain at the frequency of the heart beat is consistent with a larger estimate for the MWA for the Edwards sensor. This was the case in the majority of the 6-second segments with different MWA-values. Of the 47% of segments where the MWA values differed more than 1 mmHg, 28.2% of all segments were for this case. Discussion The main findings of the present study were that when comparing solid ICP sensors, major differences in ICP level despite close to identical ICP waveform occurred in 2 of 5 (40%) of patients. Inspection of the trend plots of static and dynamic pressure Eide et al. BioMedical Engineering OnLine 2012, 11:66 http://www.biomedical-engineering-online.com/content/11/1/66 Page 12 of 20 Page 12 of 20 Table 5 Differences in mean wave rise time (MWRT) between simultaneous ICP signals PatID MWRT (sec; average + std) Percentage of 6-s time windows with difference in MWRT: Signal 1 Signal 2 >0.1 s >0.2 s Category A 1 0.27 + 0.04 0.27 + 0.04 - - 2 0.21 + 0.02 0.21 + 0.02 - - 3 0.14 + 0.04 0.14 + 0.03 1 - 4 0.18 + 0.05 0.18 + 0.05 - - 5 0.28 + 0.06 0.28 + 0.05 - - Category B 6 0.29 + 0.04 0.24 + 0.04 13 - 7 0.25 + 0.02 0.25 + 0.04 2 - 8 0.23 + 0.02 0.21 + 0.03 2 - 9 0.14 + 0.04 0.14 + 0.04 3 1 10 0.28 + 0.03 0.24 + 0.06 14 - Category C 11 0.26 + 0.05 0.28 + 0.02 9 - 12 0.33 + 0.13 0.23 + 0.09 45 17 13 0.36 + 0.06 0.45 + 0.12 50 20 14 0.33 + 0.09 0.39 + 0.10 25 7 15 0.22 + 14.3 0.22 + 14.3 8 1 16 0.12 + 0.01 0.12 + 0.01 - - 17 0.24 + 0.05 0.25 + 0.03 7 - MWRT: mean wave rise time. Table 5 Differences in mean wave rise time (MWRT) between simultaneous ICP signals parameters showed that shifts in mean ICP (due to spontaneously changed ICP base- line pressure) occur clinically, most often in the solid sensors. The ICP waveform is reproduced with variable quality by the fluid-filled and air pouch sensors. The ICP sensors applied in this study are widely used. The solid Codman sensor [8–12] was introduced in the 1980’s while the air-pouch Spiegelberg ICP sensor [13,14] has been used since the 1990’s. The fluid Edward’s sensor has been used extensively in monitoring of various fluid-pressures such as arterial blood pressure, intraventricular pressure and central venous pressure. All of these technologies represent state-of-the art ICP monitoring. ICP parameters of the study The trend plots of mean ICP and mean ICP wave amplitude (MWA) of (a) Signal 1 and (b) Signal 2 of PatID 1 is shown [Average of mean ICP: 7.6 mmHg (Signal 1), 4.3 mmHg (Signal 2). Average of mean ICP wave amplitude: 4.1 mmHg (Signal 1), 4.0 mmHg (Signal 2)]. The vertical arrows indicate sudden changes in mean ICP without accompanying changes in MWA; mean ICP changed differently for the two signals, being more extensive for Signal 2. mean ICP would have been considerably higher if including all 6-s time windows. In the clinical setting, usually the ICP level (mean ICP) is determined without considering the quality of the ICP signal, i.e. whether the ICP signal contains cardiac induced ICP waves or not. However, given the retrospective design of the study, we are unable to as- sess to which degree patient management and patient outcome were influenced by the differences in the two signals. With regard to the MWA, the 6-s time window data were presented as percentage of differences >1 or >2 mmHg, mm respectively. These thresholds were chosen as our ex- perience suggests an upper normal threshold value in the MWA of 4–5 mmHg [3,15]. Accordingly, a difference in MWA >1 mmHg could have impact on patient Figure 6 Trend plots of mean ICP/mean ICP wave amplitude of Signals 1 and 2 (PatID 2). The trend plots of mean ICP and mean ICP wave amplitude (MWA) of (a) Signal 1 and (b) Signal 2 of PatID 2 is shown [Average of mean ICP: 8.4 mmHg (Signal 1), 15.2 mmHg (Signal 2). Average of mean ICP wave amplitude: 4.8 mmHg (Signal 1), 4.8 mmHg (Signal 2)]. The vertical arrow in (a) indicates a drift in mean ICP without accompanying change in MWA. While mean ICP drifted in (a), no such simultaneous change in mean ICP in (b) was seen. Figure 6 Trend plots of mean ICP/mean ICP wave amplitude of Signals 1 and 2 (PatID 2). The trend plots of mean ICP and mean ICP wave amplitude (MWA) of (a) Signal 1 and (b) Signal 2 of PatID 2 is shown [Average of mean ICP: 8.4 mmHg (Signal 1), 15.2 mmHg (Signal 2). Average of mean ICP wave amplitude: 4.8 mmHg (Signal 1), 4.8 mmHg (Signal 2)]. The vertical arrow in (a) indicates a drift in mean ICP without accompanying change in MWA. ICP parameters of the study Both static and dynamic intracranial pressure parameters were registered and data are presented as average values for the whole observation period and as well as percentage of differences between simultaneous 6-s time periods. When monitoring ICP as surveil- lance of the critically ill patient, it is the score at a certain point of time or during a shorter time period that guides patient management; here that corresponds to the 6-s time periods. Though it can be disputed what may be regarded as major differences in mean ICP, the present findings that the mean ICP differed >5 mm Hg in >20% of the 6-s time windows in 2 out of 5 patients when comparing findings in two solid sensors and in 4 out of 5 patients when comparing scores from one solid and one fluid sensor raises concerns. Since we compared only accepted 6-s time windows, the differences in Page 13 of 20 Page 13 of 20 Eide et al. BioMedical Engineering OnLine 2012, 11:66 Eide et al. BioMedical Engineering OnLine 2012, 11:66 http://www.biomedical-engineering-online.com/content/11/1/66 http://www.biomedical-engineering-online.com/content/11/1/66 Figure 5 Trend plots of mean ICP/mean ICP wave amplitude of Signals 1 and 2 (PatID 1). The trend plots of mean ICP and mean ICP wave amplitude (MWA) of (a) Signal 1 and (b) Signal 2 of PatID 1 is shown [Average of mean ICP: 7.6 mmHg (Signal 1), 4.3 mmHg (Signal 2). Average of mean ICP wave amplitude: 4.1 mmHg (Signal 1), 4.0 mmHg (Signal 2)]. The vertical arrows indicate sudden changes in mean ICP without accompanying changes in MWA; mean ICP changed differently for the two signals, being more extensive for Signal 2. Figure 5 Trend plots of mean ICP/mean ICP wave amplitude of Signals 1 and 2 (PatID 1). The trend plots of mean ICP and mean ICP wave amplitude (MWA) of (a) Signal 1 and (b) Signal 2 of PatID 1 is shown [Average of mean ICP: 7.6 mmHg (Signal 1), 4.3 mmHg (Signal 2). Average of mean ICP wave amplitude: 4.1 mmHg (Signal 1), 4.0 mmHg (Signal 2)]. The vertical arrows indicate sudden changes in mean ICP without accompanying changes in MWA; mean ICP changed differently for the two signals, being more extensive for Signal 2. Figure 5 Trend plots of mean ICP/mean ICP wave amplitude of Signals 1 and 2 (PatID 1). ICP parameters of the study While mean ICP drifted in (a), no such simultaneous change in mean ICP in (b) was seen. Page 14 of 20 Eide et al. BioMedical Engineering OnLine 2012, 11:66 http://www.biomedical-engineering-online.com/content/11/1/66 Figure 7 Trend plots of mean ICP/mean ICP wave amplitude of Signals 1 and 2 (PatID 3). The trend plots of mean ICP and mean ICP wave amplitude (MWA) of (a) Signal 1 and (b) Signal 2 of PatID 3 is shown [Average of mean ICP: 3.2 mmHg (Signal 1), 11.3 mmHg (Signal 2). Average of mean ICP wave amplitude: 6.1 mmHg (Signal 1), 6.4 mmHg (Signal 2)]. The vertical arrow in (b) indicates a shift in mean ICP without accompanying changes in MWA; no accompanying change in mean ICP in (a) was seen. Figure 7 Trend plots of mean ICP/mean ICP wave amplitude of Signals 1 and 2 (PatID 3). The trend plots of mean ICP and mean ICP wave amplitude (MWA) of (a) Signal 1 and (b) Signal 2 of PatID 3 is shown [Average of mean ICP: 3.2 mmHg (Signal 1), 11.3 mmHg (Signal 2). Average of mean ICP wave amplitude: 6.1 mmHg (Signal 1), 6.4 mmHg (Signal 2)]. The vertical arrow in (b) indicates a shift in mean ICP without accompanying changes in MWA; no accompanying change in mean ICP in (a) was seen. Figure 7 Trend plots of mean ICP/mean ICP wave amplitude of Signals 1 and 2 (PatID 3). The trend plots of mean ICP and mean ICP wave amplitude (MWA) of (a) Signal 1 and (b) Signal 2 of PatID 3 is shown [Average of mean ICP: 3.2 mmHg (Signal 1), 11.3 mmHg (Signal 2). Average of mean ICP wave amplitude: 6.1 mmHg (Signal 1), 6.4 mmHg (Signal 2)]. The vertical arrow in (b) indicates a shift in mean ICP without accompanying changes in MWA; no accompanying change in mean ICP in (a) was seen. management and patient outcome. It should be noted that MWRT differed between the solid - and the fluid sensor, and particularly between the solid - and the air-pouch sensors (Table 5). When the MWRT was shorter for the fluid sensors this is probably related to the fact that EVDs may be kept open during monitoring, giving shortened MWRT. For the air- pouch sensor, it is related to the transfer function with variable phase. ICP parameters of the study The combination of findings when comparing scores from two solid sensors with considerable difference in mean ICP but close to identity in the MWAs, are very simi- lar to results we have previously obtained using two solid sensors when monitoring neurosurgical patients [16,17], and thus should reflect reality. Others also previously Figure 8 Trend plots of mean ICP/mean ICP wave amplitude of Signals 1 and 2 (PatID 4). The trend plots of mean ICP and mean ICP wave amplitude (MWA) of (a) Signal 1 and (b) Signal 2 of PatID 4 is shown [Average of mean ICP: 8.6 mmHg (Signal 1), 1.5 mmHg (Signal 2). Average of mean ICP wave amplitude: 3.3 mmHg (Signal 1), 3.2 mmHg (Signal 2)]. The vertical arrows in (a) and (b) indicate shifts in mean ICP without accompanying changes in MWA; mean ICP shifted differently for the two signals, being more extensive for Signal 1. Figure 8 Trend plots of mean ICP/mean ICP wave amplitude of Signals 1 and 2 (PatID 4). The trend plots of mean ICP and mean ICP wave amplitude (MWA) of (a) Signal 1 and (b) Signal 2 of PatID 4 is shown [Average of mean ICP: 8.6 mmHg (Signal 1), 1.5 mmHg (Signal 2). Average of mean ICP wave amplitude: 3.3 mmHg (Signal 1), 3.2 mmHg (Signal 2)]. The vertical arrows in (a) and (b) indicate shifts in mean ICP without accompanying changes in MWA; mean ICP shifted differently for the two signals, being more extensive for Signal 1. Page 15 of 20 Eide et al. BioMedical Engineering OnLine 2012, 11:66 Eide et al. BioMedical Engineering OnLine 2012, 11:66 http://www.biomedical-engineering-online.com/content/11/1/66 http://www.biomedical-engineering-online.com/content/11/1/66 Figure 9 Trend plots of mean ICP/mean ICP wave amplitude of Signals 1 and 2 (PatID 6). The trend plots of mean ICP and mean ICP wave amplitude (MWA) of (a) Signal 1 and (b) Signal 2 of PatID 6 is shown [Average of mean ICP: 15.7 mmHg (Signal 1), 19.5 mmHg (Signal 2). Average of mean ICP wave amplitude: 6.7 mmHg (Signal 1), 6.7 mmHg (Signal 2)]. The vertical arrow in (a) indicates a shift in mean ICP without accompanying change in MWA. The shift of mean ICP in (a) was not accompanied by a simultaneous shift in mean ICP in (b). showed marked differences in level of ICP measured from two different ICP sensors though the ICP waveforms were not compared [10]. ICP parameters of the study Figure 9 Trend plots of mean ICP/mean ICP wave amplitude of Signals 1 and 2 (PatID 6). The trend plots of mean ICP and mean ICP wave amplitude (MWA) of (a) Signal 1 and (b) Signal 2 of PatID 6 is shown [Average of mean ICP: 15.7 mmHg (Signal 1), 19.5 mmHg (Signal 2). Average of mean ICP wave amplitude: 6.7 mmHg (Signal 1), 6.7 mmHg (Signal 2)]. The vertical arrow in (a) indicates a shift in mean ICP without accompanying change in MWA. The shift of mean ICP in (a) was not accompanied by a simultaneous shift in mean ICP in (b). Figure 9 Trend plots of mean ICP/mean ICP wave amplitude of Signals 1 and 2 (PatID 6). The trend plots of mean ICP and mean ICP wave amplitude (MWA) of (a) Signal 1 and (b) Signal 2 of PatID 6 is shown [Average of mean ICP: 15.7 mmHg (Signal 1), 19.5 mmHg (Signal 2). Average of mean ICP wave amplitude: 6.7 mmHg (Signal 1), 6.7 mmHg (Signal 2)]. The vertical arrow in (a) indicates a shift in mean ICP without accompanying change in MWA. The shift of mean ICP in (a) was not accompanied by a simultaneous shift in mean ICP in (b). showed marked differences in level of ICP measured from two different ICP sensors though the ICP waveforms were not compared [10]. showed marked differences in level of ICP measured from two different ICP sensors though the ICP waveforms were not compared [10]. Properties of the ICP sensors Figure 11 Trend plots of mean ICP/mean ICP wave amplitude of Signals 1 and 2 (PatID 12). The trend plots of mean ICP and mean ICP wave amplitude (MWA) of (a) Signal 1 and (b) Signal 2 of PatID 12 is shown [Average of mean ICP: 6.8 mmHg (Signal 1), 11.9 mmHg (Signal 2). Average of mean ICP wave amplitude: 5.6 mmHg (Signal 1), 5.7 mmHg (Signal 2)]. The vertical arrow in (b) indicates a shift in mean ICP without accompanying change in MWA. The shift of mean ICP in (b) was not accompanied by a simultaneous shift in mean ICP in (a). Figure 11 Trend plots of mean ICP/mean ICP wave amplitude of Signals 1 and 2 (PatID 12). The trend plots of mean ICP and mean ICP wave amplitude (MWA) of (a) Signal 1 and (b) Signal 2 of PatID 12 is shown [Average of mean ICP: 6.8 mmHg (Signal 1), 11.9 mmHg (Signal 2). Average of mean ICP wave amplitude: 5.6 mmHg (Signal 1), 5.7 mmHg (Signal 2)]. The vertical arrow in (b) indicates a shift in mean ICP without accompanying change in MWA. The shift of mean ICP in (b) was not accompanied by a simultaneous shift in mean ICP in (a). even higher cut-off. This frequency response is higher than the frequency contents anticipated in the ICP signal and should provide for good waveform reproduction. According to the manufacturers, the frequency response of the fluid Edward’s sensor is 40 Hz (response goes to > 200 Hz for the transducer alone). With regard to the air- pouch Spiegelberg sensor, its cut-off was reported by Czosnyka et al. [22] to be 4–5 Hz. We presently found it to be even lower (1–2 Hz). The time-domain method depends on as exact as possible reproduction of the time domain waveform for measuring peak values, peak excursions and latencies. This means that the transfer function between different sensors should be flat in the most Figure 12 Trend plots of mean ICP/mean ICP wave amplitude of Signals 1 and 2 (PatID 13). The trend plots of mean ICP and mean ICP wave amplitude (MWA) of (a) Signal 1 and (b) Signal 2 of PatID 13 is shown [Average of mean ICP: 9.8 mmHg (Signal 1), 5.0 mmHg (Signal 2). Average of mean ICP wave amplitude: 3.4 mmHg (Signal 1), 2.5 mmHg (Signal 2)]. Properties of the ICP sensors With regard to the quality of the ICP signal, focus has previously been on the long- term-drift of the sensors, their sensitivity to temperature changes, and sensor accuracy comparisons [18–21]. In contrast, the literature has been scarce on specifications of the ICP sensors. Czosnyka et al. [18] reported that the Codman frequency response goes to more than 30 Hz (slew rate −2200 mm Hg/s), and Piper and Miller [8] indicated an Figure 10 Trend plots of mean ICP/mean ICP wave amplitude of Signals 1 and 2 (PatID 7). The trend plots of mean ICP and mean ICP wave amplitude (MWA) of (a) Signal 1 and (b) Signal 2 of PatID 7 is shown [Average of mean ICP: 16.3 mmHg (Signal 1), 6.4 mmHg (Signal 2). Average of mean ICP wave amplitude: 4.5 mmHg (Signal 1), 4.2 mmHg (Signal 2)]. The vertical arrows in (a) indicate shifts in mean ICP without accompanying changes in MWA. The shifts of mean ICP in (a) were not accompanied by simultaneous shifts in mean ICP in (b). Figure 10 Trend plots of mean ICP/mean ICP wave amplitude of Signals 1 and 2 (PatID 7). The trend plots of mean ICP and mean ICP wave amplitude (MWA) of (a) Signal 1 and (b) Signal 2 of PatID 7 is shown [Average of mean ICP: 16.3 mmHg (Signal 1), 6.4 mmHg (Signal 2). Average of mean ICP wave amplitude: 4.5 mmHg (Signal 1), 4.2 mmHg (Signal 2)]. The vertical arrows in (a) indicate shifts in mean ICP without accompanying changes in MWA. The shifts of mean ICP in (a) were not accompanied by simultaneous shifts in mean ICP in (b). Page 16 of 20 Eide et al. BioMedical Engineering OnLine 2012, 11:66 Eide et al. BioMedical Engineering OnLine 2012, 11:66 http://www.biomedical-engineering-online.com/content/11/1/66 http://www.biomedical-engineering-online.com/content/11/1/66 Figure 11 Trend plots of mean ICP/mean ICP wave amplitude of Signals 1 and 2 (PatID 12). The trend plots of mean ICP and mean ICP wave amplitude (MWA) of (a) Signal 1 and (b) Signal 2 of PatID 12 is shown [Average of mean ICP: 6.8 mmHg (Signal 1), 11.9 mmHg (Signal 2). Average of mean ICP wave amplitude: 5.6 mmHg (Signal 1), 5.7 mmHg (Signal 2)]. The vertical arrow in (b) indicates a shift in mean ICP without accompanying change in MWA. The shift of mean ICP in (b) was not accompanied by a simultaneous shift in mean ICP in (a). Properties of the ICP sensors The vertical arrow in (b) indicates a shift in mean ICP without accompanying change in MWA. The shift of mean ICP in (b) was not accompanied by a simultaneous shift of mean ICP in (a). Figure 12 Trend plots of mean ICP/mean ICP wave amplitude of Signals 1 and 2 (PatID 13). The trend plots of mean ICP and mean ICP wave amplitude (MWA) of (a) Signal 1 and (b) Signal 2 of PatID 13 is shown [Average of mean ICP: 9.8 mmHg (Signal 1), 5.0 mmHg (Signal 2). Average of mean ICP wave amplitude: 3.4 mmHg (Signal 1), 2.5 mmHg (Signal 2)]. The vertical arrow in (b) indicates a shift in mean ICP without accompanying change in MWA. The shift of mean ICP in (b) was not accompanied by a simultaneous shift of mean ICP in (a). Page 17 of 20 Page 17 of 20 Eide et al. BioMedical Engineering OnLine 2012, 11:66 Eide et al. BioMedical Engineering OnLine 2012, 11:66 http://www.biomedical-engineering-online.com/content/11/1/66 http://www.biomedical-engineering-online.com/content/11/1/66 Figure 13 Trend plots of mean ICP/mean ICP wave amplitude of Signals 1 and 2 (PatID 15). The trend plots of mean ICP and mean ICP wave amplitude (MWA) of (a) Signal 1 and (b) Signal 2 of PatID 15 is shown [Average of mean ICP: -1.3 mmHg (Signal 1), 7.2 mmHg (Signal 2). Average of mean ICP wave amplitude: 8.7 mmHg (Signal 1), 5.5 mmHg (Signal 2)]. The vertical arrows in (a) and (b) indicate shifts in mean ICP; no apparent lasting changes in MWA were seen. The shifts of mean ICP in (a) and (b) were in opposite directions. Figure 13 Trend plots of mean ICP/mean ICP wave amplitude of Signals 1 and 2 (PatID 15). The trend plots of mean ICP and mean ICP wave amplitude (MWA) of (a) Signal 1 and (b) Signal 2 of PatID 15 is shown [Average of mean ICP: -1.3 mmHg (Signal 1), 7.2 mmHg (Signal 2). Average of mean ICP wave amplitude: 8.7 mmHg (Signal 1), 5.5 mmHg (Signal 2)]. The vertical arrows in (a) and (b) indicate shifts in mean ICP; no apparent lasting changes in MWA were seen. The shifts of mean ICP in (a) and (b) were in opposite directions. Figure 13 Trend plots of mean ICP/mean ICP wave amplitude of Signals 1 and 2 (PatID 15). Properties of the ICP sensors The trend plots of mean ICP and mean ICP wave amplitude (MWA) of (a) Signal 1 and (b) Signal 2 of PatID 15 is shown [Average of mean ICP: -1.3 mmHg (Signal 1), 7.2 mmHg (Signal 2). Average of mean ICP wave amplitude: 8.7 mmHg (Signal 1), 5.5 mmHg (Signal 2)]. The vertical arrows in (a) and (b) indicate shifts in mean ICP; no apparent lasting changes in MWA were seen. The shifts of mean ICP in (a) and (b) were in opposite directions. important frequency range (up to 4-5x normal heart rate) and the phase as linear as possible. The present comparisons between sensors showed that the phase was linear over the most important frequency range which is a good feature. With regard to the magnitude of the transfer function, it corresponded poorly with the only past statement about the frequency response of the air pouch Spiegelberg sensor known [22]. Based on our analysis, we conclude that the transfer function between the solid Codman sen- sor and either the fluid Edwards sensor or the air pouch Spiegelberg sensor is quite variable, affecting both the MWA and the MWRT measures. The variation in MWA and MWRT estimates must be caused by variability in the fluid Edward´s and air- pouch Spiegelberg sensors. This may result in either higher or lower values for both the MWAs and MWRTs of these sensors when compared to the reference solid Codman sensor. Impact of hospital environment on ICP scores There are three major factors related to hospital environment that may affect the ICP scores: Human factors, technical issues and technology issues. Among the human factors, erroneous zeroing of the ICP sensor is of paramount im- portance. Some solid sensors, such as the Codman and the Camino, can be zeroed only prior to implantation. The Spiegelberg sensor is the only ICP sensor that performs in vivo zeroing, while the Raumedic sensor performs post-implantation electrical zero- ing, though not a true in-vivo (atmospheric pressure) check of the catheter-tip sensor. Using a fluid sensor, the zero pressure level depends on the selected zero point relative to the head or the heart; furthermore, with regard to the head there is no consensus whether the zero point should be at the level of the tragus, the eye, vertex, or the frontal bone. Several technical issues are involved. Sensor damage may occur at any point of time. The sensor may further respond to electrostatic discharges (ESD’s) [6]. Using a fluid system, the sensor is placed at some distance from the patient, requiring that a fluid Eide et al. BioMedical Engineering OnLine 2012, 11:66 http://www.biomedical-engineering-online.com/content/11/1/66 Eide et al. BioMedical Engineering OnLine 2012, 11:66 http://www.biomedical-engineering-online.com/content/11/1/66 Page 18 of 20 filled catheter serves as a connection line between the measurement site (the ventricu- lar fluid) and the sensor site. Air bubbles and/or partial/total occlusion of the fluid catheter by blood cloths or brain tissue allows for erroneous pressure reproduction. Mal-positioning of the catheter (outside of the brain fluid) may also give erroneous pressure readings. The physical properties - and the length of the fluid-filled catheter as well as sensor location relative to the patient also influence pressure readings. Technology issues relate to the properties of the ICP sensor itself. Presently, we showed that the transfer function varied between solid and fluid/air pouch sensors, in particular, the technology of the air pouch sensor made this sensor type less useful for reproduction of pressure waveforms. The Spiegelberg sensor presently used utilizes an air pouch system that lacks the frequency characteristics needed for proper waveform reproductions. On the other hand, the ability of the Spiegelberg sensor to perform true post-implantation zeroing might be one explanation of the relatively smaller differences in mean ICP we observed when using this sensor. Both human errors and technical issues may affect the static pressure parameter mean ICP. Impact of hospital environment on ICP scores This is related to the current approach of determining the mean ICP as an absolute value relative to the atmospheric pressure (usually referred to as baseline pres- sure, reference pressure, or zero pressure level). Accordingly, the ICP sensors are zer- oed against the atmospheric pressure, and the mean ICP that is revealed on the screen is a pressure value relative to the atmospheric pressure. The present data show that baseline pressure shifts and drifts in deed represent a clinical issue (Table 3; Figure 5-13). The spontaneous sudden shifts in mean ICP thus resemble those we have previously seen in an experimental setting where the ESD’s impact the zero pressure level of ICP sensors were extensive [6]. They are also similar to those and may explain plain the sudden shifts in the mean ICP that can occur during long-term monitoring within the intensive care unit (ICU) [23]. Conclusions Simultaneous monitoring of ICP using two solid sensors may show marked differences in static ICP (mean ICP or ICP level) but close to identity in dynamic ICP waveforms. This indicates that shifts in ICP baseline pressure (sensor zero level) occur during ICP monitoring. Trend plots of the ICP parameters confirm that shift in mean ICP (due to spontaneously changed ICP baseline pressure) in deed occur clinically. Solid sensors are superior to fluid – and air pouch sensors when evaluating the dynamic ICP parameters. Quality control during ICP monitoring Managing patients according to erroneous ICP scores could be fatal; thus the issue of quality control during ICP monitoring is important. Today’s quality control is usually as follows: In situations where an external drain had been placed for CSF drainage, the static intracranial pressure (mean ICP) can be “semi-quantitatively” assessed by measur- ing the height of the fluid in the fluid-filled catheter. Also, in patients treated with decompressive craniectomy, the ICP may be crudely evaluated by slight compression at the craniectomy site. Quality assessment of the pressure waveform is done by visual in- spection of the ICP waveform on the screen of the vital signs monitor or by inspection of the fluid pulsations within a fluid-filled catheter of a CSF drain. However, as men- tioned above even partial occlusion of a CSF drain may give erroneous dynamic pres- sure reproductions. Also, qualitative assessment of the pressure waveform provides minimal information whether the waveform is normal or not. A quality control beyond what is mentioned above can be obtained by the use of computer software that in an intelligent manner identifies “noise” in the ICP signal, and also automatically identifies the cardiac induced ICP waves [7]. Increasing propor- tion of “noisy” waves indicates poor signal quality. In such a setting, the sensor system would need the proper ability to retrieve the ICP waveform. Incorporating ICP Eide et al. BioMedical Engineering OnLine 2012, 11:66 http://www.biomedical-engineering-online.com/content/11/1/66 Eide et al. BioMedical Engineering OnLine 2012, 11:66 http://www.biomedical-engineering-online.com/content/11/1/66 Page 19 of 20 waveform evaluation in patient management would also identify situations of erroneous static ICP scores [16,17,24]. Based on how the static mean ICP and the dynamic MWA relates, finding of a pathological high mean ICP together with a normal (low) MWA indicates an erroneous static ICP. From this, monitoring the dynamic intracranial pres- sure parameters ought to be the primary type of intracranial pressure monitoring. Pres- ently, the solid Codman sensor was superior for measuring ICP waveforms when compared with that of the fluid Edward’s sensor and particularly the air-pouch Spiegelberg sensor. Abbreviations ICP: Intracranial pressure; MWA: Mean ICP wave amplitude; SW: Single wave. ICP: Intracranial pressure; MWA: Mean ICP wave amplitude; SW: Single wave. Authors' contributions All authors contributed to conception and design, acquisition and interpretation of data. SH contributed with determining transfer functions of the different sensors. PKE contributed the bulk of the drafting of the manuscript and SH and WSO contributed with thorough editing of the manuscript. All authors have read and approved the final manuscript. Competing interests p g SH and WSO report no conflicts of interest. PKE has financial interest in the software company (dPCom A/S) that manufactures the software (SensometricsW Software), which was used for digital recording of the continuous pressure signals in this study. Author details 1 1Department of Neurosurgery, Oslo University Hospital, Rikshospitalet, Oslo, Norway. 2Faculty of Medicine, University of Oslo, Oslo, Norway. 3Department of Informatics, University of Oslo, Oslo, Norway. Received: 14 March 2012 Accepted: 2 July 2012 Published: 7 September 2012 Received: 14 March 2012 Accepted: 2 July 2012 Published: 7 September 2012 Additional files Additional file 1: Category A. Animation of measurements of Signals 1 and 2 in PatID 2. The animation shows the simultaneous continuous ICP waveform of Signals 1 (Codman; lower signal) and 2 (Codman; upper signal) in PatID 2. Note that the ICP waveform is identical while the baseline pressure is different. Additional file 2: Category B. Animation of measurements of Signals 1 and 2 in PatID 8. The animation shows the simultaneous continuous ICP waveform of Signals 1 (Codman; lower signal) and 2 (Edwards; upper signal) in PatID 8. Note that the ICP waveform is quite similar while the baseline pressure is different. Additional file 3: Category C. Animation of measurements of Signals 1 and 2 in PatID 11. The animation shows the simultaneous continuous ICP waveform of Signals 1 (Codman; lower signal) and 2 (Spiegelberg; upper signal) in PatID 11. Note that both the ICP waveform and the baseline pressure are different. Additional file 1: Category A. Animation of measurements of Signals 1 and 2 in PatID 2. The animation shows the simultaneous continuous ICP waveform of Signals 1 (Codman; lower signal) and 2 (Codman; upper signal) in PatID 2. Note that the ICP waveform is identical while the baseline pressure is different. Additional file 2: Category B. Animation of measurements of Signals 1 and 2 in PatID 8. The animation shows the simultaneous continuous ICP waveform of Signals 1 (Codman; lower signal) and 2 (Edwards; upper signal) in PatID 8. Note that the ICP waveform is quite similar while the baseline pressure is different. Additional file 3: Category C. Animation of measurements of Signals 1 and 2 in PatID 11. The animation shows the simultaneous continuous ICP waveform of Signals 1 (Codman; lower signal) and 2 (Spiegelberg; upper signal) in PatID 11. Note that both the ICP waveform and the baseline pressure are different. References References 1. Czosnyka M, Pickard JD: Monitoring and interpretation of intracranial pressure. J Neurol Neurosurg Psychiatry 2004, 75:813–821. 1. Czosnyka M, Pickard JD: Monitoring and interpretation of intracranial pressure. J Neurol Neurosurg Psychiatry 2004, 75:813–821. 2. Eide PK, Bentsen G, Sorteberg AG, Marthinsen PB, Stubhaug A, Sorteberg W: A randomized and blinded single- center trial comparing the effect of intracranial pressure and intracranial pressure wave amplitude-guided intensive care management on early clinical state and 12-month outcome in patients with aneurysmal subarachnoid hemorrhage. Neurosurgery 2011, 69:1105–1115. Page 20 of 20 Eide et al. BioMedical Engineering OnLine 2012, 11:66 http://www.biomedical-engineering-online.com/content/11/1/66 3. Marmarou A, Anderson RL, Ward JD, Choi SC, Young HF, Eisenberg HM, Foulkes MA, Marshall LF, Jane JA: Impact of ICP instability and hypotension on outcome in patients with severe head trauma. J Neurosurg 1991, 75:S59–S66. 4. Zhong J, Dujovny M, Park HK, Perez E, Perlin AR, Diaz FG: Advances in ICP monitoring techniques. Neurol Res 2003, 25:339–350. 5. Allen R: Intracranial pressure: A review of clinical problems, measurement techniques and monitoring methods. J Med Engin Tech 1986, 10:299–320. 6. Eide PK, Bakken A: The baseline pressure of intracranial pressure (ICP) sensors can be altered by electrostatic discharges. BioMedical Engineering OnLine 2011, 10:75. 6. Eide PK, Bakken A: The baseline pressure of intracran discharges. BioMedical Engineering OnLine 2011, 10:75. 6. Eide PK, Bakken A: The baseline pressure of intracrania discharges. BioMedical Engineering OnLine 2011, 10:75. g g g 7. Eide PK: A new method for processing of continuous intracranial pressure signals. Med Eng Physics 2006, 28:579–587. 8. Piper IR, Miller JD: The evaluation of the wave-form analysis capability of a new strain-gauge intracranial pressure microsensor. Neurosurgery 1995, 36:1142–1145. 9. Gray WP, Palmer JD, Gill J, Gardner M, Iannotti F: A clinical study of parenchymal and subdural miniature strain- gauge transducers for monitoring intracranial pressure. Neurosurgery 1996, 39:927–932. 10. Fernandes HM, Bingham K, Chambers IR, Mendelow AD: Clinical evaluation of the Codman microsensor inracranial pressure monitoring system. Acta Neurochir 1998 71(Suppl):44–46 10. Fernandes HM, Bingham K, Chambers IR, Mendelow AD: Clinical evaluation of the Codman microsensor p g y , ( pp ) 11. Signorini DF, Shad A, Piper IR, Statham PFX: A clinical evaluation of the Codman Microsensor for intracranial pressure monitoring. Br J Neurosurg 1998, 1:223–227. 12. Koskinen L-O, Olivecrona M: Clinical experience with the intraparenchymal intracranial pressure monitoring Codman Microsensor system. Neurosurgery 2005, 56:693–698. 13. References BioMedical Engineering OnLine 2012 11:66. Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color ﬁgure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color ﬁgure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit Eide et al. BioMedical Engineering OnLine 2012, 11:66 http://www.biomedical-engineering-online.com/content/11/1/66 References Yau YH, Piper I, Clutton RE, Whittle IR: Experimental evaluation of the Spiegelberg intracranial pressure and intracranial compliance monitor. J Neurosurg 2000, 93:1072–1077. p g 14. Lang J-M, Beck J, Zimmermann M: Clinical evaluation of intraparenchymal Spiegelberg pressure sensor. Neurosurgery 2003, 52:1455–1459. g y 15. Eide PK, Sorteberg W: Diagnostic intracranial pressure monitoring and surgical management in idiopathic normal pressure hydrocephalus: a 6-year review of 214 patients. Neurosurgery 2010, 66:80–91. 16. Eide PK: Comparison of simultaneous continuous intracranial pressure (ICP) signals from a Codman and a Camino ICP sensor. Med Eng Physics 2006, 28:542–549. Camino ICP sensor. Med Eng Physics 2006, 28:542–549. 17. Eide PK, Sorteberg W: Simultaneous measurements of intracranial pressure parameters in the epidural space and in brain parenchyma in patients with hydrocephalus. J Neurosurg 2010, 113:1317–1325. y y g 18. Czosnyka M, Czosnyka Z, Pickard JD: Laboratory testing of three intracranial pressure microtransducers: Technical report. Neurosurgery 1996, 38:219–224. p g y 19. Morgalla MH, Mettenleiter H, Bitzer M, Fretschner R, Grote EH: ICP measurement control: Laboratory test of 7 types of intracranial pressure transducers. J Medical Engineer Techn 1999, 23:144–151. 20. Morgalla MH, Krasznai L, Dietz K, Mettenleiter H, Deininger M, Grote EH: Methods of experimental and clinical assessment of the relative measurement accuracy of an intracranial pressure transducer. Technical note. J Neurosurg 2001, 95:529–532. 21. Morgalla MH, Dietz K, Deininger M, Grote EH: The problem of long-term ICP drift assessment: Improvement by use of the ICP drift index. Acta Neurochir 2002, 144:57–61. 22. Czosnyka M, Czosnyka Z, Pickard JD: Laboratory testing of the Spiegelberg brain pressure monitor: a technical report. J Neurol Neurosurg Psychiatr 1997, 63:732–735. p g y 23. Eide PK, Rapoport BI, Gormley WB, Madsen JR: A dynamic nonlinear relationship between the static and pulsatile components of intracranial pressure in patients with subarachnoid hemorrhage. J Neurosurg 2010, 112:616–625. 23. Eide PK, Rapoport BI, Gormley WB, Madsen JR: A dynamic nonlinear relationship between the static and pulsatile components of intracranial pressure in patients with subarachnoid hemorrhage. J Neurosurg 2010, 112:616–625. 24. Eide PK: Assessment of quality of continuous intracranial pressure recordings in children. Pediatr Neurosurg 2006, 42:28–34. 24. Eide PK: Assessment of quality of continuous intracranial pressure recordings in children. Pediatr Neurosurg 2006, 42:28–34. doi:10.1186/1475-925X-11-66 Cite this article as: Eide et al.: Simultaneous monitoring of static and dynamic intracranial pressure parameters from two separate sensors in patients with cerebral bleeds: comparison of findings. Submit your next manuscript to BioMed Central and take full advantage of: Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color ﬁgure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit • Convenient online submission
https://openalex.org/W2099653715	https://wjso.biomedcentral.com/counter/pdf/10.1186/1477-7819-12-87	English	null	Suprarenal solitary fibrous tumor associated with a NF1 gene mutation mimicking a kidney neoplasm: implications for surgical management	World journal of surgical oncology	2,014	cc-by	3,255	Conzo et al. World Journal of Surgical Oncology 2014, 12:87 http://www.wjso.com/content/12/1/87 Conzo et al. World Journal of Surgical Oncology 2014, 12:87 http://www.wjso.com/content/12/1/87 WORLD JOURNAL OF SURGICAL ONCOLOGY Suprarenal solitary fibrous tumor associated with a NF1 gene mutation mimicking a kidney neoplasm: implications for surgical management Giovanni Conzo1, Ernesto Tartaglia1, Claudio Gambardella1, Claudio Mauriello1, Daniela Esposito1, Massimo Mascolo2, Daniela Russo2, Gianfranca Stornaiuolo3, Giovan Battista Gaeta3 and Luigi Santini1 Background reported. Neurofibromatosis type 1 (von Recklinghausen’s disease-NF1) is caused by an alteration of the NF1 gene, a tumor suppressor located on the long arm of chromosome 17 (17q11.2) [5]. Loss of the gene function, due to a point mutation, leads to an increase in cell proliferation and to development of tumors [5]. g Solitary fibrous tumor (SFT) is a rare spindle cell neo- plasm, firstly described in the1930s [1]. Prognosis seems to be favorable, but large series with long-term follow-up are still lacking. This kind of tumor, usually occurring in the pleura, the so called ‘localized fibrous mesothelioma’, has recently been described in different and multiple extra- pleural sites including orbit, nasal cavity, breast, adrenal or thyroid gland, liver and lung [2]. Although the retroperi- toneum is frequently involved, according to a computed search of the medical literature, only 39 cases of renal or pararenal SFT have been reported. Tumor size ranged from 2 to 25 cm, and nephrectomy was the standard treat- ment. The origin of these neoplasms is still controversial [3,4]. They are considered as slow-growing tumors with a favorable prognosis, although malignant cases, showing locoregional recurrence or distant metastases, have been We describe a right suprarenal SFT in a female patient carrying a NF1 gene mutation. Preoperative diagnostic work-up was unable to determine its primal tissue, and therefore an open radical right nephrectomy was per- formed. Only the immunohistochemical assay allowed a correct pathological diagnosis. To better evaluate the role of radical nephrectomy, in a case of pararenal tumor with no clear signs of renal infiltration, more detailed therapeutic guidelines could be useful to establish the recommended extension of resection. According to our literature search, this is the first case in which SFT was associated to a NF1 gene mutation. Correspondence: giovanni.conzo@unina2.it 1Department of Anesthesiologic, Surgical and Emergency Science VII Division of General Surgery, Second University of Naples Italy, Via Pansini 5, 80131 Naples, Italy Full list of author information is available at the end of the article Abstract Solitary fibrous tumor (SFT) is a rare spindle cell neoplasm, usually occurring in the pleura. Pararenal SFT, mimicking an adrenal gland or renal tumor, as here described, is extremely rare. We report a case of a right suprarenal SFT, incidentally discovered by abdominal ultrasound in a 54-year-old woman carrying a point neurofibromatosis 1 (NF1) gene mutation. Preoperative diagnostic work-up was ineffective in evaluating its origin, and an open radical right nephrectomy was therefore undertaken. Immunohistochemical assay showed a positivity for CD34, CD99 and Bcl-2, so suggesting a diagnosis of SFT. According to our knowledge, the association between this type of tumor and NF1 gene mutation has never been described. In cases of pararenal tumors, a more detailed preoperative diagnosis could be useful to better plan the extension of resection, allowing, in selected cases, nephron-sparing surgery. More studies are needed to better analyze the relationship between NF1 gene mutation and SFT. Keywords: Solitary fibrous tumor, Spindle cells, Renal neoplasm, Immunohistochemical assay, Neurofibromatosis gene mutation © 2014 Conzo et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. * Correspondence: giovanni.conzo@unina2.it 1Department of Anesthesiologic, Surgical and Emergency Science VII Division of General Surgery, Second University of Naples Italy, Via Pansini 5, 80131 Naples, Italy Full list of author information is available at the end of the article © 2014 Conzo et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. Case presentation In July 2012, a 52-year-old woman was referred to our observation for the presence of a solid mass in her right Page 2 of 5 Conzo et al. World Journal of Surgical Oncology 2014, 12:87 http://www.wjso.com/content/12/1/87 Figure 1 Computerized tomographic scan depicting large superior pole mass of the right kidney. Figure 1 Computerized tomographic scan depicting large superior pole mass of the right kidney. kidney, detected during a routine follow-up for chronic hepatitis B virus infection. Computed tomography scan showed a well-delineated, encapsulated tumor arising from the upper pole of the right kidney and measuring 96 × 63 mm, contrast-enhanced and with no evidence of vessel infiltration (Figure 1). The ipsilateral adrenal gland was not clearly distinguishable from the neoplasm, and its adrenal origin was apparent. Magnetic resonance imaging (MRI) scan showed a mass of low intensity on T1-weighted images and of irregular high intensity on T2-weighed images. Neither renal vein nor inferior vena cava thrombosis were present, and suspected enlarged lymph nodes were also not identified. Due to the pres- ence of a daughter with a previous diagnosis of NF1 gene mutation, and the presence of small neurofibromas over the patient’s body, a genetic test for NF1 using reverse transcription (RT)-PCR and high throughput- denaturing high performance liquid chromatography (HT- DHPLC) was performed, confirming the diagnosis of von Recklinghausen’s disease. Considering that a pheochromo- cytoma (PCC) has been identified in 0.1 to 5.7% of pa- tients with von Recklinghausen’s disease, a PCC was considered in the differential diagnosis [6]. The absence of clinical signs and of arterial pressure lability, associated with a normal level of urinary metanephrine concentra- tions and the absence of pathological uptake following MIBG (meta iodo benzyl guanidine) scintigraphy, allowed the exclusion of a PCC neoplasm. Since a diagnosis of suspected renal or adrenal gland mass was suspected, the patient underwent an explorative laparotomy. During the procedure, a large neoplasm, strongly adherent to the upper renal pole and to the ipsilateral suprarenal space, was observed and, presuming a renal cell carcinoma, we undertook a surgical resection including a radical right nephrectomy. No macroscopic enlarged lymph nodes were observed. The postoperative course was uneventful Figure 2 The tumor showed an admixture of irregularly distributed highly cellular and hypocellular zones, composed of spindled cells with vesicular nuclei, which surrounded thin- and thick-walled vessels with a characteristic ‘stag horn’ appearance. Case presentation Figure 2 The tumor showed an admixture of irregularly distributed highly cellular and hypocellular zones, composed of spindled cells with vesicular nuclei, which surrounded thin- and thick-walled vessels with a characteristic ‘stag horn’ appearance. Conzo et al. World Journal of Surgical Oncology 2014, 12:87 http://www.wjso.com/content/12/1/87 Page 3 of 5 Conzo et al. World Journal of Surgical Oncology 2014, 12:87 http://www.wjso.com/content/12/1/87 Page 3 of 5 ureter, adrenal gland, perinephric tissue and a mass meas- uring 90 × 60 × 40 mm in overall dimensions. At micros- copy, the tumor showed an admixture of irregularly and the patient was discharged on postoperative day four. Diagnostic work-up was negative during the eight-month follow-up. The gross specimen included: right kidney, and the patient was discharged on postoperative day four. Diagnostic work-up was negative during the eight-month follow-up. The gross specimen included: right kidney, Figure 3 A panoramic view of the solitary fibrous tumor. (a) note the cellular proliferation of spindled cells, the dense collagen and the ‘stag horn’ vessels (H&E, x100); (b) The tumor cells surround a vascular component with a ‘stag horn’ appearance ((b) H&E, x150); (c) The tumor cells show a widespread immunoreactivity for CD34 (immunoperoxidase stain for CD34, x200); (d) The tumor cells exhibit a strong positivity for CD99 (immunoperoxidase stain for CD99, x200); (e) The tumor cells exhibit a diffuse staining for bcl-2 (immunoperoxidase stain for bcl-2, x200); (f) The tumor cells show a proliferation index to a maximum of about 10% of neoplastic cells (immunoperoxidase stain for Ki67/MIB-1, x150). Figure 3 A panoramic view of the solitary fibrous tumor. (a) note the cellular proliferation of spindled cells, the dense collagen and the ‘stag horn’ vessels (H&E, x100); (b) The tumor cells surround a vascular component with a ‘stag horn’ appearance ((b) H&E, x150); (c) The tumor cells show a widespread immunoreactivity for CD34 (immunoperoxidase stain for CD34, x200); (d) The tumor cells exhibit a strong positivity for CD99 (immunoperoxidase stain for CD99, x200); (e) The tumor cells exhibit a diffuse staining for bcl-2 (immunoperoxidase stain for bcl-2, x200); (f) The tumor cells show a proliferation index to a maximum of about 10% of neoplastic cells (immunoperoxidase stain for Ki67/MIB-1, x150). Figure 3 A panoramic view of the solitary fibrous tumor. Case presentation (a) note the cellular proliferation of spindled cells, the dense collagen and the ‘stag horn’ vessels (H&E, x100); (b) The tumor cells surround a vascular component with a ‘stag horn’ appearance ((b) H&E, x150); (c) The tumor cells show a widespread immunoreactivity for CD34 (immunoperoxidase stain for CD34, x200); (d) The tumor cells exhibit a strong positivity for CD99 (immunoperoxidase stain for CD99, x200); (e) The tumor cells exhibit a diffuse staining for bcl-2 (immunoperoxidase stain for bcl-2, x200); (f) The tumor cells show a proliferation index to a maximum of about 10% of neoplastic cells (immunoperoxidase stain for Ki67/MIB-1, x150). Conzo et al. World Journal of Surgical Oncology 2014, 12:87 http://www.wjso.com/content/12/1/87 Page 4 of 5 Conzo et al. World Journal of Surgical Oncology 2014, 12:87 http://www.wjso.com/content/12/1/87 aggressive behavior, and vice versa. Therefore, all SFT patients need to be on long-term follow-up [4,7]. Turning to treatment, complete surgical resection is recommended [3], but because SFTs exhibit relative chemo-radio tumor resistance, no data are reported regarding the role of adju- vant treatment. In case of a preoperative suspicion of a well-encapsulated pararenal SFT, in the absence of clear signs of locoregional infiltration, the recommended exten- sion of resection could be controversial. We hypothesize that, in selected cases, a nephrectomy could be avoided with comparable outcomes [11]. The rarity of pararenal SFT does not allow randomized prospective trials con- cerning this issue. Preoperative diagnostic efforts must be made to achieve detailed data about its origin in order to plan a tailored surgery. The association of STF and the NF1 gene mutation is intriguing. However, since NF1 gene mutation has not been investigated systematically in STF cases, and the available evidence in the literature data is insufficient, it is hard to hypothesize how a gene mutation might determine the development of a suprarenal neo- plasm. More studies are needed to better analyze the rela- tionships between SFT and NF1 gene mutation. distributed highly cellular and hypocellular areas, com- posed of spindled cells with vesicular nuclei, which sur- rounded thin- and thick-walled vessels with a characteristic ‘stag horn’ appearance. A prominent hyalinized collagen was present (Figure 2). Most tumor cells expressed CD34, CD99, and bcl-2, whereas cytokeratin, CD117 (c-Kit), S- 100 protein and EMA were negative. The Ki-67 (MIB-1) staining, a marker of cellular proliferation, was irregularly expressed, exhibiting positivity in up to about 10% of neo- plastic cells. Author details 1 1Department of Anesthesiologic, Surgical and Emergency Science VII Division of General Surgery, Second University of Naples Italy, Via Pansini 5, 80131 Naples, Italy. 2Department of Advanced Biomedical Sciences, University of Naples Federico II, Via Pansini 5, 80131 Naples, Italy. 3Department of Infectious Diseases, Viral Hepatitis Unit, Second University of Naples, Via Pansini 5, 80131 Naples, Italy. Received: 24 May 2013 Accepted: 15 March 2014 Published: 7 April 2014 Conclusions In most cases of SFT, described as renal or pararenal tumor, it is very difficult to determine the true origin from the renal capsule [4], the interstitial tissue, or the peripelvic connective tissue. Therefore, in planning a therapeutic protocol, extension of resection may be controversial and ‘classical’ nephrectomy remains the most common treatment. In the reported case, only de- finitive pathology identified a well-encapsulated SFT not infiltrating the kidney and the ipsilateral adrenal gland, and radical nephrectomy could probably be con- sidered an overtreatment. Histopathological examin- ation, immunohistochemical and ultrastructural studies are the cornerstone of SFT diagnosis. Morphologically, it is characterized by spindle cell proliferation, and about 70% of cases expresses CD 34, CD99, and Bcl-2; only between 20 and 35% of cases are variably positive for epithelial membrane antigen and smooth muscle actin. Focal and limited reactivity of S-100 protein, cytokeratins and/or desmin has also occasionally been reported [6]. Strong CD34 reactivity is currently regarded as characteristic, and an indispensable finding in the diag- nosis of SFT [7]. Authors’ contributions CG P i i d b i CG: Participated substantially in conception, design, and execution of the study; also participated substantially in the drafting and editing of the manuscript. TE: Participated substantially in conception, design, and execution of the study; also participated substantially in the drafting and editing of the manuscript. GC: Participated substantially in conception, design, and execution of the study. MC: Participated substantially in conception, design, and execution of the study. ED: Participated substantially in conception, design, and execution of the study. MM: Participated substantially in conception, design, and execution of the study. RD: Participated substantially in conception, design, and execution of the study. SG: Participated substantially in conception, design, and execution of the study. GGB: Participated substantially in conception, design, and execution of the study. SL: Participated substantially in conception, design, and execution of the study. All authors read and approved the final manuscript. Since these tumors typically show hemangio- pericytomatous patterns, differential diagnosis includes sarcomatoid renal cell carcinoma or renal adenoma and other benign spindle cell tumors such as angiomyoli- poma, fibroma, or fibrosarcoma [4]. In presence of a NF1 gene mutation, neurofibroma should also be con- sidered. SFT is a rare tumor sometimes mimicking renal cell or adrenal gland neoplasms [8,9], and it must be considered in cases of renal tumors comprising mesen- chymal elements. In roughly 10 to 15% of cases, they behave aggressively, showing local recurrence or distant metastases [4]. According to England et al. [10], in- creased cellularity with crowded/overlapping nuclei, cellular pleomorphism, and a mitotic count of more than four figures per ten high-power fields are consid- ered criteria for malignancy. However, clinical progres- sion of tumors with a benign appearance cannot be predicted on histopathological basis, which can exhibit Case presentation Based on morphological features and immu- nohistochemical profile, a diagnosis of a solitary fibrous tumor was made (Figure 3). Consent Written informed consent was obtained from the patient for the publication of this report and any accompanying images. Competing interests The authors declare that they have no competing interests. Klemperer P, Rabin CB: Primary neoplasm of the pleura: a report of five cases. Arch Pathol 1931, 11:385–412. References l Klemperer P, Rabin CB: Primary neoplasm of the pleura: a report of five cases. Arch Pathol 1931, 11:385–412. Page 5 of 5 Page 5 of 5 Conzo et al. World Journal of Surgical Oncology 2014, 12:87 http://www.wjso.com/content/12/1/87 2. Lo Muzio L, Mascolo M, Capodiferro S, Favia G, Maiorano E: Solitary fibrous tumor of the oral cavity: the need for an extensive sampling for a correct diagnosis. J Oral Pathol Med 2007, 36(9):538–542. 2. Lo Muzio L, Mascolo M, Capodiferro S, Favia G, Maiorano E: Solitary fibrous tumor of the oral cavity: the need for an extensive sampling for a correct diagnosis. J Oral Pathol Med 2007, 36(9):538–542. 3. Morimitsu Y, Nakajima M, Hisaoka M, Hashimoto H: Extrapleural solitary fibrous tumor: clinicopathologic study of 17 cases and molecular analysis of the p53 pathway. APMIS 2000, 108:617–625. 3. Morimitsu Y, Nakajima M, Hisaoka M, Hashimoto H: Extrapleural solitary fibrous tumor: clinicopathologic study of 17 cases and molecular analysis of the p53 pathway. APMIS 2000, 108:617–625. y y 4. Fine SW, McCarthy DM, Chan TY, et al: Malignant solitary fibrous tumor of the kidney: report of a case and comprehensive review of the literature. Arch Pathol Lab Med 2006, 130:857–861. 5. Zinnamosca L, Petramala L, Cotesta D, Marinelli C, Schina M, Cianci R, Giustini S, Sciomer S, Anastasi E, Calvieri S, De Toma G, Letizia C: Neurofibromatosis type 1 (NF1) and pheochromocytoma: prevalence, clinical and cardiovascular aspects. Arch Dermatol Res 2011, 303(5):317–325. 6. Walther MM, Herring J, Enquist E, Keiser HR, Linehan WMJ: von Recklinghausen’s disease and pheochromocytomas. Urol 1999, 162(5):1582–1586. 7. Hirato D, Mashiko A, Murata Y, Satoh K, Ichinose T, Takahashi S, Jike T, Sugitani M: Case of solitary fibrous tumor of the kidney: an immunohistochemical and ultrastructural study with a review of the literature. Med Mol Morphol 2009, 42:239–244. 7. Hirato D, Mashiko A, Murata Y, Satoh K, Ichinose T, Takahashi S, Jike T, Sugitani M: Case of solitary fibrous tumor of the kidney: an immunohistochemical and ultrastructural study with a review of the literature. Med Mol Morphol 2009, 42:239–244. 8. Conzo G, Musella M, Corcione F, De Palma M, Ferraro F, Palazzo A, Napolitano S, Milone M, Pasquali D, Sinisi AA, Colantuoni V, Santini L: Laparoscopic adrenalectomy, a safe procedure for pheochromocytoma. A retrospective review of clinical series. Int J Surg 2013, 11(2):152–156. 8. Conzo G, Musella M, Corcione F, De Palma M, Ferraro F, Palazzo A, Napolitano S, Milone M, Pasquali D, Sinisi AA, Colantuoni V, Santini L: Laparoscopic adrenalectomy, a safe procedure for pheochromocytoma. A retrospective review of clinical series. Int J Surg 2013, 11(2):152–156. 9. Conzo et al. World Journal of Surgical Oncology 2014, 12:87 http://www.wjso.com/content/12/1/87 Conzo G, Tricarico A, Belli G, Candela S, Corcione F, Delgenio G, Ferulano GP, Giardiello C, Livrea A, Marzano LA, Porcelli A, Sperlongano P, Vincenti R, Palazzo A, De Martino C, Musella M: Adrenal incidentalomas in the laparoscopic era and the role of correct surgical indications: observations from 255 consecutive adrenalectomies in an Italian series. Can J Surg 2009, 52:281–285. 10. England DM, Hochholzer L, McCarthy MJ: Localized benign and malignant fibrous tumors of the pleura. A clinicopathologic review of 223 cases. Am J Surg Pathol 1989, 13(8):640–658. 10. England DM, Hochholzer L, McCarthy MJ: Localized benign and malignant fibrous tumors of the pleura. A clinicopathologic review of 223 cases. Am J Surg Pathol 1989, 13(8):640–658. 11. Conzo G, Sciascia V, Palazzo A, Stanzione F, Della Pietra C, Insabato L, Natella V, Radice L, Santini L: Radiofrequency-assisted partial nephrectomy for metanephric adenoma: a case report and literature review. Surg Innov 2013, 20(1):55–58. 11. Conzo G, Sciascia V, Palazzo A, Stanzione F, Della Pietra C, Insabato L, Natella V, Radice L, Santini L: Radiofrequency-assisted partial nephrectomy for metanephric adenoma: a case report and literature review. Surg Innov 2013, 20(1):55–58. doi:10.1186/1477-7819-12-87 doi:10.1186/1477-7819-12-87 Cite this article as: Conzo et al.: Suprarenal solitary fibrous tumor associated with a NF1 gene mutation mimicking a kidney neoplasm: implications for surgical management. World Journal of Surgical Oncology 2014 12:87. Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color ﬁgure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color ﬁgure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit Submit your next manuscript to BioMed Central and take full advantage of: Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission
W4220844449.txt	https://bmcgenomics.biomedcentral.com/track/pdf/10.1186/s12864-022-08417-8	en		Transcriptional response of murine microglia in Alzheimer’s disease and inflammation	BMC genomics	2,022	cc-by	6,747	(2022) 23:183 Shippy et al. BMC Genomics https://doi.org/10.1186/s12864-022-08417-8 Open Access RESEARCH Transcriptional response of murine microglia in Alzheimer’s disease and inflammation Daniel C. Shippy1, Jyoti J. Watters2 and Tyler K. Ulland1* Abstract Background: Alzheimer’s disease (AD) is a neurodegenerative disorder and is the most common cause of late-onset dementia. Microglia, the primary innate immune cells of the central nervous system (CNS), have a complex role in AD neuropathology. In the initial stages of AD, microglia play a role in limiting pathology by removing amyloid-β (Aβ) by phagocytosis. In contrast, microglia also release pro-inflammatory cytokines and chemokines to promote neuroinflammation and exacerbate AD neuropathology. Therefore, investigating microglial gene networks could identify new targets for therapeutic strategies for AD. Results: We identified 465 differentially expressed genes (DEG) in 5XFAD versus wild-type mice by microarray, 354 DEG in lipopolysaccharide (LPS)-stimulated N9 microglia versus unstimulated control cells using RNA-sequencing (RNA-seq), with 32 DEG common between both datasets. Analyses of the 32 common DEG uncovered numerous molecular functions and pathways involved in Aβ phagocytosis and neuroinflammation associated with AD. Furthermore, multiplex ELISA confirmed the induction of several cytokines and chemokines in LPS-stimulated microglia. Conclusions: In summary, AD triggered multiple signaling pathways that regulate numerous genes in microglia, contributing to Aβ phagocytosis and neuroinflammation. Overall, these data identified several regulatory factors and biomarkers in microglia that could be useful in further understanding AD neuropathology. Keywords: Alzheimer’s disease, Microglia, RNA-seq, Microarray, Transcriptomics Background Alzheimer’s disease (AD) is a neurodegenerative disorder and is the most common cause of late-onset dementia. In the United States, approximately 6.2 million people are living with Alzheimer’s dementia, a number estimated to grow to 13.8 million by 2060 unless medical intervention strategies are developed for AD [1]. AD neuropathology is defined by the aggregation of extracellular amyloid-β (Aβ) plaques followed by the development of intracellular neurofibrillary tangles (NFTs) composed of hyperphosphorylated tau [2]. In addition to Aβ plaques and tau NFTs, neuroinflammation plays a key role in AD Correspondence: tulland@wisc.edu 1 Department of Pathology and Laboratory Medicine, University of Wisconsin, Madison, WI, USA Full list of author information is available at the end of the article neuropathology, promoting numerous inflammatory processes in the central nervous system (CNS) [3]. Microglia, the primary innate immune cells of the CNS, have a complex role in AD neuropathology. In the early stages of AD, microglia reduce Aβ accumulation by phagocytosis, and act as a defense barrier to protect plaque adjacent neurons from neurotoxic effects [4, 5]. Alternatively, microglia can contribute to neuroinflammation by the release of pro-inflammatory cytokines and chemokines, reactive oxygen species, and other molecules associated with increased AD neuropathology [6]. Although the role of microglia in AD is still not entirely understood, it is clear microglia play a key role in the development of AD neuropathology. Since microglia appear to be an important factor in AD development, investigating microglial gene networks © The Author(s) 2022. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativeco mmons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Shippy et al. BMC Genomics (2022) 23:183 could lead to new therapies to treat AD. Several studies have already highlighted the importance of specific genes involved in microglial metabolism [7] and response to Aβ plaque pathology [8]. In this study, we investigated the transcriptional response of microglia in an AD versus non-AD state using microarray and RNA-sequencing (RNA-seq). We identified 465 differentially expressed genes (DEG) in 5XFAD versus wild-type mice, and 354 DEG in lipopolysaccharide (LPS)-stimulated N9 microglia versus unstimulated control cells, with 32 DEG common to both experiments. Of the 32 DEG, functional enrichment analyses identified numerous processes and pathways in which microglia are potentially involved during AD development. Furthermore, multiplex ELISA confirmed the induction of several cytokines and chemokines in LPS-stimulated microglia that were also differentially expressed in the microarray and RNA-seq datasets. Overall, these data identify novel potential regulatory factors and biomarkers in the microglial response to AD. Page 2 of 12 Results Transcriptome analyses of microglia in AD Transcriptional analysis of LPS-stimulated N9 microglia versus non-stimulated control cells was performed using RNA-seq. A total of 354 significant DEG (log2FC > 1.5, FDR-adjusted P-value < 0.05) were identified with 323 upregulated genes and 31 down-regulated genes (Fig. 1A and Additional File 1). Interleukin-1 alpha (Il1α) was the most up-regulated gene (log2FC = 9.70) and albumin (Alb) was the most down-regulated gene ( log2FC = -7.62) (Table 1). We compared this data to publicly available data derived from sorted microglia from female 8-month-old wild-type and 5XFAD mice, a mouse model of AD which accumulates Aβ plaques [9]. Transcriptional analysis of microglia from 5XFAD versus wild-type mice was performed by microarray [10]. A total of 465 significant gene transcripts (FC > 2, FDR-adjusted P-value < 0.05) were identified with 337 up-regulated gene transcripts and 128 down-regulated gene transcripts (Fig. 1B and Additional File 2). Glycoprotein (transmembrane) nmb (Gpnmb) was Fig. 1 AD transcriptional alterations in microglia. A Scatter plot of significantly altered genes ( log2FC > 1.5, FDR-adjusted P-value < 0.05) by RNA-seq in N9 microglia stimulated with LPS (1 µg/ml) for 6 h versus unstimulated control cells. B Scatter plot of significantly altered transcripts (FC > 2, FDR-adjusted P-value < 0.05) by microarray in microglia isolated from the brains of female 5XFAD mice versus wild-type mice (8 months old). For both scatter plots, up-regulated genes are shown in red and down-regulated genes are shown in green. C Venn diagram showing the 354 significantly altered genes in N9 microglia by RNA-seq, 465 significantly altered transcripts in murine microglia by microarray, and 32 significantly altered genes found in both the RNA-seq and microarray data Shippy et al. BMC Genomics (2022) 23:183 Page 3 of 12 Table 1 Top 10 up-regulated and down-regulated genes in LPS-stimulated N9 microglia versus control cells determined by RNA-seq Gene Gene IDa Gene Description Log2FC Il1a 27399 interleukin 1 alpha 9.70 Ifit3b 62488 interferon-induced protein with tetratricopeptide repeats 3B 8.08 Nos2 20826 nitric oxide synthase 2, inducible 8.07 Col5a3 04098 collagen, type V, alpha 3 7.44 Rsad2 20641 radical S-adenosyl methionine domain containing 2 7.39 Gm19410 09372 predicted gene, 19,410 7.39 Ifit3 74896 interferon-induced protein with tetratricopeptide repeats 3 7.21 Il1f9 44103 interleukin 1 family, member 9 7.15 Trim30c 78616 tripartite motif-containing 30C 7.13 Ifit1 34459 interferon-induced protein with tetratricopeptide repeats 1 7.06 Alb 29368 albumin -7.62 Gc 35540 vitamin D binding protein -6.04 Ttr 61808 transthyretin -4.66 BC021767 85006 cDNA sequence BC021767 -2.68 Tlr8 40522 toll-like receptor 8 -2.65 4930473A02Rik 60029 RIKEN cDNA 4930473A02 gene -2.55 Fsbp 94595 fibrinogen silencer binding protein -2.51 Nlrp1a 69830 NLR family, pyrin domain containing 1A -2.45 Cgn 68876 cingulin -2.41 Abcd2 55782 ATP-binding cassette, sub-family D (ALD), member 2 -2.37 a All gene IDs start with ENSMUSG000000 Table 2 Top 10 up-regulated and down-regulated gene transcripts in 5XFAD versus wild-type mice determined by microarray Gene Gene IDa Gene Description FC Gpnmb 031840 glycoprotein (transmembrane) nmb 46.43 Ddx3y 091190 DEAD (Asp-Glu-Ala-Asp) box polypeptide 3, Y-linked 27.16 Spp1 112747 secreted phosphoprotein 1 26.34 Mamdc2 036069 MAM domain containing 2 24.41 Cst7 089200 cystatin F (leukocystatin) 24.03 Fabp3 070532 fatty acid binding protein 3, muscle and heart 23.89 Fabp5 029046 fatty acid binding protein 5, epidermal 22.76 Bhlhe40 032194 basic helix-loop-helix family, member e40 21.89 Hpse 045617 heparanase 19.26 18.88 Igf1 121161 insulin-like growth factor 1 Xlr4b 114506 X-linked lymphocyte-regulated 4B -9.76 Gpr165 033554 G protein-coupled receptor 165 -4.84 Snord35b 082833 small nucleolar RNA, C/D box 35B -4.83 Xist 127786 inactive X specific transcripts -4.70 4933434E20Rik 159064 RIKEN cDNA 4933434E20 gene -4.35 Snord61 083176 small nucleolar RNA, C/D box 61 -4.25 Bank1 041577 B-cell scaffold protein with ankyrin repeats 1 -4.08 Ttr 075312 transthyretin -3.90 Fam71a 171798 family with sequence similarity 71, member A -3.76 Il7r 003981 interleukin 7 receptor -3.73 a All gene IDs start with ENSMUST00000 Shippy et al. BMC Genomics (2022) 23:183 Page 4 of 12 Table 3 Genes found in both the RNA-seq and microarray experiments Gene Gene IDa Gene Description RNA-seq Log2FC Microarray FC Ifit3 74896 interferon-induced protein with tetratricopeptide repeats 3B 7.21 2.21 Cxcl10 34855 chemokine (C-X-C motif ) ligand 10 7.04 3.08 Cxcl2 58427 chemokine (C-X-C motif ) ligand 2 6.60 3.72 Il1b 27398 interleukin 1 beta 5.94 7.76 Oasl2 29561 2’-5’ oligoadenylate synthetase-like 2 5.18 4.29 Ifit2 45932 interferon-induced protein with tetratricopeptide repeats 2 4.58 3.15 Ifi204 73489 interferon activated gene 204 4.13 2.70 Tnf 24401 tumor necrosis factor 3.87 2.41 Slfn5 54404 schlafen 5 3.67 3.57 Prdm1 38151 PR domain containing 1, with ZNF domain 3.29 2.57 Slfn2 72620 schlafen 2 3.16 2.25 Il1rn 26981 interleukin 1 receptor antagonist 2.99 4.35 Cd83 15396 CD83 antigen 2.97 2.17 Gvin1 45868 GTPase, very large interferon inducible 1 2.82 2.20 Phlda1 20205 pleckstrin homology like domain, family A, member 1 2.75 3.85 Gm1966 73902 predicted gene 1966 2.44 2.19 Gpr84 63234 G protein-coupled receptor 84 2.39 2.02 Cd69 30156 CD69 antigen 2.36 9.28 Slc7a11 27737 solute carrier family 7 (cationic amino acid transporter, y + system), member 11 2.31 2.59 Cd274 16496 CD274 antigen 2.09 5.72 Irak3 20227 interleukin-1 receptor-associated kinase 3 2.06 2.20 Serpine1 37411 serine (or cysteine) peptidase inhibitor, clade E, member 1 2.03 4.16 Mmp12 49723 matrix metallopeptidase 12 1.77 2.73 Tlr2 27995 toll-like receptor 2 1.77 2.70 Olr1 30162 oxidized low density lipoprotein (lectin-like) receptor 1 1.75 2.70 Rab11fip1 31488 RAB11 family interacting protein 1 (class I) 1.66 2.42 Cd300lf 47798 CD300 molecule like family member F 1.63 3.45 C3 24164 complement component 3 1.59 2.12 Itga5 00555 integrin alpha 5 (fibronectin receptor alpha) 1.57 4.85 Bcl2a1d 99974 B cell leukemia/lymphoma 2 related protein A1d 1.53 4.05 Plaur 46223 plasminogen activator, urokinase receptor 1.51 3.60 Ttr 61808 transthyretin -4.66 -3.9 a All gene IDs start with ENSMUSG000000 the most up-regulated gene transcript (FC = 46.43) and X-linked lymphocyte-regulated 4B (Xlr4b) was the most down-regulated gene transcript (FC = -9.76) (Table 2). In total, 32 genes overlapped between the N9 RNA-seq experiment and 5XFAD microarray dataset (Fig. 1C). Of the 32 total genes, 31 were up-regulated and only one was downregulated (Table 3). The majority of the most up-regulated genes were cytokines and chemokines involved in inflammation (Cxcl10, Cxcl2, Il1β, Tnf). Transthyretin (Ttr) was the only down-regulated gene common to both datasets. Pathway and functional prediction of microglia in AD Geno ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), and Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) analyses were performed on the 32 genes found in both the RNA-seq and microarray datasets. Biological Process (BP) GO indicated the DEG participated in immune system process (Oasl2, Cd300lf, Prdm1, C3, Ifit2, Ifit3, Olr1, Tlr2), immune response (Oasl2, C d274, Cxcl10, Cxcl2, Il1β, Tlr2, Tnf), positive regulation of interleukin-8 (Il1β, Serpine1, Tlr2, Tnf), response to LPS (Cxcl10, Cxcl2, Il1β, Irak3, Tlr2, Tnf), inflammatory response (Cxcl10, Cxcl2, C3, Il1β, Olr1, Tlr 2, Tnf), positive regulation of gene expression (Prdm1, Il1β, Plaur, Serpine1, Tlr2, Tnf), innate immune response (Oasl2, Prdm1, C3, Ifit2, Ifit3, Tlr2), regulation of cell proliferation (Prdm1, Cxcl10, Cxcl2, Serpine1, Tnf), positive Shippy et al. BMC Genomics (2022) 23:183 regulation of NF-kappaB transcription factor activity (Il1β, Irak3, Tlr2, Tnf), positive regulation of apoptotic process (Bcl2a1d, Ifit2, Il1β, Phlda1, Tnf), positive regulation of protein phosphorylation (C3, Il1β, Plaur, Tnf), negative regulation of cell proliferation (Ifit3, Il1β, Slfn2, Tlr2, Tnf), cellular response to LPS (Cxcl10, Cxcl2, Serpine1, Tnf), negative regulation of gene expression (Prdm1, Il1β, Serpine1, Tnf), and signal transduction (Cd274, Cd83, Gpr84, Irak3, Tlr2, Tnf) (Fig. 2A). Cellular Component (CC) GO indicated the DEG were located in the external side of plasma membrane (Cd274, Cd69, Cd83, Cxcl10, Itga5, Tlr2, Tnf), extracellular region (Cxcl10, Cxcl2, C3, Il1β, Il1rn, Mmp12, Olr1, Plaur, Serpine1, Ttr, Tnf), extracellular space (Cxcl10, Cxcl2, C3, Il1β, Il1rn, Serpine1, Ttr, Tnf), cell surface (Cd274, Itga5, Slc7a11, Tlr2, Tnf), and cytoplasmic vesicle (Rab11fip1, Itga5, Il1β, Phlda1, Tlr2) (Fig. 2B). Molecular Function (MF) GO indicated the DEG were involved in cytokine activity (Cxcl10, Cxcl2, Il1β, Il1rn, Tnf), CXCR chemokine receptor binding (Cxcl10, Cxcl2), protein binding (Cd274, Cd300lf, Prdm1, C3, Itga5, Ifi204, Ifit3, Irak3, Plaur, Serpine1, Tlr2, Ttr, Tnf), interleukin-1 receptor binding (Il1β, Il1rn), and protein heterodimerization activity (Bcl2a1d, Irak3, Tlr2, Ttr) (Fig. 2C). KEGG analysis was performed on the 32 common genes found in both the microarray and RNA-seq Page 5 of 12 datasets to identify pathways associated with AD. KEGG identified a total of seven pathways (P < 0.05) associated with the 32 genes (Fig. 3). The pathways included toll-like receptor signaling (Tlr2, Tnf, Il1β, Cxcl10), TNF signaling (Cxcl10, Cxcl2, Il1β, Tnf), phagosome (C3, Itga5, Olr1, Tlr2), NOD-like receptor signaling (Cxcl2, Il1β, Tnf), cytokine-cytokine receptor interaction (Cxcl10, Cxcl2, Il1β, Tnf), complement and coagulation cascades (C3, Plaur, Serpine1), and NF-kappa B signaling (Bcl2ald, Il1β, Tnf). To further understand the interactions of the DEG, we performed protein–protein interaction (PPI) analysis on the 32 genes found in both the RNA-seq and microarray datasets using STRING. For the analysis, text mining, experiments, and databases were chosen for the interaction sources, and the high confidence value of 0.700 was selected as the minimum required interaction score threshold. Of the proteins encoded by the 32 identified DEG, two distinct networks emerged with 15 proteins in one network (IL1β, Irak3, Il1rn, Cxcl2, Cxcl10, Tnf, Tlr2, Olr1, Cd83, Cd69, Cd274, Serpine1, Plaur, Itga5, Mmp12), three proteins in the second network (Ifit2, Ifit2, Oasl2), and 13 of the proteins not clustering (Fig. 4). The database did not recognize predicted gene 1966 (Gm1966). The results suggest that the given proteins were highly enriched (P < 1 × 10–16). Fig. 2 GO enrichment analysis. Biological function analyses for the 32 genes found in both the RNA-seq and microarray experiments was performed using DAVID. Analyses were performed for Biological Process (BP) (A), Cellular Component (CC) (B), and Molecular Function (MF) (C). Pathways are shown in descending order based on –log10 P-values. The number of genes associated with each GO term is shown above each bar. Only GO terms with a P-value < 0.05 are shown Shippy et al. BMC Genomics (2022) 23:183 Fig. 3 KEGG pathway enrichment analysis. KEGG pathway analysis was performed on the 32 genes found in both the RNA-seq and microarray data using DAVID. Pathways are shown in descending order based on –log10 P-values. The number of genes associated with each pathway is shown above each bar. Only pathways with a P-value < 0.05 are shown Secretion of cytokines and chemokines by microglia Since so many DEG were cytokines and chemokines associated with inflammation, we performed multiplex ELISAs (22-plex) on supernatants from LPS-stimulated bone marrow-derived macrophages (BMDM), primary microglia, and N9 microglia. In BMDM, IL-α, IL-1β, IL-2, IL-5, IL-6, IL-10, IL-12, MCP-1, IFN-γ, TNF-α, MIP-1α, GM-CSF, RANTES, KC, MDC, TARC, and TCA-3 showed significant induction when stimulated with LPS versus non-stimulated control cells (Additional File 3). In primary microglia, IL-6, IL-12, TNF-α, MIP-1α, GM-CSF, RANTES, KC, and MCP-1 all showed significant induction in LPS-stimulated primary microglia versus non-stimulated control cells (Fig. 5). In N9 microglia, IL-6, MCP-1, TNF-α, MIP-1α, and RANTES all showed significant induction in LPS-stimulated N9 microglia versus non-stimulated control cells (Fig. 6). Cytokines and chemokines from BMDM, primary microglia, and N9 microglia that showed no significant difference between LPS-stimulated and control cells are shown in Additional File 4. Page 6 of 12 Discussion A thorough understanding of the molecular characteristics and regulatory pathways of microglia is essential to develop therapeutic strategies for AD. Accordingly, numerous studies have performed transcriptome analyses to elucidate the role of specific genes in microglia important in AD neuropathology, including Trem2 and ApoE [7, 10–13]. Other studies have investigated the transcriptome of microglia under distinct biological activities, such as Aβ plaque phagocytosis [14], tau pathology [15] and metabolism [16]. In this study, we used a two-pronged approach to determine DEG in AD and inflammation. First, we determined the DEG in microglia from 5XFAD mice versus wild-type mice by microarray. We then determined the DEG in LPS-stimulated microglia using RNA-seq. Altogether, 32 DEG overlapped between the two datasets. Neuroinflammation by microglia is hypothesized to exacerbate AD neuropathology [17]. In our study, several immunological pathways associated with inflammation were identified in AD-associated microglia that correlated with DEG of well-known pro-inflammatory cytokines and chemokines, such as IL-1β, TNF, CXCL10, and CXCL2. Caspase-1 dependent IL-1β secretion occurs during NLRP3 inflammasome activation, and activation of the inflammasome has emerged as an important mechanism in chronic neuroinflammation in AD [18]. Furthermore, immunotherapies blocking inflammasome activation are being investigated for use in AD and other inflammatory diseases [19–21]. BP GO analysis also identified Olr1 as an inflammatory and immune system response gene. Several studies suggest a role for Olr1 in lipid metabolism, and genetic variation in Olr1 as a risk factor for AD [22–24], but little is known about Olr1 in AD-associated neuroinflammation. Furthermore, Olr1 was recognized as a gene involved in the phagosome pathway by KEGG, suggesting an important role in multiple AD pathogenic processes. Genes not previously characterized in AD were identified by KEGG to be involved in molecular signaling pathways important in AD development. The B cell leukemia/ lymphoma 2 related protein A1d (Bcl2a1d) was identified as being involved in the nuclear factor kappa-light-chainenhancer of activated B cells (NF-κB) signaling pathway. Activation of NF-κB is an important mechanism in chronic neuroinflammation that significantly increases AD pathology [25]. Furthermore, the plasminogen activator, urokinase receptor (Plaur) was recognized as being associated with complement and coagulation cascades. Complement and coagulation cascades are associated with blood–brain barrier dysfunction and AD progression in mice [26] and humans [27]. Characterization of these genes in the context of AD could further our Shippy et al. BMC Genomics (2022) 23:183 Page 7 of 12 Fig. 4 PPI analysis using STRING. STRING analysis was performed on the 32 genes found in both the RNA-seq and microarray data. For the analysis, text mining, experiments, and databases were chosen for active interaction sources, and a high value of 0.700 was selected as the minimum required interaction score. Line colors represent known interactions from curated databases (blue), experimentation (purple) and text mining (yellow) understanding of the underlying regulatory mechanisms in AD pathogenesis. Biomarkers can be a useful tool to predict and diagnosis a variety of neurological disorders, including AD [28]. In our study, the only down-regulated gene amongst the 32 DEG was transthyretin (Ttr). Indeed, Ttr has been implicated in numerous processes associated with AD, including Aβ binding [29], Aβ transport at the blood– brain barrier (BBB) [30], toxicity [31, 32], and neuroprotection by interfering with Aβ formation [29, 33, 34]. Ttr has been previously identified as being decreased in the cerebral spinal fluid (CSF) [35] and blood [36–38] of AD patients, and is being investigated as a blood biomarker for AD [38, 39]. Several other DEG in our study could also potentially be used as biomarkers due to their highlevel of differential expression between groups (greater than threefold in both assays), including Ifit3, Cxcl10, Cxcl2, Il1β, Oasl2, Ifit2, and Slfn5. Further investigation will be needed, as clinically useful biomarkers for AD should be inexpensive, non-invasive, reliably detectable and able to distinguish AD from other forms of dementia [40]. There are several limitations of our study which are common to most studies directed at identifying dynamic alterations in gene expression networks in AD. Microgliosis is a hallmark of AD neuropathology, resulting in large numbers of microglia concentrated around Aβ plaques [41]. Since the role of microglia in AD is not fully understood, it is difficult to ascertain whether the altered genes identified in our study are involved in disease progression or in the neuroprotective microglia response. Another caveat to our study is that murine microglia were used in both experiments. Although the value of murine microglial research in AD cannot be overstated, the transcriptional signature of microglia in human AD is drastically different than that of murine microglia [11, 42]. When comparing human and mouse AD models, transgenic mice overexpress Aβ in a non-physiological manner, resulting in rapid Aβ accumulation and a higher plaque burden when compared to human AD [43, 44]. Shippy et al. BMC Genomics (2022) 23:183 Page 8 of 12 Fig. 5 Cytokines and chemokines induced by LPS in primary microglia. Primary microglia were stimulated with LPS (50 ng/ml) for 6 h. Supernatants were collected and assayed for 22 cytokines and chemokines using a multiplex assay. Experiments were performed in triplicate. Results are from three independent experiments. Data are shown as mean ± SEM. *P ≤ 0.0001, P ≤ 0.01, P ≤ 0.05 Although the 5XFAD mouse model does not perfectly recapitulate human disease studies, comparing the microglial response in humans and mice have been mixed, with some indicating that the response may be different [11, 42] and some studies indicating that the microglial response in both humans and mice may be similar [7, 45]. This is likely due to differences in the complexity of human disease where both plaques and neurofibrillary tangles are present compared to the 5XFAD mice in which only plaques form. Therefore, in vitro and in vivo models of AD remain powerful tools to begin to dissect underlying microglial gene regulatory mechanisms in early AD pathology. Conclusions In summary, we identified 465 DEG in 5XFAD versus wild-type mice by microarray, 354 DEG in LPS-stimulated N9 microglia versus unstimulated control cells using RNA-seq, with 32 DEG common between both data sets. Analyses of the 32 DEG uncovered numerous molecular functions and pathways involved in Aβ phagocytosis and neuroinflammation associated with AD that may be further investigated. Furthermore, multiplex ELISA confirmed the induction of several cytokines and chemokines in LPS-stimulated microglia. Overall, these data identified several regulatory factors and biomarkers in microglia that could be useful in further understanding AD neuropathology. Methods Ethical approval and consent to participate All methods were carried out in accordance with relevant local and University of Wisconsin guidelines and regulations. All animals were handled in accordance with the Animal Research: Reporting of in vivo Experiments (ARRIVE) guidelines and the University of Wisconsin’s Institutional Animal Care and Use Committee policies and our approved protocols. Cell culture assays BMDM were prepared as previously described [19]. Briefly, tibias were removed from wild-type C57BL/6J mice (6–10 months old) (The Jackson Laboratory), flushed with complete RPMI media supplemented with 20% L-cell conditioned media (LCCM) [19], and cultured for 4–7 days prior to use. Differentiated BMDM were seeded at a cell density of 400,000 cells/well in a 24-well tissue culture plate. Cells were stimulated with ultrapure LPS (50 ng/ml) from Escherichia coli O111:B4 Shippy et al. BMC Genomics (2022) 23:183 Page 9 of 12 Fig. 6 Cytokines and chemokines secreted by N9 microglia following LPS stimulation. N9 microglia were stimulated with LPS (1 µg/ml) for 6 h. Supernatants were collected and assayed for 22 cytokines and chemokines using a multiplex assay. Experiments were performed in triplicate. Results are from three independent experiments. Data are shown as mean ± SEM. *P ≤ 0.0001, P ≤ 0.001 (InvivoGen) for 6 h. Supernatants were flash frozen on dry ice and stored at -80℃ until use. Microglia were prepared from wild-type C57BL/6J mice (5–8 months old) (The Jackson Laboratory) as previously described [7]. Microglia were isolated using Magnetic Activated Cell Sorting (MACS, Miltenyi Biotec) according to manufacturer’s instructions. Briefly, mice were perfused with cold PBS containing 0.1% heparin. Brains were collected in C-tubes (Miltenyi Biotec, Cat. No. 130–096-334) and dissociated using a Neural Tissue Dissociation Kit (T) (Miltenyi Biotec, Cat. No. 130–093-231). Cell suspensions were filtered through a 70 µM cell strainer (Corning Falcon™, Cat. No. 352350). Microglia were labeled with anti-mouse CD45 magnetic beads (Miltenyi Biotec, Cat. No. 130–052-301) and isolated on LS columns (Miltenyi Biotec, Cat. No. 130–042401). LS columns were washed three times with MACS buffer (PBS containing 0.5% BSA and 1 mM EDTA) before elution. This method allows for preparation of microglial cultures of high purity (> 95%). Purity of isolated microglia are routinely confirmed by flow cytometry. Approximately 50,000 microglia/well were seeded on poly-L-lysine coated 24-well plates in complete RPMI media supplemented with 20% LCCM [19] and human TGF-β (10 ng/ml) (PeproTech, Cat. No. 100–21). Media was changed three days after plating, and cells were used in the week following the media change. Microglia were stimulated with LPS (50 ng/ml) for 6 h. Supernatants were flash frozen on dry ice and stored at -80℃ until use. Immortalized murine N9 microglia were cultured as previously described [46]. N9 microglia were seeded at a cell density of 250,000 cells/well in a 24-well tissue culture plate. Cells were stimulated with LPS (1 µg/ml) [47] for 6 h. Supernatants were flash frozen on dry ice and stored at -80℃ until use. ELISA BMDM, primary microglia, and N9 microglial supernatants were assayed for IL-1α, IL-1β, IL-2, IL-3, IL-4, IL-5, IL-6, IL-10, IL-12, IL-13, IL-17, MCP-1, IFN-γ, TNF-α, MIP-1α, RANTES, GM-CSF, Eotaxin, KC, MDC, TARC, and TCA-3 using a multiplex assay (Quansys Biosciences). Shippy et al. BMC Genomics (2022) 23:183 RNA‑seq RNA was extracted from immortalized N9 microglia stimulated with LPS (1 µg/ml) [47] for 6 h using a RNeasy Plus Mini Kit (Qiagen, Cat. No. 74134). The N9 microglial cell line is derived from mouse brain and shares numerous phenotypic traits with primary mouse microglia [48]. Quality and quantity of RNA was assessed using an Agilent 2100 Bioanalyzer (Agilent Technologies) and a Nanodrop spectrophotometer (Thermo Scientific). All samples had an RNA integrity number (RIN) of 9.7 or higher. RNA library preparation and transcriptome sequencing were performed by Novogene using the Illumina NovaSeq 6000 Sequencing System. Bioinformatics analysis was performed by Novogene with differential expression analysis performed using the DESeq2 R package (1.20.0). The resulting P-values were adjusted using the Benjamini and Hochberg method for controlling the false discovery rate [49]. Genes with FDR-adjusted P-values < 0.05 and log2FC > 1.5 were considered differentially expressed. Page 10 of 12 and MF respectively. Each GO and KEGG pathway with a P-value < 0.05 was considered significant. Additionally, PPI analysis was performed to identify interactions of the selected proteins based on their gene IDs using the STRING database [52]. For the analysis, text mining, experiments, and databases were chosen for active interaction sources using the high confidence (0.700) threshold setting. Statistical analyses Statistical analysis was performed using Prism 9.0.0 (GraphPad). Data are presented as mean ± SEM. Comparison between two groups was performed using a Student’s t-test. A P-value ≤ 0.05 (P ≤ 0.05, P ≤ 0.01, P ≤ 0.001 and ***P ≤ 0.0001) was used as the significance cutoff. The Venn diagram demonstrating overlap in DEG amongst the two datasets was generated using InteractiVenn [53]. Microarray Supplementary Information The microarray has been published in a previous study [10] and the publicly available dataset was used (GSE65067). Briefly, microglia from female 8 month old wild-type (n = 3) and 5XFAD (n = 5) mice (The Jackson Laboratory) were FACS-sorted directly into RTL-plus lysis buffer. RNA extraction from microglia was performed using a RNeasy Plus Micro Kit (Qiagen, Cat. No. 74034). Microarray hybridization (Affymetrix MoGene 1.0 ST array) and data processing were performed at the Washington University Genome Center. Raw data were normalized using the Robust MultiArray (RMA) method and genes were pre-filtered for expression value greater than or equal to 120 expression units. This method provides a cut-off above which genes have a 95% chance of expression demonstrated in Immgen dataset, which uses the same array platform [10]. P-values were calculated using a Student’s t-test and FDR-adjusted P-values were calculated using the Benjamini and Hochberg method [49]. Transcripts with FDR-adjusted P-values < 0.05 and FC > 2 were considered differentially expressed. The online version contains supplementary material available at https://doi. org/10.1186/s12864-022-08417-8. Functional enrichment analyses Genes found to be differentially expressed in both the RNA-seq and microarray experiments were selected for biological function and pathway analysis. The gene list was uploaded into the Database for Annotation, Visualization and Integrated Discovery (DAVID, v. 6.8) [50, 51] for GO and KEGG pathway analysis. GO gene count thresholds of 4, 2, and 2 were used for BP, CC, Additional file1. Additional file2. Additional file3. Additional file 4. Acknowledgements We thank Paula Ricciardi-Castagnoli for the N9 microglial cells. Authors’ contributions DCS and TKU were responsible for conceptualizing the project, designing, executing, analyzing the results from, and ensuring the validity of the experiments, drafting, and editing the manuscript. JJW supplied reagents, assisted with experimental design, and the drafting and editing of the manuscript. All authors read and approved the final manuscript. Funding This work was funded by NIH R21AG068652 awarded to TKU and R01HL142752-03S1 awarded to TKU and JJW. Availability of data and materials The datasets generated and/or analyzed during the current study are available in the Gene Expression Omnibus (GEO) repository, GSE65067 and GSE183038. Declarations Ethics approval and consent to participate All methods were carried out in accordance with relevant local and University of Wisconsin guidelines and regulations. All animals were handled in accordance with the Animal Research: Reporting of in vivo Experiments (ARRIVE) guidelines and the University of Wisconsin’s Institutional Animal Care and Use Committee policies and our approved protocols. Consent for publication Not applicable. Shippy et al. BMC Genomics (2022) 23:183 Page 11 of 12 Competing interests The authors declare that they have no competing interests. Author details 1 Department of Pathology and Laboratory Medicine, University of Wisconsin, Madison, WI, USA. 2 Department of Comparative Biosciences, University of Wisconsin, Madison, WI, USA. 18. 19. Received: 4 October 2021 Accepted: 21 February 2022 20. References 1. 2021 Alzheimer’s disease facts and figures. Alzheimers Dement. 2021;17(3):327–406. 2. Holtzman DM, Morris JC, Goate AM. Alzheimer’s disease: the challenge of the second century. Sci Transl Med. 2011;3(77):77sr71. 3. Leng F, Edison P. Neuroinflammation and microglial activation in Alzheimer disease: where do we go from here? Nat Rev Neurol. 2021;17(3):157–72. 4. Frackowiak J, Wisniewski HM, Wegiel J, Merz GS, Iqbal K, Wang KC. Ultrastructure of the microglia that phagocytose amyloid and the microglia that produce beta-amyloid fibrils. Acta Neuropathol. 1992;84(3):225–33. 5. Condello C, Yuan P, Schain A, Grutzendler J. Microglia constitute a barrier that prevents neurotoxic protofibrillar Abeta42 hotspots around plaques. Nat Commun. 2015;6:6176. 6. Wang WY, Tan MS, Yu JT, Tan L. Role of pro-inflammatory cytokines released from microglia in Alzheimer’s disease. Ann Transl Med. 2015;3(10):136. 7. Ulland TK, Song WM, Huang SC, Ulrich JD, Sergushichev A, Beatty WL, Loboda AA, Zhou Y, Cairns NJ, Kambal A, et al. TREM2 Maintains Microglial Metabolic Fitness in Alzheimer’s Disease. Cell. 2017;170(4):649-663 e613. 8. Ulrich JD, Ulland TK, Mahan TE, Nystrom S, Nilsson KP, Song WM, Zhou Y, Reinartz M, Choi S, Jiang H, et al. ApoE facilitates the microglial response to amyloid plaque pathology. J Exp Med. 2018;215(4):1047–58. 9. Oakley H, Cole SL, Logan S, Maus E, Shao P, Craft J, Guillozet-Bongaarts A, Ohno M, Disterhoft J, Van Eldik L, et al. Intraneuronal beta-amyloid aggregates, neurodegeneration, and neuron loss in transgenic mice with five familial Alzheimer’s disease mutations: potential factors in amyloid plaque formation. J Neurosci. 2006;26(40):10129–40. 10. Wang Y, Cella M, Mallinson K, Ulrich JD, Young KL, Robinette ML, Gilfillan S, Krishnan GM, Sudhakar S, Zinselmeyer BH, et al. TREM2 lipid sensing sustains the microglial response in an Alzheimer’s disease model. Cell. 2015;160(6):1061–71. 11. Zhou Y, Song WM, Andhey PS, Swain A, Levy T, Miller KR, Poliani PL, Cominelli M, Grover S, Gilfillan S, et al. Human and mouse single-nucleus transcriptomics reveal TREM2-dependent and TREM2-independent cellular responses in Alzheimer’s disease. Nat Med. 2020;26(1):131–42. 12. Keren-Shaul H, Spinrad A, Weiner A, Matcovitch-Natan O, Dvir-Szternfeld R, Ulland TK, David E, Baruch K, Lara-Astaiso D, Toth B, et al. A Unique Microglia Type Associated with Restricting Development of Alzheimer’s Disease. Cell. 2017;169(7):1276-1290 e1217. 13. Krasemann S, Madore C, Cialic R, Baufeld C, Calcagno N, El Fatimy R, Beckers L, O’Loughlin E, Xu Y, Fanek Z, et al. The TREM2-APOE Pathway Drives the Transcriptional Phenotype of Dysfunctional Microglia in Neurodegenerative Diseases. Immunity. 2017;47(3):566-581 e569. 14. Grubman A, Choo XY, Chew G, Ouyang JF, Sun G, Croft NP, Rossello FJ, Simmons R, Buckberry S, Landin DV, et al. Transcriptional signature in microglia associated with Abeta plaque phagocytosis. Nat Commun. 2021;12(1):3015. 15. Wang H, Li Y, Ryder JW, Hole JT, Ebert PJ, Airey DC, Qian HR, Logsdon B, Fisher A, Ahmed Z, et al. Genome-wide RNAseq study of the molecular mechanisms underlying microglia activation in response to pathological tau perturbation in the rTg4510 tau transgenic animal model. Mol Neurodegener. 2018;13(1):65. 16. Baik SH, Kang S, Lee W, Choi H, Chung S, Kim JI, Mook-Jung I. A Breakdown in Metabolic Reprogramming Causes Microglia Dysfunction in Alzheimer’s Disease. Cell Metab. 2019;30(3):493-507 e496. 17. Heneka MT, Carson MJ, El Khoury J, Landreth GE, Brosseron F, Feinstein DL, Jacobs AH, Wyss-Coray T, Vitorica J, Ransohoff RM, 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. et al. Neuroinflammation in Alzheimer’s disease. Lancet Neurol. 2015;14(4):388–405. Heneka MT, Kummer MP, Stutz A, Delekate A, Schwartz S, Vieira-Saecker A, Griep A, Axt D, Remus A, Tzeng TC, et al. NLRP3 is activated in Alzheimer’s disease and contributes to pathology in APP/PS1 mice. Nature. 2013;493(7434):674–8. Shippy DC, Wilhelm C, Viharkumar PA, Raife TJ, Ulland TK. beta-Hydroxybutyrate inhibits inflammasome activation to attenuate Alzheimer’s disease pathology. J Neuroinflammation. 2020;17(1):280. Goldberg EL, Asher JL, Molony RD, Shaw AC, Zeiss CJ, Wang C, MorozovaRoche LA, Herzog RI, Iwasaki A, Dixit VD. beta-Hydroxybutyrate Deactivates Neutrophil NLRP3 Inflammasome to Relieve Gout Flares. Cell Rep. 2017;18(9):2077–87. Youm YH, Nguyen KY, Grant RW, Goldberg EL, Bodogai M, Kim D, D’Agostino D, Planavsky N, Lupfer C, Kanneganti TD, et al. The ketone metabolite beta-hydroxybutyrate blocks NLRP3 inflammasome-mediated inflammatory disease. Nat Med. 2015;21(3):263–9. Lambert JC, Luedecking-Zimmer E, Merrot S, Hayes A, Thaker U, Desai P, Houzet A, Hermant X, Cottel D, Pritchard A, et al. Association of 3’-UTR polymorphisms of the oxidised LDL receptor 1 (OLR1) gene with Alzheimer’s disease. J Med Genet. 2003;40(6):424–30. Wang ZT, Zhong XL, Tan MS, Wang HF, Tan CC, Zhang W, Zheng ZJ, Kong LL, Tan L, Sun L. Association of lectin-like oxidized low density lipoprotein receptor 1 (OLR1) polymorphisms with late-onset Alzheimer disease in Han Chinese. Ann Transl Med. 2018;6(10):172. Luedecking-Zimmer E, DeKosky ST, Chen Q, Barmada MM, Kamboh MI. Investigation of oxidized LDL-receptor 1 (OLR1) as the candidate gene for Alzheimer’s disease on chromosome 12. Hum Genet. 2002;111(4–5):443–51. Granic I, Dolga AM, Nijholt IM, van Dijk G, Eisel UL. Inflammation and NF-kappaB in Alzheimer’s disease and diabetes. J Alzheimers Dis. 2009;16(4):809–21. Ujiie M, Dickstein DL, Carlow DA, Jefferies WA. Blood-brain barrier permeability precedes senile plaque formation in an Alzheimer disease model. Microcirculation. 2003;10(6):463–70. Ryu JK, McLarnon JG. A leaky blood-brain barrier, fibrinogen infiltration and microglial reactivity in inflamed Alzheimer’s disease brain. J Cell Mol Med. 2009;13(9A):2911–25. Mayeux R. Biomarkers: potential uses and limitations. NeuroRx. 2004;1(2):182–8. Schwarzman AL, Gregori L, Vitek MP, Lyubski S, Strittmatter WJ, Enghilde JJ, Bhasin R, Silverman J, Weisgraber KH, Coyle PK, et al. Transthyretin sequesters amyloid beta protein and prevents amyloid formation. Proc Natl Acad Sci U S A. 1994;91(18):8368–72. Alemi M, Gaiteiro C, Ribeiro CA, Santos LM, Gomes JR, Oliveira SM, Couraud PO, Weksler B, Romero I, Saraiva MJ, et al. Transthyretin participates in beta-amyloid transport from the brain to the liver–involvement of the low-density lipoprotein receptor-related protein 1? Sci Rep. 2016;6:20164. Cascella R, Conti S, Mannini B, Li X, Buxbaum JN, Tiribilli B, Chiti F, Cecchi C. Transthyretin suppresses the toxicity of oligomers formed by misfolded proteins in vitro. Biochim Biophys Acta. 2013;1832(12):2302–14. Costa R, Goncalves A, Saraiva MJ, Cardoso I. Transthyretin binding to A-Beta peptide–impact on A-Beta fibrillogenesis and toxicity. FEBS Lett. 2008;582(6):936–42. Li X, Zhang X, Ladiwala AR, Du D, Yadav JK, Tessier PM, Wright PE, Kelly JW, Buxbaum JN. Mechanisms of transthyretin inhibition of beta-amyloid aggregation in vitro. J Neurosci. 2013;33(50):19423–33. Nilsson L, Pamren A, Islam T, Brannstrom K, Golchin SA, Pettersson N, Iakovleva I, Sandblad L, Gharibyan AL, Olofsson A. Transthyretin Interferes with Abeta Amyloid Formation by Redirecting Oligomeric Nuclei into Non-Amyloid Aggregates. J Mol Biol. 2018;430(17):2722–33. Serot JM, Christmann D, Dubost T, Couturier M. Cerebrospinal fluid transthyretin: aging and late onset Alzheimer’s disease. J Neurol Neurosurg Psychiatry. 1997;63(4):506–8. Ribeiro CA, Santana I, Oliveira C, Baldeiras I, Moreira J, Saraiva MJ, Cardoso I. Transthyretin decrease in plasma of MCI and AD patients: investigation of mechanisms for disease modulation. Curr Alzheimer Res. 2012;9(8):881–9. Shippy et al. BMC Genomics (2022) 23:183 Page 12 of 12 37. Han SH, Jung ES, Sohn JH, Hong HJ, Hong HS, Kim JW, Na DL, Kim M, Kim H, Ha HJ, et al. Human serum transthyretin levels correlate inversely with Alzheimer’s disease. J Alzheimers Dis. 2011;25(1):77–84. 38. Velayudhan L, Killick R, Hye A, Kinsey A, Guntert A, Lynham S, Ward M, Leung R, Lourdusamy A, To AW, et al. Plasma transthyretin as a candidate marker for Alzheimer’s disease. J Alzheimers Dis. 2012;28(2):369–75. 39. Bradley-Whitman MA, Abner E, Lynn BC, Lovell MA. A Novel Plasma Based Biomarker of Alzheimer’s Disease. J Alzheimers Dis. 2015;47(3):761–71. 40. Consensus report of the Working Group on: "Molecular and Biochemical Markers of Alzheimer’s Disease". The Ronald and Nancy Reagan Research Institute of the Alzheimer’s Association and the National Institute on Aging Working Group. Neurobiol Aging. 1998;19(2):109–16. 41. Sarlus H, Heneka MT. Microglia in Alzheimer’s disease. J Clin Invest. 2017;127(9):3240–9. 42. Olah M, Menon V, Habib N, Taga MF, Ma Y, Yung CJ, Cimpean M, Khairallah A, Coronas-Samano G, Sankowski R, et al. Single cell RNA sequencing of human microglia uncovers a subset associated with Alzheimer’s disease. Nat Commun. 2020;11(1):6129. 43. Drummond E, Wisniewski T. Alzheimer’s disease: experimental models and reality. Acta Neuropathol. 2017;133(2):155–75. 44. Liu P, Reichl JH, Rao ER, McNellis BM, Huang ES, Hemmy LS, Forster CL, Kuskowski MA, Borchelt DR, Vassar R, et al. Quantitative Comparison of Dense-Core Amyloid Plaque Accumulation in Amyloid-beta Protein Precursor Transgenic Mice. J Alzheimers Dis. 2017;56(2):743–61. 45. Sobue A, Komine O, Hara Y, Endo F, Mizoguchi H, Watanabe S, Murayama S, Saito T, Saido TC, Sahara N, et al. Microglial gene signature reveals loss of homeostatic microglia associated with neurodegeneration of Alzheimer’s disease. Acta Neuropathol Commun. 2021;9(1):1. 46. Righi M, Mori L, De Libero G, Sironi M, Biondi A, Mantovani A, Donini SD, Ricciardi-Castagnoli P. Monokine production by microglial cell clones. Eur J Immunol. 1989;19(8):1443–8. 47. Nikodemova M, Watters JJ. Outbred ICR/CD1 mice display more severe neuroinflammation mediated by microglial TLR4/CD14 activation than inbred C57Bl/6 mice. Neuroscience. 2011;190:67–74. 48. Stansley B, Post J, Hensley K. A comparative review of cell culture systems for the study of microglial biology in Alzheimer’s disease. J Neuroinflammation. 2012;9:115. 49. Benjamini Y, Hochberg Y. Controlling the False Discovery Rate - a Practical and Powerful Approach to Multiple Testing. J R Stat Soc B. 1995;57(1):289–300. 50. da Huang W, Sherman BT, Lempicki RA. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protoc. 2009;4(1):44–57. 51. da Huang W, Sherman BT, Lempicki RA. Bioinformatics enrichment tools: paths toward the comprehensive functional analysis of large gene lists. Nucleic Acids Res. 2009;37(1):1–13. 52. Szklarczyk D, Gable AL, Lyon D, Junge A, Wyder S, Huerta-Cepas J, Simonovic M, Doncheva NT, Morris JH, Bork P, et al. STRING v11: protein-protein association networks with increased coverage, supporting functional discovery in genome-wide experimental datasets. Nucleic Acids Res. 2019;47(D1):D607–13. 53. Heberle H, Meirelles GV, da Silva FR, Telles GP, Minghim R. InteractiVenn: a web-based tool for the analysis of sets through Venn diagrams. BMC Bioinformatics. 2015;16:169. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Ready to submit your research ? Choose BMC and benefit from: • fast, convenient online submission • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations • maximum visibility for your research: over 100M website views per year At BMC, research is always in progress. Learn more biomedcentral.com/submissions
https://openalex.org/W4206822208	https://jist.publikasiindonesia.id/index.php/jist/article/download/317/579	Indonesian	null	Implementasi Aplikasi collabsproject.com di Masa Pandemi Covid-19	Jurnal Indonesia Sosial Teknologi	2,021	cc-by-sa	1,938	Abstrak Artikel ini bertujuan untuk membahas permasalahan para pekerja di masa pandemi yang banyak di PHK/Di rumahkan dan memberi solusi untuk para pekerja lewat platform collabsproject.com. Dimasa pandemi seperti sekarang ini, para pekerja di hadapkan pada situasi sulit dimana perusahaan perusahaan banyak melakukan pemutusan hubungan kerja akibat dampak dari pandemi yang berkepanjangan dan belum bisa untuk di prediksi pandemi ini akan selesai. Dengan kebutuhan sehari hari yang tetap harus terpenuhi banyak karyawan – karyawan yang memutuskan untuk bekerja freelance di dunia digital, dengan adanya aplikasi collabsproject.com yang berbasis website maka para pencari kerja paruh waktu/freelance dapat dengan mudah mendapatkan pekerjaan yang mereka cari sesuai dengan keahlian dan kompetensi mereka masing – masing. Artikel ini ditulis dengan pendekan penelitian kualitatif. Hasil analisis menunjukan bahawa pertumuhan ekonomi indonesia dapat diukur melalui peningkatan atau penurunan produk domestik bruto (PDB/GDP) karena indikator yang berhubungan dengan jumlah pengangguran adalah GDP. Kata kunci: Pandemi; collabsproject.com; Pertumbuhan Ekonomi. IMPLEMENTASI APLIKASI COLLABSPROJECT.COM DI MASA PANDEMI COVID-19 Muhammad Ilyas Manajemen Informatika, Politeknik Piksi Ganesha Bandung, Indonesia Email: muhammadilyas071@gmail.com Muhammad Ilyas Manajemen Informatika, Politeknik Piksi Ganesha Bandung, Indonesia Email: muhammadilyas071@gmail.com Jurnal Indonesia Sosial Teknologi:p–ISSN: 2723 - 6609 e-ISSN :2745-5254 Vol. 2, No.12 Desember 2021 Jurnal Indonesia Sosial Teknologi:p–ISSN: 2723 - 6609 e-ISSN :2745-5254 Vol. 2, No.12 Desember 2021 Jurnal Indonesia Sosial Teknologi:p–ISSN: 2723 - 6609 e-ISSN :2745-5254 Vol. 2, No.12 Desember 2021 Keywords: Pandemic; collabsproject.com; Growth Economic. Pendahuluan Perkembangan teknologi komunikasi dan informasi sekarang ini perlahan mengubah segala aktivitas manusia, segala jenis pekerjaan dapat dilakukan dengan efisien dan efektif karena kemajuan teknologi (Riwayadi, 2013). Salah satu fitur yang menjadi favorit masayarakat adalah internet, kehadiran internet inilah yang membawa perubahan yang signifikan kepada keseharian manusia itu sendiri. Bahkan di era digital seperti ini, semua dapat dilakukan melalui sebuah platform digital seperti berkomunikasi, berbelanja, berdiskusi dan masih banyak lagi pekerjaan yang memanfaatkan internet (Sitanggang & Manalu, 2018). Di masa pandemi seperti sekarang ini, Platform digital sangat berperan untuk menabah atau mencari penghasilan dalam memenuhu kebutuhan sehari – hari, masih terbatasnya platform platform yang berbasis freelance menjadi kan untuk membuat salah satu platform digital yang bertujuan untuk membantu dan memudahkan para pekerja yang terkena dampak akibat covid 19 (Aristi, 2021). Collabsproject.com sendiri dibuat oleh salah satu startup yang mempunyai visi dan misi membantu para pekerja yang terkena dampak pandemi itu sendiri. Aplikasi collabsproject.com sendiri berbasis website agar mudah di akses oleh semua orang. Dengan hadirnya platform ini diharapkan dapat membantu pertumbuhan ekonomi di indonesia yang terlebih di masa pandemi ini yang makin banyaknya para pekerja yang di PHK atau di rumahkan, menurut hasil riset yang ada pada awal tahun 2020 pertumbuhan ekonomi dunia mengalami penurunan, termasuk indonesia. Perihal tersebut karena wabah yang melanda seluruh indonesia dikarenakan wabah coronavirus (Fitriani, 2020). Berdasarkan kutipan dari sumber kontan.co.id beberapa dampak coronavirus terhadap ekonomi indonesia antara lain adalah angka impor indonesia merosot sebesar 3,7% year to date (ytd) januari sampai maret 2020. Dikutip dari cnnindonesia.com, bahwa menteri keuangan, sri mulyani indrawati dalam video conference di jakarta pada rabu, 1 April 2020, mengatakan jika pertumbuhan ekonomi indonesia diperkirakan turun sebesar 2.3% bahkan dapat menembus angka - 0.4% akibat pandemi ini. Kondisi di negara berkembang seperti indonesia, pertumbuhan ekonomi yang dicapai juga diiringi dengan meningkatnya jumlah penduduk indonesia (Kalsum, 2017). Jumlah penduduk indonesia masih banyak yang menjadi pengangguran. Berdasarkan dari data Badan Pusat Statistik (BPS) tingkat pengangguran indonesia pada tahun 2020 bulan Febuari tercatat sebesar 05,01%. Tingkat pengangguran yang tinggi merupakan masalah mendasar dalam ketenagakerjaan di indonesia. Ihwal tersebut dikarenakan oleh tenaga kerja baru yang bertambah jauh lebih besar dibandingkan dengan pertambahnannya lapangan pekerjaan. Artinya , lapangan pekerjaan yang tersedia belum memenuhi tingkat tenaga kerja yang ada di indonesia. Sehingga menimbulkan tingkat pengangguran yang tinggi. Pengangguran menjadi masalah bagi negara yang tingkat pertumbuhannya tinggi seperti indonesia. Pengangguran juga dapat mempengaruhi tingkat perekonomian indonesia (Hanoatubun, 2020). Implementasi Aplikasi collabsproject.com di Masa Pandemi Covid-19 Implementasi Aplikasi collabsproject.com di Masa Pandemi Covid-19 Abstract This article aims to discuss the problems of workers during the pandemic who are often laid off/at home and provide solutions for workers through the collabsproject.com platform. During a pandemic like today, workers are faced with a difficult situation where many companies have laid off their work relationships due to the impact of a prolonged pandemic and it is impossible to predict that this pandemic will end. With daily needs that still have to be met, many employees decide to work freelance in the digital world, with the website-based collabsproject.com application, part-time/freelance job seekers can easily get the job they are looking for according to their skills and abilities. their respective competencies. This article was written with a qualitative research approach. The results of the analysis show that Indonesia's economic growth can be measured through an increase or decrease in gross domestic product (GDP/GDP) because the indicator related to the number of unemployed is GDP. Keywords: Pandemic; collabsproject.com; Growth Economic. 2250 Pendahuluan Hal itu menyebabkan pendapatan negara juga tidak maksimal, karena pendapatan nasional negara diukur dari persentase jumlah seluruh pendapatan masyarakat indonesia yang berpenghasilan dibagi jumlah Jurnal Indonesia Sosial Teknologi, Vol. 2, No. 12, Desember 2021 2 2251 Muhammad Ilyas seluruh penduduk indonesia. Jika jumlah penghasilan lebih sedikit maka rata-rata pendapatan nasional negara akan lebih kecil karena jumlah penduduk belum sebanding dengan jumlah pendapatan nasional. Pengangguran merupak permasalahan jangka panjang suatu negara (Indayani & Hartono, 2020). Setiap hari angka covid-19 di indonesia bertambah. Bertambahnya kasus tersebut menyebabkan pemerintah memberlakukan kebijakan baru yaitu pembatasan sosial bersekala besar (PSBB). Kebijakan PSBB ini diterapkan di beberapa wilayah di indonesia yang mengalami kasus covid-19 tinggi, seperti wilayah jakarta. Selain itu, covid – 19 menyebabkan tenaga kerja yang tidak memiliki pekerjaan tetap atau sebagai buruh harian berhenti bekerja. Hal itu, mendapatkan pendapatan masyarakat menurun, maka pendapatan nasional juga akan menurun. Oleh karena itu, pertumbuhan ekonomi juga melemah (Munandar, 2017). Kondisi saat pandemi ini dengan gambaran di atas, indonesia belum bisa meningkatkan pertumbuhan ekonomi secara signifikan. Tujuan penelitian adalah untuk menganalisis tingkat pengangguran terhadap pertumbuhan ekonomi indonesia dimasa pandemi dan solusi yang penulis berikan (Pramesthi, 2013). Berdasarkan latar belakang masalah pada penelitian ini, maka rumusan masalah yang dijadikan sebagai batasan masalah bagi penulis dalam penelitian ini adalah, bagai mana analisi dampak pandemi terhadap perekonomian dan solusi apa yang bisa di berikan oleh penulis (Santoso, 2020). Hasil dan Pembahasan Analisa Kebutuhan Sistem Penulis mengimplementasikan kebutuhan sistem dalam web ini yaitu sistem harus bisa menyediakan informasi tentang pekerjaan atau project. Jurnal Indonesia Sosial Teknologi, Vol. 2, No. 12, Desember 2 Metode Penelitian Jenis penelitian ini merupakan sebuah metode deskriptif dengan pendekatan kualitatif. Metode penelitian kualitatif adalah metode penelitian yang berlandaskan pada filsafat postpositivisme, digunakan untuk meneliti pada kondisi objek alamiah, dimana peneliti adalah sebagai instrumen kunci, teknik pengumpulan data dengan triangulasi (gabungan), analisis data bersifat induktif atau kualitatif, dan hasil penelitian kualitatif lebih menekankan makna daripada generalisasi (Sugiyono, 2012). Metode kualitatif digunakan untuk mendapatkan data yang mendalam, suatu data yang mengandung makna. Makna adalah data yang sebenarnya, data yang pasti yang merupakan suatu nilai di batik data yang tampak. Oleh karena itu dalam penelitian kualitatif tidak menekankan pada generalisasi, tetapi lebih menekankan pada makna. Selain itu, penelitian deskriptif tidak memberikan perlakuan, manipulasi atau pengubahan pada variabel-variabel yang diteliti, melainkan menggambarkan suatu kondisi yang apa adanya melalui observasi, wawancara, dan dokumentasi. Perancangan Perangkat Lunak Rancangan Antar Muka Perancangan Perangkat Lunak Rancangan Antar Muka 2021 2252 Implementasi Aplikasi collabsproject.com di Masa Pandemi Covid-19 Menjelaskan rancangan antar muka (interface) yang terdapat pada website collabsproject.com. Sosial Teknologi, Vol. 2, No. 12, Desember 2021 an rancangan antar muka (interface) yang terdapat m. Gambar 1. Halaman Beranda Gambar 1. Halaman Beranda Jurnal Indonesia Sosial Teknologi, Vol. 2, No. 12, Desember 2021 2253 Gambar 1. Halaman Beranda Gambar 1. Halaman Beranda Jurnal Indonesia Sosial Teknologi, Vol. 2, No. 12, Desember 2021 2253 Muhammad Ilyas Muhammad Ilyas urnal Indonesia Sosial Teknologi Vol 2 No 12 Desem Gambar 2. Halaman registrasi partner Gambar 3. Halaman store Gambar 4. Halaman News sis Data Tabel 1. Spesifikasi File Tabel Contact ld Tipe Size ke t int 11 Pr urnal Indonesia Sosial Teknologi, Vol. 2, No. 12, Desem Gambar 2. Halaman registrasi partner Gambar 3. Halaman store Gambar 4. Halaman News sis Data Tabel 1. Spesifikasi File Tabel Contact ld Tipe Size ke t int 11 Pri Jurnal Indonesia Sosial Teknologi, Vol. 2, No. 12, Desember 2021 2254 Implementasi Aplikasi collabsproject.com di Masa Pandemi Covid-19 Name varchar 255 email varchar 255 subject varchar 255 Message varchar 255 Tabel 2. Spesifikasi File Tabel perm_proj Nama Field Tipe Size keterangan Id_proj int 11 Primary key email varchar 255 Keterangan_proj varchar 255 Kendala_proj varchar 255 Tabel 3. Spesifikasi File Tabel registpartner Nama Field Tipe Size keterangan Id_registpartner int 11 Primary key email varchar 255 name varchar 255 alamat varchar 255 nohp text Ket_skill varchar 255 Tabel 4. Spesifikasi File Tabel Store Nama Field Tipe Size keterangan Id_brg int 11 Primary key Jenis_brg varchar 255 Tipe_img varchar 255 Data_img longbob 255 Rancangan Struktur Navigasi Implementasi Aplikasi collabsproject.com di Masa Pandemi Covid-19 Rancangan Struktur Navigasi Rancangan Struktur Navigasi Pengujian Unit Pengujian Unit Pengujian Unit Pengujian terhadap program yang dibuat menggunakan black box testing yang fokus terhadap proses masukan dan keluaran program. No Skenario Pengujian Test Case Hasil Yang Diharapkan Hasil Pengujian Kesimpulan 1. Email tidak di isi kemudian klik tombol send Email : (kosong) Sistem akan error dan akan menampilkan pesan “Data Belum Di Isi” Sesuai Harapan Valid 2. project description project descriptio n : (kosong) Sistem akan error dan akan menampilkan pesan “Data Belum Di Isi” Sesuai Harapan Valid g j Pengujian terhadap program yang dibuat menggunakan black box testing yang fokus terhadap proses masukan dan keluaran program. Rancangan Struktur Navigasi Struktur navigasi pada website ini digunakan untuk menggambarkan secara garis besar isi website dan menggambarkan bagaimana hubungan antara isi – isi dari website tersebut. 2255 Jurnal Indonesia Sosial Teknologi, Vol. 2, No. 12, Desember 2021 22 Muhammad Ilyas Pengujian Unit Kesimpulan Dapat di simpulkan dari urain di atas bahwa masalah yang sedang dihadapi sekarang ini di masa pandemi adalah penyerapan tenaga kerja yang belum benar benar terserap dan penyaluran keahlian dari masing masing pencari kerja yang belum tepat, maka dari itu collabsproject.com mencoba membantu membuat atau mencarikan pekerjaan atau project-project yang sesuai dengan keahlian dari masing masing pencari kerja atau mitra itu sendiri, melalui sebuah platform berbasis website. Jurnal Indonesia Sosial Teknologi, Vol. 2, No. 12, Desember 2021 2256 Implementasi Aplikasi collabsproject.com di Masa Pandemi Covid-19 Bibliografi Aristi, Nina Mia. (2021). Peran Digital Freelance Marketplace dan Media Sosial untuk Jual Beli Jasa Profesional Lepas dalam Gig Economy. Fitriani, Feni Freycineta. (2020). Parah! Pertumbuhan Ekonomi Indonesia Hanya 2; 9 Persen Kuartal I/2020. Bisnis. Com. Https://Ekonomi. Bisnis. Com/Read/20200, 505(9), 1236510. Hanoatubun, Silpa. (2020). Dampak Covid–19 terhadap Prekonomian Indonesia. EduPsyCouns: Journal of Education, Psychology and Counseling, 2(1), 146–153. Indayani, Siti, & Hartono, Budi. (2020). Analisis Pengangguran Dan Pertumbuhan Ekonomi Sebagai Akibat Pandemi Covid-19. Jurnal Perspektif, 18(2), 201–208. Kalsum, Umi. (2017). Pengaruh pengangguran dan inflasi terhadap pertumbuhan ekonomi di Sumatera Utara. Jurnal Ekonomikawan, 17(1), 163065. Munandar, Aris. (2017). Analisis Regresi Data Panel Pada Pertumbuhan Ekonomi Di Negara–Negara Asia. Jurnal Ilmiah Ekonomi Global Masa Kini, 8(1), 59–67. Pramesthi, Rovia Nugrahani. (2013). Pengaruh Pengangguran dan Inflasi Terhadap Pertumbuhan Ekonomi di Kabupaten Trenggalek. Jurnal Pendidikan Ekonomi (JUPE), 1(3). Riwayadi, Purwo. (2013). Pemanfaatan Perkembangan Teknologi Informasi Dan Komunikasi Untuk Kemajuan Pendidikan Di Indonesia. Available at PLS-UM Database. Santoso, Y. I. (2020). Ini delapan dampak negatif bagi perekonomian Indonesia akibat wabah virus corona. Kontan. Co. Id. Sitanggang, March Hot Asi, & Manalu, S. Rouli. (2018). Memahami mekanisme Crowdfunding dan Motivasi Berpartisipasi dalam platform kitabisa. com. Interaksi Online, 6(3), 24–34. Sugiyono. (2012). Metode Penelitian Kuantitatif, Kualitatif dan R & D.Bandung:Alfabeta. Metode Penelitian Kuantitatif, Kualitatif Dan R & D.Bandung:Alfabeta. https://doi.org/10.1017/CBO9781107415324.004 2257 Jurnal Indonesia Sosial Teknologi, Vol. 2, No. 12, Desember 2021
https://openalex.org/W2784248529	http://hrcak.srce.hr/file/283072	English	null	The Effect of Heat Modification and Densification on Physical Properties of Poplar Wood	Drvna industrija	2,017	cc-by	5,831	The Effect of Heat Modification and Densification on Physical Properties of Poplar Wood Original scientiﬁ c paper • Izvorni znanstveni rad Received – prispjelo: 24. 3. 2017. Accepted – prihvaćeno: 1. 12. 2017. UDK: 630812.421; 630812.46; 674.031.623.23 doi:10.5552/drind.2017.1719 ABSTRACT • In this study, density, volumetric swelling, mass loss, volume weight, ﬁ ber saturation point and water absorption of the poplar wood (Populus usbekistanica) were investigated with the effect of heat modiﬁ ca- tion (HM) and heat-modiﬁ ed densiﬁ cation (HMD). Poplar samples were modiﬁ ed with steam at 120 °C, 160 °C and 200 °C for 1 and 3 h. After heat modiﬁ cation, the samples were compressed in hot press at a temperature of 120 °C, press pressure of 5 MPa and press time of 30 minutes for densiﬁ cation. Physical properties of the samples were determined according to Turkish standards. The results showed that heat modiﬁ cation affected densiﬁ cation and increased density. Densiﬁ cation had a positive effect on oven dry density (D0), oven dry density after soaking (D0S), volume weight (R) and ﬁ ber saturation point (FSP), except for volumetric swelling. Similarly, the densiﬁ ca- tion process had an increasing effect on the water absorption, excluding 360 hours. Keywords: poplar wood, densiﬁ cation, heat modiﬁ cation, physical properties SAŽETAK • U ovoj je studiji istražen utjecaj toplinske modiﬁ kacije (HM) i ugušćivanja (HMD) na gustoću, volumno bubrenje, gubitak mase, volumnu težinu, točku zasićenosti vlakanaca i upijanje vode drva topole (Popu- lus usbekistanica). Uzorci topolovine modiﬁ cirani su parom pri 120, 160 i 200 °C tijekom jednoga i tri sata. Nakon toplinske modiﬁ kacije uzorci su radi ugušćivanja stlačeni u vrućoj preši, pri temperaturi 120 °C, tlaku prešanja 5 MPa i vremenu prešanja 30 minuta. Fizikalna svojstva uzoraka utvrđena su prema Turskim standardima. Re- zultati su pokazali da je toplinska modiﬁ kacija utjecala na proces ugušćivanja i pridonijela povećanju gustoće. Ugušćivanje je imalo pozitivan učinak na gustoću topolovine u apsolutno suhom stanju (D0), na njezinu gustoću u apsolutno suhom stanju nakon natapanja (D0S), na volumnu težinu (R) i na točku zasićenosti vlakanaca (FSP), ali ne i na volumno bubrenje. Također, ugušćivanje je znatnije utjecalo na sposobnost upijanja veće količine vode, osim za vrijeme od 360 sati. Ključne riječi: topolovina, ugušćivanje, toplinska modiﬁ kacija, ﬁ zikalna svojstva 1 Authors are Ph.D. and professor at Department of Forest Industry Engineering, Faculty of Forestry, Kahramanmaras Sutcu Imam University, Kahramanmaras, Turkey. Düzkale Sözbir, Bektaş: The Effect of Heat Modiﬁ cation and Densiﬁ cation on Physical... Düzkale Sözbir, Bektaş: The Effect of Heat Modiﬁ cation and Densiﬁ cation on Physical... Gonca Düzkale Sözbir, İbrahim Bektaş 1 Authors are Ph.D. and professor at Department of Forest Industry Engineering, Faculty of Forestry, Kahramanmaras Sutcu Imam University, Kahramanmaras, Turkey. 1 Autori su doktor znanosti i professor Odjela za inženjerstvo proizvoda od drva, Šumarski fakultet, Sveučilište Kahramanmaras Sutcu Imam, Kahramanmaras, Turska. DRVNA INDUSTRIJA 68 (4) 315-321 (2017) 315 1 Authors are Ph.D. and professor at Department of Forest Industry Engineering, Faculty of Forestry, Kahramanmaras Sutcu Imam University, Kahramanmaras, Turkey. 1 Autori su doktor znanosti i professor Odjela za inženjerstvo proizvoda od drva, Šumarski fakultet, Sveučilište Kahramanmaras Sutcu Imam, Kahramanmaras, Turska. 2 MATERIALS AND METHODS 2. MATERIJALI I METODE Populus trees were obtained from Kahramanmaraş province in Turkey. Wood samples used in the study were prepared in dimensions 20×20×30 mm. For each group of modiﬁ cation, 27 samples were prepared from logs. Before testing, each sample was conditioned at 20 °C temperature and at 65 % relative humidity until reaching a moisture content of 12 %. The samples were ﬁ rst heat modiﬁ ed by using laboratory drying oven at a temperature of 120 °C, 160 °C and 200 °C and modiﬁ cation time of either 60 or 180 min (Table 1). During heat modiﬁ cation, 100 ml water vapor was heated up to 100 °C degrees in the oven, and then the vapor was removed and kept still until the desired temperatures reached 120 °C, 160 °C and 200 °C. After the heat modiﬁ cation, samples were hot pressed by using a laboratory hot press at 120 °C temperature and 5 MPa press pressure for 30 minutes (Table 1). In 1980s, densiﬁ ed wood products were pro- duced from low density wood species, especially for utilization of some fast growing trees (Wang et al., 2000). The utilization of hardwood species for ﬂ ooring and furniture reduced their supply all over the world. Various efforts have been made to modify the surface of low-quality softwood and planted fast-growing spe- cies. The surface densiﬁ cation technology has been developed and several researches have studied this technology (Rautkari et al., 2008; Gong and Lamason, 2007; Diouf et al., 2011). The thermal compression process might affect the drying duration, dimensional The obtained results were statistically analyzed by using one-way ANOVA and Duncan’s mean separa- Table 1 Parameters used in tests of samples - undensiﬁ ed and densiﬁ ed Tablica 1. Düzkale Sözbir, Bektaş: The Effect of Heat Modiﬁ cation and Densiﬁ cation on Physical... ..... Düzkale Sözbir, Bektaş: The Effect of Heat Modiﬁ cation and Densiﬁ cation on Physical... ..... 1 INTRODUCTION 1. UVOD stability, surface quality, hygroscopicity, durability, and mechanical properties (Welzbacher et al., 2008). stability, surface quality, hygroscopicity, durability, and mechanical properties (Welzbacher et al., 2008). The aim of this research is to determine the inter- active effect of heat modiﬁ cation and post-heat densiﬁ - cation on the poplar wood. The speciﬁ c objective of the study was to analyze volumetric swelling and other chosen properties of heat threated and densiﬁ ed sam- ples compared with samples densiﬁ ed without heat modiﬁ cation. These properties are: density, mass loss, volume weight, ﬁ ber saturation point and water uptake. The aim of this research is to determine the inter- active effect of heat modiﬁ cation and post-heat densiﬁ - cation on the poplar wood. The speciﬁ c objective of the study was to analyze volumetric swelling and other chosen properties of heat threated and densiﬁ ed sam- ples compared with samples densiﬁ ed without heat modiﬁ cation. These properties are: density, mass loss, volume weight, ﬁ ber saturation point and water uptake. There have been many studies on heat modiﬁ ca- tion recently, due to its many advantages such as im- proving the dimensional stability and durability of wood, without using chemicals (Tjeerdsma et al.; 1998; Weiland et al.; 2003; Garcia et al., 2012). This technology was applied to wood at a temperature of about 200ºC for several hours at low oxygen level, a non-inﬂ ammable gas like oil or nitrogen. The main aim of this technique is to convert the chemical composi- tion of wood using heat, which results in a change of physical properties. The process of thermal modiﬁ ca- tion is based on high temperatures, which cause the decomposition of basic wood component like hemicel- luloses and celluloses (Tjeerdsma and Militz, 2005; Rowell et al., 2009). Schneider (1973) noted that treat- ment, at temperatures above 200 ºC and of longer dura- tion, decreased swelling and shrinkage up to 50 %. Besides, he stated that sorption and desorption charac- teristics also changed in heat modiﬁ ed wood. The Effect of Heat Modification and Densification on Physical Properties of Poplar Wood 1 Autori su doktor znanosti i professor Odjela za inženjerstvo proizvoda od drva, Šumarski fakultet, Sveučilište Kahramanmaras Sutcu Imam, Kahramanmaras, Turska. 315 DRVNA INDUSTRIJA 68 (4) 315-321 (2017) Düzkale Sözbir, Bektaş: The Effect of Heat Modiﬁ cation and Densiﬁ cation on Physical... Mass loss (ML), volumet- ric swelling (%) and water absorption (WA) were calcu- lated by equations (2), (3) and (4), respectively, (2) (2) ML – mass loss (%), Mo – mass of oven-dry sam- ple before heat modiﬁ cation (gr), M – mass of the sam- ple after heat modiﬁ cation (gr). ML – mass loss (%), Mo – mass of oven-dry sam- ple before heat modiﬁ cation (gr), M – mass of the sam- ple after heat modiﬁ cation (gr). (g ) VS(%) (3) The average volumetric swelling is given in Fig- ure 1 for undensiﬁ ed and densiﬁ ed poplar samples with and without heat modiﬁ cation. The undensiﬁ ed volu- metric swelling values at 120 and 160 ºC are higher than those of control samples but at 200 ºC they are lower. These differences are insigniﬁ cant (p<0.062). Besides, the duration of heat modiﬁ cation did not indi- cate signiﬁ cant differences in all parameters, except for control samples. On the other hand, in densiﬁ ed sam- ples, all temperatures and durations had a signiﬁ cant effect (p<0.001) on respective values. Moreover, as the duration of heat modiﬁ cation rose, volumetric swelling decreased. The mean volumetric swelling of undensi- ﬁ ed and modiﬁ ed samples (10.1 %) decreased by 3.7 % in comparison with the undensiﬁ ed control group (10.49 %). The value of modiﬁ ed and densiﬁ ed sam- ples (180.35 %) increased by 85.68 % compared to their control samples (97.13 %). (3) VS – volumetric swelling (radial, tangential and longitudinal), L – wet dimension after immersion in water (mm), Lo– oven-dry dimension (mm), WA(%) (4) (4) WA – water absorption, W – weight after immer- sion in water (gr), Wo – oven-dry weight before immer- sion in water (gr). 2 MATERIALS AND METHODS 2. MATERIJALI I METODE Parametri obrade toplinski modiﬁ ciranih uzoraka, neugušćenih i ugušćenih Treatment Obrada Heat temperature Temperatura zagrijavanja °C Heat time Vrijeme zagrijavanja min Pressure Tlak MPa Pressure temperature Temperatura prešanja °C Pressure time Vrijeme prešanja min Undensiﬁ ed / Neugušćeni uzorci Control / kontrolni uzorak - - - - - Heat modiﬁ ed (HM) Toplinski modiﬁ cirani uzorci HM11 120 60 - - - HM13 120 180 - - - HM21 160 60 - - - HM23 160 180 - - - HM31 200 60 - - - HM33 200 180 - - - Densiﬁ ed / Ugušćeni uzorci Control / kontrolni uzorak - - 5 120 30 Heat modiﬁ ed (HM) Toplinski modiﬁ cirani uzorci HM11 120 60 5 120 30 HM13 120 180 5 120 30 HM21 160 60 5 120 30 HM23 160 180 5 120 30 HM31 200 60 5 120 30 HM33 200 180 5 120 30 Table 1 Parameters used in tests of samples - undensiﬁ ed and densiﬁ ed Tablica 1. Parametri obrade toplinski modiﬁ ciranih uzoraka, neugušćenih i ugušćenih DRVNA INDUSTRIJA 68 (4) 315-321 (2017) 316 ..... Düzkale Sözbir, Bektaş: The Effect of Heat Modiﬁ cation and Densiﬁ cation on Physical... Düzkale Sözbir, Bektaş: The Effect of Heat Modiﬁ cation and Densiﬁ cation on Physical... tion test to populate homogeneity groups that showed signiﬁ cant differences at the 95 % conﬁ dence level. The effect of heat modiﬁ cation on density is given in Table 2. According to this Table, it can be seen that oven dry density of densiﬁ ed samples dif- fers from that of undensiﬁ ed samples. Heat modiﬁ ca- tion has not shown signiﬁ cant effect on undensiﬁ ed poplar and has caused 1 % decrease in density. The highest decrease in density among undensiﬁ ed poplar was observed at 200 ºC heat modiﬁ cation (HT31 and HT33). Means of samples densiﬁ ed with heat modiﬁ - cation (897 kg/m3) were compared to density of den- siﬁ ed control samples (668 kg/m3), and the results clearly showed that heat modiﬁ cation increased den- sity by 34 %. The density of samples densiﬁ ed with and without heat modiﬁ cation was increased by 102 % and 174 % on average, respectively. Gong et al. (2010) studied the effect of heat treatment on density between undensiﬁ ed and densiﬁ ed poplar (Populus tremuloides) and concluded that wood density in den- siﬁ ed poplar decreased to 11 % during the heat treat- ment, while density decreased to 2 % in undensiﬁ ed poplar wood. In another words, they concluded that the heat treatment of densiﬁ ed poplar increased den- sity. Wang and Cooper (2005) found that the density of densiﬁ ed wood was affected by compress tempera- ture, press duration time, press closing time and hu- midity of samples before compressing Oven dry density (D0): This term expresses the amount of substance in the full dry unit volume and it is calculated according to TS 2472. Volume weight value (R): This value describes the amount of substance in the full wet unit volume and it is determined according to TS 2472. Fiber saturation point (FSP) was calculated ac- cording to Turkish standards TS 2371 and Equation 1 given below: (1) (1) where VS is volumetric swelling (%). where VS is volumetric swelling (%). Oven dry density after soaking (D0S): After soak- ing for 15 days in water, samples were dried to 0 % moisture content and oven dry densities were deter- mined according to TS 2472. DRVNA INDUSTRIJA 68 (4) 315-321 (2017) 317 3 RESULTS AND DISCUSSION 3. REZULTATI I RASPRAVA Average oven density values of densiﬁ ed and un- densiﬁ ed poplar wood samples with and without heat modiﬁ cation are shown in Table 2. ble 2 Average oven density (kg/m3) values of undensiﬁed and densiﬁed poplar wood; control and heat modiﬁ ed blica 2. Srednja vrijednost gustoće neugušćenih i ugušćenih uzoraka topolovine u apsolutno suhom stanju; vrijednost ntrolnih i toplinski modiﬁ ciranih uzoraka Table 2 Average oven density (kg/m3) values of undensiﬁed and densiﬁed poplar wood; control and heat modiﬁ ed Tablica 2. Srednja vrijednost gustoće neugušćenih i ugušćenih uzoraka topolovine u apsolutno suhom stanju; vrijednosti kontrolnih i toplinski modiﬁ ciranih uzoraka Table 2 Average oven density (kg/m3) values of undensiﬁed and densiﬁed poplar wood; control and heat modiﬁ ed Tablica 2. Srednja vrijednost gustoće neugušćenih i ugušćenih uzoraka topolovine u apsolutno suhom stanju; vrijednosti kontrolnih i toplinski modiﬁ ciranih uzoraka Treatment Obrada Densiﬁ ed / Ugušćeni Undensiﬁ ed / Neugušćeni Increase / Povećanje % Control / kontrolni uzorak 668 331 102 Heat modiﬁ ed / Toplinski modiﬁ cirani uzorci HM11 950 334 184 HM13 859 335 156 HM21 944 331 185 HM23 884 330 168 HM31 878 310 183 HM33 869 321 171 Average / prosječna vrijednost 897 327 174 HM H t diﬁd / t li ki diﬁi i i 317 Düzkale Sözbir, Bektaş: The Effect of Heat Modiﬁ cation and Densiﬁ cation on Physical... ..... 10.49 11.33 11.48 11.56 10.92 7.30 8.01 97.13 185.84 176.11 186.93 175.81182.16 175.25 0 50 100 150 200 Volumetric swelling, % volumno bubrenje, % Undensified / QHXJXãüHQLX]RUFL Densified / XJXãüHQLX]RUFL Figure 1 Average volumetric swelling of undensiﬁed and densiﬁed poplar wood with and without heat-modiﬁ cation Slika 1. Prosječno volumno bubrenje neugušćenih i ugušćenih te nemodiﬁ ciranih i toplinski modiﬁ ciranih uzoraka topolovine e Sözbir, Bektaş: The Effect of Heat Modiﬁ cation and Densiﬁ cation on Physical... ..... Undensified / QHXJXãüHQLX]RUFL Undensified / QHXJXãüHQLX]RUFL Densified / XJXãüHQLX]RUFL Densified / XJXãüHQLX]RUFL Figure 1 Average volumetric swelling of undensiﬁed and densiﬁed poplar wood with and without heat-modiﬁ cation Slika 1. Prosječno volumno bubrenje neugušćenih i ugušćenih te nemodiﬁ ciranih i toplinski modiﬁ ciranih uzoraka topolovi Comparable results were recorded in other stud- ies. For instance, Gong et al. (2010) studied the effect of heat modiﬁ cation on densiﬁ cation and stated that heat modiﬁ ed wood is more resistant to swelling than densiﬁ ed wood. 3 RESULTS AND DISCUSSION 3. REZULTATI I RASPRAVA Also, Bal and Bektaş (2012) noted that heat treatment without steam decreased thickness swelling specially above 180 °C. In some studies, it was noted that the relationship between volumetric swelling and oven dried density was potently positive (Table 3) (Kord et al., 2010; Kurt, 2010). Candan et al. (2013) noted that all thermally compressed boards in- creased thickness swelling values. This result could be explained by springback behavior of wood due to densiﬁcation (Abraham et al., 2010). The results of physical properties of undensiﬁ ed and densiﬁ ed poplar samples without and with heat modiﬁ cation are given in Table 3. Densiﬁ cation process with and without heat modiﬁ cation increased density values according to Ta- ble 3. The highest density for undensiﬁ ed poplar was observed for heat modiﬁ cation temperature of 120 °C compared to control group. Table 3 shows that the den- sities measured at 200 °C in undensiﬁ ed samples seem to be lower, but this decrease is not signiﬁ cant. As seen from statistical results in Table 3, while heat tempera- ture duration had no effect on density in undensiﬁ ed samples, a signiﬁ cant increase in density was observed when the duration of heat modiﬁ cation decreased from Table 3 Comparison of physical properties of undensiﬁed and densiﬁed poplar wood Tablica 3. Usporedba ﬁ zikalnih svojstava neugušćenih i ugušćenih uzoraka topolovine ble 3 Comparison of physical properties of undensiﬁed and densiﬁed poplar wood blica 3. Usporedba ﬁ zikalnih svojstava neugušćenih i ugušćenih uzoraka topolovine Tablica 3. D0 - Oven dry density / gustoća u apsolutno suhom stanju; DOS - Oven dry density after soaking / gustoća u apsolutno suhom stanju nakon natapanja; ML - Mass loss / gubitak mase; R - Volume weight / volumna težina; FSP - Fiber saturation point / točka zasićenosti vlakanaca. Means with the same small letter are not signiﬁ cantly different in Duncan’s mean separation test. / Srednje vrijednosti označene istim malim slovom nisu signiﬁ kantno različite prema Duncanovu testu. 3 RESULTS AND DISCUSSION 3. REZULTATI I RASPRAVA After the samples were heat modiﬁ ed at 200 °C (the mean of HT31 and HT33), the FSP values decreased to 23.75 % in poplar wood. Furthermore, the FSP values at 120 °C and 160 °C in- creased by 7.3 %. The FSP value of the control group (densiﬁ ed and control) was 146.02 %. After the densi- ﬁ cation, the FSP values increased to 40.47 %. Bal and Bektas (2012) noted that heat treatment at higher tem- peratures (180 °C) decreased values of FSP. Addition- ally, the volumetric swelling is very high due to the back-spring effect of wood. Abraham et al. (2010) also noted that the densiﬁ ed wood exhibited this behavior. For this reason, the high springback effect has caused the FPS value to increase up to 223 %. Comparing the mean oven dry density after soak- ing (D0S) with oven dry density (D0), the density de- creased 1 % in undensiﬁ ed samples and decreased 55 % in densiﬁ ed samples as seen in Table 3. This reduc- tion occurred due to the permanent springback effect of the compressed wood. Densiﬁ ed and undensiﬁ ed pop- lar oven-dry densities after soaking (Dos) decreased compared to D0. However, D0S of densiﬁ ed poplar wood was still higher after soaking compared to un- densiﬁ ed samples. Values given in Table 3 demonstrate that densiﬁ - cation process did not cause mass loss but heat modiﬁ - cation increased mass loss with increased temperature. ML values of poplar at 200 °C were greater than at 120 and 160 °C. The average mass loss at 200, 160 and 120 °C was 6 %, 1.33 % and 0.45 %, respectively for all samples. A notably increased dimensional stability fol- lowing thermal modiﬁ cation above 180 °C was noted in many studies (Welzbacher et al., 2008). This could be explained by the decreased hygroscopicity of com- ponents on wood cell walls (Fang et al., 2012). Decom- position of hemicelluloses and lignin at elevated tem- peratures and stress relaxation in samples caused a reduction in the hygroscopicity of wood (Cai et al., 2013). The value of volume weight of densiﬁ ed poplar Table 4 shows that the effect of heat modiﬁ cation on water absorption (%) differed between undensiﬁ ed and densiﬁ ed samples. 3 RESULTS AND DISCUSSION 3. REZULTATI I RASPRAVA Usporedba ﬁ zikalnih svojstava neugušćenih i ugušćenih uzoraka topolovine Treatment / Obrada Do, g/cm3 Dos, g/cm3 ML, % R, g/cm3 FSP, % Undensiﬁ ed / Neugušćeni uzorci Control / kontrolni uzorak 0.331a() 0.328ab __ 0.299a 31.74a Heat modiﬁ ed / Toplinski modiﬁ cirani uzorci HM11 0.334a 0.333 0.30a 0.299a 33.92a HM13 0.335a 0.324b 0.44a 0.299a 34.33a HM21 0.331a 0.329ab 1.05ab 0.296a 34.89a HM23 0.330a 0.328ab 1.82bc 0.297a 33.11a HM31 0.310a 0.303a 8.40e 0.289a 23.41a HM33 0.321a 0.316ab 6.61d 0.296a 24.99a Densiﬁ ed / Ugušćeni uzorci Control / kontrolni uzorak 0.668b 0.376cd __ 0.320b 146.02b Heat modiﬁ ed / Toplinski modiﬁ cirani uzorci HM11 0.950d 0.425e 0.42a 0.335b 195.58c HM13 0.859c 0.394d 0.66ab 0.323b 205.06c HM21 0.944d 0.386cd 0.97ab 0.337b 197.98c HM23 0.884c 0.366c 1.49abc 0.301a 223.36d HM31 0.878c 0.396d 6.36d 0.322b 205.99c HM33 0.869c 0.393cd 2.60c 0.327b 202.56c ANOVA p<0.001 D O d d i / ć l h j D O d d i f ki / ć l h j k DRVNA INDUSTRIJA 68 (4) 315-321 (2017) 318 Düzkale Sözbir, Bektaş: The Effect of Heat Modiﬁ cation and Densiﬁ cation on Physical... 3 to 1h for densiﬁ ed poplar wood. As for densiﬁ ed samples, density decreased depending on the increase of the process duration (from 1 to 3h), except for the temperature of 200 °C. No signiﬁ cant difference was observed. This is in accordance with literature data. For example, Metsä-Kortelainen and Viitanen (2011) determined that 120 °C heat modiﬁ cation increased density compared to control samples, while 200 °C heat modiﬁ cation decreased density due to degradation of hemicellulose and cellulose. In another study, Cao and Huang (2012) noted that steam-heat-modiﬁ cation increased wood thermal conductivity and density. (R) increased because density of densiﬁ ed poplar was higher than that of undensiﬁ ed poplar. The volume weight of densiﬁ ed wood increased by 9.31 %. The value of volume weight of undensiﬁ ed poplar (R) de- creased at 200 °C with heat modiﬁ cation compared to other temperatures. Furthermore, FSP is directly relat- ed to volumetric swelling and oven dry density accord- ing to Equation (1) mentioned in the Method section. As is known, there is a linear relationship between VS and FSP. The FSP value of the control group (undensi- ﬁ ed and control) was 31.74 %. HM – Heat-modiﬁ ed / toplinski modiﬁ cirani uzorci. Means with the same small letter are not signiﬁ cantly different in Duncan’s mean separa- tion test. / Srednje vrijednosti označene istim malim slovom nisu signiﬁ kantno različite prema Duncanovu testu. DRVNA INDUSTRIJA 68 (4) 315-321 (2017) 319 4 CONCLUSIONS 4. ZAKLJUČAK 3. Candan, Z.; Korkut, S.; Unsal, O., 2013: Thermally com- pressed poplar wood (TCW): Physical and mechanical properties. Drvna industrija, 64 (2): 107-211. http://dx.doi.org/10.5552/drind.2013.1216. The main outputs of this study are summarized below: 4. Cai, W.; Yang, J.; Liu, Z.; Hu, Y.; Weisberg, P. J., 2013: Post-ﬁ re tree recruitment of a boreal larch forest in Northeast China. Forest Ecol Manag. 307: 20-29. http://dx.doi.org/10.1016/j.foreco.2013.06.056. 1. Density of undensiﬁ ed poplar wood was minimally (1 %) affected by heat modiﬁ cation. However, density was inﬂ uenced by heat modiﬁ cation before densiﬁ ca- tion and it increased density by 34 %. 5. Cao, Y.; Lu, J.; Huang, R.; Jiang, J., 2012: Increased di- mensional stability of Chinese ﬁ r through steam-heat- threatment. Eur. J. Wood Prod. 70: 441-444. http://dx.doi.org/10.1007/s00107-011-0570-y. 2. Volumetric swelling of undensiﬁ ed poplar wood de- creased signiﬁ cantly only at 200 °C heat modiﬁ cation. Moreover the densiﬁcation increased the volumetric 2. Volumetric swelling of undensiﬁ ed poplar wood de- creased signiﬁ cantly only at 200 °C heat modiﬁ cation. Moreover, the densiﬁ cation increased the volumetric swelling due to the springback effect. Moreover, the densiﬁ cation increased the volumetric swelling due to the springback effect. 6. Diouf, P. N.; Stevanovich T.; Cloutier, A.; Fang, C. H.; Blanchet, P.; Koubaa, A.; Mariotti, N., 2011: Effects of thermo-hygro-mechanical densiﬁcation on the surface characteristics of trembling aspen and hybrid poplar wood veneers. Applied Surface Science, 257: 3558-3564. http://dx.doi.org/10.1016/j.apsusc.2010.11.074. 3. In comparison to oven dry density (D0), the oven dry density after soaking (D0S) decreased in undensi- ﬁ ed and densiﬁ ed samples at the rate of 1 % and 55 %, respectively. In addition, the mass loss increased with the increase of heat modiﬁ cation temperature in all samples, notwithstanding the densiﬁ cation process. Densiﬁ cation process increased the volume weight and ﬁ ber saturation point compared to heat modiﬁ cation. 7. Fang, C. H.; Cloutier, A.; Blanchet, P.; Koubaa, A., 2012: Densiﬁ cation of wood veneers combined with oil-heat treatment. Part II: Hygroscopicity and mechanical prop- erties. BioResources, 7 (1): 925-935. http://dx.doi.org/10.15376/biores.7.1.0925-0935. 4. No signiﬁ cant correlation was found between water absorption time and heat modiﬁ cation process for un- densiﬁ ed samples. On the other hand, in densiﬁ ed sam- ples, water absorption retention improved as the heat modiﬁ cation temperature and duration increased. Be- sides, the effect of the application time increased de- pending on temperature increase. Acknowledgement – Zahvala This work was supported by a grant from the Kahramanmaras Sutcu Imam University Scientiﬁ c Re- search Projects Unit, Project Number: 2013/2-40M. 5 REFERENCES 5. LITERATURA 1. Abraham, J.; Nemeth, R.; Molnar, S., 2010: Thermome- chanical densiﬁcation of Pannonia Poplar. In: Proceed- ings of the ﬁ nal conference of COST Action E53: ‘Qual- ity control for wood & wood products’ 4-7th May 2010, Edinburgh. 2010.05.04-2010.05.10. pp. 282-292. g pp 2. Bal, B. C.; Bektaş, İ., 2013: The Effects of Heat Treat- ment on Some Mechanical Properties of Juvenile Wood and Mature Wood of Eucalyptus grandis. Drying Tech- nology, 31: 479-485. http://dx.doi.org/10.1080/07373937.2012.742910. 3 RESULTS AND DISCUSSION 3. REZULTATI I RASPRAVA The lowest water absorption in the densiﬁ ed wood in the in- terval from 1 hour to 360 hours was determined in HT31 compared to the densiﬁ ed control. The water absorption in the ﬁ rst 24 hours in the densiﬁ ed wood continued to be water intake without heat treatment temperature and duration effective, and after 24 hours HT23, HT31 and HT33 showed lower water absortion than the densiﬁ ed control sample. When compared to the undensiﬁ ed con- trol sample (129.62 %), the lowest water absorption per- centage was determined in HM23 (122.71 %). Likewise, in densiﬁ ed samples, the lowest decrease in HM31 (89.56 %) was calculated according to control samples (119.02 %).Also, Cai et al. (2013) noted that the water absorption of densiﬁ ed wood with and without heat treatment was affected by compression temperature, compression ratio and pressure holding time. Similar ﬁ ndings were obtained by Unsal et al. (2009), and they indicated that the compressed pine decreased water ab- sorption for 5MPa at 120 °C compared to uncompressed pine, while higher compression temperature and com- pression ratio increased water absorption. niﬁ cant difference in water absorption compared to the control sample. As water absorption (%) changed with time, the lowest water absorption in undensiﬁ ed group was obtained in HT23 (122.71 %) after 360 hours. The lowest water absorption in the densiﬁ ed wood in the in- terval from 1 hour to 360 hours was determined in HT31 compared to the densiﬁ ed control. The water absorption in the ﬁ rst 24 hours in the densiﬁ ed wood continued to be water intake without heat treatment temperature and duration effective, and after 24 hours HT23, HT31 and HT33 showed lower water absortion than the densiﬁ ed control sample. When compared to the undensiﬁ ed con- trol sample (129.62 %), the lowest water absorption per- centage was determined in HM23 (122.71 %). Likewise, in densiﬁ ed samples, the lowest decrease in HM31 (89.56 %) was calculated according to control samples (119.02 %).Also, Cai et al. (2013) noted that the water absorption of densiﬁ ed wood with and without heat treatment was affected by compression temperature, compression ratio and pressure holding time. Similar ﬁ ndings were obtained by Unsal et al. 3 RESULTS AND DISCUSSION 3. REZULTATI I RASPRAVA (2009), and they indicated that the compressed pine decreased water ab- sorption for 5MPa at 120 °C compared to uncompressed pine, while higher compression temperature and com- pression ratio increased water absorption. 6. One of the remarkable characteristics of this study is that no chemicals have been used during the process and there have been no side effects that could adverse- ly affect human and environmental health during and after the process. 7. In order to make the above physical properties meaningful and useful at the desired level, mechanical properties should also be researched in future as a part of this study. 3 RESULTS AND DISCUSSION 3. REZULTATI I RASPRAVA It can be seen from Table 4 that, in the densiﬁ ed samples, the heat modiﬁ cation time had no statistically signiﬁ cant effect on water absorption at 120 ºC (HM11, HM13), whereas at 200 ºC (HM31, HM33) the respective effect was signiﬁ cant (p<0.000). As shown in Table 4, the modiﬁ cation temperature and duration have no statistically signiﬁ cant effect in the ﬁ rst 48 hours only in heat modiﬁ ed (not densiﬁ ed) wood. The water absorption values of heat modiﬁ ed samples decreased after 48 hours compared to control group. However, in 72 h and 96 h there was no statistically sig- Table 4 Water absorption (%) of undensiﬁed and densiﬁed poplar wood without and with heat-modiﬁ cation Tablica 4. Upijanje vode (%) ugušćenih i neugušćenih te nemodiﬁ ciranih i toplinski modiﬁ ciranih uzoraka to Table 4 Water absorption (%) of undensiﬁed and densiﬁed poplar wood without and with heat-modiﬁ cation Tablica 4. Upijanje vode (%) ugušćenih i neugušćenih te nemodiﬁ ciranih i toplinski modiﬁ ciranih uzoraka topolovine Treatment Obrada 1 h 2 h 3 h 24 h 48 h 72 h 96 h 360 h Undensiﬁ ed Neugušćeni uzorci Control / kontrolni uzorci 7.64a() 13.77a 21.24a 51.49a 81.94bcd 92.05b 99.19b 129.62bcd HM11 8.16a 13.48a 21.74a 48.26a 77.78bc 88.92b 95.13b 129.19bc HM13 8.95a 15.75a 21.16a 51.67a 80.55bcd 91.00b 99.83b 130.42bcde HM21 8.95a 15.04a 21.02a 49.43a 79.05bcd 90.78b 99.53b 131.44bcde HM23 8.61a 14.42a 18.79a 45.03a 74.90bc 87.81b 94.32b 122.71bc HM31 10.00a 17.57a 21.70a 50.57a 77.74bc 100.74b 112.24b 150.44def HM33 6.25a 11.50a 17.39a 43.28a 68.51ab 87.52b 95.32b 131.78bcde Densiﬁ ed / Ugušćeni uzorci Control / kontrolni uzorci 30.98cd 54.44b 56.96b 84.12b 96.54d 105.50b 108.78b 119.02b HM11 31.21cd 49.44b 58.55b 112.48cd 126.23ef 132.82c 135.64cd 151.00ef HM13 23.32bc 50.15b 62.84b 107.05c 122.02e 127.69c 130.33c 149.29cdef HM21 36.20cde 59.74b 80.25c 126.09d 141.08f 156.15d 150.80d 167.31f HM23 48.24e 62.70b 69.70bc 78.80b 90.08cd 94.59b 98.23b 111.81b HM31 11.54ab 21.08a 25.60a 48.87a 58.09a 67.76a 68.01a 89.56a HM33 41.33de 50.73b 58.65b 81.33b 91.55cd 98.68b 100.85b 118.49b ANOVA p<0.001 * DRVNA INDUSTRIJA 68 (4) 315-321 (2017) 319 319 Düzkale Sözbir, Bektaş: The Effect of Heat Modiﬁ cation and Densiﬁ cation on Physical.. niﬁ cant difference in water absorption compared to the control sample. As water absorption (%) changed with time, the lowest water absorption in undensiﬁ ed group was obtained in HT23 (122.71 %) after 360 hours. Prof. Dr. İBRAHIM BEKTAŞ, Ph.D. Prof. Dr. İBRAHIM BEKTAŞ, Ph.D. Kahramanmaras Sutcu Imam University, Faculty of Forestry Department of Forest Industry Engineering 46100 Kahramanmaras, TURKEY e-mail: ibtas@ksu.edu.tr Kahramanmaras Sutcu Imam University, Faculty of Forestry Department of Forest Industry Engineering 46100 Kahramanmaras, TURKEY e-mail: ibtas@ksu.edu.tr Kahramanmaras Sutcu Imam University, Faculty of Forestry Department of Forest Industry Engineering 46100 Kahramanmaras, TURKEY e-mail: ibtas@ksu.edu.tr 18. Tjeerdsma, B. F.; Militz, H., 2005: Chemical changes in hydrothermal treated wood: FTIR analysis of combined hydrothermal and dry heat-treated wood. Holz als Roh- und Werkstoff, 63 (2): 102-111. http://dx.doi.org/10.1007/s00107-004-0532-8. Department of Forest Industry Engineering 46100 Kahramanmaras, TURKEY e-mail: ibtas@ksu.edu.tr Düzkale Sözbir, Bektaş: The Effect of Heat Modiﬁ cation and Densiﬁ cation on Physical... Düzkale Sözbir, Bektaş: The Effect of Heat Modiﬁ cation and Densiﬁ cation on Physical... 11. Kord, B.; Kialashaki, A.; Kord, B., 2010: The within-tree variation in wood density and shrinkage, and their rela- tionship in Populus euramericana. Turk. J. Agric. For., 34: 121-126. http://dx.doi.org/10.3906/tar-0903-14. 19. Ünsal, Ö.; Kartal, S. N.; Candan, Z.; Arango, R. A.; Clausen, C. A.; Green, F., 2009: Decay and termite resist- ance, water absorption and swelling of thermally com- pressed wood panels. International Biodeterioration, 63: 548-552. http://dx.doi.org/10.1016/j.ibiod.2009.02.001. 12. Kurt, R., 2010: Suitability of three hybrid poplar clones for laminated veneer lumber manufacturing using mela- mine urea formaldehyde adhesive. BioResources, 5 (3): 1868-1878. http://dx.doi.org/10.15376/biores.5.3.1868-1878. 20. Wang, J.; Cooper, P., 2005: Effect of oil type, tempera- ture and time on moisture properties of hot oil-treated wood. Holz Roh Werkst, 63: 417-422. http://dx.doi.org/10.1007/s00107-005-0033-4. http://dx.doi.org/10.1007/s00107-005-0033-4. 13. Metsä-Kortelainen, S.; Viitanen, H., 2011: Wettability of sapwood and heartwood of thermally modiﬁ ed Norway spruce and Scots pine Eur. J. Wood Prod., 70 (1-3): 135- 139. http://dx.doi.org/10.1007/s00107-011-0523-5. 21. Wang, J. M.; Zhao, G. J.; Lida, I., 2000: Effect of oxida- tion on heat ﬁ xation compressed wood of China ﬁ r. For- estry studies in China, 2: 73-79. 22. Weiland, J. J.; Guyonnet, R., 2003: Study of chemical modiﬁcations and fungi degradation of thermally modiﬁed wood using DRIFT spectroscopy. Holz als Roh- und Werkstoff, 61: 216-220. http://dx.doi.org/10.1007/s00107-003-0364-y. 14. Rautkari, L.; Properzi, M.; Pichelin, F.; Hughes, M., 2008: An innovative thermo densiﬁ cation method for wooden surfaces. In: Proceedings of the 10 th World Conference on Timber Engineering, Miyazaki, Japan, pp.177-184. 15. Rowell, R.; Ibach, E. E.; McSweeny, J.; Nilsson, T., 2009: Understanding decay resistance, dimensional sta- bility and strength changes in heat treated and acetylated wood. Proceedings of 4th European conference on wood modiﬁ cation, Stockholm, 489-502. https://doi.org/10.1080/17480270903261339 23. Welzbacher, C. R.; Wehsener, J.; Rapp, A. O.; Haller, P., 2008: Thermo-mechanical densiﬁ cation combined with thermal modiﬁ cation of Norway spruce in industrial scale- dimensional stability and durability aspects. Holz als Roh-und Werkstoff, 66: 39-49. http://dx.doi.org/10.1007/s00107-007-0198-0. 16. Schneider, A., 1973: Investigation on the convection dry- ing of lumber at extremely high temperatures. Part II: Drying degrade, changes in sorption, colour and strength of pine sapwood and beech wood at drying temperatures from 110 to 180 °C. Holz. als Roh-und Werkstoff, 31: 198-206. https://doi.org/10.1007/BF02613842 4 CONCLUSIONS 4. ZAKLJUČAK 8. Garcia, R. A.; Carvalho, A. M.; Latorraca, J. V. F.; Matos, J. L. M.; Santos, W. A.; Silva, R. F. M., 2012: Nonde- structive evaluation of heat-treated Eucalyptus grandis Hill ex Maiden wood using stress wave method. Wood Sci. Technol., 46: 41-52. http://dx.doi.org/10.1007/s00226-010-0387-6. 9. Gong, M.; Lamason, C., 2007: Improvement of surface properties of low density wood: Mechanical modiﬁ ca- tion with heat treatment. Research report, Project No: UNB57, Value to Wood No: UNB57 Natural Resources Canada. 5. Poplar trees are fast growing and have low durabil- ity. However, poplar wood has become a material with high density and lower water absorption by heat modi- ﬁ cation and densiﬁ cation process. Thus, species with low durability can be converted to wood material of high density and extended service life by heat modiﬁ - cation and densiﬁ cation process. 10. Gong, M.; Lamason, C.; Li, L., 2010: Interactive effect of surface densiﬁ cation and postheat-treatment on aspen wood. J Mater Process Tech., 210: 293-296. http://dx.doi.org/10.1016/j.jmatprotec.2009.09.013. DRVNA INDUSTRIJA 68 (4) 315-321 (2017) DRVNA INDUSTRIJA 68 (4) 315-321 (2017) 320 Corresponding address: 17. Tjeerdsma, B. F.; Boonstra, M.; Militz, H., 1998: Ther- mal modiﬁcation of non-durable wood species 2. Im- proved wood properties of thermal treated wood. In: Pro- ceedings of 29th Annual meeting, Maastricht-The Low Countries, 14-19 May. Doc. No. IRG/WP/98-40124. DRVNA INDUSTRIJA 68 (4) 315-321 (2017) 321
https://openalex.org/W2518455214	https://biblio.ugent.be/publication/8540095/file/8540096	English	null	Longitudinal associations of lifestyle factors and weight status with insulin resistance (HOMA-IR) in preadolescent children: the large prospective cohort study IDEFICS	The international journal of behavioural nutrition and physical activity	2,016	cc-by	9,724	Longitudinal associations of lifestyle factors and weight status with insulin resistance (HOMA-IR) in preadolescent children: the large prospective cohort study IDEFICS Jenny Peplies1,2, Claudia Börnhorst1, Kathrin Günther1, Arno Fraterman3, Paola Russo4, Toomas Veidebaum5, Michael Tornaritis6, Stefaan De Henauw7, Staffan Marild8, Dénes Molnar9, Luis A. Moreno10, Wolfgang Ahrens1,2,11* and on behalf of the IDEFICS consortium * Correspondence: ahrens@bips.uni-bremen.de 1Leibniz Institute for Prevention Research and Epidemiology – BIPS, Bremen, Germany 2Faculty of Human and Health Sciences, Institute for Public Health and Nursing Research, Bremen University, Bremen, Germany Full list of author information is available at the end of the article © 2016 The Author(s). Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Peplies et al. International Journal of Behavioral Nutrition and Physical Activity (2016) 13:97 DOI 10.1186/s12966-016-0424-4 Peplies et al. International Journal of Behavioral Nutrition and Physical Activity (2016) 13:97 DOI 10.1186/s12966-016-0424-4 RESEARCH Open Access Longitudinal associations of lifestyle factors and weight status with insulin resistance (HOMA-IR) in preadolescent children: the large prospective cohort study IDEFICS Jenny Peplies1,2, Claudia Börnhorst1, Kathrin Günther1, Arno Fraterman3, Paola Russo4, Toomas Veidebaum5, Michael Tornaritis6, Stefaan De Henauw7, Staffan Marild8, Dénes Molnar9, Luis A. Moreno10, Wolfgang Ahrens1,2,11* and on behalf of the IDEFICS consortium DOI 10.1186/s12966-016-0424-4 Open Access Background for monitoring purposes (with HOMA-IR ≥90th per- centile amongst other criteria) and one to define a cut point for clinical action (with HOMA-IR ≥95th percent- ile amongst other criteria). When HOMA-IR alone is considered prevalence proportions are usually higher than those for MS [8, 21], but data from population based studies in children are still limited. Insulin resistance (IR), a reduced physiological response of the peripheral tissues to normal levels of insulin, is a growing concern in childhood obesity, although not all obese people are insulin resistant and IR may also occur in nonobese children and adults [1]. IR is also a central as- pect of the metabolic syndrome (MS) and most likely a link between obesity and type 2 diabetes [2–6]. Given the constantly growing prevalence of metabolic disorders in children and adolescents [7], it appears to be important to identify children at risk before clinical symptoms occur. The importance of metabolic health in the presence of obesity has been of growing research interest in the last years. Metabolically healthy obesity is apparently associ- ated with lower levels of adiposity in youth [22, 23] and a high level of physical activity (PA) in adults [22, 23]. Increased PA and cardiorespiratory fitness are also be- lieved to attenuate the risk of cardiovascular disease, type 2 diabetes, and MS, independent of weight status [24]. Physical fitness and physical activity are closely re- lated and have been shown to correlate well in children [25]. The Lancet Physical Activity Series working group estimated that worldwide in 2008, physical inactivity caused 6 % of the burden of disease from coronary heart disease and 7 % of type 2 diabetes, as well as 9 % of pre- mature mortality [26]. A meta-analysis on the influence of cardiorespiratory fitness and weight status on mortal- ity from all causes showed that overweight and obese fit individuals had similar mortality risks as normal weight fit individuals. Compared to normal weight fit individ- uals, unfit individuals had twice the risk of mortality re- gardless of their body mass index (BMI) [27]. Several risk factors for IR have been suggested, e.g. Abstract Background: This study investigates prospective associations of anthropometrical and lifestyle indices with insulin resistance (IR) in European children from the IDEFICS cohort. Insulin resistance (IR) is a growing concern in childhood obesity and a central aspect of the metabolic syndrome (MS). It most likely represents the link between obesity and type 2 diabetes. Methods: This longitudinal study included 3348 preadolescent children aged 3 to 10.9 years from 8 European countries who were observed from 2007/2008 to 2009/2010. The main outcome measure in the present analysis is HOMA-IR (homeostasis model assessment as a common proxy indicator to quantify IR) at follow-up and in its longitudinal development. Anthropometrical measures and lifestyle indices, including objectively determined physical activity, were considered, among others factors, as determinants of IR. Prospective associations between IR at follow-up and anthropometrical and lifestyle indices were estimated by logistic regression models. Results: Country-specific prevalence rates of IR in the IDEFICS cohort of European children showed a positive trend with weight category. Prospective multivariate analyses showed the strongest positive associations of IR with BMI z-score (OR = 2.6 for unit change from the mean, 95 % CI 2.1–3.1) and z-score of waist circumference (OR = 2.2 for unit change from the mean, 95 % CI 1.9–2.6), which were analysed in separate models, but also for sex (OR = 2.2 for girls vs. boys, 95 % CI 1.5–3.1 up to OR 2.5, 95 % CI 1.8–3.6 depending on the model), audio-visual media time (OR = 1.2 for an additional hour per day, 95 % CI 1.0–1.4 in both models) and an inverse association of objectively determined physical activity (OR = 0.5 for 3rd compared to 1st quartile, 95 % CI 0.3–0.9 in both models). A longitudinal reduction of HOMA-IR was accompanied with a parallel decline in BMI. (Continued on next page) © 2016 The Author(s). Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Page 2 of 12 Peplies et al. Abstract International Journal of Behavioral Nutrition and Physical Activity (2016) 13:97 (Continued from previous page) Conclusions: This study is, to our knowledge, the first prospective study on IR in a preadolescent children’s population. It supports the common hypothesis that overweight and obesity are the main determinants of IR. Our data also indicate that physical inactivity and a sedentary lifestyle are likewise associated with the development of IR, independent of weight status. The promotion of physical activity should thus be considered as an equal option to dietary intervention for the treatment of IR in the paediatric practice. Keywords: Insulin resistance, HOMA-IR, Physical activity, Accelerometer, Obesity, Pre-pubertal children, Cohort study Abbreviations: BIA, Bioelectrical impedance analysis; BMI, Body mass index; HOMA-IR, Homeostasis model assessment to quantify insulin resistance; IDEFICS (acronym), Identification and prevention of dietary- and lifestyle-induced health effects in children and infants; IR, Insulin resistance; ISCED, International Standard Classification of Education; MS, Metabolic syndrome; MVPA, Time spent in moderate to vigorous physical activity; OR, Odds ratio; PA, Physical activity; SACINA, Self-administered children and infants nutrition assessment (a standardized 24-h recall method) Background weight status [8], measures of central and peripheral adi- posity [9, 10], dietary factors such as intake of total fat or saturated fat [11, 12], physical inactivity [13], poor physical fitness [14], low or high birth weight for gesta- tional age [7, 15], and maternal factors like gestational diabetes [16] or unbalanced maternal nutrition [17]. Nevertheless, in children, population-based epidemio- logical data on the determinants of IR are still rare and mostly available from cross-sectional studies. A recent review [18] on the clustering of obesogenic behaviours in children or adolescents concluded that further re- search is needed particularly in younger children and from longitudinal studies. The prevalence of MS in children has recently been described in a systematic review [19] that related to IR as one of several possible criteria of MS. Estimation of prevalence in this review was not straightforward be- cause many different criteria had been used to define MS in children. The median prevalence was overall 3.3 % (range 0–19.2 %), 11.9 % (range 2.8–29.3 %) in overweight children, and 29.2 % (range 10–66 %) in obese children, when studies in all ethnicities were con- sidered. For European children only, prevalence tended to be slightly lower. Two new definitions of paediatric MS were also suggested by the IDEFICS (“Identification and prevention of dietary- and lifestyle-induced health effects in children and infants”) study group [20], one The present study investigates the prospective associa- tions between anthropometrical and lifestyle indices with IR in preadolescent European children. It also shows the development of HOMA-IR-values in relation to the lon- gitudinal development of weight status between the two IDEFICS surveys. Insulin resistance Fasting insulin and HOMA-IR (homeostasis model as- sessment to quantify IR) have been suggested among others as surrogate markers for screening purposes in adults [33]. The gold standard method to measure insu- lin sensitivity, the hyperinsulinemic euglycemic clamp, is invasive and very labour- and time-intensive, and thus not feasible in epidemiological research. Estimates of IR from HOMA-IR have been shown to correlate well with estimates from the clamp-technique [34], especially in healthy populations [35]. IR was defined as HOMA-IR ≥95th percentile, calculated for half year age groups [29]. Study design IDEFICS is an Integrated Project within the 6th Frame- work Programme of the European Commission aimed at Peplies et al. International Journal of Behavioral Nutrition and Physical Activity (2016) 13:97 Page 3 of 12 Page 3 of 12 the investigation of diet- and lifestyle-related diseases and their prevention. The baseline survey was conducted in the school year 2007/2008 and included 16,228 pre- school and primary school children aged 2 to 9 years from eight European countries. The survey comprised anthropometrical measurements and examinations of children as well as parental self-completion question- naires on lifestyle habits and dietary intakes of children. Biomarkers were analysed in blood, urine, and saliva samples. Standardised procedures were used by all sur- vey centres. Venous blood was collected from 9185 of the IDEFICS children. The majority of children were re- examined after 2 years during a follow-up survey (n = 11,292 plus 2517 newly recruited children). The back- ground of the study, its research goals and instruments have been described elsewhere in detail [28]. Study sample y After exclusion of non-fasting children, children with diagnosed diabetes, children who had missing data for waist circumference or one of the laboratory analyses of interest, and children from Cyprus (due to the very small sample size that fulfilled the requirements), 6970 and 6708 children remained in the data sets for baseline and follow-up survey, respectively. The present analyses are based on the 3348 children who were part of both groups. For the prospective ana- lyses, also children with IR at baseline (HOMA-IR ≥95th percentile, according to Peplies et al. [29]) were excluded, which further reduced the sample to 3125 children. Due to missing information for some of the potential risk factors or confounders, some analyses were conducted in smaller samples (numbers are indi- cated in the tables). A selection effect could be ex- cluded as characteristics of subgroups (children with data on sleep duration or children with accelerometer data) only differed by the first or second decimal compared to the whole study sample (data not shown). Blood samples p Children participating in the IDEFICS surveys were asked to donate fasting venous blood samples. If con- sent was not given for venous blood withdrawal, ca- pillary blood was taken with the consent of the parents and children. For the present analyses, only children with venous blood samples were included. Pre-analytical processing of blood samples was done at the local survey centres or at local laboratories. Samples were frozen at −80 °C and shipped to a cen- tral laboratory (accredited according to ISO 15189) for later analysis. Details on the biological sampling pro- cedures can be obtained from a previous publication [36]. Blood glucose was assessed on site at each study centre by point-of-care analysis using the Cholestech LDX analyser (Cholestech®, Hayward, CA, USA) either in venous or capillary blood. Insulin was determined by electrochemilu- minescence immunoassay in the central laboratory. HOMA-IR was calculated as fasting insulin (μIU/ml) × fasting glucose (mg/dl)/405. Physical examination The physical examination programme during the two IDEFICS surveys covered standard anthropometric mea- sures, i.e. height (SECA 225), weight (TANITA BC 420 SMA), and circumferences of waist, hip, upper arm, and neck (SECA 200), as well as skinfold thicknesses (sub- scapular and triceps) (Holtain skinfold caliper), foot-to- foot bioelectrical impedance (TANITA BC 420 SMA), and the measurement of blood pressure and pulse rate (Welch Allyn 4200B) [28]. BMI was calculated as weight (in kg) divided by height squared (in m). Sex-specific BMI categories were interpolated for continuous age ac- cording to the extended IOTF criteria [30]. Sleep duration Information on sleep duration was collected in the con- text of a standardised 24-h recall called SACINA (self- administered children and infants nutrition assessment). SACINA is a computer-based instrument filled out by the parents/guardians of participating children with the assistance of a study nurse. Next to questions on dietary intakes, parents were asked about their children’s get up and bed time (hour/minute) of the previous day. Sleep duration on weekday nights was calculated resulting in a continuous estimate of sleep hours per night as de- scribed previously [31, 32]. Questionnaire data Data on education, lifestyle habits and dietary intakes of children was retrieved from parental self-completion questionnaires. Parental education was used as a proxy of socioeconomic status. It was coded country-by- country according to the International Standard Classifi- cation of Education (ISCED) [37]. For the analyses, the Page 4 of 12 Page 4 of 12 Peplies et al. International Journal of Behavioral Nutrition and Physical Activity (2016) 13:97 median (25th, 75th percentile). Normality of distribution was assessed by the Kolmogorov-Smirnov test. Preva- lence of IR was calculated using the age- and sex- specific 95th percentile of HOMA-IR derived from normal weight IDEFICS children as cut off value for each half year age group [29]. Furthermore, an age- and sex-specific z-score was calculated for HOMA based on this healthy paediatric population. Delta z-IR was calcu- lated as the difference between the z-scores of HOMA- IR-values at T1 and T0 to depict the development of HOMA-IR between the two surveys. Logistic regression analyses was limited to children without IR at T0, i.e. to all children with a HOMA-IR <95th percentile in their corresponding half year age groups (cut-off ranged from 1.5 to 2.9 for 3–9 year old girls and 1.3–2.7 for the re- spective boys) [29]. An indicator variable for presence of HOMA-IR at follow-up (HOMA-IR above or below 95th percentile) was defined as dependent variable for the lo- gistic regression analyses. Univariate logistic regression analyses were conducted for sex, age (continuous), ISCED-level, BMI, BMI z-score, z-score of waist circum- ference (continuous and in quartiles), audio-visual media time (in quartiles), number of media in bedroom (in quartiles), propensity scores for sugar and fat consump- tion (in quartiles), sleep duration (continuous), and for accelerometer data as time spent in moderate to vigor- ous physical activity (MVPA, in quartiles) both, for all children and children with normal weight only. Multi- variate mixed logistic models were calculated separately for the exposures BMI z-score and z-score of waist cir- cumference to avoid collinearity. Covariables were added to the model if they were significant in the univariate analysis for normal weight children, i.e. audio-visual media time (as continuous variable), sex, age, ISCED level and time spend in MVPA (in quartiles). A random country effect was included in the model to account for the clustered study design. Quality assurance All measurements followed detailed standard operation procedures (SOPs) which were pre-tested before the baseline survey [43]. Field personnel from each study centre participated in a central training course. Site visits were conducted at all study locations during field sur- veys to check adherence to the SOPs. Questionnaires were developed in English, translated to local languages, and checked for translation errors after back-translation. All technical equipment and laboratory materials were purchased centrally to maximise comparability of data. Laboratory analyses were conducted at the central la- boratory which was accredited according to ISO 15189. Questionnaire data ‘Number of media in bed- room’ was dropped in favour of audio-visual media time as both measure a similar construct, thus to avoid collin- earity. Even though no significant effect was seen for the nutritional covariables in the univariate analyses, fat con- sumption score (as continuous marker) was included into the model due to the biological connection of the children’s nutrition and the biological markers. Odds ra- tios (ORs) and 95 % confidence limits were calculated for all children and for boys and girls separately. All ana- lyses were performed using SAS® statistical software ver- sion 9.3 (SAS Institute, Inc., Cary, NC). maximum ISCED level of both parents was considered. Media use was used as a proxy of sedentary behavior. It was described by the time spent with audiovisual media (hours/week) and the number of audiovisual media de- vices (TV, Computer, Internet connection, DVD player, Playstation, Game console) located in the child’s bed- room. For dietary assessment, propensity scores for sugar and fat consumption of children were developed from a parental food frequency questionnaire [38, 39]. This questionnaire recorded the child’s consumption of certain obesity-promoting or -inhibiting food items from a given list of foods [40] on a typical week, excluding foods provided in school or day care setting. Based on the food frequency questionnaire, the estimation of en- ergy intake or total food intake was thus not possible. A continuous index was developed, using the total weekly frequency for high-sugar or high-fat items divided by the individual’s total consumed food frequencies. Physical activity d ( Activity data (time-varying accelerations) of a subset of participating children was recorded by uniaxial accelerom- eters (ActiGraph® GT1M or Actitrainer, LLC, Pensacola, FL, USA). Accelerometers were mounted on the right hip with an elastic belt ensuring close contact with the body. Activity data was analysed on the basis of a 60 s epoch. Included children had at least three measurement days and a minimum of 8 h of valid ac- celerometer wear time per day. Non-wear time was assumed for 20 min of consecutive zero counts. Time spent performing either moderate or vigorous inten- sity of PA (average minutes over all valid recording days) was calculated according to the cut points pro- posed by Evenson [41]. A detailed description of IDE- FICS accelerometer data can be found elsewhere [42]. Statistical analyses International Journal of Behavioral Nutrition and Physical Activity (2016) 13:97 Page 5 of 12 Table 1 Characteristics of study population Baseline survey T0 Follow-up survey T1 All Boys Girls All Boys Girls N 3348 1743 1605 3348 1743 1605 Age (years) a 6.4 (± 1.7) 6.3 (± 1.7) 6.4 (± 1.6) 8.4 (± 1.7) 8.3 (± 1.7) 8.4 (± 1.6) Anthropometry Weight (kg) a 23.9 (± 6.9) 24.0 (± 6.9) 23.9 (± 6.9) 30.6 (± 9.1) 30.7 (± 9.1) 30.6 (± 9.1) Height (cm) a 120.0 (± 12.0) 120.1 (± 12.0) 119.9 (± 12.0) 132.3 (± 11.3) 132.4 (± 11.2) 132.3 (± 11.4) BMI (kg/m2) a 16.3 (± 2.4) 16.3 (± 2.3) 16.3 (± 2.4) 17.2 (± 3.0) 17.1 (± 3.0) 17.2 (± 3.0) Waist circumference (cm) a 54.7 (± 6.7) 55.0 (± 6.8) 54.4 (± 6.6) 59.1 (± 8.4) 59.4 (± 8.5) 58.9 (± 8.3) Biochemical markers Fasting glucose (mmol/l) a 4.6 (± 0.5) 4.7 (± 0.5) 4.6 (±0.5) 4.8 (± 0.5) 4.8 (± 0.5) 4.7 (± 0.5) Fasting insulin (pmol/l) b 25.2 (15.6, 38.4) 23.9 (14.4, 37.1) 26.7 (17.4, 39.7) 38.0 (26.1, 55.4) 35.1 (23.6, 50.6) 42.1 (28.7, 61.6) HOMA-IR b 0.7 (0.4, 1.2) 0.7 (0.4, 1.1) 0.8 (0.5, 1.2) 1.1 (0.8, 1.7) 1.1 (0.7, 1.6) 1.3 (0.8, 1.8) Other variables Time spent with audio-visual media (h/day) b c 1.5 (1.0, 2.2) 1.6 (1.0, 2.2) 1.4 (1.0, 2.0) 1.9 (1.2, 2.5) 1.9 (1.2, 2.8) 1.8 (1.0, 2.4) Number of media in the child’s bedroom a c 0.8 (± 1.2) 0.9 (± 1.3) 0.7 (± 1.2) 1.0 (± 1.4) 1.1 (± 1.4) 1.0 (± 1.4) Fat consumption propensity score a c d 25.5 (± 9.3) 25.4 (± 9.4) 25.5 (± 9.2) not available not available not available Sugar consumption propensity score a c d 25.0 (± 11.3) 25.3 (± 11.5) 24.7 (± 11.2) not available not available not available Highest parental education (ISCED-level) a c 4.0 (± 1.2) 4.0 (± 1.2) 3.9 (± 1.2) 4.0 (± 1.2) 4.0 (± 1.2) 4.0 (± 1.2) N with data on sleep duration 1836 961 875 1450 763 687 Sleep duration on weekdays 10.2 (± 0.9) 10.2 (± 1.0) 10.1 (± 0.9) 9.8 (± 1.0) 9.9 (± 1.0) 9.8 (± 1.0) N with accelerometer data 1967 1029 938 1793 920 873 Time spent in MVPA (average minutes/day) a e 41.2 (± 21.2) 46.0 (± 22.8) 36.0 (± 18.0) 43.4 (± 22.6) 48.5 (± 24.6) 38.0 (± 19.0) aData are presented as mean (± standard deviation), b Data are presented as median (25th, 75th percentile), cVariables with missing data, descriptive statistics based on slightly smaller numbers of children (N ≥3055) dPropensity to consume items high in fat or sugar resp., relative to frequency of all items on food frequency questionnaire eMVPA moderate to vigorous physical activity Table 1 Characteristics of study population aData are presented as mean (± standard deviation), b Data are presented as median (25th, 75th percentile), cVariables with missing data, descriptive statistics based on slightly smaller numbers of children (N ≥3055) dPropensity to consume items high in fat or sugar resp., relative to frequency of all items on food frequency questionnaire eMVPA moderate to vigorous physical activity Longitudinal data were analysed for time varying ex- posure of weight status on IR (Table 3). Statistical analyses Characteristics of the study population at baseline (T0) and during the follow-up survey (T1) are presented in Table 1. Values for all anthropometrical measurements and biological markers of study participants were higher Data of parameters with normal distribution are pre- sented as mean values (± standard deviation); data of pa- rameters with skewed distributions are presented as Peplies et al. Statistical analyses IR was consid- ered as time-varying outcome (delta z IR), which was negative (a lower HOMA-IR value at T1) for children with weight loss between the surveys and highest (with a mean delta z of 0.64) for children with substantial weight gain between T0 and T1. Both, children with a low BMI (thin or normal weight) at both measurements and chil- dren with a high BMI (overweight or obese) in both sur- veys, also showed increased values of HOMA-IR in T1, with a higher increase for the overweight or obese children. at T1. Highest parental ISCED level was about equal between the surveys and among the sexes. Girls and boys exhibited no differences as to age and anthropo- metrical measures. There were small sex differences for the biochemical markers: insulin values were higher in girls (12 % at T0 and 20 % at T1) and glu- cose levels were slightly higher in boys (3 %). Also media consumption, i.e. time spent with audiovisual media and number of media in bedroom, was higher in boys, and time spent in MVPA was 30 % higher in boys. Prevalence rates of IR at follow-up were determined for children in the different BMI categories and are shown in Table 2, stratified by country. IR prevalence clearly showed an increasing trend with BMI, from an overall 2.2 % among thin (underweight) children and 10.9 % in normal weight, via 26.5 % in overweight, reaching a remarkable 66.7 % in obese children. This trend can be seen, both, in countries with a high amount of overweight and obese children like Italy, and in coun- tries with very low rates of overweight and obesity like Sweden. Possible baseline determinants of IR at follow-up were analysed in univariate logistic regression models (Table 4). Crude ORs are shown for all children and for normal weight children only. IR at T1 was positively as- sociated with female sex, increasing age, low SES (max- imum parental ISCED of three or lower), overweight and obesity, waist circumference, sleep duration ≤9 h/ night, media consumption (more than 7 h/week of audio-visual media time) and number of audio-visual media in bedroom (any media). No associations were Peplies et al. Statistical analyses International Journal of Behavioral Nutrition and Physical Activity (2016) 13:97 Page 6 of 12 Table 2 Prevalence of insulin resistance at follow-up by BMI categories according to Cole & Lobstein [30] Country Insulin resistance Thin Normal weight Overweight Obese All Italy HOMA-IR ≥p95a 1 27 44 64 136 8.3 % 13.9 % 35.5 % 68.0 % 32.1 % All 12 (2.8 %) 194 (45.8 %) 124 (29.3 %) 94 (22.2 %) 424 (100.0 %) Estonia HOMA-IR ≥p95a 3 35 18 11 67 7.1 % 11.9 % 40.9 % 84.6 % 17.0 % All 42 (10.7 %) 295 (74.9 %) 44 (11.2 %) 13 (3.3 %) 394 (100.0 %) Belgium HOMA-IR ≥p95a 2 11 11 5 29 4.2 % 3.7 % 42.3 % 55.6 % 7.6 % All 48 (12.5 %) 300 (78.3 %) 26 (6.8 %) 9 (2.4 %) 383 (100.0 %) Sweden HOMA-IR ≥p95a 2 30 19 5 56 3.8 % 7.7 % 36.5 % 62.5 % 11.1 % All 52 (10.3 %) 391 (77.7 %) 52 (10.3 %) 8 (1.6 %) 503 (100.0 %) Germany HOMA-IR ≥p95a 2 27 16 8 53 5.4 % 13.1 % 39.0 % 88.9 % 18.1 % All 37 (12.6 %) 206 (70.3 %) 41 (14.0 %) 9 (3.1 %) 293 (100.0 %) Hungary HOMA-IR ≥p95a 8 89 46 30 173 6.7 % 18.2 % 44.7 % 69.8 % 22.9 % All 120 (15.9 %) 488 (64.7 %) 103 (13.7 %) 43 (5.7 %) 754 (100.0 %) Spain HOMA-IR ≥p95a 0 31 33 17 81 0.0 % 7.5 % 27.1 % 50.0 % 13.6 % All 28 (4.7 %) 413 (69.2 %) 122 (20.4 %) 34 (5.7 %) 597 (100.0 %) All countries HOMA-IR ≥p95a 8 250 187 140 595 2.2 % 10.9 % 36.5 % 66.7 % 17.8 % All 339 (10.1 %) 2287 (68.3 %) 512 (15.3 %) 210 (6.3 %) 3348 (100.0 %) aAge- and sex-specific 95th percentiles (p95) from Peplies et al. [29] markers (skinfolds, fat free mass from BIA, weight-to- height-ratio, waist-to-height-ratio) but there were only little differences in the associations of these markers with HOMA-IR (data not shown) and the strongest as- sociations were seen for BMI and waist circumference. observed for the consumption of sugar or fat. MVPA at baseline (upper two quartiles) showed a protective effect on the development of IR 2 years later. bdelta z-IR = z-score IR (T1) - z-score IR(T0) Statistical analyses This remained unchanged when the analysis was limited to 5–8 year old children, the age range that the applied cut-off- points by Evenson [41] were calibrated for (data not shown). When only children with normal weight at base- line were considered, an association with IR at follow-up was still evident for sex, low SES, media consumption and audio-visual media in bedroom and MVPA (only 3rd quartile). We also looked at different anthropometric Results from multiple logistic regression analysis are presented in Table 5. Sex (being female), BMI (z-score) and waist circumference (z-score) at baseline were the strongest predictors of HOMA-IR at T1. MVPA showed the same pattern of association as in the univariate ana- lysis: results were strongest for the 3rd quartile. A small Table 3 Two-year change of HOMA-IR by changes of weight status BMIa at baseline BMIa at follow-up N Two-year change of HOMA-IR delta z-IRb (Stdc) Overweight/obese Thin/normal weight 75 −0.19 (1.1) Thin/normal weight Thin/normal weight 2539 0.20 (1.2) Overweight/obese Overweight/obese 513 0.46 (1.1) Thin/normal weight Overweight/obese 221 0.64 (1.2) All 3348 0.26 (1.2) aweight status according to extended IOTF criteria (Cole, 2012) bdelta z-IR = z-score IR (T1) - z-score IR(T0) cstandard deviation Table 3 Two-year change of HOMA-IR by changes of weight status BMIa at baseline BMIa at follow-up N Peplies et al. International Journal of Behavioral Nutrition and Physical Activity (2016) 13:97 Page 7 of 12 Table 4 Determinants of IR (HOMA-IR ≥p95a) from univariate logistic regression analyses (children without IR at T0) Children of all weight groups Normal weight children only N % OR 95 % CI N % OR 95 % CI Sex 3152 (all) 2297 (all) Female 1508 47.8 1.5 1.3–1.9 1078 46.9 1.4 1.1–1.9 Male 1644 52.2 Ref. 1219 53.1 Ref. Age (continuous, unit change from mean) 3152 (all) 1.2 1.1–1.3 2297 (all) 1.1 1.0–1.2 ISCED 3114 (all) 2269 (all) Low (1–2) 215 6.9 2.6 1.9–3.6 135 6.0 2.2 1.4–3.5 Medium (3–4) 1563 50.2 1.8 1.4–2.3 1121 49.4 1.6 1.2–2.2 High (5–6) 1336 52.9 Ref. 1013 44.7 Ref. BMI (Cole) 3152 (all) Thin 362 11.5 0.5 0.3–0.9 Normal weight 2297 72.9 Ref. Overweight 345 11.0 4.1 3.3–5.0 Obese 148 4.7 10.8 8.2–14.0 BMI z-score (Cole) 3152 (all) 1st Quartile (≤−0.49) 813 26.2 0.7 0.5–1.1 2nd Quartile (−0.49 - ≤0.19) 802 25.6 Ref. Statistical analyses 2nd Quartile (27 - ≤38.7 min./day) 256 24.6 0.8 0.5–1.2 193 25.0 0.9 0.5–1.4 3rd Quartile (38.7 - ≤54.6 min./day) 264 25.3 0.4 0.2–0.7 193 25.0 0.4 0.2–0.9 4th Quartile (≥54.6 min./day) 263 25.2 0.6 0.4–1.0 182 23.6 0.7 0.3–1.0 Figures in bold indicate a significant effect. Confidence bounds of 1.0 are caused by rounding aAge- and sex-specific 95th percentiles (p95) from Peplies et al. [29] bMVPA moderate to vigorous physical activity cPropensity to consume items high in sugar or fat resp relative to frequency of all items on food frequency questionnaire ts of IR (HOMA-IR ≥p95a) from univariate logistic regression analyses (children without IR at T0) (Continued) Table 4 Determinants of IR (HOMA-IR ≥p95a) from univariate logistic regression analyses (children without IR Figures in bold indicate a significant effect. Confidence bounds of 1.0 are caused by rounding aAge- and sex-specific 95th percentiles (p95) from Peplies et al. [29] Table 5 Determinants of IR (HOMA-IR ≥p95a) - results from multivariate mixed logistic regression models N = 1083 Model 1b (with BMI z-score) Model 2b (with waist z-score) OR 95 % CI OR 95 % CI BMI z-score (Cole) (unit change from mean) 2.6 2.1–3.1 Waist z-score (Cole) (unit change from mean) 2.2 1.9–2.6 Time spent in MVPAc 1st Quartile (≤27 min./day) Ref. Ref. 2nd Quartile (27 - ≤38.7 min./day) 0.9 0.6–1.5 1.1 0.7–1.7 3rd Quartile (38.7 - ≤54.6 min./day) 0.5 0.3–0.9 0.5 0.3–0.9 4th Quartile (≥54.6 min./day) 0.7 0.5–1.1 0.7 0.5–1.1 Sex (female versus male) 2.2 1.5–3.1 2.5 1.8–3.6 Age (unit change from mean) 1.0 0.9–1.2 1.0 0.9–1.1 ISCED Low (1–2) 1.2 0.7–2.2 1.3 0.7–2.3 Medium (3–4) 1.2 0.9–1.8 1.2 0.8–1.8 High (5–6) Ref. Ref. Ref. Ref. Audio-visual media time (h/d) (unit change from mean) 1.2 1.0–1.4 1.2 1.0–1.4 Fat consumption propensity scored (unit change from mean) 1.2 1.0–1.4 1.2 1.0–1.4 Figures in bold indicate a significant effect. Confidence bounds of 1.0 are caused by rounding aAge- and sex-specific 95th percentiles (p95) from Peplies et al. [29] bAnalyses were adjusted for all parameters in the respective column, including country as a random effect to account for the clustered study design cMVPA moderate to vigorous physical activity dPropensity to consume items high in or fat, relative to frequency of all items on food frequency questionnaire. Statistical analyses 3rd Quartile (0.19 - ≤0.97) 811 23.1 1.2 0.9–1.8 4th Quartile (≥0.97) 726 25.1 4.9 3.7–6.6 BMI z-score (Cole) (unit change from mean) 3152 (all) 2.5 2.2–2.8 Waist z-score (Cole) 3152 (all) 1st Quartile (≤−0.66) 826 26.2 0.8 0.5–1.2 2nd Quartile (−0.66 - ≤0.09) 791 25.1 Ref. 3rd Quartile (0.09 - ≤0.95) 808 25.6 1.5 1.0–2.1 4th Quartile (≥0.95) 727 23.1 5.7 4.1–7.7 Waist z-score (Cole) (unit change from mean) 3152 (all) 2.2 2.0–2.4 Audio-visual media time 3152 (all) 2297 (all) ≤1 h/day 1062 33.7 Ref. 774 33.7 Ref. 1 - ≤2 h/day 1221 38.7 1.4 1.1–1.9 899 39.1 1.4 1.0–2.1 2 - ≤3 h/day 603 19.1 1.7 1.3–2.3 439 19.1 1.8 1.2–2.6 >3 h/day 266 8.4 2.1 1.5–2.9 185 8.1 2.3 1.5–3.6 Media in bedroom 3090 (all) 2250 (all) 0 media 1902 61.5 Ref. 1429 63.5 Ref. 1–2 media 868 28.1 1.8 1.4–2.2 609 27.1 1.8 1.3–2.3 3 media 320 10.4 2.6 2.0–3.4 212 9.4 2.4 1.7–3.4 Sugar consumption propensity score (N = 3125)c 3152 (all) 2297 (all) 1st Quartile (≤16.9) 842 26.7 Ref. 625 27.2 Ref. 2nd Quartile (16.9 - ≤24.1) 832 26.4 1.0 0.8–1.4 557 24.3 1.0 0.7–1.5 3rd Quartile (24.1 - ≤32.4) 766 24.3 1.1 0.8–1.4 562 24.5 1.0 0.7–1.5 4th Quartile (≥32.4) 713 22.6 1.2 0.9–1.6 553 24.1 1.4 1.0–1.9 3rd Quartile (24.1 - ≤32.4) 4th Quartile (≥32.4) Peplies et al. International Journal of Behavioral Nutrition and Physical Activity (2016) 13:97 Page 8 of 12 Table 4 Determinants of IR (HOMA-IR ≥p95a) from univariate logistic regression analyses (children without IR at T0) (Continued) Fat consumption propensity score (N = 3125) c 3152 (all) 2297 (all) 1st Quartile (≤19.0) 901 28.6 Ref. 595 25.9 Ref. 2nd Quartile (19.0 - ≤24.8) 790 25.1 0.8 0.6–1.0 543 23.6 1.1 0.8–1.6 3rd Quartile (24.8 - ≤31.3) 748 23.7 0.8 0.6–1.0 574 25.0 1.1 0.8–1.6 4th Quartile (≥31.3) 713 22.6 0.9 0.7–1.2 585 25.5 1.1 0.8–1.6 Sleep duration on weekdays (N = 1730) 1730 (all) 1253 (all) Average sleep time <9 h 104 6.0 1.8 1.2–2.7 58 4.6 1.5 0.8–2.8 Average sleep time ≥9 h 1626 94.0 Ref. 1195 95.4 Ref. Time spent in MVPAb 1042 (all) 771 (all) 1st Quartile (≤27 min./day) 259 24.9 Ref. 203 26.3 Ref. bMVPA moderate to vigorous physical activity cPropensity to consume items high in sugar or fat resp., relative to frequency of all items on food frequency questionnaire Discussion This study shows a strong prospective association be- tween weight status and HOMA-IR in preadolescent European children and a protective effect of MVPA. It indicates that a longitudinal reduction of BMI also leads to a decline in HOMA-IR and thus a favourable change in metabolic status. The study describes country-specific prevalence rates of IR in the IDEFICS cohort, showing an increasing trend of IR prevalence with BMI category. g g y Prevalence proportions of IR found in our study (10.9 % in normal weight, 26.5 % in overweight, 66.7 % in obese children) were in agreement with those re- ported in the literature, even though comparability is limited, because previous studies were based on older children and used different definitions of IR. A Chilean cohort of 10–15 year old children displayed prevalence values of 13 % in normal weight, 37.1 % in overweight, and 61.6 % in obese children, using HOMA-IR ≥p90 as cut point for IR [21]. An US-American survey on 12–19 year old adolescents [8] used a cut point of HOMA-IR ≥p97.5, which resulted in prevalence values of about 4 % in normal weight, 16 % in overweight and 52 % in obese US-adolescents aged 12–19 years (data taken from fig- ure). As it was previously shown that insulin levels and thus also HOMA-IR levels show a peak round the age of 13 to 14 years [44], slightly lower prevalence proportions for older adolescents are plausible. When longitudinal change of HOMA-IR (delta z-IR) was linked to overweight and obesity, children who retained their low or normal weight until the follow-up survey showed a moderate increase of their IR z-score, while children who were overweight or obese in both surveys showed a stronger increase of their IR z-score. In both cases, the mean BMI increased within the de- fined weight groups. In the small group of children (N = 75) who had changed to a lower weight group at follow- up, the delta z-IR values also declined. The highest gain in IR was seen for children who changed to a higher weight group. In our study, weight status and waist circumference at baseline appeared as the main risk factors for IR at follow-up, but sex and lifestyle indices (objectively deter- mined PA, fat consumption score, audio-visual media time, and media in bedroom) were also associated with incidence of IR. Statistical analyses The score was included into the model in a modified form: it was divided by 10 to obtain meaningful effect estimates – one unit in the multivariate model thus represents 10 units of the original score used in the univariate model Peplies et al. International Journal of Behavioral Nutrition and Physical Activity (2016) 13:97 Peplies et al. International Journal of Behavioral Nutrition and Physical Activity (2016) 13:97 Peplies et al. International Journal of Behavioral Nutrition and Physical Activity (2016) 13:97 Page 9 of 12 The associations observed in this study confirm previ- ous findings, which however are mainly based on studies in older children/ adolescents and on cross-sectional data. A Spanish cross-sectional study on cardiovascular risk factors in 6 to 8 year-old schoolchildren found some of the same metabolic consequences of obesity as in adults (elevated triglycerides, insulin, HOMA-IR, and lower HDL-cholesterol) [46]. In a representative sub- sample of diabetes-free US-adolescents aged 12–19 years who participated in the National Health and Nutrition Examination Survey (NHANES), obesity was by far the most important determinant of IR, independent of sex, age, or race/ethnicity, but data on PA was not included in the analysis [8]. Subcutaneous adiposity was also the most significant covariate for HOMA-IR in a family- based US-American study, including children and ado- lescents from 6 years on [9]. A large insulin screening study conducted on 10–14 year old children in a meth- odological context, using a blood spot assay on filter paper, indicated associations of pubertal stage and mea- sures of central and peripheral adiposity with insulin level [10]. In the IDEFICS baseline survey, a low amount of physical activity was shown to be associated with a cluster of CVD risk factors including IR [13]. In accord- ance with the evidence from our analysis, a few other studies published data on the association of dietary fac- tors like intake of total energy, total fat or saturated fat with IR in children and adolescents [11, 12]. but significant increase of risk was also seen for audio- visual media time and fat consumption score. ISCED, age and country showed no influence on HOMA-IR in the multivariate model. Discussion The age effect which was seen in the univariate analysis can probably be attributed in large part to the increase of overweight and obesity with age. The higher risk of IR in girls compared to boys also per- sisted when older children (>7 years at baseline) were excluded from the analysis to avoid possible influences of early puberty. Fat consumption, as expressed by a propensity score [38, 39], was also connected with IR risk in the multivariate model. The PA level of children in our study was rather low, and only a small proportion of children (24.2 % of the boys and 9.7 % of the girls) reached the daily activity level recommended by the World Health Organization (≥60 min/ day) [45]. Con- sidering this and the fact that accelerometer data were only available for about one third of the study popula- tion, the protective effect on IR found for MVPA is cer- tainly noteworthy, despite the lack of a clear trend. There is a solid body of evidence for the association of obesity and physical inactivity with IR, especially in adults, but there is a controversy on whether the influ- ence of sedentary lifestyle on IR is mediated by obesity, whether both are independent predictors of this condi- tion, or whether the risk for IR involved with obesity is modified by PA. In our study, the association of MVPA with IR is attenuated only slightly when only children with normal weight at baseline are considered, i.e. MVPA reduces the risk of developing IR, also for chil- dren with normal weight at baseline which indicates that the effect of missing physical activity is not just mediated by obesity. This is also confirmed by the results of the multivariate model. Page 10 of 12 Page 10 of 12 Peplies et al. International Journal of Behavioral Nutrition and Physical Activity (2016) 13:97 Other studies have also shown that increased PA can re- duce the risk of cardiovascular disease and metabolic risk factors including insulin, already in children [47–49]. Obesity might thus be the main determinant of IR, but its effect can be attenuated by a sufficient amount of PA. An US-American cross-sectional study [50] included 8–17 year olds with a BMI ≥85th percentile who were enrolled in a multidisciplinary paediatric weight management clinic. MVPA was the strongest independent predictor of metabolic health in these overweight and obese children. Conclusions This study is, to our knowledge, the first study to show longitudinal data on IR in a preadolescent children’s population. It supports the available evidence, that over- weight and obesity are the main determinants of IR, while PA seems to ameliorate the risk, independent of weight status. Reduction of weight is thus an important measure in the fight against IR in children, but children should above all be encouraged to engage in regular PA, as this will keep them metabolically healthy even under the presence of overweight/obesity. In an analysis on NHANES data from obese adoles- cents and adults [52], self-reported PA was associated with a metabolically healthy phenotype in adults, but not in adolescents. A large cross-sectional study among Spanish adults [23] provided evidence that PA is one of the main factors responsible for a healthy phenotype among the obese. Only few studies published results for adolescents and these showed no association between PA and metabolically healthy obesity [22, 52]. Further re- search in children is needed, as there are no studies on metabolically healthy obesity in children. A cross-sectional analysis of IDEFICS baseline data showed the importance of PA to protect against clustering of CVD risk factors [13]. In preschool and even more so in school children (≤9 years), CVD risk was elevated for children in the lower quintiles of MVPA. Availability of data and materials Due to ethical restrictions and participant confidentiality, data cannot be made publicly available. Competing interests Th h d l h The authors declare that they have no competing interests. Discussion A review [51] on the therapeutic power of PA in children suggests that PA may have greater influence on body com- position and cardiovascular risk factors than dieting as it possibly modulates the fuel metabolism. The increased fat oxidation by PA might be the basis for prevention and res- toration of insulin sensitivity and reduction of MS in obese children. portion sizes and were thus calculated from average quantities which might reduce the truly existing differ- ences. Presented prevalence rates only apply for the study population at hand. These should neither be trans- ferred to the respective countries in general nor to other study populations. Acknowledgement Th k d g This work was done as part of the IDEFICS Study (http:// www.ideficsstudy.eu). The authors also wish to thank the IDEFICS children and their parents for their participation. p y p www.ideficsstudy.eu). The authors also wish to thank the IDEFICS children and their parents for their participation. Funding Funding We gratefully acknowledge the financial support of the European Community within the Sixth RTD Framework Programme Contract No. 016181 (FOOD). The European Community had no influence on the design of the study, on collection, analysis, and interpretation of data nor on writing the manuscript. Authors’ contributions JP conceived the study, conducted the analyses, interpreted the data and drafted the paper with advice from CB, WA and KG. WA; PR, TV, MT, SDH, SM, DM and LAM, were responsible for data collection, AF was in charge of biological sample analyses. All authors contributed to revisions of the paper. All authors read and approved the final manuscript. The main strengths of this study are the large study size, its longitudinal design, the highly standardised data collection across different European countries, and the young age of the examined children, as well as the fact that PA was measured objectively by activity monitors. There are on the other hand also some limitations that should be mentioned: Pubertal stage was not assessed in the IDEFICS surveys and it can be suspected that a con- siderable proportion of the older children might have already started into puberty at follow-up. During the German Health Interview and Examination Survey for Children and Adolescents (KiGGS) [53], at an age of 10 years, 42.4 % of girls and 35.7 % of boys reported the development of pubic hair. Early puberty is thus more likely in girls and especially in girls, insulin concentra- tions have been shown to have a distinct pubertal peak [44]. Nevertheless, as mentioned above, the elevated IR risk seen in girls remained unchanged when older chil- dren (>7 years) where excluded. Food consumption scores used in this study were based on a food frequency questionnaire which neither included school meals nor References 1. Levy-Marchal C, Arslanian S, Cutfield W, et al. Insulin resistance in children: consensus, perspective, and future directions. J Clin Endocrinol Metab. 2010;95(12):5189–98. doi:10.1210/jc.2010-1047. 1. Levy-Marchal C, Arslanian S, Cutfield W, et al. Insulin resistance in children: consensus, perspective, and future directions. J Clin Endocrinol Metab. 2010;95(12):5189–98. doi:10.1210/jc.2010-1047. 24. Duncan GE. Exercise, fitness, and cardiovascular disease risk in type 2 diabetes and the metabolic syndrome. Curr Diab Rep. 2006;6(1):29–35. 25. Armstrong N. Aerobic fitness and physical activity in children. Pediatr Exerc Sci. 2013;25(4):548–60. 2. Weiss R, Kaufman FR. Metabolic complications of childhood obesity: identifying and mitigating the risk. Diabetes Care. 2008;31 Suppl 2:S310–316. doi:10.2337/dc08-s273. 2. Weiss R, Kaufman FR. Metabolic complications of childhood obesity: identifying and mitigating the risk. Diabetes Care. 2008;31 Suppl 2:S310–316. doi:10.2337/dc08-s273. 26. Lee I-M, Shiroma EJ, Lobelo F, et al. Effect of physical inactivity on major non-communicable diseases worldwide: an analysis of burden of disease and life expectancy. Lancet Lond Engl. 2012;380(9838):219–29. doi:10.1016/ S0140-6736(12)61031-9. 3. Martin BC, Warram JH, Krolewski AS, Bergman RN, Soeldner JS, Kahn CR. Role of glucose and insulin resistance in development of type 2 diabetes mellitus: results of a 25-year follow-up study. Lancet. 1992;340(8825):925–9. 4 International Obesity Task Force Obesity in Europe the Case for Action 2002 27. Barry VW, Baruth M, Beets MW, Durstine JL, Liu J, Blair SN. Fitness vs. fatness on all-cause mortality: a meta-analysis. Prog Cardiovasc Dis. 2014;56(4):382– 90. doi:10.1016/j.pcad.2013.09.002. International Obesity Task Force. Obesity in Europe, the Case for Action. 5. Cruz ML, Shaibi GQ, Weigensberg MJ, Spruijt-Metz D, Ball GDC, Goran MI. Pediatric obesity and insulin resistance: chronic disease risk and implications for treatment and prevention beyond body weight modification. Annu Rev Nutr. 2005;25:435–68. doi:10.1146/annurev.nutr.25.050304.092625. 28. Ahrens W, Bammann K, Siani A, et al. The IDEFICS cohort: design, characteristics and participation in the baseline survey. Int J Obes 2005. 2011;35 Suppl 1:S3–15. doi:10.1038/ijo.2011.30. 6. Berenson GS, Srinivasan SR, Bao W, Newman WP, Tracy RE, Wattigney WA. Association between multiple cardiovascular risk factors and atherosclerosis in children and young adults. The Bogalusa Heart Study. N Engl J Med. 1998;338(23):1650–6. doi:10.1056/NEJM199806043382302. 29. Peplies J, Jiménez-Pavón D, Savva SC, et al. Percentiles of fasting serum insulin, glucose, HbA1c and HOMA-IR in pre-pubertal normal weight European children from the IDEFICS cohort. Int J Obes 2005. 2014;38 Suppl 2:S39–47. doi:10.1038/ijo.2014.134. 7. Maclaren NK, Gujral S, Ten S, Motagheti R. Childhood obesity and insulin resistance. Cell Biochem Biophys. 2007;48(2-3):73–8. 30. Ethics approval and consent to participate We certify that all applicable institutional and governmental regulations concerning the ethical use of human volunteers were followed during this research, and that the IDEFICS project passed the Ethics Review process of the Sixth Framework Programme (FP6) of the European Commission. Ethical approval was obtained from the relevant local or national ethics committees by each of the 8 study centers, namely from the Ethics Committee of the University Hospital Ghent (Belgium), the National Bioethics Committee of Cyprus (Cyprus), the Tallinn Medical Research Ethics Committee of the National Institutes for Health Development (Estonia), the Ethics Committee of the University Bremen (Germany), the Scientific and Research Ethics Committee of the Medical Research Council Budapest (Hungary), the Ethics Committee of the Health Office Avellino (Italy), the Ethics Committee for Clinical Research of Aragon (Spain), and the Regional Ethical Review Board of Gothenburg (Sweden). All parents or legal guardians of the participating children gave written informed consent to data collection, examinations, collection of samples, subsequent analysis, and storage of personal data and collected samples. Additionally, each child gave oral consent after being orally informed about the modules by a study nurse immediately before every examination using a simplified text. This procedure was chosen due to the young age of the children. The oral consenting process was not further Page 11 of 12 Page 11 of 12 Peplies et al. International Journal of Behavioral Nutrition and Physical Activity (2016) 13:97 Peplies et al. International Journal of Behavioral Nutrition and Physical Activity (2016) 13:97 documented, but it was subject to central and local training and quality control procedures of the study. Study participants and their parents / legal guardians could consent to single components of the study while abstaining from others. All procedures were approved by the above-mentioned Ethics Committees. documented, but it was subject to central and local training and quality control procedures of the study. Study participants and their parents / legal guardians could consent to single components of the study while abstaining from others. All procedures were approved by the above-mentioned Ethics Committees. 15. Guerrero-Romero F, Aradillas-García C, Simental-Mendia LE, Monreal- Escalante E, de la Cruz Mendoza E, Rodríguez-Moran M. Birth weight, family history of diabetes, and metabolic syndrome in children and adolescents. J Pediatr. 2010;156(5):719–23. doi:10.1016/j.jpeds.2009.11.043. 723.e1. ediatr. 2010;156(5):719–23. doi:10.1016/j.jpeds.2009.11.043. 723.e1. 16. Boney CM, Verma A, Tucker R, Vohr BR. References Cole TJ, Lobstein T. Extended international (IOTF) body mass index cut-offs for thinness, overweight and obesity. Pediatr Obes. 2012;7(4):284–94. doi:10.1111/j.2047-6310.2012.00064.x. resistance. Cell Biochem Biophys. 2007;48(2-3):73–8. 8. Lee JM, Okumura MJ, Davis MM, Herman WH, Gurney JG. Prevalence and determinants of insulin resistance among U.S. adolescents: a population- based study. Diabetes Care. 2006;29(11):2427–32. doi:10.2337/dc06-0709. 31. Hense S, Barba G, Pohlabeln H, et al. Factors that influence weekday sleep duration in European children. Sleep. 2011;34(5):633–9. 9. Ali O, Cerjak D, Kent JW, James R, Blangero J, Zhang Y. Obesity, central adiposity and cardiometabolic risk factors in children and adolescents: a family- based study. Pediatr Obes. 2014;9(3):e58–62. doi:10.1111/j.2047-6310.2014.218.x. 32. Börnhorst C, Hense S, Ahrens W, et al. From sleep duration to childhood obesity–what are the pathways? Eur J Pediatr. 2012;171(7):1029–38. doi:10.1007/s00431-011-1670-8. 10. Martin RM, Patel R, Zinovik A, et al. Filter paper blood spot enzyme linked immunoassay for insulin and application in the evaluation of determinants of child insulin resistance. PLoS One. 2012;7(10):e46752. doi:10.1371/journal. pone.0046752. 33. Singh B, Saxena A. Surrogate markers of insulin resistance: a review. World J Diabetes. 2010;1(2):36–47. doi:10.4239/wjd.v1.i2.36. 34. Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, Turner RC. Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia. 1985;28(7):412–9. 11. Zimmermann MB, Aeberli I. Dietary determinants of subclinical inflammation, dyslipidemia and components of the metabolic syndrome in overweight children: a review. Int J Obes 2005. 2008;32 Suppl 6:S11–18. doi: 10.1038/ijo.2008.202. 35. Gungor N, Saad R, Janosky J, Arslanian S. Validation of surrogate estimates of insulin sensitivity and insulin secretion in children and adolescents. J Pediatr. 2004;144(1):47–55. doi:10.1016/j.jpeds.2003.09.045. 12. Aeberli I, Spinas GA, Lehmann R, l’Allemand D, Molinari L, Zimmermann MB. Diet determines features of the metabolic syndrome in 6- to 14-year-old children. Int J Vitam Nutr Res. 2009;79(1):14–23. doi:10.1024/0300-9831.79.1.14. 36. Peplies J, Fraterman A, Scott R, Russo P, Bammann K. Quality management for the collection of biological samples in multicentre studies. Eur J Epidemiol. 2010;25(9):607–17. doi:10.1007/s10654-010-9481-1. 13. Jiménez-Pavón D, Konstabel K, Bergman P, et al. Physical activity and clustered cardiovascular disease risk factors in young children: a cross- sectional study (the IDEFICS study). BMC Med. 2013;11:172. doi:10.1186/ 1741-7015-11-172. 37. UNESCO Institute for Statistics. International Standard Classification for Education ISCED 1997. 2006. 14. Leite SA, Monk AM, Upham PA, Chacra AR, Bergenstal RM. Low cardiorespiratory fitness in people at risk for type 2 diabetes: early marker for insulin resistance. Author details 1 1Leibniz Institute for Prevention Research and Epidemiology – BIPS, Bremen, Germany. 2Faculty of Human and Health Sciences, Institute for Public Health and Nursing Research, Bremen University, Bremen, Germany. 3MVZ Dortmund Dr. Eberhard und Partner, Dortmund, Germany. 4Epidemiology & Population Genetics, Institute of Food Sciences, CNR, Avellino, Italy. 5National Institute for Health Development, Tallinn, Estonia. 6Research & Education Institute of Child Health, Strovolos, Cyprus. 7Department of Public Health, Ghent University, Ghent, Belgium. 8Department of Public Health and Community Medicine, The Queen Silvia Children’s Hospital, Göteborg, Sweden. 9National Institute of Health Promotion, University of Pécs, Gyermekklinika, Pécs, Hungary. 10GENUD (Growth, Exercise, Nutrition and Development) Research Group, School of Health Sciences, University of Zaragoza, Zaragoza, Spain. 11Epidemiological Methods and Etiologic Research, Leibniz Institute for Prevention Research and Epidemiology – BIPS, Achterstraße 30, 28359 Bremen, Germany. 17. Bruce KD, Hanson MA. The developmental origins, mechanisms, and implications of metabolic syndrome. J Nutr. 2010;140(3):648–52. doi:10.3945/jn.109.111179. 18. Leech RM, McNaughton SA, Timperio A. The clustering of diet, physical activity and sedentary behavior in children and adolescents: a review. Int J Behav Nutr Phys Act. 2014;11:4. doi:10.1186/1479-5868-11-4. 19. Friend A, Craig L, Turner S. The prevalence of metabolic syndrome in children: a systematic review of the literature. Metab Syndr Relat Disord. 2013;11(2):71–80. doi:10.1089/met.2012.0122. 20. Ahrens W, Moreno LA, Mårild S, et al. Metabolic syndrome in young children: definitions and results of the IDEFICS study. Int J Obes 2005. 2014; 38 Suppl 2:S4–14. doi:10.1038/ijo.2014.130. 21. Mardones F, Arnaiz P, Barja S, et al. Nutritional status, metabolic syndrome and insulin resistance in children from Santiago (Chile). Nutr Hosp. 2013; 28(6):1999–2005. doi:10.3305/nutrhosp.v28in06.6770. Research, Leibniz Institute for Prevention Research and Epidemiology – BIPS, Achterstraße 30, 28359 Bremen, Germany. 22. Sénéchal M, Wicklow B, Wittmeier K, et al. Cardiorespiratory fitness and adiposity in metabolically healthy overweight and obese youth. Pediatrics. 2013;132(1):e85–92. doi:10.1542/peds.2013-0296. Received: 25 March 2016 Accepted: 24 August 2016 Received: 25 March 2016 Accepted: 24 August 2016 23. Lopez-Garcia E, Guallar-Castillon P, Leon-Muñoz L, Rodriguez-Artalejo F. Prevalence and determinants of metabolically healthy obesity in Spain. Atherosclerosis. 2013;231(1):152–7. doi:10.1016/j.atherosclerosis.2013.09.003 Ethics approval and consent to participate Metabolic syndrome in childhood: association with birth weight, maternal obesity, and gestational diabetes mellitus. Pediatrics. 2005;115(3):e290–296. doi:10.1542/peds.2004-1808. References Diabetol Metab Syndr. 2009;1:8. doi:10.1186/1758-5996-1-8. 38. Lanfer A, Knof K, Barba G, et al. Taste preferences in association with dietary habits and weight status in European children: results from the IDEFICS study. Int J Obes 2005. 2012;36(1):27–34. doi:10.1038/ijo.2011.164. Page 12 of 12 Peplies et al. International Journal of Behavioral Nutrition and Physical Activity (2016) 13:97 39. Lissner L, Lanfer A, Gwozdz W, et al. Television habits in relation to overweight, diet and taste preferences in European children: the IDEFICS study. Eur J Epidemiol. 2012;27(9):705–15. doi:10.1007/s10654-012-9718-2. 40. Lanfer A, Hebestreit A, Ahrens W, et al. Reproducibility of food consumption frequencies derived from the Children’s Eating Habits Questionnaire used in the IDEFICS study. Int J Obes 2005. 2011;35 Suppl 1:S61–68. doi:10.1038/ijo.2011.36. 41. Evenson KR, Catellier DJ, Gill K, Ondrak KS, McMurray RG. Calibration of two objective measures of physical activity for children. J Sports Sci. 2008;26(14): 1557–65. doi:10.1080/02640410802334196. 42. Konstabel K, Veidebaum T, Verbestel V, et al. Objectively measured physical activity in European children: the IDEFICS study. Int J Obes 2005. 2014;38 Suppl 2:S135–143. doi:10.1038/ijo.2014.144. 42. Konstabel K, Veidebaum T, Verbestel V, et al. Objectively measured physical activity in European children: the IDEFICS study. Int J Obes 2005. 2014;38 Suppl 2:S135–143. doi:10.1038/ijo.2014.144. j 43. Suling M, Hebestreit A, Peplies J, et al. Design and results of the pretest of the IDEFICS study. Int J Obes 2005. 2011;35 Suppl 1:S30–44. doi:10.1038/ijo.2011.33. 43. Suling M, Hebestreit A, Peplies J, et al. Design and results of the pretest of the IDEFICS study. Int J Obes 2005. 2011;35 Suppl 1:S30–44. doi:10.1038/ijo.2011.33. 44. Mellerio H, Alberti C, Druet C, et al. Novel modeling of reference values of cardiovascular risk factors in children aged 7 to 20 years. Pediatrics. 2012; 129(4):e1020–1029. doi:10.1542/peds.2011-0449. 44. Mellerio H, Alberti C, Druet C, et al. Novel modeling of reference values of cardiovascular risk factors in children aged 7 to 20 years. Pediatrics. 2012; 129(4):e1020–1029. doi:10.1542/peds.2011-0449. 45. World Health Organization. WHO: Global Recommendations on Physical Activity for Health. 2010. 46. Garcés C, Gutierrez-Guisado J, Benavente M, et al. Obesity in Spanish schoolchildren: relationship with lipid profile and insulin resistance. Obes Res. 2005;13(6):959–63. doi:10.1038/oby.2005.111. 46. Garcés C, Gutierrez-Guisado J, Benavente M, et al. Obesity in Spanish schoolchildren: relationship with lipid profile and insulin resistance. Obes Res. 2005;13(6):959–63. doi:10.1038/oby.2005.111. 47. Ekelund U, Luan J’a, Sherar LB, et al. Moderate to vigorous physical activity and sedentary time and cardiometabolic risk factors in children and adolescents. JAMA. References 2012;307(7):704–12. doi:10.1001/jama.2012.156. 47. Ekelund U, Luan J’a, Sherar LB, et al. Moderate to vigorous physical activity and sedentary time and cardiometabolic risk factors in children and adolescents. JAMA. 2012;307(7):704–12. doi:10.1001/jama.2012.156. 48. Väistö J, Eloranta A-M, Viitasalo A, et al. Physical activity and sedentary behaviour in relation to cardiometabolic risk in children: cross-sectional findings from the Physical Activity and Nutrition in Children (PANIC) Study. Int J Behav Nutr Phys Act. 2014;11:55. doi:10.1186/1479-5868-11-55. 49. Van Buren DJ, Tibbs TL. Lifestyle interventions to reduce diabetes and cardiovascular disease risk among children. Curr Diab Rep. 2014;14(12):557. doi:10.1007/s11892-014-0557-2. 50. Prince RL, Kuk JL, Ambler KA, Dhaliwal J, Ball GDC. Predictors of metabolically healthy obesity in children. Diabetes Care. 2014;37(5):1462–8. doi:10.2337/dc13-1697. 51. Brambilla P, Pozzobon G, Pietrobelli A. Physical activity as the main therapeutic tool for metabolic syndrome in childhood. Int J Obes 2005. 2011;35(1):16–28. doi:10.1038/ijo.2010.255. 52. Camhi SM, Waring ME, Sisson SB, Hayman LL, Must A. Physical activity and screen time in metabolically healthy obese phenotypes in adolescents and adults. J Obes. 2013;2013:984613. doi:10.1155/2013/984613. 53. Kahl H, Schaffrath Rosario A, Schlaud M. Sexual maturation of children and adolescents in Germany. Results of the German Health Interview and Examination Survey for Children and Adolescents (KiGGS). Bundesgesundheitsblatt Gesundheitsforschung Gesundheitsschutz. 2007; 50(5-6):677–85. doi:10.1007/s00103-007-0229-3. Submit your next manuscript to BioMed Central and we will help you at every step: • We accept pre-submission inquiries • Our selector tool helps you to ﬁnd the most relevant journal • We provide round the clock customer support • Convenient online submission • Thorough peer review • Inclusion in PubMed and all major indexing services • Maximum visibility for your research Submit your manuscript at www.biomedcentral.com/submit Submit your next manuscript to BioMed Central and we will help you at every step: • We accept pre-submission inquiries
https://openalex.org/W2572907923	https://europepmc.org/articles/pmc5243910?pdf=render	English	null	An efficient multi-stage fermentation strategy for the production of microbial oil rich in arachidonic acid in Mortierella alpina	Bioresources and bioprocessing	2,017	cc-by	5,557	© The Author(s) 2017. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. Correspondence: xiaojunji@njtech.edu.cn 4 Jiangsu National Synergetic Innovation Center for Advanced Materials (SICAM), No. 5 Xinmofan Road, Nanjing 210009, People’s Republic of China Full list of author information is available at the end of the article Wu et al. Bioresour. Bioprocess. (2017) 4:8 DOI 10.1186/s40643-017-0138-8 Wu et al. Bioresour. Bioprocess. (2017) 4:8 DOI 10.1186/s40643-017-0138-8 Open Access An efficient multi‑stage fermentation strategy for the production of microbial oil rich in arachidonic acid in Mortierella alpina Jia Wu1, Ai‑Hui Zhang1, Chao Peng5, Lu‑Jing Ren1,4, Ping Song1, Ya‑Dong Yu4, He Huang2,3,4 X J J 1 4 Abstract Background: Fungal morphology and aeration play a significant role in the growth process of Mortierella alpina. The production of microbial oil rich in arachidonic acid (ARA) in M. alpina was enhanced by using a multi-stage fermenta‑ tion strategy which combined fed-batch culture with precise control of aeration and agitation rates at proper times. Results: The fermentation period was divided into four stages according to the cultivation characteristics of M. alpina. The dissolved oxygen concentration was well suited for ARA biosynthesis. Moreover, the ultimate dry cell weight (DCW), lipid, and ARA yields obtained using this strategy reached 41.4, 22.2, 13.5 g/L, respectively. The respec‑ tive values represent 14.8, 25.8, and 7.8% improvements over traditional fed-batch fermentation processes. Conclusions: This strategy provides promising control insights for the mass production of ARA-rich oil on an indus‑ trial scale. Pellet-like fungal morphology was transformed into rice-shaped particles which were beneficial for oxygen transfer and thus highly suitable for biomass accumulation. rds: Arachidonic acid, Mortierella alpina, Multi-stage fermentation, Aeration, Agitation, Morphology use of ARA in infant formula. On the other hand, ARA- rich oil derived from the oleaginous fungus Mortierella alpina has received GRAS status from the US FDA in 2001 (Ryan et al. 2010), and M. alpina is regarded as one of the most promising candidates for the mass produc- tion of ARA-rich oil (Ji et al. 2014a). The ARA biosyn- thesis pathway in M. alpina proceeds via the formation of C16 or C18 saturated fatty acids, which are further modified through a series of elongation and desaturation steps, culminating in the formation of ARA. It is known that these reactions require NADPH, an electron trans- port system, a terminal desaturase, and molecular oxygen (Ward and Singh 2005). Fermentation methods Pellets of M. alpina were used to inoculate the PDA slants which were cultivated at 25 °C. After 12–15 days of incubation, a loop was used to transfer mycelial material into deep 250 mL baffled flasks containing 50 mL inocu- lation medium, and the cultures were subsequently incu- bated for 2 days at 25 °C under constant orbital shaking at 125 rpm. Fed-batch fermentations were carried out in a 7.5 L bioreactor (New Brunswick Scientific, USA) con- taining 5 L of fermentation medium.h The multi-stage process was carried out according to our proposed stepwise aeration and agitation control strategy. The aeration rate was set at 6 L/min to achieve an aeration rate of 1.2 volumes of air per volume of liquid per minute (vvm), without agitation in stage I (0–48 h). The agitation rate was increased stepwise from 50 to 150 revolutions per minute (rpm) in stage II and subsequently kept constant at 200 rpm until the end of the fermenta- tion. The aeration rate was set at 1.0 vvm from step II to step IV. Glucose (500 g/L stock solution) was fed into the fermentation broth during the entire fermentation pro- cess to maintain the glucose concentration at 5–20 g/L. Samples comprising 100 mL of the fermentation broth were taken periodically for further examination. i In this study, an innovative multi-stage strategy was investigated to optimize ARA productivity in bioreac- tors. We thereby aimed at balancing the contradiction between the aeration and agitation controls required for optimal dissolved oxygen concentration and fun- gal morphology, respectively. The strategy was further assessed regarding its effectiveness in improving the bio- mass yield, which reached >40 g/L. This approach gives a detailed insight into the mycelial morphology control of oil-producing filamentous fungi and will provide guid- ance for the large-scale production of ARA and similar polyunsaturated fatty acids. Background Arachidonic acid, (5, 8, 11, 14-cis-eicosatetraenoic acid, ARA), a representative of the omega-6 group of essential polyunsaturated fatty acids (PUFAs), acts as a precursor for eicosanoid hormones such as prostaglandins, leukot- rienes, and thromboxanes (Ji et al. 2014a). The applica- tion of ARA as the active ingredient in drugs and food additives thus has great potential. Owing to its unique physiological functions, it has been widely applied in the food industry as well as cosmetics, medicine, and many other fields (Ward and Singh 2005). For a long time, egg yolk, animal liver, and adrenal glands were the main sources of ARA. However, their low intrinsic ARA con- tent (Higashiyama et al. 2002) restricts their application, and it is not possible to source sufficient material for In general, fungal mycelia are brittle and physically weak. Therefore, the agitation rate in mechanically stirred bio- reactors, which are normally used for the production of ARA-rich oil, has to be controlled within a very precise range. High agitation rates increase the shear forces, which can break mycelial integrity and influence the broth char- acteristics. Low agitation rates, on the other hand, lead to Wu et al. Bioresour. Bioprocess. (2017) 4:8 Page 2 of 9 (Angel Yeast Co., Ltd, China) 6; glucose 30; KH2PO4 3; NaNO3 3; MgSO4·7H2O 3. Fermentation medium (g/L): yeast extract 10; glucose 80; KH2PO4 4; NaNO3 3; MgSO4·7H2O 0.6; initial pH 6.0. a low dissolved oxygen concentration insufficient for ARA biosynthesis. Overall, mycelial morphology has a strong effect on the physical properties of the broth and often leads to a number of different problems in large bioreac- tors with respect to gas dispersion, as well as mass and heat transfer (Higashiyama et al. 2002). There are many reports that discuss the size and shape of fungal mycelial pellets (Xu et al. 2010; Tai et al. 2010), but little is known about the true features of the internal pellet structure, including geometry and mycelial viability (Hamanaka et al. 2001). Interestingly, pellets with a moderate compactness are the more productive morphological form for the pro- duction of ARA-rich oil, compared to free filamentous mycelia. Therefore, controlling proper aeration and agita- tion rates in the whole process to balance the contradic- tion between these two factors is vitally important for the fermentation of fungal producers of ARA-rich oils. Microorganism Mortierella alpina R807 (CCTCC M 2012118), preserved in the China Center for Type Culture Collection, was used in the present study. It was maintained on potato dextrose agar (PDA) slants by culturing for 10 days at 25 °C, and transferred every 3 months. Total lipids (TLs)h The dry cell material was ground into a fine powder for lipid extraction and fatty acid determination. A 2 g ali- quot of the resulting powder was loaded onto a Soxhlet extractor with 150 mL chloroform/methanol (2:1, v/v) and extracted for 8 h at 75 °C. Finally, the solvent was removed on a rotary evaporator and recycled, with TLs remaining as evaporation residue (Ji et al. 2014b). Determination of dry cell weight (DCW) and glucose concentration Aliquots comprising 100 mL of the fermentation broth were used to determine the DCW using the filtration method. The broth samples were transferred to a suction filter under 0.1 MPa negative pressure. The cell pellet was washed twice with distilled water and dried at 60 °C to constant weight (12 h). An aliquot comprising 1 mL of fermentation broth was transferred to a centrifuge tube, centrifuged at 3000×g for 3 min, and the resulting super- natant was used to measure the glucose concentration, which was determined enzymatically using a simultane- ous Bioanalyzer (SBA-40C, Institute of Biology, Shan- dong Academy of Sciences, China). Background There have been some attempts to fulfill this objective by control- ling the aeration (Higashiyama et al. 1999; Nie et al. 2014) and agitation rates (Higashiyama et al. 1999; Peng et al. 2010), respectively. ARA yields in these reports reached 4.7 g/L by strictly monitoring the mycelial morphology and employing a two-stage control strategy for the aera- tion rate, which represents an increase of 38.2% (Gao et al. 2016). However, until now, no efforts have been made to simultaneously evaluate the aeration and agitation rate in relation to the proper mycelial morphology for increasing the biomass yield of the filamentous fungus M. alpina. Fermentation disparities between batch and fed‑batch protocolshf The effects of batch and fed-batch fermentation on DCW, lipid, ARA contents, and ARA production were investi- gated using a 7.5 L bioreactor (Fig. 1). Though batch fer- mentation was found to be optimal for growth and total lipid production, ARA was synthesized more rapidly in fed-batch cultures. The lipid concentration reached a maximum value of 17.6 g/L at 6.5 days. However, higher ARA contents (42.8%) and ARA yield (10.0 g/L) were obtained with fed-batch fermentation. These results sug- gest that the optimal culture conditions for lipid accu- mulation and ARA biosynthesis are different. A higher C/N ratio was achieved in batch fermentations, which stimulated lipid accumulation. However, it also led to lower ARA biosynthesis. Furthermore, the C/N ratio is an important fermentation parameter which can affect mycelial morphology (Koike et al. 2001; Park et al. 2001), and it has been demonstrated previously that the mor- phology of M. alpina mycelia has a strong effect on phys- ical properties of the broth, which in turn might lead to poor mass transfer performance. Due to the difficulty of controlling the mycelial morphology of M. alpina under constant aeration and agitation rates, the dry cell weight obtained in fed-batch fermentation (36.1 g/L) was not much higher than what was obtained in batch fermen- tation (31.1 g/L). However, the ARA contents and ARA yield (42.8% and 10.0 g/L, respectively) were clearly enhanced over the batch fermentation. A low initial Fig. 1 Time-course of different fermentation parameters from fermentations using M. alpina for the production of ARA-rich oil; a batch; b fed-batch glucose concentration is usually used to shorten the lag phase of fungal growth (Zhu et al. 2006). The glucose consumption during this stage was rapid, and the glu- cose was used up almost completely by day 6. The ARA productivity achieved by this method was 1.67 g/L day−1, which is 1.70-fold higher than what was obtained in batch fermentations. Culture medium Slant medium: Potato dextrose agar (PDA). The PDA medium contained (g/L): potatoes 200; glucose 25; agar 20. Inoculation medium (g/L): yeast extract Wu et al. Bioresour. Bioprocess. (2017) 4:8 Page 3 of 9 Fig. 1 Time-course of different fermentation parameters from fermentations using M. alpina for the production of ARA-rich oil; a batch; b fed-batch Fatty acid methyl esters (FAMEs) were prepared according to the established method (Ji et al. 2014b; Ren et al. 2009) as follows: 1.5 mL n-hexane and 0.2 mL 0.5 M KOH– methanol were added to a centrifuge tube containing 0.1 g of the powdered dry cells, and mixed thoroughly by vortexing for 3 min, followed by stewing for 15 min. A 0.3 mL aliquot of the resulting upper phase was combined with 0.5 mL distilled water in another cen- trifuge tube and the tube was centrifuged at 5000×g for 3 min. The upper phase containing FAMEs was applied to a Thermo Finnigan trace GC2000 DSQ gas chroma- tograph equipped with a 30 m × 0.25 mm × 0.25 µm DB-23MS capillary column (Agilent Technologies). The column temperature was increased from 80 to 200 °C at 40 °C/min, and subsequently to 300 °C at 10 °C/min. The temperature of both the injector and detector was set to 250 °C. Nitrogen was used as the carrier gas at 1 mL/min. Peaks were identified using authentic standards of the corresponding fatty acid methyl esters (Sigma-Aldrich). Fatty acids were quantified based on their corresponding peak areas relative to the peak areas of the standards. Controlling the morphology of Mortierella alpina using a multi‑stage fermentation strategyi Seed culture morphology was found to be a significant factor in fermentations producing ARA-rich oil. This was directly due to the effects of mycelial morphology on the physical properties of the fermentation broth (Higashiy- ama et al. 2002). Thus, mycelial morphology is consid- ered to be a key parameter in ARA fermentation, and the fungus must consequently be cultivated in the correct morphological form in order to obtain maximal ARA concentration (Ji et al. 2014b). Although feather-like hyphal filaments (Fig. 2a) were observed to be optimal for ARA production at low densities (Park et al. 1999), Wu et al. Bioresour. Bioprocess. (2017) 4:8 Page 4 of 9 Fig. 2 Morphology of M. alpina mycelia cultured using different fermentation strategies. a At the end of the fed-batch fermentation; b at 48 h of the multi-stage strategy; c at the end of the multi-stage fermentation strategy Fig. 2 Morphology of M. alpina mycelia cultured using different fermentation strategies. a At the end of the fed-batch fermentation; b at 48 h of the multi-stage strategy; c at the end of the multi-stage fermentation strategy a mycelia cultured using different fermentation strategies. a At the end of the fed-batch fermentation; b at 48 h of he end of the multi-stage fermentation strategy respectively. The highest ARA productivity obtained in this study, which stood at 1.81 g/L day−1, was achieved using the stepwise aeration and agitation control devel- oped here. This was the same as the highest value pub- lished for M. alpina ME-1 (Jin et al. 2008) and was much higher than the 1.50 g/L day−1 reported for M. alpina DSA-12 using conventional protocols (Hwang et al. 2005). this morphology is disadvantageous at high cell densi- ties because viscosity of the ferrmentation broth may be increased to an extent that oxygen transmission becomes limited (Wynn and Ratledge 2005). In stage I (from 0 to 48 h), pellet-like mycelia of M. alpina were formed using baffled shake flasks and trans- ferred into the nutrient-rich fermentation medium. Even though this pellet-like morphology allowed easier mix- ing and better mass transfer, pellet-like mycelia were highly sensitive to shear stress (Fig. 2b). No agitation combined with 1.2 vvm aeration was used to maintain this pellet-like morphology during the 2-day lag phase. During the fermentation period, the pellet-like particles became fragmented into both small pellets and filamen- tous mycelia. Controlling the morphology of Mortierella alpina using a multi‑stage fermentation strategyi As the cultivation processed, the pellet-like cores became smaller, and more cell material displaying the pellet-like morphology was formed. After 132 h, the cells started to autolyse, and particles with a rice-shaped morphology were formed (Fig. 2c). This rice-shaped mor- phology proved to be optimal during the stage at which a high agitation rate was used to maintain the DO level. Compared to the standard batch and fed-batch fermen- tation protocols, this multi-stage culture method pro- longs the fermentation period by nearly 24 h, and while the total consumption glucose also increased sharply from 80 to 100 g/L, an obvious increase of dry cell weight was also noticed (41.4 g/L). ARA productivity conse- quently increased to 1.81 g/L day−1, which was 1.08-fold higher than in the fed-batch fermentation. To optimally analyze the process of cell growth and ARA accumulation, as well as to understand the effects of aeration and agitation on mycelial morphology, the fer- mentation process was divided into four stages accord- ing to cell growth characteristics (Ren et al. 2010). Stage I represents the beginning of the process until the mor- phological adaptation period; stage II was the phase of high cell density fermentation; stage III encompasses the lipid biosynthesis period; and stage IV comprises the period of most efficient ARA accumulation. Development of a multi‑stage strategy for ARA fermentationf g p p High ARA content in total fatty acids is the prerequi- site for high-quality ARA-rich oil, and a high aeration rate was favorable for efficient ARA biosynthesis. It is reported that the pathway of ARA biosynthesis was most widespread in oleaginous yeasts, and fungi encompasses both desaturation and elongation steps (Ji et al. 2014a). Palmitic acid (C16:0) is the main saturated fatty acid formed by fatty acid synthase (FAS), and ARA is subse- quently produced from it via desaturase and elongase reactions. Our study showed that adequate oxygen was needed to increase the levels of unsaturated fatty acids, especially C18:3 and C20:4 (ARA). In stage IV (from 132 to 216 h), an improved mycelium-aging protocol was used to enhance ARA production (Zhang et al. 2015). When cells were cultivated under high aeration and agi- tation rates (Fig. 3), glucose was exhausted at 132 h, with dry cell weight reaching 44.2 g/L. Overall, the cells con- sumed 100 g/L of glucose, which was 20 g/L more than in fed-batch fermentation. The main increase in ARA contents was observed during stage IV, at which point the ARA contents reached 61.1%, with total yield also increasing sharply. g gy During stage I (from 0 to 48 h), glucose was consumed to below 10 g/L after a single feeding. Nitrogen was con- sidered to be exhausted at 48 h (Lu et al. 2011; Ling et al. 2016). With ample carbon and nitrogen source, the dry cell weight increased slightly to 13 g/L, albeit with only 3 g/L of lipids and nearly 10 g/L of non-lipid dry cell weight. At the same time, pellet-like morphology could be preserved better. In stage II (from 48 to 96 h), the agi- tation rate was increased stepwise from 50 to 150 rpm, while the DO concentration was maintained between 10 and 20%. The glucose consumption rate reached its maximum, which might be explained by increased con- sumption for cell maintenance. At this stage, the glu- cose consumption rate was so high that glucose needed to be fed every 12 h, and a sharp increase of biomass from 13 to 28 g/L was also noticed. In stage III (from 96 to 132 h), the dry cell weight reached its maximum value of 44.2 g/L, whereas the non-lipid dry cell weight increased slightly and remained at a constant level. Development of a multi‑stage strategy for ARA fermentationf In this paper, the effects of aeration and agitation on cell morphology, lipid accumulation, and ARA production were investigated systemically, and a multi-stage strat- egy was developed aimed at achieving a high cell den- sity, high accumulation rate, and high ARA yield. The final dry cell weight, total lipids, ARA contents, and ARA yield reached 41.41, 22.17 g/L, 61.05%, and 13.53 g/L, fi High cell density is the first precondition for high pro- duction of intracellular products, and it was obvious that high aeration was beneficial to cell growth. Our study also showed this positive effect, but with a reduction of Yx/s at stage II and a slight increase at stage III (Table 1). Wu et al. Bioresour. Bioprocess. (2017) 4:8 Page 5 of 9 Table 1 Comparison of fermentation parameters at different stages of ARA fermentation via a multi-stage fermentation strategy Yx/s: conversion of glucose to biomass Yl/s: conversion of glucose to lipids YARA/s: conversion of glucose to ARA Stage I II III IV Glucose consumption rate (g/L h−1) 0.479 ± 0.28 1.104 ± 0.19 0.694 ± 0.29 None ARA increase (%) None 10.676 ± 0.81 2.492 ± 0.13 18.862 ± 0.23 Yx/s 0.614 ± 0.21 0.373 ± 0.32 0.419 ± 0.18 None Yl/s 0.165 ± 0.33 0.208 ± 0.38 0.341 ± 0.21 None YARA/s 0.048 ± 0.33 0.098 ± 0.33 0.159 ± 0.18 None Table 1 Comparison of fermentation parameters at different stages of ARA fermentation via a multi-stage fermentation strategy Stage f fermentation parameters at different stages of ARA fermentation via a multi-stage fermentation This might be explained by the fact that cell respiration would be intensified and additional carbon flux chan- neled towards the tricarboxylic acid cycle under a high aeration rate. Consequently, dissolved oxygen can be controlled by using an appropriate aeration and agitation rate. To maintain the dissolved oxygen concentration, usually either an oxygen-enrichment method or a pres- surization method is used (Higashiyama et al. 2002). In this research, on the other hand, the DO concentration was maintained via a combined stepwise aeration and agitation control strategy. the seventh batch of glucose feed, the consumption rate began to decline at 96 h. Moreover, the ARA contents slightly decreased, and other fatty acids (such as C18:0 and C18:1) increased, which can likely be explained by the high accumulation of lipid droplets. Development of a multi‑stage strategy for ARA fermentationf At the same time, the lipid contents increased from 11 to 25 g/L (Fig. 3), which indicated that cell metabolism had shifted away from cell growth towards lipid accumulation. After Changes in PUFA distribution in response to different dissolved oxygen conditions PUFAs are produced via desaturation and elongation reactions, which involve aerobic oxygenation. There- fore, dissolved oxygen (DO) is a very significant factor Wu et al. Bioresour. Bioprocess. (2017) 4:8 Page 6 of 9 Fig. 3 Time-course of various fermentation parameters during the production of high-quality ARA-rich oil via a multi-stage fermentation strategy. a Biomass, lipid yield, non-lipid biomass, and DO; b glucose and glucose consumption rate; c percentage of various fatty acids in TFA; d concentration of various fatty acids. The shown values represent the averages ± SD from three independent experiments Fig. 3 Time-course of various fermentation parameters during the production of high-quality ARA-rich oil via a multi-stage fermentation strategy. a Biomass, lipid yield, non-lipid biomass, and DO; b glucose and glucose consumption rate; c percentage of various fatty acids in TFA; d concentration of various fatty acids. The shown values represent the averages ± SD from three independent experiments 10–20, and 0–5% DO groups. The contents of C16:0 and C18:0 were increased slightly at 30–40 and 10–20% DO, respectively, compared to 0–5%. In this optimal DO concentration range, the ARA yield was enhanced about 1.2-fold compared to that obtained at 10–20%, and ARA contents decreased drastically from 39.4 to 22.8% at 0–5% DO. This decrease was likely due to stress caused by the very limited DO concentration. This observation underscores that DO concentration is indeed one of the most important factors influencing ARA productivity. for PUFA production, as reported by previous stud- ies (Higashiyama et al. 2002; Su et al. 2016). There have also been some attempts to monitor and control the DO concentration in order to prevent DO limita- tion during ARA production (Higashiyama et al. 1999; Totani et al. 1992). Cultivations were carried out at different DO levels in the range of 30–40, 10–20, and 0–5%, respectively, and the optimum DO concentra- tion range was found to be 30–40%, as shown in Fig. 4. No significant differences were observed in the con- tents of C18:2, C18:3, and C20:3 among the 30–40, Wu et al. Bioresour. Bioprocess. (2017) 4:8 Page 7 of 9 Fig. 4 Effect of different DO concentrations on the accumulation of ARA-rich oil during aging. Changes in PUFA distribution in response to different dissolved oxygen conditions a DO kept between 30 and 40%; b 10–20%; c 0–5% Kinetic parameters of the multi‑stage strategy Kinetic parameters of the multi‑stage strategy g gy To analyze the kinetic characteristics of the multi-stage fermentation process, five parameters, including glucose consumption rate, ARA increase rate, Yx/s, Yl/s, and YARA/s, were compared at different stages. The corresponding data are summarized in Table 1. The values of these kinetic parameters all fluctuated with time and were especially influenced by glucose feeding. At the early fermentation stage (stage I), high aeration without mechanical (Table 1) agitation was able to ensure a higher Yx/s (conversion of glucose to biomass) than was observed at the other stages. This indicates that high aeration could not only improve cell growth and glucose consumption, but could also accel- erate the conversion of glucose to biomass. What is more, omitting the mechanical agitation was beneficial for main- taining a pellet-like morphology. At stage II, the value of Yx/s showed a decrease, whereas the values of ARA and YARA/s (conversion of glucose to ARA) increased sharply, indicating that a high lipid accumulation rate could be maintained under the high aeration and low agitation con- ditions found in stage II. After 96 h, in stage III, the value of lipid had increased significantly, indicating that the ARA contents were slightly decreased, whereas the other fatty acids, especially C18:0 and C18:1, increased sharply. After 132 h, the increasing ARA concentration reached the value of 18.9%, while glucose was exhausted. Although the ARA increase at the end of the multi-stage process (18.9%) was higher than during stages I to III, beginning cell autolysis resulted in less glucose consumption, and led to a slight decrease of overall dry cell weight. This further confirmed the importance of the multi-stage strategy combined with an efficient control of mycelial morphology. Fig. 4 Effect of different DO concentrations on the accumulation of ARA-rich oil during aging. a DO kept between 30 and 40%; b 10–20%; 0 5% fi To sum up, this study systematically examined the effects of aeration and agitation rates on ARA production by M. alpina and proposed a stepwise aeration and agi- tation rate control strategy to achieve a high cell growth rate and optimal overall productivity (Table 2). Acknowledgements The authors wish to acknowledge the financial support from the National Science Foundation for Distinguished Young Scholars of China (No. 21225626), the National Natural Science Foundation of China (Nos. 21376002, and 21476111), the Natural Science Foundation of Jiangsu Province (No. BK20131405), the National High-Tech R&D Program of China (No. 2014AA021703), and the Priority Academic Program Development of Jiangsu Higher Education Institutions. Nie ZK, Ji XJ, Shang JS, Zhang AH, Ren LJ, Huang H (2014) Arachidonic acid- rich oil production by Mortierella alpina with different gas distributors. Bioprocess Biosyst Eng 37:1127–1132 Park EY, Koike Y, Higashiyama K, Fujikawa S, Okabe M (1999) Effect of nitrogen source on mycelial morphology and arachidonic acid production in cultures of Mortierella alpina. J Biosci Bioeng 88:61–67 Park EY, Koike Y, Cai HJ, Higashiyama K, Fujikawa S (2001) Morphological diver‑ sity of Mortierella alpina: effect of consumed carbon to nitrogen ratio in flask culture. Bioprocess Biosyst Eng 6:161–166 Conclusionsh This paper compared the experimental results of fermen- tations using M. alpina to produce oils rich in ARA via different culture strategies. The morphology of the fungal mycelia could be maintained in an optimal state through- out the fermentation. It could be shown that the multi- stage strategy provides a favorable gas–liquid mixture, Fig. 4 Effect of different DO concentrations on the accumulation of ARA-rich oil during aging. a DO kept between 30 and 40%; b 10–20%; c 0–5% Wu et al. Bioresour. Bioprocess. (2017) 4:8 Page 8 of 9 Table 2 Comparison of parameters from different ARA fermentation strategies a This value represents the corresponding data from the multi-stage strategy divided by the highest respective value from either the batch or fed-batch fermentation Fermentation strategy Rate of increasea Batch Fed-batch Multi-stage DCW (g/L) 31.06 36.08 41.41 14.77% Mycelial specific growth rate (g/L d−1) 4.78 6.01 5.92 −1.50% Lipids (g/L) 17.63 17.09 22.17 25.75% Fermentation duration (d) 6.5 6 7.5 7.69% Total glucose (g/L) 80 80 100 25% Glucose consumption rate (g/L h−1) 0.51 0.56 0.76 35.71% ARA contents (%) 36.30 42.81 61.05 42.61% ARA yield (g/L) 6.40 10.01 13.53 35.16% ARA productivity (g/L d−1) 0.98 1.67 1.81 7.78 Table 2 Comparison of parameters from different ARA fermentation strategies and consequently increase biomass accumulation sig- nificantly. This work offers insights into the control of aeration and agitation and provides a reference for the fer- mentation of filamentous fungi at a mass industrial scale. Author details 1 1 College of Biotechnology and Pharmaceutical Engineering, Nanjing Tech University, No. 30 South Puzhu Road, Nanjing 211816, People’s Republic of China. 2 School of Pharmaceutical Sciences, Nanjing Tech University, No. 30 South Puzhu Road, Nanjing 211816, People’s Republic of China. 3 State Key Laboratory of Materials‑Oriented Chemical Engineering, Nanjing Tech University, No. 5 Xinmofan Road, Nanjing 210009, People’s Republic of China. 4 Jiangsu National Synergetic Innovation Center for Advanced Materials (SICAM), No. 5 Xinmofan Road, Nanjing 210009, People’s Republic of China. 5 Beijing Key Laboratory of Nutrition Health and Food Safety, COFCO Nutrition and Health Research Institute, Beijing 102209, People’s Republic of China. Jin MJ, Huang H, Xiao AH, Zhang K, Liu X, Li S, Peng C (2008) A novel two-step fermentation process for improved arachidonic acid production by Mortierella alpina. Biotechnol Lett 30:1087–1091 Koike Y, Cai HJ, Higashiyama K, Fujikawa S, Park EY (2001) Effect of consumed carbon to nitrogen ratio of mycelial morphology and arachidonic acid production in cultures of Mortierella alpina. J Biosci Bioeng 91:382–389 Ling XP, Zeng SY, Chen CX, Liu XT, Lu YH (2016) Enhanced arachidonic acid production using a bioreactor culture of Mortierella alpina with a com‑ bined organic nitrogen source. Bioresour Bioprocess 3:43 Lu JM, Peng C, Ji XJ, You JY, Cong LL, Ouyang PK, Huang H (2011) Fermenta‑ tion characteristics of Mortierella alpina in response to different nitrogen sources. Appl Biochem Biotechnol 164:979–990 Authors’ contributions WJW and AHZ carried out the main experiments. CP, LJR, PS, and YDY helped in the cultivation of the strain and fatty acids assay. WJW, HH, and XJJ are involved in the drafting and revision of the manuscript. XJJ has given final approval of the version to be published. All authors read and approved the final manuscript. Hwang BH, Kim JW, Park CY, Park CS, Kim YS, Ryu YW (2005) High-level produc‑ tion of arachidonic acid by fed-batch culture of Mortierella alpina using NH4OH as a nitrogen source and pH control. Biotechnol Lett 27:731–735 Ji XJ, Ren LJ, Nie ZK, Huang H, Ouyang PK (2014a) Fungal arachidonic acid-rich oil: research, development and industrialization. Crit Rev Biotechnol 34:197–214 Ji XJ, Zhang AH, Nie ZK, Wu WJ, Ren LJ, Huang H (2014b) Efficient arachidonic acid-rich oil production by Mortierella alpina through a repeated fed- batch fermentation strategy. Bioresour Technol 170:356–360 Competing interests The authors declare that they have no competing interests. Received: 30 November 2016 Revised: 28 December 2016 Accepted: 10 January 2017 Received: 30 November 2016 Revised: 28 December 2016 Accepted: 10 January 2017 Peng C, Huang H, Ji X, Liu X, Ren LJ, Yu W, You JY, Lu JM (2010) Effects of n-hexadecane concentration and a two-stage oxygen supply control strategy on arachidonic acid production by Mortierella alpina ME-1. Chem Eng Technol 33:692–697 Ren LJ, Huang H, Xiao AH, Lian M, Jin LJ, Ji XJ (2009) Enhanced docosahexae‑ noic acid production by reinforcing acetyl-CoA and NADPH supply in Schizochytrium sp. HX-308. Bioprocess Biosyst Eng 32:837–843 Gao MJ, Wang C, Zheng ZY, Zhu L, Zhan XB, Lin CC (2016) Improving arachi‑ donic acid fermentation by Mortierella alpina through multistage tem‑ perature and aeration rate control in bioreactor. Prep Biochem Biotechnol 46:360–367 Wynn JP, Ratledge C (2005) Oils from microorganisms. In: Shahidi F (ed) Bailey’s industrial oil and fat products, 6th edn. Wiley, New York, pp 121–153 Zhu M, Yu LJ, Li W, Zhou PP, Li CY (2006) Optimization of arachidonic acid production by fed-batch culture of Mortierella alpina based on dynamic analysis. Enzyme Microb Technol 38:735–740 References Ren LJ, Ji XJ, Huang H, Qu L, Feng Y, Tong QQ, Ouyang PK (2010) Develop‑ ment of a stepwise aeration control strategy for efficient docosahexae‑ noic acid production by Schizochytrium sp. Appl Microbiol Biotechnol 87:1649–1656 Wu et al. Bioresour. Bioprocess. (2017) 4:8 Page 9 of 9 Wynn JP, Ratledge C (2005) Oils from microorganisms. In: Shahidi F (ed) Bailey’s industrial oil and fat products, 6th edn. Wiley, New York, pp 121–153 Ryan AS, Zeller S, Nelson EB, Cohen Z, Ratledge C (2010) Safety evaluation of single cell oils and the regulatory requirements for use as food ingredi‑ ents. In: Cohen Z, Ratledge C (eds) Single cell oils: microbial and algal oils, 2nd edn. AOCS Press, Urbana Xu Q, Li S, Fu Y, Tai C, Huang H (2010) Two-stage utilization of corn straw by Rhizopus oryzae for fumaric acid production. Bioresour Technol 101:6262–6264 Su GM, Jiao KL, Chang JY, Li Z, Guo XY, Sun Y, Zeng XH, Lu YH, Lin L (2016) Enhancing total fatty acids and arachidonic acid production by the red microalgae Porphyridium purpureum. Bioresour Bioprocess 3:33 Zhang AH, Ji XJ, Wu WJ, Ren LJ, Yu YD, Huang H (2015) Lipid fraction and intracellular metabolite analysis reveal the mechanism of arachidonic acid-rich oil accumulation in the aging process of Mortierella alpina. J Agric Food Chem 63:9812–9819 Tai C, Li S, Xu Q, Ying H, Huang H, Ouyang PK (2010) Chitosan production from hemicellulose hydrolysate of corn straw: impact of degradation products on Rhizopus oryzae growth and chitosan fermentation. Lett Appl Micro‑ biol 51:278–284 Zhu M, Yu LJ, Li W, Zhou PP, Li CY (2006) Optimization of arachidonic acid production by fed-batch culture of Mortierella alpina based on dynamic analysis. Enzyme Microb Technol 38:735–740 Totani N, Someya K, Oba K (1992) Industrial production of arachidonic acid by Mortierella. In: Kyle DJ, Ratledge C (eds) Industrial applications of single cell oils. AOCS Press, Urbana Ward OP, Singh A (2005) Omega-3/6 fatty acids: alternative sources of produc‑ tion. Process Biochem 40:3627–3652
https://openalex.org/W4368240950	https://hal.science/hal-04087437/document	English	null	Static reinforcement and vibration reduction of structures using topology optimization	Mechanics & industry	2,023	cc-by	11,185	Static reinforcement and vibration reduction of structures using topology optimization Sylvain Burri, Antoine Legay To cite this version: Sylvain Burri, Antoine Legay. Static reinforcement and vibration reduction of structures using topol- ogy optimization. Mechanics & Industry, 2023, 24, pp.14. 10.1051/meca/2023003. hal-04087437 Received: 23 August 2022 / Accepted: 13 January 2023 Abstract. This paper presents a topology optimization formulation based on the Solid Isotropic Material with Penalization (SIMP) method and solved by the Modified Optimality Criteria (MOC) algorithm. It addresses mechanical design problems such as structural reinforcement adding elastic material or vibration reduction using viscoelastic layers. The aim is thus to attach on a pre-existing given structure a design domain in order to improve the behavior of this elastic structure, according to an objective function. This can be useful when one wants to use, for example, additive manufacturing to reinforce a pre-existing structure or to maximize structural damping. Two objective functions are tested in both linear statics and dynamics: a compliance based objective function and a displacement based one. In the dynamic case, written in the frequency domain, the two proposed objective functions include the viscoelastic material model (a Zener fractional derivative one) used to fill the design domain. The displacement criteria is developed using a general formula able to take into account as many degree of freedom as necessary. Finally, some applications based on beams and CubeSat-like structures are shown in this article. The proposed examples show that in both statics and dynamics, the optimization of a restrained design domain attached to an existing structure can improve its behavior: stiffness improvement or vibration reduction. Keywords: Static Reinforcement / Vibration Reduction / Topology optimization / Viscoelasticity * e-mail: antoine.legay@lecnam.net This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. tributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0) stricted use, distribution, and reproduction in any medium, provided the original work is properly cited. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://cre which permits unrestricted use, distribution, and reproduction in any medium, provided the original w HAL Id: hal-04087437 https://hal.science/hal-04087437v1 Submitted on 4 May 2023 L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés. HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. Static reinforcement and vibration reduction of structures using topology optimization Laboratoire de M´ecanique des Structures et des Syst`emes Coupl´es (LMSSC), Structural Mechanics and Coupled System Laboratory, Conservatoire National des Arts et M´etiers (Cnam), 292 rue Saint-Martin, 75003 Paris, France Mechanics &Industry Mechanics & Industry 24, 14 (2023) © S. Burri and A. Legay, Published by EDP Sciences, 2023 https://doi.org/10.1051/meca/2023003 Mechanics & Industry 24, 14 (2023) © S. Burri and A. Legay, Published by EDP Sciences, 2023 https://doi.org/10.1051/meca/2023003 Mechanics & Industry 24, 14 (2023) © S. Burri and A. Legay, Published by EDP Sciences, 2023 https://doi.org/10.1051/meca/2023003 Available online at: www.mechanics-industry.org Available online at: www.mechanics-industry.org Available online at: www.mechanics-industry.org REGULAR ARTICLE REGULAR ARTICLE 1 Introduction and motivation formulation) and stiffness matrices of each substructure are considered separately and assembled afterwards in the global system. During the optimization process, the discretized operators of Ω1 remain constant (stiffness and mass matrices) while they vary for Ω2 according to its material density distribution. In order to perform this topology optimization, the design variables are the elemental material density distribution in Ω2. The structural mass is often a main issue which has to be cautiously considered during the design process of a mechanical structure, for instance for a vehicle or an equipment in the aerospace industry. This is a structural optimization problem where one wants to reduce the mass regarding to another criterion such as, non-exhaustively: stiffness maximization usually expressed as a compliance minimization problem, displacement lowering or vibration reduction [1]. Such objectives can be effectively addressed by using a topology optimization approach [2–11]. One can therefore imagine to reinforce Ω1 using an efficient manufacturing process such as additive manu- facturing [19] to build Ω2. Concerning the purpose of vibration reduction, it is per- formed in the frequency domain and damping is taken into account thanks to viscoelastic materials [20–22]. Among the various rheological existing models of viscoelastic behavior, the four-parameter Zener fractional-derivative model is used in this work. It allows to efficiently repre- sent the frequency-dependence of the damping properties with only four parameters: correlations between numerical and experimental tests show a good accuracy [23]. The present article uses an algorithm based on the well known SIMP-method [12–14] in order to focus on these objectives and tries to address the situation when one wants to add to a pre-existing given structure Ω1, a sec- ond domain Ω2 with the aim of either reinforcing the pre-existing structure, lowering the displacement of a cho- sen area, or reducing vibrations of Ω1 using the design domain Ω2. Although few papers exist on the subject of structural reinforcement using topology optimization [15–18], the idea of considering separated substructures is newly addressed here. This means that mass (for dynamic [ ] The next part of the present paper is dedicated to the building of the discretized numerical model of the physical problem, including a sensitivity analysis of the solution with respect to the design variables, namely the elemental material densities in the design domain. The S. Burri and A. 1 Introduction and motivation Legay: Mechanics & Industry 24, 14 (2023) 2 2 Ω2 Ω1 xe = 0 xe = 1 Interface I ⃗F Fig. 1. Geometry and finite element mesh of the problem composed of a given structure Ω1 and a design domain Ω2. third section details the topology optimization problem formulation, introducing the two different used objective functions: compliance and displacement based objective functions. The fourth section presents a selection of appli- cations corresponding to a static reinforcement and to a vibration reduction. The numerical implementation is done using Python and Fortran languages. Gmsh [24] is used for pre- treatment (meshing) and post-treatment (plotting). 2.1 Geometry Fig. 1. Geometry and finite element mesh of the problem composed of a given structure Ω1 and a design domain Ω2. The studied problem is composed of a purely elastic given domain Ω1 and a design domain Ω2 which can be either purely elastic or viscoelastic (Fig. 1). The whole problem is assumed to be linear. The interface between the two domains is denoted by I. In the static case, the aim is to reinforce the given structure Ω1 subjected to the static load ⃗F. In the dynamic case, considered in the frequency domain, the aim is to dissipate energy in order to protect the given structure Ω1 from spurious vibrations due to the harmonic excitation ⃗F. The first case can be degenerated from the second by taking the angular frequency ω to zero. This last expression is the one that is implemented to describe the material contained in the sub-part Ω2. 2.3 Implementation of design variables For the purpose of topology optimization, the design domain Ω2 contains a set of design variables x = [x1, x2, . . . , xn]T which is the set of elemental material densities (n is the number of elements in Ω2-domain’s mesh). These densities are such that 0 ≤xe ≤1 where 0 is the lower bound of xe corresponding to an absence of material in the element and 1 is the upper bound meaning a presence of material [26] (Fig. 1). A design domain Ω2 is then attached to Ω1, in which each element e has a material density variable xe. The aim is to find the best material distribution in Ω2 according to an objective function associated to constrains functions. This optimization problem is solved using a topology optimization procedure described in Section 3.1. An efficient way to use this is the well-known penal- ization algorithm SIMP from Bendsøe [12] and Zhou & Rozvany [13]. However, a derivative from this law estab- lished by Sigmund [27] and called the Modified SIMP-law is used here. This formulation has several advantages, including the fact that it avoids stiffness (or mass) matrix to become singular. 2.5 Sensitivity analysis of the solution with respect to the design variables Using the finite element method, the dynamic discretized system of the whole problem is written in the frequency domain as: In the following work, a topology optimization procedure is used (Sect. 3.1). This algorithm needs to compute the sensibility of the solution according to the design variables xe. Since only matrix S2(ω, x) is depending on design variables, the derivative of equation (11) with respect to the design variable xe leads to the following equation: K −ω2M U = F, (6) (6) where K is the complex stiffness matrix (taking into account viscoelastic terms), M is the mass matrix, ω is the angular frequency and F represents the external nodal forces. The design variable vector x, composed of elemental material densities in design sub-domain Ω2 is introduced in order to use a topology optimization proce- dure. Using a sub-domain decomposition between domain Ω1 and Ω2, the global discretized system equation (6) is written: ∂S2(ω, x) ∂xe U(x) + S(ω, x)∂U(x) ∂xe = 0. (12) (12) The sensitivity of the solution with respect to the design variable xe is then given by: ∂U(x) ∂xe = −S−1(ω, x)∂S2(ω, x) ∂xe U(x), (13) (13) "K111 K11I 012 K1I1 K1II + K2II(ω, x) K2I2(ω, x) 021 K22I(ω, x) K222(ω, x) # −ω2 "M111 M11I 012 M1I1 M1II + M2II(x) M2I2(x) 021 M22I(x) M222(x) #! × "U 1(x) U I(x) U 2(x) # = "F 1 0I 02 # , (7) "K111 K11I 012 K1I1 K1II + K2II(ω, x) K2I2(ω, x) 021 K22I(ω, x) K222(ω, x) # −ω2 "M111 M11I 012 M1I1 M1II + M2II(x) M2I2(x) 021 M22I(x) M222(x) #! × "U 1(x) U I(x) U 2(x) # = "F 1 0I 02 # , (7) where U(x) is the solution of equation (11). In this last equation, the derivative of the S2(ω, x) matrix needs to be computed. (7) 2.2 Viscoelastic model In this work, the viscoelastic behavior of the dissipa- tive material in Ω2 is modeled using a four-parameter Zener fractional-derivative viscoelastic model, introduced by Bagley and Torvik [20]. For that purpose, the Young modulus is considered as a complex one such as: The effective Young modulus in the element e of Ω2 is expressed as a function of the elemental material densities xe such that: E⋆ eff.(ω, xe) = hEmin E0 + xp e 1 −Emin E0 i E⋆(ω), (4) E⋆(ω) = E′(ω) + iE′′(ω), (1) (1) (4) where E′(ω) is the storage modulus, E′′(ω) the loss mod- ulus and i2 = −1. Both moduli can be expressed as a function of the static stiffness E0, the dynamic stiffness E∞, the relaxation time τ and the fractional order of the derivative α [25], which represent the four parameters of the model: where E⋆(ω) is given by equation (3), Emin = ϵE0 is the imposed lower bound of the effective Young modulus (ϵ < 1) and p is a penalization factor. In the same way, the effective volumetric mass density in the element e of Ω2 is expressed as a function of the elemental material densities xe such that: E′(ω) = E0 + (E0 + E∞)(ωτ)α cos( απ 2 ) + E∞(ωτ)2α 1 + 2(ωτ)α cos( απ 2 ) + (ωτ)2α , E′′(ω) = (E∞−E0)(ωτ)α sin( απ 2 ) 1 + 2(ωτ)α cos( απ 2 ) + (ωτ)2α . (2 ρeff.(xe) = hρmin ρ0 + xm e 1 −ρmin ρ0 i ρ0, (5) (5) where ρ0 is the volumetric mass density of the material used for the design domain Ω2, ρmin = ϵρ0 is the imposed lower bound of the effective volumetric mass density and m is a penalization factor. In this work, the penalization factor p related to the Young modulus is set to 3 and the penalization factor m related to the volumetric mass is set to 1 [5]; the ϵ coefficient is chosen to be 10−3. (2) h (2) Finally, the Young modulus can be expressed using the four above parameters: E⋆(ω) = E0 + E∞(iωτ)α 1 + (iωτ)α . (3) E⋆(ω) = E0 + E∞(iωτ)α 1 + (iωτ)α . (3) S. Burri and A. Legay: Mechanics & Industry 24, 14 (2023) 3 3 2.6 Derivatives of the stiffness and mass matrices of the design domain The derivative of S2(ω, x) with respect to xe involves the derivative of the stiffness and mass matrices of Ω2, denoted respectively by K2(ω, x) and M2(x): where 1, 2 and I denote respectively sub-domain Ω1, sub- domain Ω2 and the interface I between Ω1 and Ω2. The stiffness matrix K2(ω, x) is complex and frequency depen- dent due to the viscoelastic material of sub-domain Ω2 (Eq. (3)). Moreover, this last matrix depends on the design variable vector x. where 1, 2 and I denote respectively sub-domain Ω1, sub- domain Ω2 and the interface I between Ω1 and Ω2. The stiffness matrix K2(ω, x) is complex and frequency depen- dent due to the viscoelastic material of sub-domain Ω2 (Eq. (3)). Moreover, this last matrix depends on the design variable vector x. ∂S2(ω, x) ∂xe = ∂K2(ω, x) ∂xe −ω2 ∂M2(x) ∂xe . (14) (14) The stiffness matrix K2(ω, x) can be expressed as the assembly of the elemental stiffness matrices kj(ω, xj) over the n elements of Ω2-domain mesh: In order to simplify the notations, the following matri- ces are introduced: S1(ω) = "K111 K11I 012 K1I1 K1II 0I2 021 02I 022 # −ω2 "M111 M11I 012 M1I1 M1II 0I2 021 02I 022 # , (8) S2(ω, x) = "011 01I 012 0I1 K2II(ω, x) K2I2(ω, x) 021 K22I(ω, x) K222(ω, x) # −ω2 "011 01I 012 0I1 M2II(x) M2I2(x) 021 M22I(x) M222(x) # , (9) S1(ω) = "K111 K11I 012 K1I1 K1II 0I2 021 02I 022 # −ω2 "M111 M11I 012 M1I1 M1II 0I2 021 02I 022 # , (8) K2(ω, x) = An j=1kj(ω, xj). (15) K2(ω, x) = An j=1kj(ω, xj). (15) (15) The elemental stiffness matrix kj(ω, xj) can be written as kj(ω, xj) = E⋆ eff.(ω, xj)k0 j, (16) S2(ω, x) = "011 01I 012 0I1 K2II(ω, x) K2I2(ω, x) 021 K22I(ω, x) K222(ω, x) # −ω2 "011 01I 012 0I1 M2II(x) M2I2(x) 021 M22I(x) M222(x) # , (9 S2(ω, x) = "011 01I 012 0I1 K2II(ω, x) K2I2(ω, x) 021 K22I(ω, x) K222(ω, x) # −ω2 "011 01I 012 0I1 M2II(x) M2I2(x) 021 M22I(x) M222(x) # , (9) kj(ω, xj) = E⋆ eff.(ω, xj)k0 j, (16) where k0 j is the stiffness matrix of element j computed with a unit Young modulus. 3.2.1 Dynamic case The compliance is given in terms of the complex displace- ment field U(x), and the stiffness and mass matrices of the problem as [30,31] 3 Topology optimization problem formulation c(ω, x) = \|U T (x) S1(ω) + S2(ω, x) U(x)\|, (28) (28) 3.2 Compliance based objective function The elemental stiffness and mass matrices obtained with a unit Young modulus and a unit volumetric mass density are computed once at the beginning of the com- putation and stored in order to be used later on during the topology optimization process. 2.6 Derivatives of the stiffness and mass matrices of the design domain Thus, the derivative of M2(x) with respect to xe is given by ∂M2(x) ∂xe = An j=1 ∂ρeff.(xj) ∂xe m0 j, (21) (21) The iterative process is stopped until either a conver- gence of the objective function is reached, meaning when the difference of the elemental densities between 2 con- secutive iterations is less than a fixed relative criterion (noted as ζ in this work); or when the maximum number of iterations is reached (noted as nmax in this work). with ∂ρeff.(xj) ∂xe = mxm−1 e 1 −ρmin ρ0 ρ0δej. (22) (22) Since m = 1 in this work, the simplification gives Since m = 1 in this work, the simplification gives ∂ρeff.(xj) ∂xe = (ρ0 −ρmin)δej. (23) (23) 2.6 Derivatives of the stiffness and mass matrices of the design domain Thus, the derivative of K2(ω, x) with respect to xe is given by (16) ) where k0 j is the stiffness matrix of element j computed with a unit Young modulus. Thus, the derivative of K2(ω, x) with respect to xe is given by 9) where k0 j is the stiffness matrix of element j computed with a unit Young modulus. Thus, the derivative of K2(ω, x) with respect to xe is given by (9) ∂K2(ω, x) ∂xe = An j=1 ∂E⋆ eff.(ω, xj) ∂xj k0 e, (17) (17) where S1(ω) does not depend on x and represents the con- tribution of Ω1, while S2(ω) depends on x and represents the contribution of the design domain Ω2. with g The global dynamic equation (Eq. (6)) then becomes ∂E⋆ eff.(ω, xe) ∂xj = pxp−1 e 1 −Emin E0 E⋆(ω)δej, (18) S1(ω) + S2(ω, x) U(x) = F. (10) (18) (10) where δej is the Kronecker symbol (no implicit sum- mation). The same development is applied to the mass matrix. The mass matrix M2(x) is expressed as the assem- bly of the elemental mass matrices mj(xj) over the n By denoting S(ω, x) = S1(ω) + S2(ω, x), the equation can be written as S(ω, x)U(x) = F. (11) (11) ( ) bly of the elemental mass matrices mj(xj) over the n S. Burri and A. Legay: Mechanics & Industry 24, 14 (2023) 4 4 elements of Ω2-domain mesh: elements of Ω2-domain mesh: M2(x) = An j=1mj(xj). (19) (19) A density filtering strategy [3] is used to avoid checker- board patterns. This minimization problem is solved using a modified-SIMP algorithm (see Sect. 2.3) associated to a solver called Modified Optimality Criteria (MOC) method [16] which is an extension of the Optimality Criteria (OC) method [14,28]. This algorithm has been chosen thanks to its capability to be adjustable depending on the optimiza- tion problem to solve. Also, it is much more suitable than OC due the modified parameter introduced for dealing with dynamic criteria, and relatively easier to implement and analyze compare to other complex algorithms such as MMA [29]. The elemental mass matrix mj(xj) can be written as The elemental mass matrix mj(xj) can be written as mj(xj) = ρeff.(xj)m0 j, (20) (20) where m0 j is the mass matrix of element j computed with a unit volumetric mass density. 3.2.2 Static case The compliance based objective function in the static case (denoted by cs(x)) can be derived from the previous section by taking ω = 0: This expression can be rewritten as: This expression can be rewritten as: cs(x) = c(0, x) = \|U T (x) S1(0) + S2(0, x) U(x)\|, (36) (x) = c(0, x) = \|U T (x) S1(0) + S2(0, x) U(x)\|, (36) re ∂d(ω, x) ∂xe = −U T (x)∂S2 T (x) ∂xe Y + Y T ∂S2(x) ∂xe U(x), (45) ∂d(ω, x) ∂xe = −U T (x)∂S2 T (x) ∂xe Y + Y T ∂S2(x) ∂xe U(x), (45) where where Y is the solution of the following system: where Y is the solution of the following system: S1(0) = K1, (37) (37) S(x)Y = βU(x). (46) (46) This equation has to be solved for each topology optimiza- tion iterative step, while equation (45) gives the sensitivity of the displacement based objective function: This equation has to be solved for each topology optimiza- tion iterative step, while equation (45) gives the sensitivity of the displacement based objective function: S2(0, x) = K2(0, x). (38) (38) In this case, S(0, x) = S(0, x) since S2(0, x) is a real matrix. The derivative of g(0, x) with respect to xe becomes: ∂d(ω, x) ∂xe = −2ℜ U T (x)∂S2 T (x) ∂xe Y , (47) (47) ∂g(0, x) ∂xe = −U T ∂K2(0, x) ∂xe U(x), (39) ∂g(0, x) ∂xe = −U T ∂K2(0, x) ∂xe U(x), (39) where ℜis the real part of the imaginary number. = −U T ∂K2(0, x) ∂xe U(x), (39) where ℜis the real part of the imaginary number. (39) (39) where ℜis the real part of the imaginary number. 3.3.1 Dynamic case For some applications, it can be relevant to use a local criterion (e.g. when focusing on a targeted component embedded in a the structure Ω1), such as the minimization of the displacement of a local point (or set of points). The following displacement based function given in terms of the displacement field U(x) [32,33] is then introduced: ∂g(ω, x) ∂xe = ∂U T (x) ∂xe S(ω, x)U(x) + U(x)T S(ω, x)∂U(x) ∂xe +U T (x)∂S2(ω, x) ∂xe U(x). (32) (32) Using equation (13), this equation becomes Using equation (13), this equation becomes d(ω, x) = U T (x)βU(x), (41) (41) ∂g(ω, x) ∂xe = −U T (x)∂S2(ω, x) ∂xe S −1(ω, x)S(ω, x)U(x). (33 where β is a diagonal localization matrix which can be defined by (using the Kronecker symbol δij with no implicit summation): where β is a diagonal localization matrix which can be defined by (using the Kronecker symbol δij with no implicit summation): (33) It can be rewritten as: βij = biδij, (42) (42) ∂g(ω, x) ∂xe = −U T ∂S2(ω, x) ∂xe X, (34) (34) The coefficient bi is prescribed by the user (0 or 1) depending on whether one wants to consider the degree of freedom (dof) i or not. The coefficient bi is prescribed by the user (0 or 1) depending on whether one wants to consider the degree of freedom (dof) i or not. where X is the solution of the following system: where X is the solution of the following system: ( ) The derivative of d(ω, x) with respect to xe is S(ω, x)X = S(ω, x)U(x). (35) (35) ∂d(ω, x) ∂xe = ∂U T (x) ∂xe βU(x) + U T (x)β ∂U(x) ∂xe . (43) (43) The system of equation (35) has to be solved for each iteration of the topology optimization loop, but not for each design variable xe. Once the sensitivity of the g(x) function with respect to xe is obtained, the compliance sensitivity is given by equation (31). Using equation (13), this last equation becomes ∂d(ω, x) ∂xe = −U T (x)∂S2 T (ω, x) ∂xe S −1(ω, x)βU(x) −U T (x)βS−1(ω, x)∂S2(ω, x) ∂xe U(x). (44) 3.1 Minimization problem where U T (x) denotes the transpose-conjugate of the U(x) vector and \|.\| denotes the norm of a complex number. By introducing the following g(ω, x) complex function where U T (x) denotes the transpose-conjugate of the U(x) vector and \|.\| denotes the norm of a complex number. By introducing the following g(ω, x) complex function In this work, the aim of the topology optimization is to find the material density set of design variables ˜xω mini- mizing the objective function f(ω, x) for a given angular frequency ω according to constraints: ˜xω = Argmin x f(ω, x), (24) g(ω, x) = U T (x) S1(ω) + S2(ω, x) U(x), (29) (24) g(ω, x) = U T (x) S1(ω) + S2(ω, x) U(x), (29) ˜xω = Argmin x f(ω, x), (24) (29) (24) such that the compliance can be expressed as a function of g(ω, x): 5) the compliance can be expressed as a function of g(ω, x): γv0 ≤0, (25) the compliance can be expressed as a function of g(ω, x): v(x) −γv0 ≤0, (25) 0 ≤xe ≤1, ∀e ∈[1, n], (26) the compliance can be expressed as a function of g(ω, x): c(ω, x) = (g(ω, x)g(ω, x)) 1 2 . (30) v(x) −γv0 ≤0, (25) the compliance can be expressed as a func (25) (26) c(ω, x) = (g(ω, x)g(ω, x)) 1 2 . (30) 0 ≤xe ≤1, ∀e ∈[1, n], (26) c(ω, x) = (g(ω, x)g(ω, x)) 1 2 . (30) (27) The sensitivity of c(ω, x) with respect to xe is then given by: (27) The sensitivity of c(ω, x) with respect to xe is then given by: S(ω, x)U(x) = F, where v(x) is the volume of material in Ω2 associated to the distribution of material density x, v0 is the total volume of Ω2 and γ is the final target of material volume ratio in Ω2, chosen by the user. Two different objective functions are considered in this work. The first function is based on the compliance (denoted by c(ω, x)) while ∂c(ω, x) ∂xe = 1 2(g(ω, x)g(ω, x))−1 2 ∂g(ω, x) ∂xe g(ω, x) +g(ω, x)∂g(ω, x) ∂xe . (31) (31) S. Burri and A. Legay: Mechanics & Industry 24, 14 (2023) 5 This last expression involves the sensitivity of g(ω, x) with respect to xe, which can be computed from equa- tion (29): where E⋆(0) = E0 in the expression of equation (3). 4.1.1 Reinforcement of a 3D cantilever beam This example is adapted from the cantilever beam from [34] and is composed of an elastic plate (Ω1) surrounded by two layers which compose Ω2, as shown in Figure 2. The beam is fully clamped on its left side while a verti- cal load of 100 N is applied on the right side of Ω1. The structure is discretized using 8-node hexahedral elements: 2160 in Ω1 and 8640 in Ω2, leading to a total of 10,800 ele- ments and 37,719 degrees of freedom. The chosen point for the local displacement minimization is point A (Fig. 2), so matrix β only has 3 unit terms corresponding to the 3 dofs associated to the node A. The optimum solution for both objective functions, static compliance cs(x) and displace- ment based function ds(x), are compared. The material for both domains Ω1 and Ω2 is aluminum whose properties are given in Table 1. g The considered initial structure is shown in Figure 5 and is composed of a skeleton (Ω1) standing on its 4 clamped feet. A vertical surface load is applied on the upper surface. Design domain Ω2 is chosen to be the volume resulting from the 6 faces of the cube times the thickness of the skeleton in the third dimension. The struc- ture is discretized using 8-node hexahedral elements: 2808 in Ω1 and 9504 in Ω2, leading to a total number of 12,312 elements and 49,560 degrees of freedom. The material for both domains Ω1 and Ω2 is aluminum whose properties are given in Table 1. The two criteria, compliance cs(x) and displacement based objective functions ds(x), are used. The chosen point for the local displacement minimization is one of the upper corner (point A), so matrix β only has 3 unit terms corresponding to the 3 dofs associated to this node A. The target for the final volume ratio is set to γ = 25% of the total volume of Ω2, the stop criterion ζ is empirically set to 1% and the maximum number of iterations nmax is set to 200. In this case, the target for the final volume ratio is set to γ = 25% of the total volume of Ω2, the stop criterion ζ is empirically set to 2.5% and the maximum number nmax of iterations is set to 100. 4 Applications The application addressed in this section concerns a CubeSat-like structure which is a standard format cre- ated by Puig-Suari and Twiggs [35,36] for the sake of university projects, in order to send small satellites into space. The specifications for the main configuration is called “1U” and is a 10 × 10 × 10 cm cube whose weight can not exceed 1.33 kg. Therefore, it seems appropriate to use topology optimization on an initial CubeSat skeleton in order to optimize the stiffness of the overall structure while controlling the mass of the added material. 3.3.2 Static case and the static compliance sensitivity is then given by The displacement based function in the static case (denoted by ds(x)) can be derived from the previous section by taking ω = 0: ∂cs(x) ∂xe = −U T ∂K2(0, x) ∂xe U(x). (40) (40) ds(x) = d(0, x) = U T (x)βU(x), (48) ssion of equation (3). ds(x) = d(0, x) = U T (x)βU(x), (48) where E⋆(0) = E0 in the expression of equation (3). where E⋆(0) = E0 in the expression of equation (3). where E⋆(0) = E0 in the expression of equation (3). (48) 6 S. Burri and A. Legay: Mechanics & Industry 24, 14 (2023) S. Burri and A. Legay: Mechanics & Industry 24, 14 (2023) 6 S. Burri and A. Legay: Mechanics & Industry 24, 14 (2023) 6 Ω1 Ω2 ⃗F A 30 cm 4 cm 4 cm 2 cm 10 cm Fig. 2. 3D cantilever beam (Ω1) reinforced by a design domain (Ω2). ) ) ) Ω1 Ω2 ⃗F A 30 cm 4 cm 4 cm 2 cm 10 cm Fig. 2. 3D cantilever beam (Ω1) reinforced by a design domain (Ω2). where U(x) is the solution of: (K1 + K2(0, x))U(x) = F. (49) The sensitivity of ds(x) with respect to xe becomes: ∂d(0, x) ∂xe = −2U T (x)∂K2 T (0, x) ∂xe Y (50) where Y is the solution of the following system: K(x)Y = βU(x). (51) Ω1 Ω2 ⃗F A 30 cm 4 cm 4 cm 2 cm 10 cm Fig. 2. 3D cantilever beam (Ω1) reinforced by a design domain (Ω2). where U(x) is the solution of: where U(x) is the solution of: (K1 + K2(0, x))U(x) = F. (49) The sensitivity of ds(x) with respect to xe becomes: The sensitivity of ds(x) with respect to xe becomes: ∂d(0, x) ∂xe = −2U T (x)∂K2 T (0, x) ∂xe Y (50) where Y is the solution of the following system: where Y is the solution of the following system: Fig. 2. 3D cantilever beam (Ω1) reinforced by a design domain (Ω2). K(x)Y = βU(x). (51) (51) 4.1.1 Reinforcement of a 3D cantilever beam The optimal final shapes obtained by minimizing cs(x) and ds(x) are denoted respectively by ˜xc and ˜xd; they are given in Figure 3. One can see that the final shapes are quite similar, with only few differences on a local aspect. In both cases, computations have converged towards a physical shape. On a numerical aspect, the evolution of both objective functions (Fig. 4) show that a convergence is reached around 80 iterations for both criteria. In the final shape for solution ˜xc, there are: The optimal final shapes obtained by minimizing cs(x) and ds(x) are denoted respectively by ˜xc and ˜xd; they are given in Figure 6. One can observe two clearly defined reinforcement shapes depending on the objective func- tion. The evolution of the objective functions are given in Figure 7. It shows that the criteria ζ is reached at around 100 iterations for the displacement objective func- tion while there are spurious oscillations in the material densities for the compliance objective function and the maximum number of iteration nmax is reached. Neverthe- less it does not affect the final shape of the design domain since it concerns only a few elements. Indeed, in the final shape for solution ˜xc, there are: 2140 elements such that xe > 0.99 6420 elements such that xe < 0.01 80 elements such that 0.01 < xe < 0.99 In the final shape for solution ˜xd, there are: 2140 elements such that xe > 0.99 6400 elements such that xe < 0.01 100 elements such that 0.01 < xe < 0.99 2360 elements such that xe > 0.99 The quantity of non-converged elements (material density between 0.01 and 0.99) is about 1% in both cases. 7064 elements such that xe < 0.01 80 elements such that 0.01 < xe < 0.99 S. Burri and A. Legay: Mechanics & Industry 24, 14 (2023) 7 Table 1. Aluminum material properties. Table 1. Aluminum material properties. p p Young Modulus E Poisson coefficient ν Volumetric mass density ρ 70 GPa 0.3 2700 kg m−3 (a) View of the optimal solution ˜xc (b) View of the optimal solution ˜xd (c) Side view of the optimal solution ˜xc (d) Side view of the optimal solution ˜xd Fig. 3. 4.1.1 Reinforcement of a 3D cantilever beam Optimal converged final shapes for the 3D cantilever beam: elements in Ω2 such that xe > 0.99 are in red, elements in Ω1 are in blue. (a) View of the optimal solution ˜xc (b) View of the optimal solution ˜xd (a) View of the optimal solution ˜xc (c) Side view of the optimal solution ˜xc (b) View of the optimal solution ˜xd (a) View of the optimal solution ˜xc (b) View of the optimal solution ˜xd (d) Side view of the optimal solution ˜xd (d) Side view of the optimal solution ˜xd (c) Side view of the optimal solution ˜xc (d) Fig. 3. Optimal converged final shapes for the 3D cantilever beam: elements in Ω2 such that xe > 0.99 are in red, elements in Ω1 are in blue Fig. 3. Optimal converged final shapes for the 3D cantilever beam: elements in Ω2 such that xe > 0.99 are in red, elements in Ω1 are in blue. Fig. 3. Optimal converged final shapes for the 3D cantilever beam: elements in Ω2 such that xe > 0 are in blue. Fig. 4. Evolution of the objective functions cs(xi)/cs(x1) and ds(xi)/ds(x1) over the optimization iterations for the 3D cantilever beam where x1 is the design at the end of the first iteration while xi is the design at iteration i. Fig. 4. Evolution of the objective functions cs(xi)/cs(x1) and ds(xi)/ds(x1) over the optimization iterations for the 3D cantilever beam where x1 is the design at the end of the first iteration while xi is the design at iteration i. S. Burri and A. Legay: Mechanics & Industry 24, 14 (2023) 8 8 1 cm ⃗F Ω1 Ω2 10 cm 10 cm 10 cm 1 cm 1 cm 8 cm 8 cm 7 cm 1 cm A Fig. 5. CubeSat-like structure (Ω1) reinforced by a design domain (Ω2). Note that half of Ω2 volume is represented with a translation from Ω1 in order to see its shape. 1 cm ⃗F Ω1 10 cm 10 cm 10 cm 1 cm 1 cm A 10 cm 10 cm Ω2 8 cm 8 cm 7 cm 1 cm Ω1 Ω2 Fig. 5. CubeSat-like structure (Ω1) reinforced by a design domain (Ω2). Note that half of Ω2 volume is represented with a translation from Ω1 in order to see its shape. Harmonic force Harmonic force Fig. 8. Optimization of an internal viscoelastic layer (Ω2) of a 2D sandwich beam composed of 2 external aluminum layers (Ω1). n internal viscoelastic layer (Ω2) of a 2D sandwich beam composed of 2 external aluminum layers (Ω Fig. 8. Optimization of an internal viscoelastic layer (Ω2) of a 2D sandwich beam composed of 2 ex In the final shape for solution ˜xd, there are: the length direction of the beam, there are 198 elements. There is a total of 1386 elements in Ω2. The target final volume ratio is γ = 50% of the total volume of Ω2, the stop criterion ζ is empirically set to 1% and the maximum number of iterations nmax is set to 100. 2368 elements such that xe > 0.99 7072 elements such that xe < 0.01 64 elements such that 0.01 < xe < 0.99 2368 elements such that xe > 0.99 7072 elements such that xe < 0.01 64 elements such that 0.01 < xe < 0.99 64 elements such that 0.01 < xe < 0.99 Firstly, the viscoelastic layer is optimized using the compliance objective function c(ω, x) at three different harmonic excitations ω (500 Hz, 1000 Hz and 2000 Hz). The solutions in terms of material densities are denoted respectively by ˜xc 500 Hz, ˜xc 1000 Hz and ˜xc 2000 Hz. These optimal material densities are plotted in Figure 9. The conver- gence of the compliance over the iterations is plotted in Figure 10. It can be seen that for the three harmonic excitations, there are oscillations of the objective func- tion and the optimization process is stopped at a fixed number of iteration (here nmax=100 iterations). How- ever, these spurious oscillations concern a few elements and do not affect the final solution (Tab. 3). The fre- quency response functions of the beam are then computed using these three material densities. The compliance func- tions c(ω, ˜xc 500 Hz), c(ω, ˜xc 1000 Hz) as well as c(ω, ˜xc 2000 Hz) are shown in Figure 11. The black curve is obtained using a fully filled layer of viscoelastic material (xe = 1, ∀e). It can be seen that for a given angular frequency, a The quantity of non-converged elements (material density between 0.01 and 0.99) is less than 1% in both cases. 4.1.1 Reinforcement of a 3D cantilever beam (a) Overall view of the optimal solution ˜xc (b) Overall view of the optimal solution ˜xd (a) Overall view of the optimal solution ˜xc (b) Overall view of the optimal solution ˜xd (c) Side view of the optimal solution ˜xc (d) Side view of the optimal solution ˜xd Fig. 6. Optimal converged final shapes for the 3D CubeSat-like structure: elements in Ω2 such that xe > 0.99 are in red, elements in Ω1 are in blue. (a) Overall view of the optimal solution ˜xc (b) Overall view of the optimal solution ˜xd (a) Overall view of the optimal solution ˜xc (c) Side view of the optimal solution ˜xc (b) Overall view of the optimal solution x (d) Side view of the optimal solution ˜xd (d) Side view of the optimal solution ˜xd (c) Side view of the optimal solution ˜xc Fig. 6. Optimal converged final shapes for the 3D CubeSat-like structure: elements in Ω2 such that xe > 0.99 are in red, elements i Ω i bl Fig. 6. Optimal converged final shapes for the 3D CubeSat-like structure: elements in Ω2 such that xe > 0.99 are in red, elements in Ω1 are in blue. Fig. 6. Optimal converged final shapes for the 3D CubeSat-like structure: elements in Ω2 such that xe > 0.99 are in red, elements in Ω1 are in blue. S. Burri and A. Legay: Mechanics & Industry 24, 14 (2023) 9 Fig. 7. Evolution of the objective functions cs(xi)/cs(x1) and ds(xi)/ds(x1) over the optimization iterations for the 3D CubeSat-like structure where x1 is the design at the end of the first iteration while xi is the design at iteration i. Fig. 7. Evolution of the objective functions cs(xi)/cs(x1) and ds(xi)/ds(x1) over the optimization iterations for the 3D CubeSat-like structure where x1 is the design at the end of the first iteration while xi is the design at iteration i. e = 1 mm e = 1 mm h = 1 mm Ω1 Ω1 Ω2 L = 0.1 m A Harmonic force Fig. 8. Optimization of an internal viscoelastic layer (Ω2) of a 2D sandwich beam composed of 2 external aluminum layers (Ω1). 4.2.1 Optimization of a visoelastic layer for a 2D sandwich beam Optimal solutions of the visco-elastic layer of the 2D sandwich beam (black: xe > 0.99) for bottom to top: ˜xc 500 Hz, ˜xc 1000 Hz, ˜xc 2000 Hz. g p y ( ) p , bottom to top: ˜xc 500 Hz, ˜xc 1000 Hz, ˜xc 2000 Hz. Fig. 10. Dynamic compliance c(x) evolution over the optimization iterations for the 2D sandwich beam and for the three different harmonic excitations: c(500 Hz, xi), c(1000 Hz, xi) and c(2000 Hz, xi). bottom to top: x500 Hz, x1000 Hz, x2000 Hz. Fig. 10. Dynamic compliance c(x) evolution over the optimization iterations for the 2D sandwich beam and for the three different harmonic excitations: c(500 Hz, xi), c(1000 Hz, xi) and c(2000 Hz, xi). Fig. 10. Dynamic compliance c(x) evolution over the optimization iterations for the 2D sandwich beam and for the three different harmonic excitations: c(500 Hz, xi), c(1000 Hz, xi) and c(2000 Hz, xi). Table 3. Number of elements in Ω2 in terms of their final material densities for the 2D sandwich beam. The same conclusions can be done with solution ˜xc 1000 Hz at the frequency 1000 Hz; and with solution ˜xc 2000 Hz at the frequency 2000 Hz. xe > 0.99 xe < 0.01 0.01 < xe < 0.99 ˜xc 500 Hz 623 754 9 ˜xc 1000 Hz 675 625 86 ˜xc 2000 Hz 676 617 93 ˜xd 500 Hz 693 693 0 ˜xd 1000 Hz 677 622 87 ˜xd 2000 Hz 686 663 37 Secondly, the viscoelastic layer is optimized using the displacement objective function d(ω, x) at the same three different harmonic excitations ω (500 Hz, 1000 Hz and 2000 Hz). The solutions in terms of material densities are denoted respectively by ˜xd 500 Hz, ˜xd 1000 Hz and ˜xd 2000 Hz. These optimal material densities are plotted in Figure 12. The convergence of the displacement objective function over the iterations is plotted in Figure 13. As for the compliance criteria, there are spurious oscillations for two harmonic excitations (1000 Hz and 2000 Hz) (Table 3). The frequency response functions of the beam are then computed using these three material densities. The dis- placement functions d(ω, ˜xd 500 Hz), d(ω, ˜xd 1000 Hz) as well as d(ω, ˜xd 2000 Hz) are shown in Figure 14. The black curve is obtained using a fully filled layer of viscoelastic mate- rial (xe = 1, ∀e). 4.2.1 Optimization of a visoelastic layer for a 2D sandwich beam 4.2.1 Optimization of a visoelastic layer for a 2D sandwich beam The problem is described in Figure 8. It is assumed a plane stress state. The sandwich beam is composed of 2 exter- nal aluminum layers (domain Ω1) and a viscoelastic layer (domain Ω2) in between them. The material properties are given in Table 1 (for aluminum) and 2 (for the vis- coelastic layer). The left side of the beam is clamped while an harmonic force (±10 N) is imposed at the right side at the angular velocity ω. The structure is discretized using quadrangular 4-node elements. There are 6 elements in the thickness direction of the aluminum layers and 7 elements in the thickness direction of the viscoelastic layer. Along 10 S. Burri and A. Legay: Mechanics & Industry 24, 14 (2023) Table 2. Viscoelastic material properties. E0 E∞ ν ρ α τ 4.2 MPa 1.62 GPa 0.45 1460 kg m−3 0.59 0.52×10−6 s Fig. 9. Optimal solutions of the visco-elastic layer of the 2D sandwich beam (black: xe > 0.99) for the compliance criteria, from bottom to top: ˜xc 500 Hz, ˜xc 1000 Hz, ˜xc 2000 Hz. S. Burri and A. Legay: Mechanics & Industry 24, 14 (2023) 10 Table 2. Viscoelastic material properties. E0 E∞ ν ρ α τ 4.2 MPa 1.62 GPa 0.45 1460 kg m−3 0.59 0.52×10−6 s Fig. 9. Optimal solutions of the visco-elastic layer of the 2D sandwich beam (black: xe > 0.99) for the compliance criteria, from bottom to top: ˜xc 500 Hz, ˜xc 1000 Hz, ˜xc 2000 Hz. Fig. 10. Dynamic compliance c(x) evolution over the optimization iterations for the 2D sandwich beam and for the three different harmonic excitations: c(500 Hz, xi), c(1000 Hz, xi) and c(2000 Hz, xi). Table 2. Viscoelastic material properties. E0 E∞ ν ρ α τ 4.2 MPa 1.62 GPa 0.45 1460 kg m−3 0.59 0.52×10−6 s Fig. 9. Optimal solutions of the visco-elastic layer of the 2D sandwich beam (black: xe > 0.99) for the compliance criteria, from bottom to top: ˜xc 500 Hz, ˜xc 1000 Hz, ˜xc 2000 Hz. Fig. 9. Optimal solutions of the visco-elastic layer of the 2D sandwich beam (black: xe > 0.99) for the compliance criteria, from bottom to top: ˜xc ˜xc ˜xc Fig. 9. 4.2.1 Optimization of a visoelastic layer for a 2D sandwich beam As for the compliance based criteria, solution can be better than an other in terms of mini- mizing the compliance. For instance, the material density solution ˜xc 500 Hz given by the minimization of the compli- ance at 500 Hz exhibits the minimum of compliance at 500 Hz compared to the other solutions, even the solu- tion computed with a full layer of viscoelastic material. solution can be better than an other in terms of mini- mizing the compliance. For instance, the material density solution ˜xc 500 Hz given by the minimization of the compli- ance at 500 Hz exhibits the minimum of compliance at 500 Hz compared to the other solutions, even the solu- tion computed with a full layer of viscoelastic material. S. Burri and A. Legay: Mechanics & Industry 24, 14 (2023) 11 Fig. 11. Frequency response functions of the 2D sandwich beam in terms of dynamic compliance: c(ω, ˜xc 500 Hz), c(ω, ˜xc 1000 Hz) and c(ω, ˜xc 2000 Hz). Fig. 11. Frequency response functions of the 2D sandwich beam in terms of dynamic compliance: c(ω, ˜xc 500 Hz), c(ω, ˜xc 1000 Hz) and Fig. 11. Frequency response functions of the 2D sandwich beam in terms of dynamic compliance: c(ω, ˜xc 500 Hz), c(ω, ˜xc 1000 Hz) and c(ω, ˜xc 2000 Hz). Fig. 11. Frequency response functions of the 2D sandwich beam in terms of dynamic compliance: c(ω, ˜xc 500 Hz), c(ω, ˜xc 1000 Hz) and c(ω, ˜xc 2000 Hz). Fig. 12. Optimal solutions of the visco-elastic layer of the 2D sandwich beam (black: xe > 0.99) for the displacement criteria, from bottom to top: ˜xd 500 Hz, ˜xd 1000 Hz, ˜xd 2000 Hz. bottom to top: ˜xd 500 Hz, ˜xd 1000 Hz, ˜xd 2000 Hz. Fig. 13. Displacement objective function d(x) evolution over the optimization iterations for the 2D sandwich beam and for the three different harmonic excitations: d(500 Hz, xi), d(1000 Hz, xi) and d(2000 Hz, xi). Fig. 13. Displacement objective function d(x) evolution over the optimization iterations for the 2D sandwich beam and for the three different harmonic excitations: d(500 Hz, xi), d(1000 Hz, xi) and d(2000 Hz, xi). S. Burri and A. Legay: Mechanics & Industry 24, 14 (2023) 12 Fig. 14. Frequency response functions of the 2D sandwich beam in terms of displacement objective function: d(ω, ˜xd 500 Hz), d(ω, ˜xd 1000 Hz) and d(ω, ˜xd 2000 Hz). Fig. 14. 4.2.1 Optimization of a visoelastic layer for a 2D sandwich beam Optimal solutions of the viscoelastic layer of the 2D CubeSat-like structure (black: xe > 0.99) for the compliance criteria, from left to right: ˜xc 50 Hz, ˜xc 100 Hz, ˜xc 200 Hz. 4.2.1 Optimization of a visoelastic layer for a 2D sandwich beam Frequency response functions of the 2D sandwich beam in terms of displacement objective function: d(ω, ˜xd 500 Hz), d(ω, ˜xd 1000 Hz) and d(ω, ˜xd 2000 Hz). Ω1 Ω2 Ω2 Ω2 ⃗F 45o A 96 cm 96 cm 19 cm 1 cm Fig. 15. Optimization of an internal viscoelastic layer (Ω2) for a 2D CubeSat-like structure (Ω1). 1 cm Fig. 15. Optimization of an internal viscoelastic layer (Ω2) for a 2D CubeSat-like structure (Ω1). true; the topology optimization process can not give a bet- ter solution than the one given by a fully filled layer of viscoelastic material. Nevertheless, the solution given by the optimization is lighter. the same conclusions can be made: the density optimized solution obtained for a given frequency gives the minimum of the displacement based function at this frequency. One can notice that in this case, for 500 Hz, it is not completely S. Burri and A. Legay: Mechanics & Industry 24, 14 (2023) 13 (a) Solution ˜xc 50 Hz at 50 Hz (b) Solution ˜xc 100 Hz at 100 Hz (c) Solution ˜xc 200 Hz at 200 Hz Fig. 16. Deformed structure for the three target frequencies and for the compliance objective function (scale factor of the deformed mesh is 5000; black correspond to no displacement, yellow to maximum). (a) Solution ˜xc 50 Hz at 50 Hz (b) Solution ˜xc 100 Hz at 100 Hz (c) Solution ˜xc 200 Hz at 200 Hz Fig. 16. Deformed structure for the three target frequencies and for the compliance objective function (scale factor of the deformed mesh is 5000; black correspond to no displacement, yellow to maximum). (b) Solution ˜xc 100 Hz at 100 Hz (c) Solution ˜xc 200 Hz at 200 Hz Fig. 16. Deformed structure for the three target frequencies and for the compliance objective function (scale factor of the deformed mesh is 5000; black correspond to no displacement, yellow to maximum). Fig. 16. Deformed structure for the three target frequencies and for the compliance objective function (scale factor of the deformed mesh is 5000; black correspond to no displacement, yellow to maximum). Fig. 17. Optimal solutions of the viscoelastic layer of the 2D CubeSat-like structure (black: xe > 0.99) for the compliance criteria, from left to right: ˜xc 50 Hz, ˜xc 100 Hz, ˜xc 200 Hz. Fig. 17. 4.2.2 Optimization of a viscoelastic layer for a 2D CubeSat-like structure Firstly, the viscoelastic layer is optimized using the compliance objective function c(ω, x) at three different harmonic excitations ω (50 Hz, 100 Hz and 200 Hz). The solutions in terms of material densities are denoted respectively by ˜xc 50 Hz, ˜xc 100 Hz and ˜xc 200 Hz. These optimal material densities are plotted in Figure 17. The conver- gence of the compliance over the iterations is plotted in Figure 18. As for the 2D sandwich beam, it can be seen that for the three harmonic excitations, there are oscillations of the objective function and the optimiza- tion process is stopped at a fixed number of iteration (here 100 iterations). However, these spurious oscillations concern a few elements and do not affect the final solu- tion (Table 4). These oscillations are more important for ˜xc 50 Hz, nevertheless if only elements with xe > 0.99 are considered the solution is acceptable. Moreover, for this solution, there are 369 elements with a material density less than 0.1; and so only 128 elements between 0.1 and 0.99. The frequency response functions of the structure The problem is described in Figure 15. It is assumed a plane stress state. The structure is composed of external aluminum layers (domain Ω1) and internal viscoelastic layers (domain Ω2). Its shape is the same as the 3D CubeSat-like structure studied in the previous section. The material properties are given in Table 1 (for alu- minum) and 2 (for the viscoelastic layers). The target final volume ratio is γ = 50% of the total volume of Ω2, the stop criterion ζ is empirically set to 1% and the maxi- mum number of iterations nmax is set to 100. The bottom side of the structure is clamped while an harmonic force ⃗F is imposed on the upper left corner at the angular velocity ω (∥⃗F ∥= 1 N). (∥ ∥ ) The structure is discretized with a total of 3222 quad- rangular 4-node elements, the number of elements in Ω2 is 966. The mesh can be seen in Figure 16. S. Burri and A. Legay: Mechanics & Industry 24, 14 (2023) 14 Fig. 18. Dynamic compliance c(x) evolution over the optimization iterations for the 2D CubeSat-like structure and for the three different harmonic excitations: c(50 Hz, xi), c(100 Hz, xi) and c(200 Hz, xi). Fig. 18. 4.2.2 Optimization of a viscoelastic layer for a 2D CubeSat-like structure Dynamic compliance c(x) evolution over the optimization iterations for the 2D CubeSat-like structure and for the three different harmonic excitations: c(50 Hz, xi), c(100 Hz, xi) and c(200 Hz, xi). Table 4. Number of elements in Ω2 in terms of their final material densities for the 2D Cu elements in Ω2 in terms of their final material densities for the 2D CubeSat-like structure. of elements in Ω2 in terms of their final material densities for the 2D CubeSat-like structu xe > 0.99 xe < 0.01 0.01 < xe < 0.99 ˜xc 50 Hz 469 339 158 ˜xc 100 Hz 499 411 56 ˜xc 200 Hz 508 432 26 ˜xd 50 Hz 450 0 516 ˜xd 100 Hz 499 414 53 ˜xd 200 Hz 509 435 22 Fig. 19. Frequency response functions of the 2D CubeSat-like structure in terms of dynamic compliance: c(ω, ˜xc 50 Hz), c(ω, ˜xc 100 Hz) and c(ω, ˜xc 200 Hz). Fig. 19. Frequency response functions of the 2D CubeSat-like structure in terms of dynamic compliance: c(ω, ˜xc 50 Hz), c(ω, ˜xc 100 Hz) and c(ω, ˜xc 200 Hz). S. Burri and A. Legay: Mechanics & Industry 24, 14 (2023) 15 Fig. 20. Optimal solutions of the viscoelastic layer of the 2D CubeSat-like structure (black: xe > 0.99) for the displacement criteria, from left to right: ˜xd 50 Hz, ˜xd 100 Hz, ˜xd 200 Hz. Fig. 20. Optimal solutions of the viscoelastic layer of the 2D CubeSat-like structure (black: xe > 0.99) for the displacement criteria, from left to right: ˜xd 50 Hz, ˜xd 100 Hz, ˜xd 200 Hz. Fig. 20. Optimal solutions of the viscoelastic layer of the 2D CubeSat-like structure (black: xe > 0.99) for the displacement criteria, from left to right: ˜xd 50 Hz, ˜xd 100 Hz, ˜xd 200 Hz. Fig. 21. Displacement objective function evolution over the optimization iterations for the 2D CubeSat-like structure and for the three different harmonic excitations: d(50 Hz, xi), d(100 Hz, xi) and d(200 Hz, xi). Fig. 21. Displacement objective function evolution over the optimization iterations for the 2D CubeSat-like structure and for the three different harmonic excitations: d(50 Hz, xi), d(100 Hz, xi) and d(200 Hz, xi). Secondly, the viscoelastic layer is optimized using the displacement objective function d(ω, x) at the same three different harmonic excitations ω (50 Hz, 100 Hz and 200 Hz). 4.2.2 Optimization of a viscoelastic layer for a 2D CubeSat-like structure The solutions in terms of material densities are denoted respectively by ˜xd 50 Hz, ˜xd 100 Hz and ˜xd 200 Hz. These optimal material densities are plotted in Figure 20. The convergence of the displacement criteria over the iterations is plotted in Figure 21. As previously, two of the three optimizations have spurious oscillations in terms of objective function over the iteration process (Table 4). These oscillations are more important for ˜xd 50 Hz: for this case, there are no elements with xe < 0.01 but there are 362 elements with a material den- sity less than 0.1; and so only 154 elements between 0.1 and 0.99. In order to illustrate the oscillations in are then computed using these three material densities. The compliance functions c(ω, ˜xc 50 Hz), c(ω, ˜xc 100 Hz) as well as c(ω, ˜xc 200 Hz) are shown in Figure 19. The black curve is obtained using a fully filled layer of viscoelastic mate- rial (xe = 1, ∀e). As for the 2D sandwich beam, it can be seen that for a given angular frequency, a solution can be better than another one in terms of minimizing the compliance. For instance, the material density solu- tion ˜xc 50 Hz given by the minimization of the compliance at 50 Hz exhibits the minimum of compliance at 50 Hz compared to the other solutions, even the solution com- puted with a full layer of viscoelastic material. The same conclusions can be done with solution ˜xc 100 Hz at the fre- quency 100 Hz; and with solution ˜xc 200 Hz at the frequency 200 Hz. are then computed using these three material densities. The compliance functions c(ω, ˜xc 50 Hz), c(ω, ˜xc 100 Hz) as well as c(ω, ˜xc 200 Hz) are shown in Figure 19. The black curve is obtained using a fully filled layer of viscoelastic mate- rial (xe = 1, ∀e). As for the 2D sandwich beam, it can be seen that for a given angular frequency, a solution can be better than another one in terms of minimizing the compliance. For instance, the material density solu- tion ˜xc 50 Hz given by the minimization of the compliance at 50 Hz exhibits the minimum of compliance at 50 Hz compared to the other solutions, even the solution com- puted with a full layer of viscoelastic material. 4.2.2 Optimization of a viscoelastic layer for a 2D CubeSat-like structure The same conclusions can be done with solution ˜xc 100 Hz at the fre- quency 100 Hz; and with solution ˜xc 200 Hz at the frequency 200 Hz. S. Burri and A. Legay: Mechanics & Industry 24, 14 (2023) 16 Fi 22 L t 3 it ti f th l ti ˜d f th 2D C b S t lik t t f l ft t i ht it ti 98 99 d 100 ( d Fig. 22. Last 3 iterations for the solution ˜xd 50 Hz of the 2D CubeSat-like structure, from left to right: iterations 98, 99, and 100 (red: xe = 1, blue: xe = 0). Fig. 22. Last 3 iterations for the solution ˜xd 50 Hz of the 2D CubeSat-like structure, from left to right: iterations 98, 99, and 100 (red: xe = 1, blue: xe = 0). Fig. 22. Last 3 iterations for the solution ˜xd 50 Hz of the 2D CubeSat-like structure, from left to right: iterations 98, 99, and 100 (red: x 1 blue: x 0) Fig. 23. Frequency response functions of the 2D CubeSat-like structure in terms of displacement amplitude: d(ω ˜xd 50 H ) d(ω ˜xd 100 H ) Fig. 23. Frequency response functions of the 2D CubeSat-like structure in terms of displacement amplitude: d(ω, ˜xd 50 Hz), d(ω, ˜xd 100 Hz) and d(ω, ˜xd 200 Hz). Fig. 23. Frequency response functions of the 2D CubeSat-like structure in terms of displacement amplitude: d(ω, ˜xd 50 Hz), d(ω, ˜xd 100 Hz) and d(ω, ˜xd 200 Hz). of viscoelastic material. Nevertheless, the solution given by the optimization is lighter. the solution, the last three iterations are plotted on Figure 22. Nevertheless if only elements with xe > 0.99 are considered the solution is acceptable. The frequency response functions of the structure are then computed using these three material densities. The displacement functions d(ω, ˜xd 50 Hz), d(ω, ˜xd 100 Hz) as well as d(ω, ˜xd 200 Hz) are shown in Figure 23. The black curve is obtained using a fully filled layer of viscoelastic material (xe = 1, ∀e). As for the compliance based criteria, the same conclusions can be made: the density optimized solution obtained for a given frequency gives the minimum of the displacement based function at this frequency. 5 Conclusion The aim of this work is to use a topology optimiza- tion process to design a domain Ω2 interacting through an interface with a fixed given structure Ω1 in order to improve the behavior of the initial structure according to a given criteria. The problem can be static or dynamic. In the dynamic case, the problem is treated in the frequency domain and the used material in the design domain has viscoelastic properties able to reduce the vibration amplitude. This viscoelastic material is modeled using However, sometimes, for this application, the topol- ogy optimization process does not improve so much the solution compared to the one given by a fully filled layer S. Burri and A. Legay: Mechanics & Industry 24, 14 (2023) 17 a four-parameter Zener fractional-derivative viscoelastic model. [2] K. Suzuki, N. Kikuchi, A homogenization method for shape and topology optimization, Comput. Methods Appl. Mech. Eng. 93, 291–318 (1991) Two objective functions are introduced in this paper: a global compliance based one and a local displacement based one. They are both defined for the dynamic as well as for the static case. Results show that the optimum solutions obtained for the two objective function are very often quite similar. [3] O. Sigmund, A 99 line topology optimization code written in Matlab, Struct. Multidiscipl. Optim. 21, 120–127 (2001) [4] M. Stolpe, K. Svanberg, An alternative interpolation scheme for minimum compliance topology optimization, Struct. Multidiscipl. Optim. 22, 116–124 (2001) [5] D. Tchierniak, Topology optimization of resonating struc- tures using SIMP method, Int. J. Numer. Methods Eng. 54, 1605–1622 (2002) The formulation of the problems allows to have a com- plex shape of the additional design, even a design domain separated into several volumes, as it is the case in a few presented applications. Moreover, the chosen material for the design domain can be different than the one used for the initial structure. The shape and the material of the design domain have simply to be coherent with func- tional surfaces of the system and with the manufacturing process. [6] A. Takezawa, M. Daifuku, Y. Nakano, K. Nakagawa, T. Yamamoto, M. Kitamuraa, Topology optimization of damping material for reducing resonance response based on complex dynamic compliance, J. Sound Vibr. 365, 230–243 (2016) [7] R.B. Haber, C.S. Jog, M.P. Bendsøe, A new approach to variable-topology shape design using a constraint on perimeter, Struct. Optim. 5 Conclusion 11, 1-12 (1996) The applications in the static case show that it is possible to improve a given structure by adding mate- rial around it using a topology optimization process. The strategy may be applied to a damaged structure needed to be repaired. Additive manufacturing may be used to manufacture the domain Ω2 in order to reinforce Ω1. [8] L. Shu, M.Y. Wang, Z. Fang, Z. Ma, P. Wei, Level set based structural topology optimization for minimiz- ing frequency response, J. Sound Vibr. 330, 5820–5834 (2011) ( ) [9] J. Cao, K. Cai, P. Fei Wang, D. Yan, J. Shi, Multiple materials layout optimization in a layered structure, Mech. Ind. 17, 404 (2016) The examples in the dynamic case show that the proposed strategy enables to design additive viscoelas- tic layers to damp the vibrations at a given frequency. The proposed examples deal with sandwich structures for which the viscoelastic layer in between two layers of metal since it is known to be more efficient than on the external surface of the structure. The final shape of the viscoelastic layer drastically depends on the target frequency. [10] O.M. Silva, M.M. Neves, A. Lenzi, A critical analysis of using the dynamic compliance as objective func- tion in topology optimization of one-material structures considering steady-state forced vibration problems, J. Sound Vibr. 444, 1–20 (2019) ( ) [11] S. Burri, Contributions `a l’optimisation topologique de liaisons amortissantes pour des applications spatiales. Th`ese de doctorat. CNAM, HESAM Universit´e (2020). https://tel.archives-ouvertes.fr/tel-03179845 Nevertheless, especially in the context of dynamics and viscoelasticity, the topology optimization process often reaches the maximum number of iterations, exhibiting spurious oscillations of the material densities between few elements. However, these oscillations concern a few elements, and do not affect the quality of the final solution. [12] M.P. Bendsøe, Optimal shape design as a material distri- bution problem, Struct. Optim. 1, 193-202 (1989) [13] N. Zhou, G.I.N. Rozvany, The COC algorithm. Part II: Topological, geometrical and generalized shape optimiza- tion. Comput. Methods Appl. Mech. Eng. 89, 309–336 (1991) Conclusions show that it is possible to deal with two different substructures with the aim of controlling the first one using the second one. The way it is implemented here allows for a great variety of configurations for future users. [14] M.P. Bendsøe, O. Sigmund, Topology Optimization – Theory, Methods and Applications (Springer, Berlin, Heidelberg, 2004) [15] A. Diaz, N. 5 Conclusion Kikuchi, Solutions to shape and topology eigenvalue optimization problems using a homogenization method. Int. J. Numer. Methods Eng. 35, 1487–1502 (1992) Funding information [16] Z.-D. Ma, N. Kikuchi, I. Hagiwara, Structural topology and shape optimization for a frequency response problem. Comput. Mech. 13, 157–174 (1993) ArianeGroup is gratefully acknowledged for its financial support. [17] Z.-D. Ma, H.-S. Cheng, N. Kikuchi, Structural design for obtaining desired eigenfrequencies by using the shape and topology optimization method. Comput. Syst. Eng. 5, 77– 89 (1994) Declaration of conflicting interests The authors declare no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. [18] M. Bruggi, A. Talierci, Topology optimization of the fiber- reinforcement retrofitting existing structures, Int. J. Solids Struct. 50, 121–136 (2013) ( ) [19] J.-P. Kruth, M.C. Leu, T. Nakagawa, Progress in addi- tive manufacturing and rapid prototyping, CIRP Ann. 47, 525–540 (1998) References [1] F. Rigaud, M. Charlotte, C. Kerdreux, P. Marechal, Multi- objective optimization of rotary-wing aircrafts at the predesign stage, Mech. Ind. 15, 267–277 (2014) [20] R.L Bagley, P. Torvik, Fractional calculus – a differ- ent approach to the analysis of viscoelastically damped structures, AIAA J. 21, 741–748 (1983) S. Burri and A. Legay: Mechanics & Industry 24, 14 (2023) 18 [21] A. Germant, XLV. On fractional differentials, Philos. Mag. Ser. 1 25, 540–549 (1938) [29] K. Svanberg, The method of moving asymptotes – a new method for structural optimization. Int. J. Numer. Methods Eng. 24, 359–373 (1987) [22] B. Morin, A. Legay, J.-F. De¨u, Reduced order models for dynamic behavior of elastomer damping devices, Finite Elem. Anal. Des. 143, 66–75 (2018) [30] Z.-D. Ma, N. Kikuchi, I. Hagiwara, Structural topology and shape optimization for a frequency response problem, Comput. Mech. 13, 157–174 (1993) [23] L. Rouleau, A. Legay, J.-F. De¨u, Mod´elisation vibro-acoustique de structures sandwich munies de mat´eriaux visco´elastiques. Th`ese de Doctorat CNAM (2013) [31] G.H. Yoon, Structural topology optimization for fre- quency response problem using model reduction schemes, Comput. Methods Appl. Mech. Eng. 199, 1744–1763 (2010) [24] C. Geuzaine, J.-F. Remacle, Gmsh: a 3-D finite element mesh generator with built-in pre- and post-processing facilities. Int. J. Numer. Methods Eng. 79, 1309–1331 (2019) [32] K.-S. Yun, S.-K. Youn, Topology optimization of viscoelas- tic damping layers for attenuating transient response of shell structures, Finite Element Anal. Des. 141, 154–165 (2018) [25] A.C. Galucio, J.-F. De¨u, R. Ohayon, Finite element formulation of viscoelastic sandwich beams using frac- tional derivative operators. Comput. Mech. 33, 282–291 (2004) [33] C.S. Jog, Topology design of structures subjected to periodic loading, J. Sound Vibr. 253, 687–709 (2002) [34] K. Liu, A. Tovar, An efficient 3D topology optimization code written in Matlab, Struct. Multidiscipl. Optim. 50, 1175–1196 (2014) ( ) [26] M.P. Bendsøe, N. Kikuchi, Generating optimal topolo- gies in structural design using a homogenization method, Comput. Methods Appl. Mech. Eng. 71, 197–224 (1988) ( ) [35] H. Heidt, J. Puig-Suari, A.S. Moore, S. Nakasuka, R.J. Twiggs, CubeSat: a new generation of picosatellite for education and industry low-cost space experimentation, in 14th Annual AIAA/USU Conference on Small Satellites (2000) [27] O. Sigmund, Morphology-based black and white filters for topology optimization. Struct. Multidiscipl. Optim. 33, 401–424 (2007) [28] G.I.N. Rozvany, Structural Design via Optimality Criteria: The Prager Approach to Structural Optimization. (Kluwer Academic Publishers, Springer Dordrecht, 1989) [36] R. Cite this article as: S. Burri and A. Legay, Static Reinforcement and Vibration Reduction of Structures using Topology Optimization, Mechanics & Industry 24, 14 (2023) References Hevner, W. Holemans, J. Puig-Suari, R.J. Twiggs, An advanced standard for CubeSats, in 25th Annual AIAA/USU Conference on Small Satellites (2011)
https://openalex.org/W2943466042	https://europepmc.org/articles/pmc6566403?pdf=render	English	null	A Preterm Rat Model for Immunonutritional Studies	Nutrients	2,019	cc-by	13,982	Received: 5 March 2019; Accepted: 29 April 2019; Published: 1 May 2019 Abstract: Neonates are born with an immature immune system, which develops during the ﬁrst stages of life. This early immaturity is more acute in preterm newborns. The aim of the present study was to set up a preterm rat model, in which representative biomarkers of innate and adaptive immunity maturation that could be promoted by certain dietary interventions are established. Throughout the study, the body weight was registered. To evaluate the functionality of the intestinal epithelial barrier, in vivo permeability to dextrans was measured and a histomorphometric study was performed. Furthermore, the blood cell count, phagocytic activity of blood leukocytes and plasmatic immunoglobulins (Ig) were determined. Preterm rats showed lower erythrocyte and platelet concentration but a higher count of leukocytes than the term rats. Although there were no changes in the granulocytes’ ability to phagocytize, preterm monocytes had lower phagocytic activity. Moreover, lower plasma IgG and IgM concentrations were detected in preterm rats compared to full-term rats, without aﬀecting IgA. Finally, the intestinal study revealed lower permeability in preterm rats and reduced goblet cell size. Here, we characterized a premature rat model, with diﬀerential immune system biomarkers, as a useful tool for immunonutritional studies aimed at boosting the development of the immune system. Keywords: prematurity; suckling rat; intestinal permeability; phagocytosis; plasmatic immunoglobulin; adaptive immunity; innate immunity A Preterm Rat Model for Immunonutritional Studies Blanca Grases-Pintó 1,2 , Paulina Torres-Castro 1,2, Mar Abril-Gil 1,2 , Margarida Castell 1,2 , María J. Rodríguez-Lagunas 1,2 , Francisco J. Pérez-Cano 1,2,* and Àngels Franch 1,2 1 Physiology Section, Department of Biochemistry and Physiology, Faculty of Pharmacy and Food Science, University of Barcelona, 08028 Barcelona, Spain; blancagrases@ub.edu (B.G.-P.); mtorreca29@alumnes.ub.edu (P.T.-C.); mariadelmar.abril@ub.edu (M.A.-G.); margaridacastell@ub.edu (M.C.); mjrodriguez@ub.edu (M.J.R.-L.); angelsfranch@ub.edu (À.F.) * Correspondence: franciscoperez@ub.edu; Tel.: +34-93-402-45-05 Nutrients 2019, 11, 999; doi:10.3390/nu11050999 www.mdpi.com/journal/nutrients nutrients nutrients Article A Preterm Rat Model for Immunonutritional Studies Blanca Grases-Pintó 1,2 , Paulina Torres-Castro 1,2, Mar Abril-Gil 1,2 , Margarida Castell 1,2 , María J. Rodríguez-Lagunas 1,2 , Francisco J. Pérez-Cano 1,2,* and Àngels Franch 1,2 1 Physiology Section, Department of Biochemistry and Physiology, Faculty of Pharmacy and Food Science, University of Barcelona, 08028 Barcelona, Spain; blancagrases@ub.edu (B.G.-P.); mtorreca29@alumnes.ub.edu (P.T.-C.); mariadelmar.abril@ub.edu (M.A.-G.); margaridacastell@ub.edu (M.C.); mjrodriguez@ub.edu (M.J.R.-L.); angelsfranch@ub.edu (À.F.) 2 Nutrition and Food Safety Research Institute (INSA·UB), 08921 Santa Coloma de Gramenet, Spain * Correspondence: franciscoperez@ub.edu; Tel.: +34-93-402-45-05 Article nutrients nutrients 1. Introduction The length of human pregnancy is about 37−44 weeks [1] and when the neonate is born before the 37th week of gestation the newborn is considered premature [2]. There is a wide variety of factors involved in preterm birth, such as the age of the mother, the use of assisted reproductive technologies or multiple gestations, among others [3]. Generally, a premature birth is associated with the newborn’s low weight, fragile skin [4], lower neurodevelopment [5], lungs not fully developed with an absence or insuﬃcient production of surfactant [6], immature stomach with lower acidity [4], and higher intestinal permeability with a low microbiota diversity [7]. These deﬁciencies make the preterm neonate more susceptible to infections and a higher rate of mortality and morbidity in later life [8]. The higher predisposition to suﬀering infections is due to the immaturity of their immune system, in both innate and adaptive immunity, and in the inadequate communication between these two systems [9,10]. In reference to innate immunity, preterm neonates have a lower count of neutrophils and monocytes and a lower cytokine production compared to term infants [11,12]. The defense mechanism against viral infections through the cytotoxic activity of the natural killer (NK) cells is also reduced in preterm compared to full-term neonates [11]. Furthermore, prematurity also aﬀects 2 of 16 Nutrients 2019, 11, 999 the adaptive immunity: these infants have leukopenia, and lymphopenia [12]. The reduced count of lymphocytes is accompanied by a decrease in the secretion of cytokines [13] and immunoglobulins (Ig) [2]. In addition, this immaturity not only aﬀects the systemic immune system, but also intestinal immunity. Premature infants show increased intestinal permeability and decreased levels of secretory IgA and mucus, making newborns susceptible to intestinal diseases such as infections or necrotizing enterocolitis (NEC) [14,15]. Animal models are essential to further understand the mechanisms involved in these processes and to generate new advances in this area. In this context, studies using preterm animal models could be useful to increase our knowledge of the state of prematurity, and to explore new insights in the ﬁeld of nutritional interventions in preterm neonates in order to counteract immunodeﬁciencies. To date, there are few preterm models in rodents, and most of these are only focused on the impact of prematurity on brain [16] or lung development [17] or NEC [18,19], but none study immune system maturation, which is especially aﬀected in prematurity. 2.1. Animals The study was performed with six pregnant Wistar rats from Janvier Labs (Le Genest Saint Isle, France), speciﬁcally, with two dams at 13 days of gestation (G13), two at G14 and two at G15. Animals were individually housed in cages and monitored daily. The animals were housed under controlled temperature and humidity conditions, in a 12:12 h light:dark cycle and were fed with chow and water ad libitum in the Faculty of Pharmacy and Food Science animal facilities (University of Barcelona). All the experimental procedures were performed according to the Guide for the Care and Use of Laboratory Animals. The study was reviewed and approved by the Ethical Committee for Animal Experimentation of the University of Barcelona (CEEA/UB ref. 148/18). 1. Introduction To our knowledge, the only premature animal model studying immunity was performed in pigs [20]. However, when studying the immune system in early life experimental approaches, it is highly valuable to provide models in species in which there is a remarkable immunological immaturity at birth, as in rodents [21]. For example, the supplementation with a particular strain of biﬁdobacteria has shown to have a preventive eﬀect on the experimental rat model of NEC [19]; however, its impact on immune system development in a preterm animal remains unknown. Therefore, the aim of the present work was to establish a preterm rat model to determine biomarkers of innate and adaptive immunity aﬀected by prematurity. Speciﬁcally, we focused on studying the blood cell count, phagocytic function of blood leukocytes, plasmatic Ig pattern, intestinal permeability and histomorphometry of the small intestine. 2.2. Experimental Design We disposed of three types of pregnant dams (2 dams/each), depending on their gestational stage: G13, G14 and G15. The G14 pregnant dams were allowed to deliver naturally, at day 22 of gestation, and their litters used to the term group. The G13 pregnant dams were subjected to a caesarean section at day 21 of gestation—just one day before normal delivery—and their litters made up the preterm group. As this surgery was not compatible with the mother’s survival, the preterm pups were given to the surrogate dams (G15), who had delivered naturally two days before the caesarean surgery of the G13 dams. This fact is highly important to ensure that they have enough milk in their breasts to breastfeed preterm pups. Litters were culled to 10 pups per lactating dam and they had free access to the nipples and rat diet during the 10 days of the study. Each group was made up of 40% male and 60% female pups. Suckling rats were weighed daily and the handling was done in the same time range to avoid the inﬂuence of biological rhythms. On the last day of the study, body length (nose−anus) was measured, and the 3 of 16 Nutrients 2019, 11, 999 body mass index (BMI), calculated as body weight/length2 (g/cm2), and the Lee index, calculated as 3√weight/length × 1000 (3√g/cm), were determined. body mass index (BMI), calculated as body weight/length2 (g/cm2), and the Lee index, calculated as 3√weight/length × 1000 (3√g/cm), were determined. 2.3. Caesarean Intervention To obtain premature pups, a caesarean section, which was performed at G21, was required. The procedure was based on the methodology described by Appleby and Towner [22] with some modiﬁcations. Dams, partially anesthetized, were immediately sacriﬁced by cervical dislocation. Immediately, the oﬀspring were extracted one by one by hysterectomy. Thereafter, they were randomly distributed among the two surrogate mothers’ cages, which had previously been separated from their own oﬀspring. It should be mentioned that, in this study, the viability of preterm rats was 100% and they were not rejected by the surrogate dam. 2.5. Immunoglobulin Quantiﬁcation At the end of the study, plasmatic IgG, IgM and IgA concentrations were quantiﬁed using a rat IgG, IgM and IgA ELISA quantiﬁcation set (Bethyl Laboratories, Montgomery, TX, USA), as performed in previous studies [23]. Absorbance was quantiﬁed using a microplate photometer (LabSystem Multiskan, LabX, Midland, CA, USA) and data were interpolated into standard curves using ASCENT version 2.6 software (Thermo Fisher Scientiﬁc, Barcelona, Spain). The lower limits of detection were as follows: 1 ng/mL for IgG; 1.95 ng/mL for IgM; and 1.95 ng/mL for IgA. 2.4. Sample Collection and Processing At day 10, each litter was divided into two: ﬁve pups were used to perform a histomorphometric study and permeability assay in the small intestine and the other ﬁve were used to analyze the blood cell count, the plasmatic Ig concentration and the phagocytic activity of leukocytes. Animals were intramuscularly anesthetized with ketamine (90 mg/kg) (Merial Laboratories S.A., Barcelona, Spain) and xylazine (10 mg/kg) (Bayer A.G., Leverkusen, Germany) in order to be exsanguinated. Then, the small intestine was collected, weighed, and measured. To perform the histomorphometric study, the intestine was washed with PBS (phosphate buﬀer saline, pH 7.2; 154 mM sodium chloride (NaCl), 3.99 mM sodium dihydrogen phosphate monohydrate (NaH2PO4·H2O) and 16 mM disodium hydrogen phosphate dehydrate (Na2HPO4·2H2O)) and 1 cm sections were cut from the distal jejunum, placed in cassettes and ﬁxed in 4% paraformaldehyde for 24 h. Blood samples were collected in heparin/lithium tubes (Sarstedt, Nümbrecht, Germany). Part of the whole blood was obtained to study the phagocytic activity and the blood cell count. The remaining blood was centrifuged to obtain plasma for Ig quantiﬁcation. To determine the composition of the cellular elements of the blood, an automated hematological analyzer (Spincell 3, Spinreact, Barcelona, Spain) was used. 2.7. Intestinal Permeability Assay The evaluation of the intestinal epithelial barrier functionality was performed by an in vivo permeability assay to dextrans. Speciﬁcally, 4 kDa-dextran conjugated to FITC (Sigma-Aldrich, Madrid, Spain) in PBS at a dose of 500 mg/kg were orally administered to rats using low-capacity syringes (Hamilton Bonaduz, Bonaduz, Switzerland) adapted to oral 23-gauge gavage tubes (ASICO, Westmont, IL, USA), as previously described [23]. A reference sample pool obtained by animals given an equivalent volume of PBS was used to rule out autoﬂuorescent background in each tissue sample itself. The oﬀspring were separated from their respective mothers 1 h before the administration, to allow gastric emptying and to avoid the interference of milk components with the dextran solution. Four hours after the administration, the animals were euthanized (ketamine (90 mg/kg)/xylazine (10 mg/kg)), plasma was obtained, diluted, and the ﬂuorescence emission was analyzed by triplicate at an excitation wavelength of 490 nm, in the Modulus™Microplate spectrophotometer (Turner Biosystems, Sunnyvale, CA, USA). 2.8. Intestinal Histomorphometric Study Fixed intestinal tissues were dehydrated by a growing gradient of ethanol, paraﬃn-embedded and cut into 5 µm sections using a microtome (RM2135, Leica, Wetzlar, Germany) at the CCiT-UB. Then, samples were mounted on Superfrost™Ultra Plus slides (Thermo Fisher Scientiﬁc), as previously described [24]. The samples were deparaﬃnized and rehydrated by periodic acid-Schiﬀ(PAS) staining. Brieﬂy, intestinal tissues were stained in a 0.5% periodic acid (Sigma-Aldrich) solution for 5 min and washed in distilled water. Afterwards, sections were immersed in Schiﬀ’s reagent (Sigma-Aldrich) for 7 min, washed in running tap water and counterstained with Mayer’s hematoxylin solution (Sigma-Aldrich) for 1 min and again washed in running tap water. Finally, the slides were mounted with coverslips using ProLongTM Gold Antifade Mountant (Life Technologies, Carlsbad, CA, USA). The observation of the intestinal architecture was accomplished using a bright-ﬁeld microscope at 100X and 400X (Olympus BX41, Olympus Corporation, Shinjuku, Tokyo, Japan). All the morphometric measurements were processed with the Image J. software (Image Processing and Analysis in Java, National Institute of Mental Health, Bethesda, MD, USA). The morphometric analysis was focused on intestinal villi, 6−10 villi per animal and n = 5 animals/group were evaluated. Speciﬁcally, the villi height, width, area and epithelium perimeter of the villi were measured. The villi height was measured from the base of the villi (junction of the villi to crypts) to the top of the villi. Moreover, the villi width was calculated at the middle of the villi. The epithelial perimeter of the villi was evaluated by measuring the outer layer of the villi. Goblet cells are expressed as PAS-positive. The number of goblet cells per villi were counted manually and their area was also measured. Statistical analysis was performed with the average values of each animal. 2.6. Phagocytic Function of Blood Leukocytes Phagocytosis was measured with the commercial kit Phagotest® (Glycotope, Biotechnology, Heidelberg, Germany) according to the manufacturer’s instructions. Brieﬂy, heparinized blood was incubated with opsonized ﬂuorescein isothiocyanate (FITC)-labelled Escherichia coli for 10 min at 37 ◦C. A control tube remained on ice. After 10 min, phagocytosis was stopped by placing the tubes into the ice. Then, non-phagocytized bacteria were eliminated with the addition of the quenching solution. After washing and centrifuging (250× g, 5 min, 4 ◦C), cells were resuspended and incubated in a solution for lysis of erythrocytes and ﬁxation of the leukocytes (20 min, room temperature). Finally, after an additional centrifugation and wash, the staining of the leukocyte DNA was carried out with a propidium iodide solution. Analyses were performed using a GalliosTM ﬂow cytometer (Beckman Coulter, Miami, FL, USA) at the Scientiﬁc and Technological Centers of the University of Barcelona (CCiT-UB). All samples were assessed by FlowJo, version 10 software (Tree Star Inc., Ashland, OR, USA). Monocyte and granulocyte populations were gated according to their forward-scatter and side-scatter characteristics (FSC/SSC). 4 of 16 Nutrients 2019, 11, 999 Then, their green ﬂuorescence histogram (FITC) was analyzed. The phagocytic activity was expressed as percentage of ﬂuorescent cells (monocytes or granulocytes) in the particular population studied. The mean ﬂuorescence intensity indicative of the extent of phagocyte eﬃciency was also quantiﬁed. Then, their green ﬂuorescence histogram (FITC) was analyzed. The phagocytic activity was expressed as percentage of ﬂuorescent cells (monocytes or granulocytes) in the particular population studied. The mean ﬂuorescence intensity indicative of the extent of phagocyte eﬃciency was also quantiﬁed. 3. Results 3. Results 3.1. Eﬀect of Prematurity on Body Weight and Other Morphometric Variables 1. Effect of Prematurity on Body Weight and Other Morphometric Variables Th h h d d h i h f h b 3.1. Eﬀect of Prematurity on Body Weight and Other Morphometric Variables 3.1. Effect of Prematurity on Body Weight and Other Morphometric Variables Th h h d d h h f h b Throughout the study, as expected, the weight of rats that were born in preterm conditions was lower than that of the term pups (p < 0.05, Figure 1A), indicating an important eﬀect of the length of gestation period on body weight. This decrease has an impact on the BMI at day 10 (p < 0.01 vs. term group, Figure 1B). No diﬀerences were observed in the Lee index at the same day (Figure 1C). Throughout the study, as expected, the weight of rats that were born in preterm conditions was lower than that of the term pups (p < 0.05, Figure 1A), indicating an important effect of the length of gestation period on body weight. This decrease has an impact on the BMI at day 10 (p < 0.01 vs. term group, Figure 1B). No differences were observed in the Lee index at the same day (Figure 1C). Term Preterm Body weight (g) 0 5 10 15 20 25 30 * * * 3 7 10 days A Figure 1. Box plot of the influence of prematurity on body weight and morphometric variables. Rat body weight along the study (A), BMI (B) and Lee index (C) at day 10 of life in both experimental groups. Line indicates the median, box shows the interquartile range (IQR) and the whiskers are 1.5 × IQR (n = 20 pups per group). Statistical differences: * p < 0.05 vs. term group. Term Preterm Term Preterm BMI (g/cm2) 0.0 0.1 0.2 0.3 0.4 0.5 * Term Preterm 0 100 200 300 B ((3√g/cm)) Lee index C Body weight (g) 0 5 10 15 20 25 30 * * * 3 7 10 days A X 1000) Figure 1. Box plot of the inﬂuence of prematurity on body weight and morphometric variables. Rat body weight along the study (A), BMI (B) and Lee index (C) at day 10 of life in both experimental groups. 3. Results 3. Results Line indicates the median, box shows the interquartile range (IQR) and the whiskers are 1.5 × IQR (n = 20 pups per group). Statistical diﬀerences: * p < 0.05 vs. term group. A Term Preter Body weight (g) 0 5 10 15 20 25 30 * * * 3 7 10 days A Term Preterm BMI (g/cm2) 0.0 0.1 0.2 0.3 0.4 0.5 * B Term Preterm 0 100 200 300 ((3√g/cm)) Lee index C X 1000) B C Preterm Figure 1. Box plot of the influence of prematurity on body weight and morphometric variables. Rat body weight along the study (A), BMI (B) and Lee index (C) at day 10 of life in both experimental groups. Line indicates the median, box shows the interquartile range (IQR) and the whiskers are 1.5 × IQR (n = 20 pups per group). Statistical differences: * p < 0.05 vs. term group. Figure 1. Box plot of the inﬂuence of prematurity on body weight and morphometric variables. Rat body weight along the study (A), BMI (B) and Lee index (C) at day 10 of life in both experimental groups. Line indicates the median, box shows the interquartile range (IQR) and the whiskers are 1.5 × IQR (n = 20 pups per group). Statistical diﬀerences: * p < 0.05 vs. term group. Furthermore, to study in more depth the influence of prematurity on body weight variables, the weight of the spleen, thymus, liver, small and large intestines on day 10 was measured (Table 1). Furthermore, to study in more depth the inﬂuence of prematurity on body weight variables the weight of the spleen, thymus, liver, small and large intestines on day 10 was measured (Table 1). Table 1. Absolute and relative weight of spleen, thymus, liver, small and large intestines, and length and relative length of small and large intestines from term and preterm groups at day 10. Table 1. Absolute and relative weight of spleen, thymus, liver, small and large intestines, and length and relative length of small and large intestines from term and preterm groups at day 10. 2.9. Statistical Analysis Results are shown as means ± SEM. The software package SPSS 22.0 (PASW Statistics, SPSS, Chicago, IL, USA) was used for statistical analysis. The Shapiro–Wilk and Levene’s tests were applied to assess normal distribution and variance equality, respectively. When they coexisted, the parametric Student’s t-test to compare the means between both experimental groups was carried out. To evaluate the correlation between body weight and Ig concentrations, Pearson’s coeﬃcient (ρ) was applied. Diﬀerences between groups were considered statistically signiﬁcant for p values < 0.05. 5 of 16 5 of 16 Nutrients 2019, 11, 999 Nutrients 2019, 11, x FOR Length (cm) Relative Length (%) Length (cm) Relative Length (%) The results are expressed as a mean ± S.E.M. (n = 20). Relative length percentage is expressed as cm of intestine per 100 g of animal body weight. Statistical diﬀerences: * p < 0.05 vs. term group. 3.2. Eﬀect of Prematurity on Blood Cell Count but a higher count of leukocytes in compari was no difference on the hematocrit (HC At the end of the study, the preterm group showed a lower count of erythrocytes and platelets, but a higher count of leukocytes in comparison to the term group (p < 0.05, Table 2). Although there was no diﬀerence on the hematocrit (HCT) and the mean corpuscular hemoglobin (MCH), a tendency to decrease hemoglobin (Hb) concentration was observed in premature rats (p < 0.08 vs. term group). In addition, premature erythrocytes showed a slight macrocytosis through mean corpuscular volume (MCV), which was signiﬁcantly higher than that in term rats (p < 0.05, Table 2). ( ) p g ( ), tendency to decrease hemoglobin (Hb) concentration was observed in premature rats (p < 0.08 vs. term group). In addition, premature erythrocytes showed a slight macrocytosis through mean corpuscular volume (MCV), which was significantly higher than that in term rats (p < 0.05, Table 2). Table 2. Blood cell count from both groups: term and preterm at day 10 of suckling period. T P Table 2. Blood cell count from both groups: term and preterm at day 10 of suckling period. Term Preterm Leukocytes (×109/L) 2.18 ± 0.20 4.57 ± 0.79 * Table 2. Blood cell count from both groups: term and preterm at day 10 of suckling period. Term Preterm Leukocytes (×109/L) 2.18 ± 0.20 4.57 ± 0.79 * Erythrocytes (×1012/L) 3.23 ± 0.08 2.83 ± 0.12 * Hb (g/L) 77.58 ± 1.84 72.56 ± 1.89 HCT (%) 27.16 ± 0.79 25.01 ± 1.16 MCV (fL) 84.28 ± 1.30 88.42 ± 0.60 * MCH (pg) 24.03 ± 0.42 25.82 ± 0.78 Platelets (×109/L) 529.18 ± 33.41 408.38 ± 56.36 * The results are expressed as a mean ± S.E.M. (n = 10). Statistical diﬀerences: * p < 0.05 vs. term group. e results are expressed as a mean ± S.E.M. (n = 10). Statistical differences: * p < 0.05 vs. 3. Results 3. Results g g p g p y Term Preterm Weight (g) Relative Weight (%) Weight (g) Relative Weight (%) Spleen 0.14 ± 0.00 0.55 ± 0.01 0.12 ± 0.01 * 0.58 ± 0.03 Thymus 0.09 ± 0.00 0.34 ± 0.02 0.07 ± 0.00 * 0.33 ± 0.02 Liver 0.73 ± 0.03 2.82 ± 0.08 0.67 ± 0.03 * 3.08 ± 0.09 Small intestine 0.84 ± 0.03 3.29 ± 0.08 0.77 ± 0.03 * 3.51 ± 0.08 Large intestine 0.16 ± 0.01 0.63 ± 0.03 0.12 ± 0.01 * 0.57 ± 0.03 Length (cm) Relative Length (%) Length (cm) Relative Length (%) Small intestine 41 04 ± 0 87 1 62 ± 0 03 37 20 ± 0 91 * 1 72 ± 0 05 Term Preterm Weight (g) Relative Weight (%) Weight (g) Relative Weight (%) Spleen 0.14 ± 0.00 0.55 ± 0.01 0.12 ± 0.01 * 0.58 ± 0.03 Thymus 0.09 ± 0.00 0.34 ± 0.02 0.07 ± 0.00 * 0.33 ± 0.02 Liver 0.73 ± 0.03 2.82 ± 0.08 0.67 ± 0.03 * 3.08 ± 0.09 Small intestine 0.84 ± 0.03 3.29 ± 0.08 0.77 ± 0.03 * 3.51 ± 0.08 Large intestine 0.16 ± 0.01 0.63 ± 0.03 0.12 ± 0.01 * 0.57 ± 0.03 Length (cm) Relative Length (%) Length (cm) Relative Length (%) Small intestine 41.04 ± 0.87 1.62 ± 0.03 37.20 ± 0.91 * 1.72 ± 0.05 Large intestine 6.88 ± 0.22 0.27 ± 0.01 6.19 ± 0.16 * 0.29 ± 0.01 The results are expressed as a mean ± S.E.M. (n = 20). Relative length percentage is expressed as cm of intestine per 100 g of animal body weight. Statistical diﬀerences: * p < 0.05 vs. term group. Nutrients 2019, 11, 999 Preterm rats s however the relati 6 of 16 0.05); er, the Preterm rats showed lower absolute weight in all these tissues than term ones (p < 0.05); however, the relative weights between the studied groups were similar (Table 1). Moreover, the length of both intestines was also measured on this day and the results showed that premature rats also had shorter intestines (p < 0.05 vs. term group, Table 1). g y p also had shorter intestines (p < 0.05 vs. term group, Table 1). 3.2. Effect of Prematurity on Blood Cell Count At the end of the study, the preterm group showed a lower count of erythrocytes and platelets, 3.2. Eﬀect of Prematurity on Blood Cell Count but a higher count of leukocytes in compari was no difference on the hematocrit (HC term grou garding the differential leukocyte count (Figure 2A and 2B) a tendency to de Leukocytes (×109/L) 2.18 ± 0.20 4.57 ± 0.79 * Erythrocytes (×1012/L) 3.23 ± 0.08 2.83 ± 0.12 * Hb (g/L) 77.58 ± 1.84 72.56 ± 1.89 HCT (%) 27.16 ± 0.79 25.01 ± 1.16 MCV (fL) 84.28 ± 1.30 88.42 ± 0.60 * MCH (pg) 24.03 ± 0.42 25.82 ± 0.78 Platelets (×109/L) 529.18 ± 33.41 408.38 ± 56.36 * Regarding the diﬀerential leukocyte count (Figure 2A,B), a tendency to decrease lymphocyte and increase granulocyte percentage in preterm animals was observed (p < 0.09 and p < 0.07 vs. term group, respectively), with no diﬀerences in monocyte percentage (Figure 2A). In absolute values, there was an increase in the count of lymphocytes, monocytes and granulocytes (p < 0.05, Figure 2B). lymphocyte and increase granulocyte percentage in preterm animals was observed (p < 0.09 and p < 0.07 vs. term group, respectively), with no differences in monocyte percentage (Figure 2A). In absolute values, there was an increase in the count of lymphocytes, monocytes and granulocytes (p < 0.05, Figure 2B). Figure 2. Box plot of differential leukocyte count in percentage (A) and in absolute values (B) at the end of the study (day 10) in both experimental groups. Line indicates the median, box shows the interquartile range (IQR) and the whiskers are 1.5 × IQR (n = 10 pups per group). Statistical differences: * p < 0.05 vs. term group. Differential leukocyte count (109/L) 0 1 2 3 4 * * * Differential leukocyte count (%) 0 20 40 60 80 100 * * A B Term Preterm Figure 2. Box plot of diﬀerential leukocyte count in percentage (A) and in absolute values (B) at the end of the study (day 10) in both experimental groups. Line indicates the median, box shows the interquartile range (IQR) and the whiskers are 1.5 × IQR (n = 10 pups per group). Statistical diﬀerences: * p < 0.05 vs. term group. Differential leukocyte count (%) 0 20 40 60 80 100 * * A Term Preterm Differential leukocyte count (109/L) 0 1 2 3 4 * * * B B A Figure 2. Box plot of differential leukocyte count in percentage (A) and in absolute values (B) at the end of the study (day 10) in both experimental groups. 3.3. Eﬀect of Prematurity on Plasma Ig Concentrations Effect of Prematurity on Plasma Ig Concentrations 3.3. Eﬀect of Prematurity on Plasma Ig Concentrations Effect of Prematurity on Plasma Ig Concentrations 3.3. Eﬀect of Prematurity on Plasma Ig Concentrations . Effect of Prematurity on Plasma Ig Concentrations 3.3. Eﬀect of Prematurity on Plasma Ig Concentrations . Effect of Prematurity on Plasma Ig Concentrations To assess the impact of prematurity on the antibody levels in the systemic compartment, total Ig and IgG, IgM, and IgA isotypes were quantiﬁed in plasma samples at day 10 of life (Figure 3). Preterm rats had lower concentrations of total Ig compared to term ones (p < 0.01, Figure 3A). This change was mainly due to a reduction in IgG concentration (p < 0.01, Figure 3B), but IgM isotype was also decreased (p < 0.01, Figure 3C). No changes in IgA levels were observed (Figure 3D). To assess the impact of prematurity on the antibody levels in the systemic compartment, tota d IgG, IgM, and IgA isotypes were quantified in plasma samples at day 10 of life (Figure eterm rats had lower concentrations of total Ig compared to term ones (p < 0.01, Figure 3A). T ange was mainly due to a reduction in IgG concentration (p < 0.01, Figure 3B), but IgM isotype w o decreased (p < 0.01, Figure 3C). No changes in IgA levels were observed (Figure 3D). igure 3. Box plot of plasmatic Ig concentrations at day 10 of life in both experimental groups. To g (A), IgG (B), IgM (C), IgA (D). Line indicates the median, box shows the interquartile range (IQ nd the whiskers are 1 5 × IQR (n = 10 pups per group) Statistical differences: * p < 0 05 vs te Term Preterm Total Ig (μg/mL) 0 1000 2000 3000 * Term Preterm IgG (μg/mL) 0 1000 2000 3000 * Term Preterm IgM (μg/mL) 0 20 40 60 80 100 * Term Preterm IgA (μg/mL) 0.00 0.02 0.04 0.06 0.08 0.10 C A B D Figure 3. Box plot of plasmatic Ig concentrations at day 10 of life in both experimental groups. Total Ig (A), IgG (B), IgM (C), IgA (D). Line indicates the median, box shows the interquartile range (IQR) and the whiskers are 1.5 × IQR (n = 10 pups per group). Statistical diﬀerences: * p < 0.05 vs. term group. 3.3. Eﬀect of Prematurity on Plasma Ig Concentrations Effect of Prematurity on Plasma Ig Concentrations Term Preterm Total Ig (μg/mL) 0 1000 2000 3000 * A Term Preterm IgG (μg/mL) 0 1000 2000 3000 * B B A Term Preterm IgA (μg/mL) 0.00 0.02 0.04 0.06 0.08 0.10 D Term Preterm IgM (μg/mL) 0 20 40 60 80 100 * C D C Preterm Preterm ure 3. Box plot of plasmatic Ig concentrations at day 10 of life in both experimental groups. To A), IgG (B), IgM (C), IgA (D). Line indicates the median, box shows the interquartile range (IQ d the whiskers are 1 5 × IQR (n = 10 pups per group) Statistical differences: * p < 0 05 vs te Figure 3. Box plot of plasmatic Ig concentrations at day 10 of life in both experimental groups. Total Ig (A), IgG (B), IgM (C), IgA (D). Line indicates the median, box shows the interquartile range (IQR) and the whiskers are 1.5 × IQR (n = 10 pups per group). Statistical diﬀerences: * p < 0.05 vs. term group. p p p g p p group. Considering all the pups in the study, there was a positive correlation between ncentration of total Ig, IgG and IgM, and body weight (ρ = 0.74, p < 0.001 at total Ig; ρ = 0 0.001 at IgG; ρ = 0.73, p < 0.001 at IgM). Looking at this in the preterm group, there was a posi Considering all the pups in the study, there was a positive correlation between the concentration of total Ig, IgG and IgM, and body weight (ρ = 0.74, p < 0.001 at total Ig; ρ = 0.73, p < 0.001 at IgG; ρ = 0.73, p < 0.001 at IgM). Looking at this in the preterm group, there was a positive correlation between total Ig, IgG and IgA, and body weight (ρ = 0.71, p < 0.05 at total Ig; ρ = 0.70, p < 0.05 at IgG; ρ = 0.72, p < 0.05 at IgA), whereas in the term group this correlation was not observed. 3.2. Eﬀect of Prematurity on Blood Cell Count but a higher count of leukocytes in compari was no difference on the hematocrit (HC Line indicates the median, box shows the interquartile range (IQR) and the whiskers are 1.5 × IQR (n = 10 pups per group). Statistical differences: * p < 0 05 vs term group Figure 2. Box plot of diﬀerential leukocyte count in percentage (A) and in absolute values (B) at the end of the study (day 10) in both experimental groups. Line indicates the median, box shows the interquartile range (IQR) and the whiskers are 1.5 × IQR (n = 10 pups per group). Statistical diﬀerences: * p < 0.05 vs. term group. 7 of 167 o 7 of 167 o Nutrients 2019, 11, 999 trients 2019, 11, x FOR P relation between total Ig, IgG and IgA, and body weight (ρ 0 05 at IgG; ρ = 0 72 p < 0 05 at IgA) whereas in the term grou 3.4. Eﬀect of Prematurity on Phagocytic Function of Blood Leukocytes 0.05 at IgG; ρ 0.72, p 0.05 at IgA), whereas in the term group this correlation was not observ . Effect of Prematurity on Phagocytic Function of Blood Leukocytes Regarding monocyte phagocytic activity, this was around 70% in the term group and 60% in Regarding monocyte phagocytic activity, this was around 70% in the term group and 60% in the preterm group, being lower in the latter (p < 0.05, Figure 4A). In contrast, the granulocyte population had a phagocytic activity higher than 90% without signiﬁcant diﬀerences between either experimental groups (Figure 4B). eterm group, being lower in the latter (p < 0.05, Figure 4A). In contrast, the granulocyte populat d a phagocytic activity higher than 90% without significant differences between eit perimental groups (Figure 4B). The phagocytic efficiency of these leukocytes was also quantified (Figure 4C and 4D). The phagocytic eﬃciency of these leukocytes was also quantiﬁed (Figure 4C,D).The obtained results showed a higher phagocytic eﬃciency of the granulocyte population in comparison to the monocytes in both groups (p < 0.05); however, no statistical diﬀerences were found in the preterm group compared to the term group, either in granulocytes or in monocytes (Figure 4B,D). 8 of 16 8 of 16 Nutrients 2019, 11, 999 Nutrients 2019, 11, x FO Figure 4. Box plot of phagocytic function of blood leukocytes. Phagocytic activity and efficiency in monocytes (A, C) and granulocytes (B, D). Line indicates the median, box shows the interquartile range (IQR) and the whiskers are 1.5 × IQR (n = 10 pups per group). Statistical differences: * p < 0.05 vs. term group. Term Preterm Phagocytic activity (%) 0 20 40 60 80 100 * Term Preterm Phagocytic efficiency (MFI) 0 5000 10000 15000 Term Preterm Phagocytic activity (%) 0 20 40 60 80 100 Term Preterm Phagocytic efficiency (MFI) 0 5000 10000 15000 Granulocytes Monocytes C D A B Figure 4. Box plot of phagocytic function of blood leukocytes. Phagocytic activity and eﬃciency in monocytes (A,C) and granulocytes (B,D). Line indicates the median, box shows the interquartile range (IQR) and the whiskers are 1.5 × IQR (n = 10 pups per group). Statistical diﬀerences: * p < 0.05 vs. term group. Figure 4. Box plot of phagocytic function of blood leukocytes. Phagocytic activity and efficiency in monocytes (A, C) and granulocytes (B, D). relation between total Ig, IgG and IgA, and body weight (ρ 0 05 at IgG; ρ = 0 72 p < 0 05 at IgA) whereas in the term grou 3.4. Eﬀect of Prematurity on Phagocytic Function of Blood Leukocytes Line indicates the median, box shows the interquartile range (IQR) and the whiskers are 1.5 × IQR (n = 10 pups per group). Statistical differences: * p < 0.05 vs. term group. Term Preterm Phagocytic activity (%) 0 20 40 60 80 100 * Term Preterm Phagocytic efficiency (MFI) 0 5000 10000 15000 Term Preterm Phagocytic activity (%) 0 20 40 60 80 100 Term Preterm Phagocytic efficiency (MFI) 0 5000 10000 15000 y y C D A B Term Preterm Phagocytic activity (%) 0 20 40 60 80 100 * Monocytes A Term Preterm Phagocytic activity (%) 0 20 40 60 80 100 * C A Term Preterm Phagocytic activity (%) 0 20 40 60 80 100 Granulocytes B Term Preterm Phagocytic activity (%) 0 20 40 60 80 100 D B A %) 100 B (%) 10 Term Preterm Phagocytic efficiency (MFI) 0 5000 10000 15000 C plot of phagocytic function of blo Term Preterm Phagocytic efficiency (MFI) 0 5000 10000 15000 Term Preterm Phagocytic efficiency (MFI) 0 5000 10000 15000 D leukocytes Phagocytic activity an Term Preterm Phagocytic efficiency (MFI) 0 5000 10000 15000 C ( I) D (MFI) 15000 Figure 4. Box plot of phagocytic function of blood leukocytes. Phagocytic activity and efficiency in monocytes (A, C) and granulocytes (B, D). Line indicates the median, box shows the interquartile range (IQR) and the whiskers are 1.5 × IQR (n = 10 pups per group). Statistical differences: * p < 0.05 vs term group Figure 4. Box plot of phagocytic function of blood leukocytes. Phagocytic activity and eﬃciency in monocytes (A,C) and granulocytes (B,D). Line indicates the median, box shows the interquartile range (IQR) and the whiskers are 1.5 × IQR (n = 10 pups per group). Statistical diﬀerences: * p < 0.05 vs. term group. g p p g y y g y y y monocytes (A, C) and granulocytes (B, D). Line indicates the median, box shows the interquartile range (IQR) and the whiskers are 1.5 × IQR (n = 10 pups per group). Statistical differences: * p < 0.05 vs. term group. vs. term group. 3 5 Eff f P i I i l B i F i 3.5. Eﬀect of Prematurity on Intestinal Barrier Function 3.5. Effect of Prematurity on Intestinal Barrier Function vs. term group. 3 5 Eff f P i I i l B i F i 3.5. Eﬀect of Prematurity on Intestinal Barrier Function 3.5. Effect of Prematurity on Intestinal Barrier Function Line indicates the median, box shows the interquartile range (IQR) and the whiskers are 1.5 × IQR (n Term Preterm 0 Figure 5. Box plot of intestinal permeability to 4 kDa-FITC-dextran in both experimental groups. Line indicates the median, box shows the interquartile range (IQR) and the whiskers are 1.5 × IQR (n = 10 pups per group). Statistical differences: * p < 0.05 vs. term group. Figure 5. Box plot of intestinal permeability to 4 kDa-FITC-dextran in both experimental groups. Line indicates the median, box shows the interquartile range (IQR) and the whiskers are 1.5 × IQR (n = 10 pups per group). Statistical diﬀerences: * p < 0.05 vs. term group. Figure 5. Box plot of intestinal permeability to 4 kDa-FITC-dextran in both experimental groups. Line indicates the median, box shows the interquartile range (IQR) and the whiskers are 1.5 × IQR (n Term Preterm 0 Figure 5. Box plot of intestinal permeability to 4 kDa-FITC-dextran in both experimental groups. Line indicates the median, box shows the interquartile range (IQR) and the whiskers are 1.5 × IQR (n = 10 pups per group). Statistical differences: * p < 0.05 vs. term group. Figure 5. Box plot of intestinal permeability to 4 kDa-FITC-dextran in both experimental groups. Line indicates the median, box shows the interquartile range (IQR) and the whiskers are 1.5 × IQR (n = 10 pups per group). Statistical diﬀerences: * p < 0.05 vs. term group. Figure 5. Box plot of intestinal permeability to 4 kDa-FITC-dextran in both experimental groups. Line indicates the median, box shows the interquartile range (IQR) and the whiskers are 1.5 × IQR (n Term Preterm Figure 5. Box plot of intestinal permeability to 4 kDa-FITC-dextran in both experimental groups. Line indicates the median, box shows the interquartile range (IQR) and the whiskers are 1.5 × IQR (n = 10 pups per group). Statistical differences: * p < 0.05 vs. term group. Figure 5. Box plot of intestinal permeability to 4 kDa-FITC-dextran in both experimental groups. Line indicates the median, box shows the interquartile range (IQR) and the whiskers are 1.5 × IQR (n = 10 pups per group). Statistical diﬀerences: * p < 0.05 vs. term group. , q g ( Q ) = 10 pups per group). Statistical differences: * p < 0.05 vs. term group. 3 6 Effect of Prematurity on Histomorphometric Characteristics of the Small Intestine 3.6. vs. term group. 3 5 Eff f P i I i l B i F i 3.5. Eﬀect of Prematurity on Intestinal Barrier Function 3.5. Effect of Prematurity on Intestinal Barrier Function 3.5. Effect of Prematurity on Intestinal Barrier Function To study the intestinal epithelial barrier functionality, an in vivo permeability assay was performed to determine the paracellular transport of the 4 kDa-dextran labeled with FITC. Premature animals showed lower dextran-FITC signal in plasma than full-term ones, indicating l bilit f th i t ti l ith li i t i l t 10 d ld ( < 0 05 Fi 5) To study the intestinal epithelial barrier functionality, an in vivo permeability assay was performed to determine the paracellular transport of the 4 kDa-dextran labeled with FITC. Premature animals showed lower dextran-FITC signal in plasma than full-term ones, indicating lower permeability of the intestinal epithelium in preterm animals at 10 days old (p < 0.05, Figure 5). To study the intestinal epithelial barrier functionality, an in vivo permeability assay was performed to determine the paracellular transport of the 4 kDa-dextran labeled with FITC. Premature animals showed lower dextran-FITC signal in plasma than full-term ones, indicating lower permeability of the intestinal epithelium in preterm animals at 10 days old (p < 0.05, Figure 5). lower permeability of the intestinal epithelium in preterm animals at 10 days old (p 0.05, Figure Figure 5. Box plot of intestinal permeability to 4 kDa-FITC-dextran in both experimental groups. Line indicates the median, box shows the interquartile range (IQR) and the whiskers are 1.5 × IQR (n 10 ) St ti ti l diff * 0 05 t Term Preterm FITC-Dextran (μg/mL) 0 2 4 6 8 10 12 14 * Figure 5. Box plot of intestinal permeability to 4 kDa-FITC-dextran in both experimental groups. Line indicates the median, box shows the interquartile range (IQR) and the whiskers are 1.5 × IQR (n = 10 pups per group). Statistical differences: * p < 0.05 vs. term group. Term Preterm FITC-Dextran (μg/mL) 0 2 4 6 8 10 12 14 * Figure 5. Box plot of intestinal permeability to 4 kDa-FITC-dextran in both experimental groups. Line indicates the median, box shows the interquartile range (IQR) and the whiskers are 1.5 × IQR (n = 10 pups per group). Statistical diﬀerences: * p < 0.05 vs. term group. 3 6 Eﬀt f P t it Hi t h t i Ch t i ti f th S ll I t ti Figure 5. Box plot of intestinal permeability to 4 kDa-FITC-dextran in both experimental groups. 4 Discussion 4. Discussion 4. Discussion The suckling rat born at term has been widely used as a model for immunonutrition studies in early life [21]. The effect of several dietary components on promoting the immature immune system development such as whey proteins [25], probiotics [26], adipokines [23,27], growth factors [28] or highly active immunomodulatory fatty acids such as conjugated linoleic acid (CLA) [29], have been studied. In addition, this rodent model has also been used in immunonutritional studies aiming to evaluate the protective action of prebiotics and probiotics during an acute intestinal infective process [24,30]. However, the impact of certain nutrients on the undeveloped response of the immune system in premature neonates has not been assessed, mainly due to the lack of optimal models to f hi ki d f i i The suckling rat born at term has been widely used as a model for immunonutrition studies in early life [21]. The eﬀect of several dietary components on promoting the immature immune system development such as whey proteins [25], probiotics [26], adipokines [23,27], growth factors [28] or highly active immunomodulatory fatty acids such as conjugated linoleic acid (CLA) [29], have been studied. In addition, this rodent model has also been used in immunonutritional studies aiming to evaluate the protective action of prebiotics and probiotics during an acute intestinal infective process [24,30]. However, the impact of certain nutrients on the undeveloped response of the immune system in premature neonates has not been assessed, mainly due to the lack of optimal models to perform this kind of intervention. perform this kind of intervention. The present study provides the set-up of a preterm rat model with the characterization of the immune state in the middle of the suckling period (day 10) with the purpose of establishing specific biomarkers of innate and adaptive immunity. For that, some variables have been assessed: blood cell count, plasmatic Ig, phagocytic function of blood leukocytes, intestinal permeability and intestinal The present study provides the set-up of a preterm rat model with the characterization of the immune state in the middle of the suckling period (day 10) with the purpose of establishing speciﬁc biomarkers of innate and adaptive immunity. For that, some variables have been assessed: blood cell count, plasmatic Ig, phagocytic function of blood leukocytes, intestinal permeability and intestinal morphometry. Nguyen et al. have established the development of the systemic immunity in a preterm pig model [20]. vs. term group. 3 5 Eff f P i I i l B i F i 3.5. Eﬀect of Prematurity on Intestinal Barrier Function 3.5. Effect of Prematurity on Intestinal Barrier Function Eﬀect of Prematurity on Histomorphometric Characteristics of the Small Intestine , q g ( Q ) = 10 pups per group). Statistical differences: * p < 0.05 vs. term group. 3 6 Effect of Prematurity on Histomorphometric Characteristics of the Small Intestine 3.6. Eﬀect of Prematurity on Histomorphometric Characteristics of the Small Intestine 3.6. Effect of Prematurity on Histomorphometric Characteristics of the Small Intestine The morphology of the distal jejunum was studied, focusing on intestinal villi and goblet cell characteristics on day 10 of life. Table 3 shows the studied variables: villi width, villi height, villi area, villi perimeter, number of goblet cells per villi and goblet cell area. Histological differences due to prematurity were not found in any of the studied villi variables compared to the term group. With respect to goblet cells, although there was no difference in the number of these cells present in the 3.6. Effect of Prematurity on Histomorphometric Characteristics of the Small Intestine The morphology of the distal jejunum was studied, focusing on intestinal villi and goblet cell characteristics on day 10 of life. Table 3 shows the studied variables: villi width, villi height, villi area, villi perimeter, number of goblet cells per villi and goblet cell area. Histological differences due to prematurity were not found in any of the studied villi variables compared to the term group. With respect to goblet cells, although there was no difference in the number of these cells present in the The morphology of the distal jejunum was studied, focusing on intestinal villi and goblet cell characteristics on day 10 of life. Table 3 shows the studied variables: villi width, villi height, villi area, villi perimeter, number of goblet cells per villi and goblet cell area. Histological diﬀerences due to prematurity were not found in any of the studied villi variables compared to the term group. With respect to goblet cells, although there was no diﬀerence in the number of these cells present in the villi, interestingly, preterm goblet cells had a smaller area than those found in term animals (p < 0.01, Table 3, Figure 6). 9 of 16 p , Nutrients 2019, 11, 999 g y Table 3, Figure 6) Table 3. Histomorphometric variables of small intestine: villi width, height, area, perimeter and the number of goblet/villi and their cellular area from term and preterm groups at day 10 of suckling period. Table 3. vs. term group. 3 5 Eff f P i I i l B i F i 3.5. Eﬀect of Prematurity on Intestinal Barrier Function 3.5. Effect of Prematurity on Intestinal Barrier Function Term Preterm Villi width (μm) 128.21 ± 15.91 173.54 ± 22.26 Villi height (μm) 602.03 ± 92.70 765.77 ± 136.41 Villi area (μm2) 75319.14 ± 22354.29 131378.51 ± 36991.03 Villi perimeter (μm) 1328.82 ± 191.19 1679.68 ± 278.57 Goblet cells/villi 6.79 ± 1.33 4.99 ± 0.44 Goblet cell area (μm2) 379.96 ± 26.72 239.08 ± 22.13 * The results are expressed as a mean ± S.E.M. (n = 5). Statistical diﬀerences: * p < 0.05 vs. term group. Goblet cell area (μm2) 379.96 ± 26.72 239.08 ± 22.13 * The results are expressed as a mean ± S.E.M. (n = 5). Statistical diﬀerences: * p < 0.05 vs. term group. Goblet cell area (μm2) 379.96 ± 26.72 239.08 ± 22.13 * Figure 6. Representative images of histological sections of the jejunum with PAS staining. Term group (A and C) and preterm group (B and D). Goblet cells with densely stained granules can be observed along the length of the villi (C and D). Scale bar = 100 μm for 100X and 50 μm for 400X. Magnification = 100X (A and B) and 400X (C and D). C D A B Figure 6. Representative images of histological sections of the jejunum with PAS staining. Term group (A,C) and preterm group (B,D). Goblet cells with densely stained granules can be observed along the length of the villi (C,D). Scale bar = 100 µm for 100× and 50 µm for 400×. Magniﬁcation = 100× (A,B) and 400× (C,D). B A B A C C D D Figure 6. Representative images of histological sections of the jejunum with PAS staining. Term group (A and C) and preterm group (B and D). Goblet cells with densely stained granules can be observed along the length of the villi (C and D). Scale bar = 100 μm for 100X and 50 μm for 400X Magnification = 100X (A and B) and 400X (C and D) Figure 6. Representative images of histological sections of the jejunum with PAS staining. Term group (A,C) and preterm group (B,D). Goblet cells with densely stained granules can be observed along the length of the villi (C,D). Scale bar = 100 µm for 100× and 50 µm for 400×. Magniﬁcation = 100× (A,B) and 400× (C,D). vs. term group. 3 5 Eff f P i I i l B i F i 3.5. Eﬀect of Prematurity on Intestinal Barrier Function 3.5. Effect of Prematurity on Intestinal Barrier Function Histomorphometric variables of small intestine: villi width, height, area, perimeter and th number of goblet/villi and their cellular area from term and preterm groups at day 10 of sucklin period. Term Preterm Villi width (µm) 128.21 ± 15.91 173.54 ± 22.26 Villi height (µm) 602.03 ± 92.70 765.77 ± 136.41 Villi area (µm2) 75319.14 ± 22354.29 131378.51 ± 36991.03 Villi perimeter (µm) 1328.82 ± 191.19 1679.68 ± 278.57 Goblet cells/villi 6.79 ± 1.33 4.99 ± 0.44 Goblet cell area (µm2) 379.96 ± 26.72 239.08 ± 22.13 * The results are expressed as a mean ± S.E.M. (n = 5). Statistical diﬀerences: * p < 0.05 vs. term group. Term Preterm Villi width (μm) 128.21 ± 15.91 173.54 ± 22.26 Villi height (μm) 602.03 ± 92.70 765.77 ± 136.41 Villi area (μm2) 75319.14 ± 22354.29 131378.51 ± 36991.03 Villi perimeter (μm) 1328.82 ± 191.19 1679.68 ± 278.57 Goblet cells/villi 6.79 ± 1.33 4.99 ± 0.44 Goblet cell area (μm2) 379.96 ± 26.72 239.08 ± 22.13 * Term Preterm Villi width (µm) 128.21 ± 15.91 173.54 ± 22.26 Villi height (µm) 602.03 ± 92.70 765.77 ± 136.41 Villi area (µm2) 75319.14 ± 22354.29 131378.51 ± 36991.03 Villi perimeter (µm) 1328.82 ± 191.19 1679.68 ± 278.57 Goblet cells/villi 6.79 ± 1.33 4.99 ± 0.44 Goblet cell area (µm2) 379.96 ± 26.72 239.08 ± 22.13 * The results are expressed as a mean ± S.E.M. (n = 5). Statistical diﬀerences: * p < 0.05 vs. term group. Term Preterm Villi width (μm) 128.21 ± 15.91 173.54 ± 22.26 Villi height (μm) 602.03 ± 92.70 765.77 ± 136.41 Villi area (μm2) 75319.14 ± 22354.29 131378.51 ± 36991.03 Villi perimeter (μm) 1328.82 ± 191.19 1679.68 ± 278.57 Goblet cells/villi 6.79 ± 1.33 4.99 ± 0.44 Goblet cell area (μm2) 379.96 ± 26.72 239.08 ± 22.13 * Term Preterm Villi width (µm) 128.21 ± 15.91 173.54 ± 22.26 Villi height (µm) 602.03 ± 92.70 765.77 ± 136.41 Villi area (µm2) 75319.14 ± 22354.29 131378.51 ± 36991.03 Villi perimeter (µm) 1328.82 ± 191.19 1679.68 ± 278.57 Goblet cells/villi 6.79 ± 1.33 4.99 ± 0.44 Goblet cell area (µm2) 379.96 ± 26.72 239.08 ± 22.13 * The results are expressed as a mean ± S.E.M. (n = 5). Statistical diﬀerences: * p < 0.05 vs. term group. 4 Discussion 4. Discussion However, to the best of our knowledge, this is the ﬁrst time that the eﬀect of prematurity Nutrients 2019, 11, 999 10 of 16 on the systemic and intestinal immune system in rat has been assessed. All these studies in preterm rats were compared with suckling rats born at term. It has been described that the length of the gestation period of rat is 19–22 days, whereas for pigs and humans it is 115 days and 280 days, respectively [31]. In this sense, it is estimated that in terms of translational relationship between age of rats and humans, 14 rat days are equivalent to 1 human year [32]. Thus, whereas the lymphoid architecture is formed prenatally in humans, it appears mostly postnatally in rats. Due to this, the development of the immune system in rats is delayed compared with humans, although both develop via similar schemes [21]. In addition, although the known diﬀerences between human and rodent immune responses, the uniform genetic background and the similarity to humans in terms of genetic and biochemical pathways led to an increasing interest in the suckling rat model as an approximation for immunonutritional studies. This is also the case in the context of early life and prematurity, where the short length periods of gestation and suckling in rats facilitate the nutritional intervention and in addition, as mentioned before, obtain a more immature pup than in larger species [21]. Although this work focuses on the study of the immune system, we also considered it necessary to evaluate some growth and body variables. Speciﬁcally, the body weight of the preterm group on day 3 of life was lower than that of those born at term, and this diﬀerence remained during the 10 days of study. Moreover, on day 10 of life, a lower BMI and organ weight were observed without aﬀecting the relative weight or the Lee index. These results revealed the repercussions of prematurity during the ﬁrst days of life. This feature found here in premature rodents is consistent with that previously found in humans. In fact, human preterm newborns also showed lower weight at birth [33]. In addition, at two years of age the incapacity to achieve similar weight values and BMI as those of term infants has been described [34]. 4 Discussion 4. Discussion Thus, the low weight of preterm rats observed in our study was not only similar to humans but also in line with other animal models, such as pigs [20]. Regarding the blood cell count, it was observed that prematurity causes an increase in the leukocyte count and a decrease in that of erythrocytes and platelets. Human studies focused on that point have evidenced that platelets showed an immature state not only in terms of a depletion of their count [35] but also in their functionality [27], thus, making preterm neonates more prone to hemorrhaging [36]. Focusing on erythrocytes, the mean corpuscular volume (MCV) was signiﬁcantly higher in preterm rats and tended to decrease hemoglobin concentration in blood in comparison to the term group. These results taken together suggest a state of macrocytic anemia, and are in line with the study by Hoﬀbrand [37], which showed increased MCV due to a folic acid deﬁciency in preterm infants between 4 and 8 weeks old. This deﬁciency could be attributed to the fact that the preterm newborns require higher amounts of folic acid than those provided through diet intake and therefore leads to a depletion of their reserves. Besides changes in erythrocytes, the blood immune cell count was also modiﬁed due to prematurity. In this sense, it has been suggested that some features found in term rats due to its short gestation period, such as the intestinal barrier or the immune system immaturity at birth with respect to humans, can resemble those of infants born prematurely [21,31]. Thus, the present model aimed to show variables with even more immature immune levels. In fact, our results showed that the total blood leukocyte count in premature rats was much higher than those in the animals born at term. This result disagrees with previous studies in pigs and humans, which evidenced lower numbers of leukocytes in preterm compared to term newborns [12,20]. In our study, the preterm group’s diﬀerential leukocyte count showed an increase in the count of the three main populations (lymphocytes, monocytes and granulocytes). However, it has been described that preterm human babies had a lower number and percentage of granulocytes, monocytes and lymphocytes [2,11]. The results obtained in our model could be associated with a pro-inﬂammatory state interpreted as a defense mechanism against a highly hostile and aggressive environment for a premature newborn with a poorly developed immune system. 4 Discussion 4. Discussion In this study, similar to what has been reported in humans, the preterm group had a lower Ig concentration compared to those born at term [2]. Moreover, the preterm group showed a positive 11 of 16 Nutrients 2019, 11, 999 correlation between the weight of the pups and total Ig, IgG, and IgA. In line with our results, it has been described that in human serum, the Ig concentration correlates directly with the gestational age and the weight of the baby at the time of birth [38]. In the case of IgG, low plasma levels found in the preterm group could be attributed to a deﬁcient transplacental transfer of maternal IgG to the fetus, because it was interrupted after premature birth, knowing that the majority of maternal IgG is acquired during the last gestation period in humans and rodents [39,40]. However, in comparison with humans, who have a low postnatal transfer, rodents have an important postnatal transmission of IgG acquired from maternal breastmilk; thus, premature rats were able to improve this deﬁciency during the suckling period [40]. Regarding IgM, this Ig isotype does not cross the placenta in humans and rodents and, therefore, the IgM of the newborn comes basically from the endogenous synthesis [41,42]. The lower concentration of IgM found in premature rats could reﬂect the immaturity of B lymphocytes, both in their absolute cell count and their deﬁciency in Ig synthesis, compared to those born at term, as described previously in humans [2]. In relation to IgA, in humans it has been described that this isotype, as for IgM, does not substantially cross the placenta [43], but the intestinal balance of microbiota colonization of the newborn is essential for the stimulation of its production [44]. In addition, in human plasma, cells that produce IgA are absent at the intestinal level until the second month of life [45], and it has also been described that in rats the number of IgA- and IgM-secreting cells from intestinal lamina propria during the ﬁrst 10 days is practically absent, and the number of IgA-secreting cells at the end of suckling period is far away from that of adults [21]. These facts explain the low plasmatic IgA concentration found in 10-day-old animals in both studied groups. With regard to phagocytic activity, the percentage of granulocytes of 10-day-old animals was not modiﬁed in conditions of prematurity. 4 Discussion 4. Discussion On the other hand, compared to the term group, premature rats showed a lower percentage of monocytes exerting phagocytosis, which showed once again the immature status of the immune system in prematurity. These results, suggesting higher susceptibility to infections, are in line with other studies that report a lower phagocytic activity in monocytes in human infants or pigs that were born early in gestation [20,46]. Moreover, this could be related to the increase in the blood monocyte count, observed in this study, as a mechanism to compensate for this lack of phagocytic activity. In relation to the phagocytic eﬃciency of granulocytes, this was higher than that of monocytes, but did not diﬀer between term and preterm groups, as previously described in humans [46,47]. Thus, this result is also in line with the translational potential of this preterm rat model to what happens in humans. In both rats and humans, this fact indicates that the phagocytic eﬃciency of leukocytes in relation to pathogenic bacteria is not aﬀected by prematurity. In previous studies it has been shown that the intestinal barrier in rat is immature at the time of birth [48,49], and overall, in preterm humans as in other preterm animals, it has also been described that the intestinal epithelium has an increased permeability [14,43,50–52]. It is also known that this status of immaturity of the intestinal barrier is, in part, responsible for most cases of NEC in premature infants [53,54]. On this basis, in our study the eﬀect of prematurity on the intestinal barrier was evaluated not only at the functional level, studying the permeability to 4-kDa-dextran, but also at the histomorphometric level, focusing on the study of the intestinal villi and goblet cells. Unexpectedly, in our work, a lower concentration of 4kDa-FITC-dextran in plasma was observed in conditions of prematurity, which indicates a lower permeability of the intestinal epithelium to this molecule in premature rats. However, in previous studies it has been observed that there is a rapid maturation of the intestinal barrier during the ﬁrst days of life, positively inﬂuenced by feeding with breast milk, in both term [55] and preterm babies [56,57], thus reducing the incidence of NEC [58,59]. In addition, no diﬀerences between the groups were observed in the morphometric variables studied in the villi. However, goblet cells, responsible for the mucus layer found in the lumen of the intestine, were smaller due to prematurity. 4 Discussion 4. Discussion According to Clark et al. the smaller size and density of the goblet cells of the intestinal epithelium could reﬂect the condition of immaturity that would lead to lower mucus production at the intestinal barrier level [60]. This result was in line with other studies where the immaturity of these goblet cells is also described as being responsible for the deﬁcient Nutrients 2019, 11, 999 12 of 16 12 of 16 mucus layer [61]. Moreover, in those infants that suﬀer enterocolitis, immature goblet cells were also observed [53]. Again, this model resembles the conditions in preterm babies. mucus layer [61]. Moreover, in those infants that suﬀer enterocolitis, immature goblet cells were also observed [53]. Again, this model resembles the conditions in preterm babies. This work has some experimental limitations, mainly due to the reduced number of dams and litters in each group. It was designed as a ﬁrst approach considering the minimum use of animals following ethical standards; however, further nutritional studies should be conducted with a greater number of dams. Moreover, although we studied some variables related to the immune system, there are more immune variables to evaluate in a preterm model, such as the expression and localization of intestinal tight junctions, phenotype characterization of lymphocytes from diﬀerent compartments (intestinal and systemic), functionality of lymphocytes (proliferation activity and the secretion of cytokines), activity of NK cells, intestinal microbiota, among others. Another limitation of the study was the mode of delivery: Term rats were born by vaginal delivery, whereas preterm rats were born by caesarean section, and this could be another variable inﬂuencing the diﬀerence between groups. This approach was required in order to obtain premature pups in a controlled manner without using inductor agents that could inﬂuence the ﬁnal outcome. For future studies, one possibility in order to avoid the C-section and the inﬂuence of anesthesia, a caesarean could be performed in all groups, as performed by Stelloh et al. in mice [62]. In addition, it has to be taken into account that, in this model, preterm pups are being nourished with term milk. It is well known that there are remarkable diﬀerences between term and preterm milk such as higher levels of proteins, fat, and free amino acids, but there is a lower concentration of lactose in preterm milk [33]. Funding: This research was funded by the Spanish Ministry of Economy, Industry and Competitiveness (AGL2013-48459-P). BG-P holds a fellowship from the Spanish Ministry of Economy, Industry and Competitiveness (BES-2014-068134). PT-C holds a fellowship from the National Secretary of Higher Education, Science, Technology and Innovation of Ecuador (SENESCYT-DMPF-2015-1666-CO). Author Contributions: The authors’ contributions were as follows: conceptualization, À.F., M.C., M.J.R.-L. and F.J.P.-C.; methodology, B.G.-P., P.T.-C. and M.A.-G.; writing—original draft preparation, B.G.-P., F.J.P.-C., and À.F.; writing—review and editing, B.G.-P., P.T.-C., M.A.-G., M.C., M.J.R.-L., F.J.P.-C., and À.F. All authors have read and approved the ﬁnal version of the manuscript for publication. 4 Discussion 4. Discussion In terms of immune factors, preterm milk is enriched in neutrophils, lactoferrin, cytokines (IL 6, IL-10, TNF-α), secretory IgA and growth factors [33,63]. This diﬀerential composition expressly stimulates the immature immune system of the preterm baby. Overall, this compensation through the dam’s milk has not been possible in this model, favoring a better separation of the results in the term and preterm variables studied. There are several types of bioactive agents capable of stimulating the innate and acquired immunity of newborns and infants, such as some prebiotics, probiotics and symbiotics that lead to have a role in infection and allergy prevention [64–66]. However, few immunonutritional works have been performed so far in preterm neonates; the literature is scarce or even absent in terms of immunomaturation and they are just focused on NEC without considering immune variables. To date, some agents, for example, some Biﬁdobacterium strains and lactoferrin, have been demonstrated to signiﬁcantly reduce the incidence of NEC and sepsis in preterm infants, but their impact on the premature immune system has not been evaluated [67,68]. Thus, the present model facilitates the study at preclinical level of the impact of bioactive agents able to promote the immune system development as a ﬁrst step for future clinical interventions. The above interventions with immunonutrients target directly the pup after birth; however, the nutritional intervention to promote the maturation of the immune system of the preterm neonate can also be performed on gestating or lactating rats, as in previous studies in term rats [29]. In summary, our ﬁndings conﬁrm the delayed development of innate and adaptive immunity in preterm rats compared with term ones. The premature rat model presented in this work is another step towards a better understanding of the immature immune system in premature conditions. In addition, we have described, for the ﬁrst time, biomarkers of innate and adaptive immunity that could be useful in future studies to explore new ways to modulate this delayed immunity in preterm newborns, as well as being a useful tool in immunonutritional interventions. Author Contributions: The authors’ contributions were as follows: conceptualization, À.F., M.C., M.J.R.-L. and F.J.P.-C.; methodology, B.G.-P., P.T.-C. and M.A.-G.; writing—original draft preparation, B.G.-P., F.J.P.-C., and À.F.; writing—review and editing, B.G.-P., P.T.-C., M.A.-G., M.C., M.J.R.-L., F.J.P.-C., and À.F. All authors have read and approved the ﬁnal version of the manuscript for publication. References 2011, 24, 25–31. [CrossRef] 10. Adams-Chapman, I. Long-term impact of infection on the preterm neonate. Semin. Perinatol. 2012, 36, 462–470. [CrossRef] 10. Adams-Chapman, I. Long-term impact of infection on the preterm neonate. Semin. Perinatol. 2012, 36, 462–470. [CrossRef] 11. Gasparoni, A.; Ciardelli, L.; Avanzini, A.; Castellazzi, A.M.; Carini, R.; Rondini, G.; Chirico, G. Age-related changes in intracellular Th1/Th2 cytokine production, immunoproliferative T lymphocyte response and 11. Gasparoni, A.; Ciardelli, L.; Avanzini, A.; Castellazzi, A.M.; Carini, R.; Rondini, G.; Chirico, G. Age-related changes in intracellular Th1/Th2 cytokine production, immunoproliferative T lymphocyte response and natural killer cell activity in newborns, children and adults. Biol. Neonate 2003, 84, 297–303. [CrossRef] natural killer cell activity in newborns, children and adults. Biol. Neonate 2003, 84, 297–303. [CrossRef] 12. Correa-Rocha, R.; Pérez, A.; Lorente, R.; Ferrando-Martínez, S.; Leal, M.; Gurbindo, D.; Muñoz-Fernández, M.Á. Preterm neonates show marked leukopenia and lymphopenia that are associated with increased regulatory T cell values and diminished IL 7 Pediatr Res 2012 71 590 597 [CrossRef] 12. Correa-Rocha, R.; Pérez, A.; Lorente, R.; Ferrando-Martínez, S.; Leal, M.; Gurbindo, D.; Muñoz-Fernández, M.Á. Preterm neonates show marked leukopenia and lymphopenia that are associated with increased regulatory T-cell values and diminished IL-7. Pediatr. Res. 2012, 71, 590–597. [CrossRef] 13. Currie, A.J.; Curtis, S.; Strunk, T.; Riley, K.; Liyanage, K.; Prescott, S.; Doherty, D.; Simmer, K.; Richmond, P.; Burgner, D. Preterm infants have deﬁcient monocyte and lymphocyte cytokine responses to group B streptococcus. Infect. Immun. 2011, 79, 1588–1596. [CrossRef] 14. Neu, J. Gastrointestinal development and meeting the nutritional needs of premature infants. Am. J. Clin. Nutr. 2007, 85, 629S–634S. [CrossRef] 15. Chin, A.M.; Hill, D.R.; Aurora, M.; Spence, J.R. Morphogenesis and maturation of the embryonic and postnatal intestine. Semin. Cell Dev. Biol. 2017, 66, 81–93. [CrossRef] 16. Barkhuizen, M.; Van de Berg, W.D.; De Vente, J.; Blanco, C.E.; Gavilanes, A.W.; Steinbusch, H.W. Nitric oxide production in the striatum and cerebellum of a rat model of preterm global perinatal asphyxia. Neurotox. Res. 2017, 31, 400–409. [CrossRef] 17. Remesal, A.; De Luca, D.; San Feliciano, L.; Isidoro-Garcia, M.; Minucci, A.; Pocino, K.; Casas, J.; Fabrias, G.; Capoluongo, E.D.; de la Cruz, D.L. Eﬀect of prenatal steroidal inhibition of sPLA2 in a rat model of preterm lung. Pulm. Pharmacol. Ther. 2016, 36, 31–36. [CrossRef] 18. Corsini, I.; Polvani, S.; Tarocchi, M.; Tempesti, S.; Marroncini, G.; Generoso, M.; Bresci, C.; Gozzini, E.; Bianconi, T.; Galli, A.; et al. 4 Discussion 4. Discussion Funding: This research was funded by the Spanish Ministry of Economy, Industry and Competitiveness (AGL2013-48459-P). BG-P holds a fellowship from the Spanish Ministry of Economy, Industry and Competitiveness (BES-2014-068134). PT-C holds a fellowship from the National Secretary of Higher Education, Science, Technology and Innovation of Ecuador (SENESCYT-DMPF-2015-1666-CO). 13 of 16 13 of 16 Nutrients 2019, 11, 999 Acknowledgments: The authors would like to thank Lidia Marín-Morote, Clara Duran, Caterina Mascaró and Elisa Alòs for their help with the laboratory work. We also thank the Biochemistry Section of the Faculty of Pharmacy and Food Science from the University of Barcelona for the use of Modulus™Microplate spectrophotometer. Conﬂicts of Interest: The authors declare no conﬂict of interest. Conﬂicts of Interest: The authors declare no conﬂict of interest. References 1. Rich-Edwards, J.W.; Klungsoyr, K.; Wilcox, A.J.; Skjaerven, R. Duration of pregnancy, even at term, predicts long-term risk of coronary heart disease and stroke mortality in women: A population-based study. Am. J. Obs. Gynecol. 2015, 15, 518.e1–518.e8. [CrossRef] [PubMed] 1. Rich-Edwards, J.W.; Klungsoyr, K.; Wilcox, A.J.; Skjaerven, R. Duration of pregnancy, even at term, predicts long-term risk of coronary heart disease and stroke mortality in women: A population-based study. Am. J. Obs. Gynecol. 2015, 15, 518.e1–518.e8. [CrossRef] [PubMed] 2. Melville, J.M.; Moss, T.J. The immune consequences of preterm birth. Front. Neurosci. 2013, 7, 79. [CrossRef] [PubMed] 2. Melville, J.M.; Moss, T.J. The immune consequences of preterm birth. Front. Neurosci. 2013, 7, 79. [CrossRef] [PubMed] 3. Frey, H.A.; Klebanoﬀ, M.A. The epidemiology, etiology, and costs of preterm birth. Semin. Fetal Neonatal Med. 2016, 21, 68–73. [CrossRef] 3. Frey, H.A.; Klebanoﬀ, M.A. The epidemiology, etiology, and costs of preterm birth. Semin. Fetal Neonatal Med. 2016, 21, 68–73. [CrossRef] 4. Lewis, E.D.; Richard, C.; Larsen, B.M.; Field, C.J. The importance of human milk for immunity in preterm infants. Clin. Perinatol. 2017, 44, 23–47. [CrossRef] 4. Lewis, E.D.; Richard, C.; Larsen, B.M.; Field, C.J. The importance of human milk for immunity in preterm infants. Clin. Perinatol. 2017, 44, 23–47. [CrossRef] 5. Chang, E. Preterm birth and the role of neuroprotection. BMJ 2015, 350, g6661. [CrossRef] 5. Chang, E. Preterm birth and the role of neuroprotection. BMJ 2015, 350, g6661. [CrossRef] 6. Rubarth, L.B.; Quinn, J. Respiratory development and respiratory distress syndrome. Neonatal Netw. 2015, 34, 231–238. [CrossRef] 6. Rubarth, L.B.; Quinn, J. Respiratory development and respiratory distress syndrome. Neonatal Netw. 2015, 34, 231–238. [CrossRef] Gritz, E.C.; Bhandari, V. The human neonatal gut microbiome: A brief review. Front. Pediatr. 2015, 3, 17. 8. Lawn, J.E.; Kerber, K.; Enweronu-Laryea, C.; Cousens, S. 3.6 Million neonatal deaths-what is progressing and what is not? Semin. Perinatol. 2010, 34, 371–386. [CrossRef] 9. Strunk, T.; Currie, A.; Richmond, P.; Simmer, K.; Burgner, D. Innate immunity in human newborn infants: Prematurity means more than immaturity. J. Matern. Fetal Neonatal Med. 2011, 24, 25–31. [CrossRef] 9. Strunk, T.; Currie, A.; Richmond, P.; Simmer, K.; Burgner, D. Innate immunity in human newborn infants: Prematurity means more than immaturity. J. Matern. Fetal Neonatal Med. 2011, 24, 25–31. [CrossRef] 10. Adams-Chapman, I. Long-term impact of infection on the preterm neonate. Semin. Perinatol. 2012, 36, Prematurity means more than immaturity. J. Matern. Fetal Neonatal Med. References Peroxisome proliferator-activated receptor-γ agonist pioglitazone reduces the development of necrotizing enterocolitis in a neonatal preterm rat model. Pediatr. Res. 2017, 81, 364–368. [CrossRef] 19. Satoh, T.; Izumi, H.; Iwabuchi, N.; Odamaki, T.; Namba, K.; Abe, F.; Xiao, J.Z. Biﬁdobacterium breve prevents necrotising enterocolitis by suppressing inﬂammatory responses in a preterm rat model. Benef. Microbes 2016, 7, 75–82. [CrossRef] 14 of 16 Nutrients 2019, 11, 999 20. Nguyen, D.N.; Jiang, P.; Frøkiær, H.; Heegaard, P.M.; Thymann, T.; Sangild, P.T. Delayed development of systemic immunity in preterm pigs as a model for preterm infants. Sci. Rep. 2016, 6, 36816. [CrossRef] 21. Pérez-Cano, F.J.; Franch, À.; Castellote, C.; Castell, M. The suckling rat as a model for immunonutrition studies in early life. Clin. Dev. Immunol. 2012, 2012, 537310. [CrossRef] [PubMed] 22. Appleby, C.J.; Towner, R.A. Magnetic resonance imaging of pulmonary damage in the term and premature rat neonate exposed to hyperoxia. Pediatr. Res. 2001, 50, 502–507. [CrossRef] [PubMed] 23. Grases-Pintó, B.; Abril-Gil, M.; Rodríguez-Lagunas, M.J.; Castell, M.; Pérez-Cano, F.J.; Franch, À. Leptin and adiponectin supplementation modiﬁes mesenteric lymph node lymphocyte composition and functionality in suckling rats. Br. J. Nutr. 2018, 119, 486–495. [CrossRef] [PubMed] 24. Azagra-Boronat, I.; Massot-Cladera, M.; Knipping, K.; Van’t Land, B.; Stahl, B.; Garssen, J.; Rodríguez-Lagunas, M.J.; Franch, À.; Castell, M.; Pérez-Cano, F.J. Supplementation with 2′-FL and scGOS/lcFOS ameliorates rotavirus-induced diarrhea in suckling rats. Front. Cell. Infect. Microbiol. 2018, 8, 372. [CrossRef] [PubMed] 25. Pérez-Cano, F.J.; Marín-Gallén, S.; Castell, M.; Rodríguez-Palmero, M.; Rivero, M.; Franch, À.; Castellote, C. Bovine whey protein concentrate supplementation modulates maturation of immune system in suckling rats. Br. J. Nutr. 2007, 98 Suppl 1, S80–S84. [CrossRef] 26. Rigo-Adrover, M.M.; Franch, À.; Castell, M.; Pérez-Cano, F.J. Preclinical immunomodulation by the p Biﬁdobacterium breve M-16V in early life. PLoS One 2016, 11, e0166082. [CrossRef] 27. Grases-Pintó, B.; Abril-Gil, M.; Castell, M.; Pérez-Cano, F.J.; Franch, À. Enhancement of immune maturation in suckling rats by leptin and adiponectin supplementation. Sci. Rep. 2019, 9, 1786. [CrossRef] 28. Torres-Castro, P.; Abril-Gil, M.; Rodríguez-Lagunas, M.J.; Castell, M.; Pérez-Cano, F.J.; Franch, À. TGF-β2, EGF, and FGF21 growth factors present in breast milk promote mesenteric lymph node lymphocytes maturation in suckling rats. Nutrients 2018, 10, 1171. [CrossRef] 29. Ramírez-Santana, C.; Pérez-Cano, F.J.; Castellote, C.; Castell, M.; Rivero, M.; Rodríguez-Palmero, M.; Franch, À. Higher immunoglobulin production in conjugated linoleic acid-supplemented rats during gestation and suckling. Br. J. Nutr. 2009, 102, 858–868. [CrossRef] 30. References Rigo-Adrover, M.M.; van Limpt, K.; Knipping, K.; Garssen, J.; Knol, J.; Costabile, A.; Franch, À.; Castell, M.; Pérez-Cano, F.J. Preventive eﬀect of a synbiotic combination of galacto- and fructooligosaccharides mixture with Biﬁdobacterium breve M-16V in a model of multiple rotavirus infections. Front. Immunol. 2018, 9, 1318. [CrossRef] 31. Puiman, P.; Stoll, B. Animal models to study neonatal nutrition in humans. Curr. Opin. Clin. Nutr. Metab. Care 2008, 11, 601–606. [CrossRef] [PubMed] 2. Sengupta, P. The laboratory rat: Relating its age with human’s. Int. J. Prev. Med. 2013, 4, 624–630. [PubM 32. Sengupta, P. The laboratory rat: Relating its age with human’s. Int. J. Prev. Med. 2013, 4, 624–630. [PubMed] 33. Underwood, M.A. Human milk for the premature infant. Pediatr. Clin. North Am. 2013, 60, 189–207. [CrossRef] 33. Underwood, M.A. Human milk for the premature infant. Pediatr. Clin. North Am. 2013, 60, 189–207. [CrossRef] 34. Migraine, A.; Nicklaus, S.; Parnet, P.; Lange, C.; Monnery-Patris, S.; Des Robert, C.; Darmaun, D.; Flamant, C.; Amarger, V.; Rozé, J.C. Eﬀect of preterm birth and birth weight on eating behavior at 2 years of age. Am. J. Clin. Nutr. 2013, 97, 1270–1277. [CrossRef] 35. Andres, O.; Schulze, H.; Speer, C.P. Platelets in neonates: Central mediators in haemostasis, antimicrobial defence and inﬂammation. Thromb. Haemost. 2015, 113, 3–12. 36. Cowman, J.; Quinn, N.; Geoghegan, S.; Müllers, S.; Oglesby, I.; Byrne, B.; Somers, M.; Ralph, A.; Voisin, B.; Ricco, A.J.; et al. Dynamic platelet function on von Willebrand factor is diﬀerent in preterm neonates and full-term neonates: Changes in neonatal platelet function. J. Thromb. Haemost. 2016, 14, 2027–2035. [CrossRef] 37. Hoﬀbrand, A.V. Folate deﬁciency in premature infants. Arch. Dis. Child. 1970, 45, 441–444. [CrossRef] 37. Hoﬀbrand, A.V. Folate deﬁciency in premature infants. Arch. Dis. Child. 1970, 45, 441–444. [CrossRef] 38. Alkan Ozdemir, S.; Ozer, E.A.; Kose, S.; Ilhan, O.; Ozturk, C.; Sutcuoglu, S. Reference values of serum IgG and IgM levels in preterm and term newborns. J. Matern. Fetal Neonatal Med. 2016, 29, 972–976. [CrossRef] 39. Palmeira, P.; Quinello, C.; Silveira-Lessa, A.L.; Zago, C.A.; Carneiro-Sampaio, M. IgG placental transfer in healthy and pathological pregnancies. Clin. Dev. Immunol. 2012, 2012, 985646. [CrossRef] 40 W Y Ji X H J Di i T Zh X Q tit ti i ti ti l ti b t I G 37. Hoﬀbrand, A.V. Folate deﬁciency in premature infants. Arch. Dis. Child. 1970, 45, 441–444. [CrossRef] 38. References Walthall, K.; Cappon, G.D.; Hurtt, M.E.; Zoetis, T. Postnatal development of the gastrointestinal system: A species comparison. Birth Defects Res. B Dev. Reprod. Toxicol. 2005, 74, 132–156. [CrossRef] [PubMed] 49. Patel, R.M.; Myers, L.S.; Kurundkar, A.R.; Maheshwari, A.; Nusrat, A.; Lin, P.W. Probiotic bacteria induce maturation of intestinal claudin 3 expression and barrier function. Am. J. Pathol. 2012, 180, 626–635. [CrossRef] [PubMed] 50. van Elburg, R.M.; Fetter, W.P.; Bunkers, C.M.; Heymans, H.S. Intestinal permeability in relation to birth weight and gestational and postnatal age. Arch. Dis. Child. Fetal Neonatal Ed. 2003, 88, F52–F55. [CrossRef] [PubMed] 51. Weaver, L.T.; Laker, M.F.; Nelson, R. Intestinal permeability in the newborn. Arch. Dis. Child. 1984, 59, 236–241. [CrossRef] 52. Sangild, P.T.; Thymann, T.; Schmidt, M.; Stoll, B.; Burrin, D.G.; Buddington, R.K. Invited review: The preterm pig as a model in pediatric gastroenterology. J. Anim. Sci. 2013, 91, 4713–4729. [CrossRef] [PubMed] 53. Anand, R.J.; Leaphart, C.L.; Mollen, K.P.; Hackam, D.J. The role of the intestinal barrier in the pathogenesis of necrotizing enterocolitis. Shock 2007, 27, 124–133. [CrossRef] [PubMed] 54. Bergmann, K.R.; Liu, S.X.; Tian, R.; Kushnir, A.; Turner, J.R.; Li, H.L.; Chou, P.M.; Weber, C.R.; De Plaen, I.G. Biﬁdobacteria stabilize claudins at tight junctions and prevent intestinal barrier dysfunction in mouse necrotizing enterocolitis. Am. J. Pathol. 2013, 182, 1596–1606. [CrossRef] g 55. Le Huërou-Luron, I.; Blat, S.; Boudry, G. Breast- v. formula-feeding: Impacts on the digestive tract and immediate and long-term health eﬀects. Nutr. Res. Rev. 2010, 23, 23–36. [CrossRef] 56. Taylor, S.N.; Basile, L.A.; Ebeling, M.; Wagner, C.L. Intestinal permeability in preterm infants by feeding type: mother’s milk versus formula. Breastfeed Med. 2009, 4, 11–15. [CrossRef] [PubMed] 57. Saleem, B.; Okogbule-Wonodi, A.C.; Fasano, A.; Magder, L.S.; Ravel, J.; Kapoor, S.; Viscardi, R.M. Intestinal barrier maturation in very low birthweight infants: Relationship to feeding and antibiotic exposure. J. Pediatr. 2017, 183, 31.e1–36.e1. [CrossRef] [PubMed] 58. Colaizy, T.T.; Bartick, M.C.; Jegier, B.J.; Green, B.D.; Reinhold, A.G.; Schaefer, A.J.; Bogen, D.L.; Schwarz, E.B.; Stuebe, A.M.; Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network. Impact of optimized breastfeeding on the costs of necrotizing enterocolitis in extremely low birthweight infants. J. Pediatr. 2016, 175, 100–105.e2. [CrossRef] [PubMed] 59. Sullivan, S.; Schanler, R.J.; Kim, J.H.; Patel, A.L.; Trawöger, R.; Kiechl-Kohlendorfer, U.; Chan, G.M.; Blanco, C.L.; Abrams, S.; Cotten, C.M.; et al. References Alkan Ozdemir, S.; Ozer, E.A.; Kose, S.; Ilhan, O.; Ozturk, C.; Sutcuoglu, S. Reference values of serum IgG and IgM levels in preterm and term newborns. J. Matern. Fetal Neonatal Med. 2016, 29, 972–976. [CrossRef] 39 Palmeira P; Quinello C ; Silveira Lessa A L ; Zago C A ; Carneiro Sampaio M IgG placental transfer in 38. Alkan Ozdemir, S.; Ozer, E.A.; Kose, S.; Ilhan, O.; Ozturk, C.; Sutcuoglu, S. Reference values of serum IgG and IgM levels in preterm and term newborns. J. Matern. Fetal Neonatal Med. 2016, 29, 972–976. [CrossRef] 39. Palmeira, P.; Quinello, C.; Silveira-Lessa, A.L.; Zago, C.A.; Carneiro-Sampaio, M. IgG placental transfer in healthy and pathological pregnancies. Clin. Dev. Immunol. 2012, 2012, 985646. [CrossRef] 40. Wang, Y.; Jiang, X.; He, J.; Diraviyam, T.; Zhang, X. Quantitative investigation on correlation between IgG and FcRn during gestation and lactating periods in rat. Am. J. Reprod. Immunol. 2016, 75, 81–85. [CrossRef] 15 of 16 15 of 16 Nutrients 2019, 11, 999 41. Desesso, J.M.; Williams, A.L.; Ahuja, A.; Bowman, C.J.; Hurtt, M.E. The placenta, transfer of immunoglobulins, and safety assessment of biopharmaceuticals in pregnancy. Crit. Rev. Toxicol. 2012, 42, 185–210. [CrossRef] [PubMed] 42. Borghesi, J.; Mario, L.C.; Rodrigues, M.N.; Favaron, P.O.; Miglino, M.A. Immunoglobulin transport gestation in domestic animals and humans — A Review. Open J. Anim. Sci. 2014, 4, 323–336. [Cross 43. Collins, A.; Weitkamp, J.H.; Wynn, J.L. Why are preterm newborns at increased risk of infection? Arch. Dis. Child. Fetal Neonatal Ed. 2018, 103, F391–F394. [CrossRef] 44. Mirpuri, J.; Raetz, M.; Sturge, C.R.; Wilhelm, C.L.; Benson, A.; Savani, R.C.; Hooper, L.V.; Yarovinsky, F. Proteobacteria-speciﬁc IgA regulates maturation of the intestinal microbiota. Gut Microbes. 2014, 5, 28–39. [CrossRef] 45. Gustafson, C.E.; Higbee, D.; Yeckes, A.R.; Wilson, C.C.; De Zoeten, E.F.; Jedlicka, P.; Janoﬀ, E.N. Limited expression of APRIL and its receptors prior to intestinal IgA plasma cell development during human infancy. Mucosal Immunol. 2014, 7, 467–477. [CrossRef] 46. Prosser, A.; Hibbert, J.; Strunk, T.; Kok, C.H.; Simmer, K.; Richmond, P.; Burgner, D.; Currie, A. Phagocytosis of neonatal pathogens by peripheral blood neutrophils and monocytes from newborn preterm and term infants. Pediatr. Res. 2013, 74, 503–510. [CrossRef] [PubMed] 7. de Jong, E.; Strunk, T.; Burgner, D.; Lavoie, P.M.; Currie, A. The phenotype and function of preterm in monocytes: Implications for susceptibility to infection. J. Leukoc. Biol. 2017, 102, 645–656. [CrossRef] 48. References An exclusively human milk-based diet is associated with a lower rate of necrotizing enterocolitis than a diet of human milk and bovine milk-based products. J. Pediatr. 2010, 156, 562–567. [CrossRef] 60. Clark, J.A.; Doelle, S.M.; Halpern, M.D.; Saunders, T.A.; Holubec, H.; Dvorak, K.; Boitano, S.A.; Dvorak, B. Intestinal barrier failure during experimental necrotizing enterocolitis: Protective eﬀect of EGF treatment. Am. J. Physiol. Liver Physiol. 2006, 291, G938–G949. [CrossRef] 16 of 16 Nutrients 2019, 11, 999 16 of 16 61. Hackam, D.J.; Upperman, J.S.; Grishin, A.; Ford, H.R. Disordered enterocyte signaling and intestinal barrier dysfunction in the pathogenesis of necrotizing enterocolitis. Semin. Pediatr. Surg. 2005, 14, 49–57. [CrossRef] [PubMed] 62. Stelloh, C.; Allen, K.P.; Mattson, D.L.; Lerch-Gaggl, A.; Reddy, S.; El-Meanawy, A. Prematurity in mi to reduction in nephron number, hypertension, and proteinuria. Transl. Res. 2012, 159, 80–89. [Cros 63. Castellote, C.; Casillas, R.; Ramírez-Santana, C.; Pérez-Cano, F.J.; Castell, M.; Moretones, M.G.; López-Sabater, M.C.; Franch, A. Premature delivery inﬂuences the immunological composition of colostrum and transitional and mature human milk. J. Nutr. 2011, 1181–1187. [CrossRef] 64. Amaral, M.A.; Guedes, G.H.B.F.; Epifanio, M.; Wagner, M.B.; Jones, M.H.; Mattiello, R. Network meta-analysis of probiotics to prevent respiratory infections in children and adolescents. Pediatr. Pulmonol. 2017, 52, 833–843. [CrossRef] 65. Wang, H.T.; Anvari, S.; Anagnostou, K. The role of probiotics in preventing allergic disease. Children 2019, 6, 24. [CrossRef] 66. Nauta, A.J.; Ben Amor, K.; Knol, J.; Garssen, J.; van der Beek, E.M. Relevance of pre- and postnatal nutrition to development and interplay between the microbiota and metabolic and immune systems. Am. J. Clin. Nutr. 2013, 98, 586S–593S. [CrossRef] 67. Hagen, P.C.; Skelley, J.W. Eﬃcacy of Biﬁdobacterium species in prevention of necrotizing enterocolitis in very-low birth weight infants. A systematic review. J. Pediatr. Pharmacol. Ther. 2019, 24, 10–15. [CrossRef] [PubMed] 68. Sherman, M.P.; Miller, M.M.; Sherman, J.; Niklas, V. Lactoferrin and necrotizing enterocolitis. Curr. Opin. Pediatr. 2014, 26, 146–150. [CrossRef] © 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
https://openalex.org/W2903041604	https://www.frontiersin.org/articles/10.3389/fnut.2018.00115/pdf	English	null	Safety of Creatine Supplementation in Active Adolescents and Youth: A Brief Review	Frontiers in nutrition	2,018	cc-by	11,020	Edited by: Bruno Gualano, University of São Paulo, Brazil Reviewed by: Scott Forbes, Brandon University, Canada Lem W. Taylor, University of Mary Hardin–Baylor, United States Keywords: creatine, safety, adverse events, supplementation, adolescents, youth, performance Correspondence: Chad M. Kerksick ckerksick@lindenwood.edu orcid.org/0000-0003-0458-7294 †Andrew R. Jagim orcid.org/0000-0002-6651-5096 Patrick S. Harty orcid.org/0000-0003-0004-2940 Correspondence: Chad M. Kerksick ckerksick@lindenwood.edu orcid.org/0000-0003-0458-7294 Safety of Creatine Supplementation in Active Adolescents and Youth: A Brief Review Andrew R. Jagim 1,2†, Richard A. Stecker 1, Patrick S. Harty 1†, Jacob L. Erickson 2 and Chad M. Kerksick 1* 1 Exercise and Performance Nutrition Laboratory, Department of Exercise Science, Lindenwood University, St. Charles, MO, United States, 2 Mayo Clinic Health Systems, Onalaska, WI, United States Creatine has been extensively researched and is well-supported as one of the most effective dietary supplements available. There is overwhelming support within the literature regarding the ability of creatine to augment performance following short term (5–7 days) and long-duration supplementation periods. There is also strong support for creatine regarding its safety proﬁle and minimal risk for adverse events or any negative inﬂuence on markers of clinical health and safety. Recent research has also highlighted the ability of creatine to confer several health-related beneﬁts in select clinical populations in addition to offering cognitive beneﬁts. Creatine is also a popular supplement of choice for adolescent athletes; however, research in this area is extremely limited, particularly when examining the safety and efﬁcacy of creatine supplementation in this population. Therefore, the purpose of this review was to highlight the limited number of studies available in adolescent populations and systematically discuss the topic of safety of creatine supplementation in a younger population. REVIEW REVIEW published: 28 November 2018 doi: 10.3389/fnut.2018.00115 published: 28 November 2018 doi: 10.3389/fnut.2018.00115 INTRODUCTION †Andrew R. Jagim orcid.org/0000-0002-6651-5096 Patrick S. Harty orcid.org/0000-0003-0004-2940 †Andrew R. Jagim orcid.org/0000-0002-6651-5096 Patrick S. Harty orcid.org/0000-0003-0004-2940 Ergogenic aids are broadly deﬁned as any training technique, mechanical device, nutritional agent or practice, pharmacological method, or psychological technique that can improve exercise performance capacity or enhance training adaptations (1). As a dietary supplement, creatine is regarded as one of the more well-researched and eﬃcacious nutritional ergogenic aids available to athletes (2). When supplemented in the human diet, creatine increases intramuscular creatine stores and can improve exercise capacity and training adaptations. More recently, creatine has been established as a legitimate nutritional adjunct in areas related to rehabilitation and neuromuscular disorders with growing evidence for some therapeutic eﬃcacy in thermoregulation, concussions, head trauma, autism, and neuroprotection (2). Specialty section: This article was submitted to Sport and Exercise Nutrition, a section of the journal Frontiers in Nutrition Specialty section: This article was submitted to Sport and Exercise Nutrition, a section of the journal Frontiers in Nutrition Received: 26 September 2018 Accepted: 12 November 2018 Published: 28 November 2018 Creatine is a naturally occurring guanidino compound composed of two amino acids (arginine-glycine) found primarily in the ﬂesh of animals, with the majority (∼95%) being present in skeletal muscle (2). Approximately two-thirds of intramuscular creatine is phosphocreatine (PCr) while the remaining proportion is free creatine (Cr). Harris et al. (3) demonstrated that a 70-kg individual averages around 120 mM/kg dry muscle mass of total creatine (PCr + Cr) with an upper limit (after supplementation) of up to 160 mM/kg dry muscle mass. Ongoing cellular metabolism of creatine results in 1–2% being degraded, leading to a daily total (exogenous + endogenous) of 1–3 g of creatine required per day to maintain normal phosphocreatine levels. Depending on food Supplementation Regimens and Considerations Efﬁcacy of Creatine as an Ergogenic Aid Use of creatine in athletes can be traced back to the 1990s and since that time, hundreds of papers have been published examining the impact of creatine supplementation on physical performance. In this respect, comprehensive reviews exist on this topic and the interested reader is encouraged to read them (2, 36–39). However, it is beyond the scope of this current review to discuss this literature in detail. Brieﬂy, several studies have indicated that supplementation periods as short as 3–5 days are suﬃcient to confer an ergogenic beneﬁt with improvements in exercise capacity, anaerobic capacity, power, and sport-speciﬁc tasks consistently being observed (36, 40). When supplementation stretches to several weeks, augmented training adaptations such as greater improvements in strength, lean body mass and anaerobic performance, when used in conjunction with a structured training program, are commonly reported (2, 37, 41–43). Finally, and as summarized by Kreider et al. (2), these outcomes are consistently reported across genders as well as in adolescents (44–48), younger adults (32, 49–59) and older individuals (38, 41, 60–68). Table 2 highlights many examples of sporting events that may be enhanced by creatine supplementation. As an ergogenic aid and in summary, creatine supplementation is commonly one of the most highly recommended and advocated by researchers and professional organizations (1, 2, 79). Creatine is stored within skeletal muscle as both free creatine and as phosphocreatine serving as a key substrate for substrate-level phosphorylation and the resynthesis of ATP (4). Collectively, these high-energy phosphates play a vital role in several metabolic processes within the body, particularly during the ﬁrst 10 s of high-intensity exercise such that depletion of PCr is accepted as a major source of fatigue (5, 6). Creatine can be found in animal food sources and typically daily ingestion equates to 1–2 g per day; however ∼1–2% of intramuscular creatine is non-enzymatically metabolized into creatinine daily (2). It is well-supported within the literature that exogenous supplementation of creatine is an eﬀective strategy to increase intramuscular phosphocreatine stores by ∼20–40% (2) depending on baseline levels. Commonly, a loading dose of 0.3 g/kg of body weight per day for 5–7 days is used as an eﬀective “loading protocol,” with a subsequent daily dose of 0.03 g/kg of body weight (3–5 g/day) required for the maintenance of elevated PCr levels (3, 7, 8). Alternatively, Hultman et al. Citation: Jagim AR, Stecker RA, Harty PS, Erickson JL and Kerksick CM (2018) Safety of Creatine Supplementation in Active Adolescents and Youth: A Brief Review. Front. Nutr. 5:115. doi: 10.3389/fnut.2018.00115 November 2018 \| Volume 5 \| Article 115 Frontiers in Nutrition \| www.frontiersin.org Creatine Use and Adolescents Jagim et al. tastes and preferences, the human diet commonly provides up to half of this amount with foods such as ﬁsh and meats being the main sources with uncooked beef and salmon delivering around 1–2 g of creatine per pound. The remaining amount of the total rate of appearance is provided by endogenous synthesis by the kidney and liver (2). eﬀect of creatine supplementation is weight gain (primarily fat-free mass), which is often a desired outcome in athletes, primarily ones with an emphasis placed on strength, power and body size, and clinical patients with any type of muscle wasting disorders (2). A summary of these studies can be found in Table 1. Supplementation Regimens and Considerations (7) demonstrated that a supplementation regimen of 3 g/day for 28 days can also result in similar levels of intramuscular creatine and phosphocreatine. Finally, once supplementation ceases, studies suggest that it may take as long 4–6 weeks before intramuscular phosphocreatine levels return to baseline (7, 9, 10). Likely due to the long half- life in muscle, studies have failed to highlight instances where endogenous production is reduced or dampened by previous creatine supplementation after supplementation is stopped. In this respect, it was commonly suggested that creatine users should cycle on and oﬀ, but continued evidence suggests that this is not necessary (11, 12) particularly when the multitude of beneﬁts resulting from ongoing creatine supplementation are considered. PREVALENCE OF CREATINE USE IN ADOLESCENTS A key question to answer relative to the premise of this paper is, “Are adolescent athletes using creatine?” In this respect, the use of dietary supplements to enhance performance or health is not limited to adult populations and is an increasingly popular strategy among young, active individuals (80), as outlined in Table 3. When surveyed, ∼5–20% of middle school and high school aged individuals reported taking creatine at some point. For example, Metzl and associates (70) surveyed 1,103 US girls and boys in grades 6–12 and reported that 8.8% of boys and 1.8% of girls reported supplementing with creatine. Interestingly, creatine use was consistent at 3.4% in grades 6–10 while substantially increasing to 12 and 44% use in grades 11 and 12, respectively. Similarly, Kayton et al. (71) surveyed 270 US high school students 13–18 years of age and found that 21% of boys and 3% of girls reported using creatine. When prevalence rates among adolescent athletes are evaluated, similar trends emerge as creatine is often listed as one of the more commonly used dietary supplements among this population (76, 81). Regardless, there does still appear to be variability in prevalence rates, which Frontiers in Nutrition \| www.frontiersin.org Safety of Creatine Use In adults, a growing number of published randomized controlled trials are available that support the safety of creatine supplementation. These studies have been conducted in both athletic and general populations and range from as short as a few days to as long as 5 years without any adverse changes in markers of clinical health (12, 13). Multiple studies have assessed and reported that creatine supplementation has no adverse impact on clinical health markers in competitive athletes (13–17), non-athletic populations (18–25), and in clinical populations (26–29). Furthermore, recent evidence suggests that creatine supplementation is unrelated to the formation of carcinogenic heterocyclic amines in humans, which was a long-standing concern due to creatine’s potential role as a precursor of the compounds (30). Generally, the only clinically-relevant side November 2018 \| Volume 5 \| Article 115 Frontiers in Nutrition \| www.frontiersin.org 2 Jagim et al. Creatine Use and Adolescents \| y Study Subjects Design Duration Dosing protocol Clinical Safety measures Presence of signiﬁcant adverse events References ATHLETIC POPULATIONS Armentano et al. 35 (male: 20; female: 15) active duty US army volunteers Double-blind, placebo-controlled. 14 Days 5 g/day on Days 1–6, 20 g/day on Days 7–14 Blood and Urine Markers, Blood Pressure None (31) Cancela et al. 14 male football players Randomized, double-blind, placebo controlled 8 Weeks 15 g/day (1 week) 3 g/day (7 weeks) Blood and Urine Markers, Hemodynamic variables, None (14) Galvan et al. Study 1: 13 recreationally-active males Study 2: 48 recreationally-active males Study 1: Randomized, double-blind, placebo controlled, crossover. Study 2: Randomized, double-blind, placebo controlled Study 1: 1 day per condition Study 2: 28 Days Study 1: 1.5, 3, or 5 g in an acute dose Study 2: 6, 12, or 20 g/day for 7 days; 1.5 , 3 , or 5 g/day for 21 days Blood Markers, Hemodynamic variables, Side effects. None (32) Greenwood et al. 72 NCAA Division 1A football players Open-label supplement intervention 120 Days 0.3 g/kg/day for 5 day, 0.03 g/kg/day for 115 day Injury Rates, Cramping Injury rates and cramping were signiﬁcantly lower in creatine users. (15) Greenwood et al. Approximately 130 Open-label supplement intervention Mixed duration of use 15.75 g/day for 5 days, 5 g/day for remainder of intervention Injury Rates, Cramping Injury rates and cramping were generally lower or proportional in creatine users. (16) Joy et al. 58 healthy males and females Randomized, double-blind, placebo controlled. Frontiers in Nutrition \| www.frontiersin.org November 2018 \| Volume 5 \| Article 115 Safety of Creatine Use 28 Days 1 g/day, 2 g/day Blood Markers None (19) Kreider et al. 116 healthy NCAA Division 1A football players Longitudinal, open-label intervention. 2 Years 15.75 g/day for 5 days, 5 g/day for up to 2 years. Blood, Serum, and Urinary Markers None (12) Lugaresi et al. 26 healthy, resistance-trained males Randomized, double-blind, placebo controlled 12 Weeks 20 g/d for 5 days, 5 g/day for remainder of trial. Blood Markers, Markers of Kidney Function None (33) Mayhew et al. 23 healthy NCAA Division 2 football players Retrospective Design Mixed duration of use prior to data collection Spontaneous use Blood Markers None (17) Poortmans and Francaux 9 (male: 8; female: 1) healthy athletes Retrospective Design Mixed duration of use prior to data collection (10 months to 5 year) Spontaneous use (1–80 g/day) Blood Markers None (13) Robinson et al. 48 (23 male; 25 female) Randomized, double-blind, placebo controlled 5 days−9 weeks. 20 g/day for 5 days, 20 g/day for 5 days followed by 3 g/day for 8 weeks Blood Markers None (34) (C ti d) November 2018 \| Volume 5 \| Article 115 3 Jagim et al. Creatine Use and Adolescents Study Subjects Design Duration Dosing protocol Clinical Safety measures Presence of signiﬁcant adverse events References NON-ATHLETIC POPULATIONS Groeneveld et al. 175 ALS patients (male: 120; female: 55) healthy athletes Randomized, double-blind, placebo-controlled. Mixed duration of use; approximately 310 days. 10 g/day Blood Markers, Side effects, Markers of Kidney Function None (35) Gualano et al. 18 healthy sedentary males Randomized, double-blind, placebo-controlled. 3 Months 10 g/day for 3 months Blood Markers, Markers of Kidney Function None (18) Gualano et al. 25 (male: 16; female: 9) Type II Diabetic Patients Randomized, double-blind, placebo-controlled. 12 Weeks 5 g/day Blood Markers None (27) Lobo et al. 109 Osteopenic, postmenopausal females Randomized, double-blind, placebo-controlled. 1 Year 1 g/day Blood Markers, Adverse events. None (20) Mihic et al. 30 (male: 15; female: 15) healthy adults Randomized, double-blind, placebo-controlled. 5 Days 20 g/day for 5 days Blood Markers, Hemodynamic variables None (21) Neves et al. 26 postmenopausal females diagnosed with knee osteoarthritis Randomized, double-blind, placebo-controlled. 12 Weeks 20 g/day for 7 days, 5 g/day for 11 weeks. Markers of Kidney Function None (22) Poortmans et al. 5 healthy males Placebo-controlled 5 Days 20 g/day Blood and Urine Markers, Markers of Kidney Function None (25) Ropero-Miller et al. November 2018 \| Volume 5 \| Article 115 Frontiers in Nutrition \| www.frontiersin.org Safety of Creatine Use 4 (male: 2; female: 2) healthy subjects Uncontrolled intervention 10 Days 20 g/day for 5 days, 5 g/day for 5 days Urine Markers None (23) TABLE 1 \| Continued Study Subjects Design Duration Dosing protocol Clinical Safety measures Presence of signiﬁcant adverse events References NON-ATHLETIC POPULATIONS Groeneveld et al. 175 ALS patients (male: 120; female: 55) healthy athletes Randomized, double-blind, placebo-controlled. Mixed duration of use; approximately 310 days. 10 g/day Blood Markers, Side effects, Markers of Kidney Function None (35) Gualano et al. 18 healthy sedentary males Randomized, double-blind, placebo-controlled. 3 Months 10 g/day for 3 months Blood Markers, Markers of Kidney Function None (18) Gualano et al. 25 (male: 16; female: 9) Type II Diabetic Patients Randomized, double-blind, placebo-controlled. 12 Weeks 5 g/day Blood Markers None (27) Lobo et al. 109 Osteopenic, postmenopausal females Randomized, double-blind, placebo-controlled. 1 Year 1 g/day Blood Markers, Adverse events. None (20) Mihic et al. 30 (male: 15; female: 15) healthy adults Randomized, double-blind, placebo-controlled. 5 Days 20 g/day for 5 days Blood Markers, Hemodynamic variables None (21) Neves et al. 26 postmenopausal females diagnosed with knee osteoarthritis Randomized, double-blind, placebo-controlled. 12 Weeks 20 g/day for 7 days, 5 g/day for 11 weeks. Markers of Kidney Function None (22) Poortmans et al. 5 healthy males Placebo-controlled 5 Days 20 g/day Blood and Urine Markers, Markers of Kidney Function None (25) Ropero-Miller et al. 4 (male: 2; female: 2) healthy subjects Uncontrolled intervention 10 Days 20 g/day for 5 days, 5 g/day for 5 days Urine Markers None (23) NCAA, National college athletic association. tiers in Nutrition \| www.frontiersin.org 4 November 2018 \| Volume 5 \| Article 115 November 2018 \| Volume 5 \| Article 115 4 Creatine Use and Adolescents Jagim et al. TABLE 2 \| Dietary supplement use in youth and adolescent populations. References Population Key ﬁndings Smith and Dahm (69) U.S. High Schools Population: 328 students (55% boys, 45% girls, 15.2 ± 1.3 years) Usage Statistics: Creatine use of 8.2% (26 boys, 1 girl) Sports: Football (29%), soccer and hockey Perceived Efﬁcacy: 79% felt it improved their performance Frequency: 35% daily, 35% weekly, 30% rate usage Dosing: 55% could not recall dosage, 23% (5 g/day), 23% (5–10 g/day) Primary Information Source: Friends Metzl et al. (70) U.S.—Middle and High School Population: 1,103 (55% boys, 45% girls, Grades 6–12) Usage Statistics: Creatine use (8.8% boys, 1.8% girls). Safety of Creatine Use Creatine use was stable (3.4%) from grades 6–, 12% use in 11th grade and 44% use in 12th grade. Sports: Strength and power sports (football, wrestling, hockey) Reasons to Take: Improve performance (72%), Improve appearance (61%), and improve speed (40%) Reasons Not Taking: Safety concerns (46%), lack of perceived beneﬁt (20%), expense (13%) Kayton et al. (71) U.S.—High School Population: 270 athletic high school boys (45%) and girls (55%), 13–18 years Usage Statistics: Sports drinks were most common (59%), vitamin/minerals (46%), creatine (21% in boys, 3% in girls), amino acids (8% in boys, 1% girls) Reasons to Take: Gain muscle, increase energy, prevent illness Primary Information Source: Coaches, doctors and parents were primary sources of nutrition and dietary supplement information, parents have largest inﬂuence on use. O’Dea (72) Australia—Middle & High School Population: 78 students, grades 7–11, 11–18 years Usage Statistics: Sports drink (56%), vitamin/minerals (49%), energy drinks (42%), creatine (5%), protein supplements (4%). Creatine was taken only by boys, all others were consumed by both gender. Reasons for Taking: Energy production or boost Misc.: Athletes had little to no knowledge of adverse events Bell et al. (73) Canada—High School Population: 333 high school boys (57%) and girls (42%), 15.4 ± 1.1 years Usage Statistics: Vitamin/minerals (43%), protein (14%), creatine (5.3%). Boys reported similar use of vitamin/minerals as girls, greater protein and creatine use when compared to girls Reasons for Taking: Students taking protein (43%) and creatine (42%) believed it would help their performance Hoffman et al. (74) U.S.—High School Population: 3,248 students in grades 8–12 Usage Statistics: Vitamin/minerals (59%), energy drinks (32%), protein (15%), and creatine (7%). Boys reported greater use of energy drinks, protein and creatine with progressively higher levels of protein (40% of 12th grade boys) and creatine (22% of 12th grade boys) use occurring with age. Primary Information Source: Teachers (36%) and parents (16%). As grade levels increased, parents, friends, coaches, athletic trainers, and internet sites take on larger roles Petroczi and Naughton (75) U.K.—Young Elite Athletes Population: 403 elite athletes (12–21 years, 17.7 ± 2.0 years) Usage Statistics: Energy drinks (87%), vitamin/minerals (47%), whey protein (44%), and creatine (28%) Sports: Rugby, soccer and swimming Misc.: Large majority (78%) did not believe nutritional supplementation was needed to achieve success in sports Diehl et al. Frontiers in Nutrition \| www.frontiersin.org Safety of Creatine Use (76) Germany—Young Elite Athletes Population: 1,138 Olympic level competitors (14–18 years, 56% boys, 44% girls) Usage Statistics: Vitamin/minerals (34–69%), energy drinks (64%), protein (38%), creatine (12%) Evans et al. (77) U.S.—Youth Population: 73.7 million U.S. children (10.8 ± 0.2 years, 57% older than age of 10) Usage Statistics: 1.64% (1.2 million) used some form of supplement to enhance sport performance in last 30 days. 94.5% used vitamin/minerals, 44% used ﬁsh oils, 34% used creatine, 26% used ﬁber. Misc.: Boys were 2x as likely to use something. Independent of gender, usage increased with age (47% of 9–12 graders) Adapted from Kerksick and Fox (78) TABLE 2 \| Dietary supplement use in youth and adolescent populations. TABLE 2 \| Dietary supplement use in youth and adolescent populations. References Population Key ﬁnding Population: 333 high school boys (57%) and girls (42%), 15.4 ± 1.1 years Population: 333 high school boys (57%) and girls (42%), 15.4 ± 1.1 years Usage Statistics: Vitamin/minerals (43%), protein (14%), creatine (5.3%). Boys repo vitamin/minerals as girls, greater protein and creatine use when compared to girls vitamin/minerals as girls, greater protein and creatine use when compared to girls vitamin/minerals as girls, greater protein and creatine use when compared to girls Reasons for Taking: Students taking protein (43%) and creatine (42%) believed it would help their performance Population: 3,248 students in grades 8–12 Usage Statistics: Vitamin/minerals (59%), energy drinks (32%), protein (15%), and creatine (7%). Boys reported greater use of energy drinks, protein and creatine with progressively higher levels of protein (40% of 12th grade boys) and creatine (22% of 12th grade boys) use occurring with age. Primary Information Source: Teachers (36%) and parents (16%). As grade levels increased, parents, friends, coaches, athletic trainers, and internet sites take on larger roles Primary Information Source: Teachers (36%) and parents (16%). Adapted from Kerksick and Fox (78). Efﬁcacy of Creatine Use in Adolescents Efﬁcacy of Creatine Use in Adolescents Despite the overwhelming supportive body of literature regarding the eﬃcacy of creatine supplementation in adult athletes, limited data are available in adolescent athletes. This lack of available literature is likely attributable to ethical restrictions, safety concerns and methodological challenges. With that being said, Unnithan et al. (88) eloquently described the physiological basis for creatine use in adolescents and came to the conclusion that in anaerobic athletes, there exists a metabolic rationale as to how and why creatine may provide an ergogenic beneﬁt. Additionally, in a 2017 position stand published by the International Society of Sports Nutrition (2), it was recommended that “younger athletes should consider a creatine supplement if the following conditions are met...” and then went on to describe certain criteria surrounding approval from parents, choosing quality supplements, abiding by recommended dosing instructions and optimizing diet prior to supplementation. However, recent evidence suggests that exogenous creatine supplementation may be less eﬀective in children and adolescents compared to adults due to limited tissue creatine uptake, particularly in brain tissues (89, 90). MMA, Mixed martial arts; PCr, Phosphocreatine. Safety of Creatine Use Increased PCr • Track sprints: 60–200 m • Swim sprints: 50 m • Pursuit cycling Increased PCr Resynthesis • Basketball • Field Hockey • American Football • Ice Hockey • Lacrosse • Volleyball Reduced Muscle Acidosis • Downhill skiing • Water sports (e.g., Rowing, Canoe, Kayak, Stand-Up Paddling) • Swim events: 100, 200 m • Track events: 400, 800 m • Combat Sports (e.g., MMA, Wrestling, Boxing, etc.) Oxidative Metabolism • Basketball • Soccer • Team Handball • Tennis • Volleyball • Interval training in endurance athletes Increased Body Mass/Muscle Mass • American Football • Bodybuilding • Combat Sports (e.g., MMA, Wrestling, Boxing, etc.) • Powerlifting • Rugby • Track/Field events (Shot put; javelin; discus; hammer throw) • Olympic Weightlifting MMA, Mixed martial arts; PCr, Phosphocreatine. Reproduced from Kreider et al. (2) under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/). Safety of Creatine Use As grade levels increased, parents, friends, coaches, athletic trainers, and internet sites take on larger roles Population: 403 elite athletes (12–21 years, 17.7 ± 2.0 years) Population: 403 elite athletes (12–21 years, 17.7 ± 2.0 years) Usage Statistics: Energy drinks (87%), vitamin/minerals (47%), whey protein (44%), and creatine (28%) Sports: Rugby, soccer and swimming Misc.: Large majority (78%) did not believe nutritional supplementation was needed to achieve success in sports Misc.: Large majority (78%) did not believe nutritional supplementation was needed to achieve success in sports Population: 1,138 Olympic level competitors (14–18 years, 56% boys, 44% girls) Usage Statistics: Vitamin/minerals (34–69%), energy drinks (64%), protein (38%), c Population: 1,138 Olympic level competitors (14–18 years, 56% boys, 44% girls) Usage Statistics: Vitamin/minerals (34–69%), energy drinks (64%), protein (38%), c Population: 73.7 million U.S. children (10.8 ± 0.2 years, 57% older than age of 10) Usage Statistics: 1.64% (1.2 million) used some form of supplement to enhance sport performance in last 30 days. 94.5% used vitamin/minerals, 44% used ﬁsh oils, 34% used creatine, 26% used ﬁber. Misc.: Boys were 2x as likely to use something. Independent of gender, usage increased with age (47% of 9–12 graders) Adapted from Kerksick and Fox (78). is likely attributable to diﬀerences in gender, sport-type and associated body composition or training-related goals (70, 75, 76, 82, 83). Males appear to be more likely than females to report using creatine and the most commonly reported reasons for supplementation often include a desire to increase lean body mass and for increased energy production. As a result, strength and power or anaerobic-type sports such as football, wrestling, and hockey appear to have the highest rates of use. For example, McGuine et al. (81) reported that 16.7% of 4,000 surveyed high school athletes reported using creatine, with prevalence rates as high as 30.1% in football players and as low as 1.3% in female cross-country runners. In a similar study, Smith and Dahm (69) reported 8.2% of surveyed high school athletes reporting taking creatine, though reported use was as high as 21% in all football players surveyed. These prevalence rates certainly highlight the fact that creatine is a popular dietary supplement choice of adolescents, highlighting the need for more research in this area. November 2018 \| Volume 5 \| Article 115 Frontiers in Nutrition \| www.frontiersin.org 5 Jagim et al. Frontiers in Nutrition \| www.frontiersin.org CREATINE USE IN YOUTH As illustrated, the number of published studies that have been completed in youth or adolescent populations is quite small (n = 8), particularly when compared to the number of studies in adults. All of the studies outlined in Table 4 used an adolescent participant and sought to examine some aspect of eﬃcacy of creatine use in reference to sporting performance. Further, as can be seen from our PRISMA diagram (Figure 1), no published studies to date have been completed with a priori intent to examine the safety of creatine use in adolescents and youth. For these reasons, we have chosen to ﬁrst discuss the eﬃcacy studies available in adolescent populations before attempting to ascertain as much information as possible about the safety of creatine use in young populations. Safety of Creatine Use Creatine Use and Adolescents TABLE 3 \| Sports and sporting events where performance may be enhanced by creatine supplementation. determine eligibility. For the purpose of this review, we have used the terms “youth” in reference to individuals between 7–12 years of age and “adolescent” for those between 13–18 years of age. TABLE 3 \| Sports and sporting events where performance may be enhanced by creatine supplementation. Increased PCr • Track sprints: 60–200 m • Swim sprints: 50 m • Pursuit cycling Increased PCr Resynthesis • Basketball • Field Hockey • American Football • Ice Hockey • Lacrosse • Volleyball Reduced Muscle Acidosis • Downhill skiing • Water sports (e.g., Rowing, Canoe, Kayak, Stand-Up Paddling) • Swim events: 100, 200 m • Track events: 400, 800 m • Combat Sports (e.g., MMA, Wrestling, Boxing, etc.) Oxidative Metabolism • Basketball • Soccer • Team Handball • Tennis • Volleyball • Interval training in endurance athletes Increased Body Mass/Muscle Mass • American Football • Bodybuilding • Combat Sports (e.g., MMA, Wrestling, Boxing, etc.) • Powerlifting • Rugby • Track/Field events (Shot put; javelin; discus; hammer throw) • Olympic Weightlifting MMA, Mixed martial arts; PCr, Phosphocreatine. Reproduced from Kreider et al. (2) under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/). creatine supplementation. Increased PCr • Track sprints: 60–200 m • Swim sprints: 50 m • Pursuit cycling Increased PCr Resynthesis • Basketball • Field Hockey • American Football • Ice Hockey • Lacrosse • Volleyball Reduced Muscle Acidosis • Downhill skiing • Water sports (e.g., Rowing, Canoe, Kayak, Stand-Up Paddling) • Swim events: 100, 200 m • Track events: 400, 800 m • Combat Sports (e.g., MMA, Wrestling, Boxing, etc.) Oxidative Metabolism • Basketball • Soccer • Team Handball • Tennis • Volleyball • Interval training in endurance athletes Increased Body Mass/Muscle Mass • American Football • Bodybuilding • Combat Sports (e.g., MMA, Wrestling, Boxing, etc.) • Powerlifting • Rugby • Track/Field events (Shot put; javelin; discus; hammer throw) • Olympic Weightlifting MMA, Mixed martial arts; PCr, Phosphocreatine. Reproduced from Kreider et al. (2) under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/). creatine supplementation. Is Creatine Safe for Youth? soccer-speciﬁc skills following 7 days of creatine supplementation (30 g/day) in 20 young (16.6 ± 1.9 years), male soccer players. Juhasz et al. (47) examined the eﬀects of 5 days of creatine supplementation (4 × 5 g/day) on mechanical power output and swim performance in highly trained junior (15.9 ± 1.6 years) competitive swimmers. The researchers observed signiﬁcant improvements in sprint swimming performance and dynamic strength following creatine supplementation. Mohebbi et al. (84) also examined the eﬀects of creatine supplementation (20 g/day) on repeat sprint performance, dribbling and shooting accuracy in 17 young (17.2 ± 0.1 years), soccer players. Following 7 days of supplementation, signiﬁcant improvements in repeat sprint performance and dribbling abilities were observed. Most recently, in 2017 a similar study (86) with elite youth (17.0 ± 0.5 years) soccer players found signiﬁcant improvements in power output following a low-dose creatine supplementation regimen (0.03 g/kg/day) for 7 days. In three of the eight studies mentioned, no adverse events or side-eﬀects were reported by the participants following supplementation, and the remaining ﬁve studies did not report adverse event occurrences. Although not an extensive list, a precedent has been set regarding creatine supplementation interventions in adolescent athletes, warranting further research in this area examining both eﬃcacy and safety. Table 4 outlines relevant details of these studies. soccer-speciﬁc skills following 7 days of creatine supplementation (30 g/day) in 20 young (16.6 ± 1.9 years), male soccer players. Juhasz et al. (47) examined the eﬀects of 5 days of creatine supplementation (4 × 5 g/day) on mechanical power output and swim performance in highly trained junior (15.9 ± 1.6 years) competitive swimmers. The researchers observed signiﬁcant improvements in sprint swimming performance and dynamic strength following creatine supplementation. Mohebbi et al. (84) also examined the eﬀects of creatine supplementation (20 g/day) on repeat sprint performance, dribbling and shooting accuracy in 17 young (17.2 ± 0.1 years), soccer players. Following 7 days of supplementation, signiﬁcant improvements in repeat sprint performance and dribbling abilities were observed. Most recently, in 2017 a similar study (86) with elite youth (17.0 ± 0.5 years) soccer players found signiﬁcant improvements in power output following a low-dose creatine supplementation regimen (0.03 g/kg/day) for 7 days. In three of the eight studies mentioned, no adverse events or side-eﬀects were reported by the participants following supplementation, and the remaining ﬁve studies did not report adverse event occurrences. REVIEW METHODOLOGY To illustrate the paucity of literature directly examining the safety of creatine supplementation in youth, a systematic review was performed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines (Figure 1). PubMed, MEDLINE, and SportDiscus databases were each searched using the following terms: “creatine supplementation” AND “safety” AND “humans” AND “adolescents.” A second search was conducted using the terms “creatine supplementation” AND “safety” AND “humans” AND “youth.” The ﬁnal search and accession date using these parameters was 08/31/2018. Grindstaﬀet al. (46) were one of the ﬁrst groups to examine the eﬀects of creatine supplementation on performance in adolescent athletes. For the study, 18 male and female swimmers (15.3 ± 0.6 years) were randomly assigned to one of two groups to ingest either 21 g/day of creatine or placebo over a period of 9 days. The researchers noted signiﬁcant improvements in repeat sprint performance in swimmers after creatine supplementation. Shortly after, Dawson et al. (45) tried to replicate these ﬁndings in young, elite swimmers (16.4 ± 1.8 years) using an extended creatine supplementation period of 28 days. The authors did not observe any signiﬁcant improvement in single sprint performance following 4 weeks of creatine supplementation (20 g/day for 5 days; followed by 5 g/day for 22 days) but did note a signiﬁcant improvement in swim bench test performance. Later in 2004, Ostojic et al. (85) observed signiﬁcant improvements in Studies were eligible for inclusion if they met all of the following criteria: (1) original research conducted in humans under 18 years of age; (2) published in peer-reviewed academic journals; (3) implemented an intervention of at least 3 days using creatine only; (4) reported a clinical measure of safety as a primary outcome measure; (5) recruited populations that were not receiving creatine as a treatment for any diagnosed medical condition; (6) utilized a double-blind, placebo-controlled design. Two independent reviewers (CK and PH) assessed all articles to November 2018 \| Volume 5 \| Article 115 6 Jagim et al. Creatine Use and Adolescents FIGURE 1 \| PRISMA ﬂow chart. FIGURE 1 \| PRISMA ﬂow chart. Is Creatine Safe for Youth? Although not an extensive list, a precedent has been set regarding creatine supplementation interventions in adolescent athletes, warranting further research in this area examining both eﬃcacy and safety. Table 4 outlines relevant details of these studies. As can be seen in Figure 1, the results of our systematic review revealed that no studies to date have been completed that sought to directly examine the safety of creatine use in an adolescent or youth population. Subsequently, each eﬃcacy study in adolescents (Table 4) was closely reviewed by the authors to ascertain any information that might be present regarding any clinical side eﬀects resulting from creatine use in adolescents. In this respect, none of these studies observed any gastrointestinal discomfort or changes in hemodynamic, urine, or any blood markers of clinical health and safety following the supplementation periods. What is important to highlight is that several studies are available that have used creatine supplementation in children as some form of a medical therapy. The most common application of creatine in clinical pediatric populations involve children born with one form of a creatine deﬁciency syndrome. This class of syndromes results in a reduction or inability to endogenously synthesize creatine leading low levels of creatine and phosphocreatine in the muscle and brain. The physical presentation of this scenario includes muscle myopathies, gyrate atrophy, movement disorders, speech delay, autism, mental development challenges, epilepsy, and other developmental problems (91–93), as reviewed by Kreider et al. (2). Similarly, Battini et al. (94) reported on a child born with a creatine November 2018 \| Volume 5 \| Article 115 Frontiers in Nutrition \| www.frontiersin.org 7 Jagim et al. Creatine Use and Adolescents \| y Study Subjects Design Duration Dosing protocol Primary variables Results Adverse events References Juhasz et al. 16 male Fin swimmers (15.9 ± 1.6 years) Randomized, double-blind, placebo controlled 5 days 20 g/day Average power, dynamic strength (swim-based tests) ↑anaerobic performance ↑dynamic strength. NR (47) Claudino et al. 14 male Brazilian elite soccer players (18.3 ± 0.9 years) Randomized, double-blind, placebo controlled 7 Weeks 20 g/day (1 week) 5 g/day (6 weeks) Lower limb muscle power via CMVJ ↔lower body power. NR (54) Dawson et al. 10 males, 10 females (16.4 ± 1.8 years) Match, placebo-controlled 4 Weeks 20 g/day (5 days); 5 g/day (22 days) Sprint swim performance and Biokinetic Swim Bench Test ↑swim bench test performance NR (45) Grindstaff et al. November 2018 \| Volume 5 \| Article 115 Frontiers in Nutrition \| www.frontiersin.org CONCLUSIONS A major driving force of this article is to clearly focus upon the available scientiﬁc information involving creatine supplementation in youth and adolescent populations. A signiﬁcant amount of concern and caution have been present within the media and sporting world up to this point regarding creatine use in younger populations. Even within the scientiﬁc literature creatine supplementation in adolescents has inappropriately been classiﬁed as “unhealthy behavior” (101), “disturbing trends” (70), or has been compared to illegal performance enhancing drug use (100) without any supporting evidence of its dangers or scientiﬁc rationale for such a classiﬁcation. Unfortunately, such concern at times has resulted in multiple scenarios where dramatized accounts of creatine’s impact or its associations next to other anabolic agents are made that are at best, inappropriate, and at worst, unethical. Evidence continues to accumulate; however, that clearly highlights the fact that high school-aged individuals and younger are using creatine. While prevalence statistics of creatine use in youth do not warrant an endorsement by anyone, the lack of consistent medical reports involving clinically compromising situations in combination with years of medical applications of creatine in children with inborn errors of metabolism or neurological diseases certainly opens the door for people to understand that creatine supplementation in healthy adolescent populations has the potential to be well-tolerated with little incident. Furthermore, an emerging body of literature in adolescent athletes using creatine has suggested that, ﬁrst, creatine use in adolescent athletes appears to be well-tolerated with no reported adverse events and, second, that creatine use in this population can operate in an ergogenic fashion [see Table 4, also extensively reviewed by (2)]. Finally, one should not dismiss the now 25+ years of research that continues to highlight that creatine use in a multitude of populations is safe and eﬀective means to improve both clinical and ergogenic outcomes (1). It is worth mentioning that some individuals point to the warnings provided on product labels that individuals younger than 18 years of age should not take creatine and inappropriately extrapolate this as evidence that creatine supplementation is unsafe in younger populations, rather than acknowledging there are insuﬃcient data to conﬁrm the need for such warnings. These warnings are not scientiﬁcally-based and are provided more as a legal precaution. Is Creatine Safe for Youth? Creatine supplementation has also been shown to be an eﬀective therapy to treat gyrate atrophy of the retina. Vannas-Sulonen et al. (29)reported on 13 patients who ranged in age from 6 to 31 years old. In particular, ﬁve patients were <18 years of age (6–16 years, median: 12 years; 119–174 cm, median: 165 cm, 21–76 kg, median: 56 kg) and were treated with creatine for 40–72 months, median treatment duration: 60 months). deﬁciency syndrome who, at 4 months of age, was treated with creatine supplementation. Moreover, Stocker-Ipsiroglu et al. (95) administered creatine monohydrate (0.3–0.8 g/kg/day, equivalent to 13.5–62 g of creatine per day for an individual weighing 45–77 kg [100–160 pounds]) to patients ranging from 25.5 months to 11 years (median age: 51 months) for a treatment period of 11–192 months (median treatment duration: 48 months). The researchers found that creatine supplementation increased brain creatine levels and stabilized other clinical outcomes. Creatine supplementation is also commonly used as a therapeutic agent for improving musculoskeletal function in patients suﬀering from muscular dystrophy (96–98). Most notable is a creatine supplementation study by Tarnopolsky et al. (96) who observed signiﬁcant improvements in fat-free mass and handgrip strength after 4 months of creatine monohydrate supplementation (0.10 g/kg/day) in 30 young boys (mean age: 10 ± 3 years) with Duchenne muscular dystrophy. Similarly, Hayashi et al. (99) administered 0.1 g/kg creatine per day for 12 weeks to patients with childhood systemic lupus erythematosus and detected no deleterious changes in laboratory parameters of inﬂammation, hematology, skeletal muscle enzymes, or kidney and liver function. Creatine supplementation has also been shown to be an eﬀective therapy to treat gyrate atrophy of the retina. Vannas-Sulonen et al. (29)reported on 13 patients who ranged in age from 6 to 31 years old. In particular, ﬁve patients were <18 years of age (6–16 years, median: 12 years; 119–174 cm, median: 165 cm, 21–76 kg, median: 56 kg) and were treated with creatine for 40–72 months, median treatment duration: 60 months). knowledge of using performance enhancing substances. These situations are troubling because ﬁrst, creatine is not a steroid nor does it act like one, as it has a completely diﬀerent molecular structure and physiological mechanism of action; and second, these unsubstantiated comments cloud the ability of individuals to understand key evidence-based information that exists on creatine while also instilling complications with the research process regarding adolescents. CONCLUSIONS Regardless, the point remains that no published studies are available to date that have used a rigorous study design to examine the impact of creatine supplementation on markers of health and safety in healthy populations, speciﬁcally athletes. Is Creatine Safe for Youth? Furthermore, the toxicity from anabolic steroids are well-documented and inappropriately discussing creatine within this category implies “guilt by association.” It is important to acknowledge that a seemingly large proportion of adolescent athletes are using or have tried creatine (see previous sections). Furthermore, it is also worth noting that creatine is not banned by any major athletic governing body or organization. While these facts are not to be intended as an endorsement of its use in young athletes, they further solidify the need for scientiﬁcally controlled investigations that seek to determine the safety of creatine use in adolescents. Frontiers in Nutrition \| www.frontiersin.org Is Creatine Safe for Youth? 18 (male: 7; female: 11) (15.3 ± 0.6 years) Randomized, double-blind, placebo controlled 9 Days 21 g/day Sprint swim performance, arm ergometer ↑sprint swimming performance None (46) Mohebbi et al. 17 Young soccer players (17.18 ± 01.37 years) Randomized, double-blind, placebo controlled 7 Days 20 g/day Repeated sprint test, dribbling performance and shooting accuracy ↑repeat sprint performance ↑dribbling NR (84) Ostojic 20 Young, male soccer players (16.6 ± 1.9 years) Matched, placebo-controlled 7 Days 30 g/day Soccer speciﬁc skills tests ↑dribble test and endurance times ↑sprint-power test and CMVJ None (85) Yanez-Silva et al. Elite youth soccer players (17.0 ± 0.5 years) Matched, double blind, placebo-controlled 7 Days 0.03 g/kg/day Muscle power output (WAnT) ↑PPO and MPO ↑total work None (86) Theodorou et al. 22 Elite (12 males: 17.7 ± 2.3 years; 10 female: 17.7 ± 2.0 years) swimmers Randomized, double-blind, placebo controlled 11 weeks 25 g/day (4 days) 5 g/day (2 months) Swimming interval performance ↑Interval Performance following loading phase ↔long-term improvements after maintenance dosing. NR (87) NR, Not Reported; ↔, Creatine supplementation resulted in no signiﬁcant (p < 0.05) change; ↑, Creatine supplementation resulted in a signiﬁcant increase (p < 0.05) over control. November 2018 \| Volume 5 \| Article 115 8 Creatine Use and Adolescents Jagim et al. deﬁciency syndrome who, at 4 months of age, was treated with creatine supplementation. Moreover, Stocker-Ipsiroglu et al. (95) administered creatine monohydrate (0.3–0.8 g/kg/day, equivalent to 13.5–62 g of creatine per day for an individual weighing 45–77 kg [100–160 pounds]) to patients ranging from 25.5 months to 11 years (median age: 51 months) for a treatment period of 11–192 months (median treatment duration: 48 months). The researchers found that creatine supplementation increased brain creatine levels and stabilized other clinical outcomes. Creatine supplementation is also commonly used as a therapeutic agent for improving musculoskeletal function in patients suﬀering from muscular dystrophy (96–98). Most notable is a creatine supplementation study by Tarnopolsky et al. (96) who observed signiﬁcant improvements in fat-free mass and handgrip strength after 4 months of creatine monohydrate supplementation (0.10 g/kg/day) in 30 young boys (mean age: 10 ± 3 years) with Duchenne muscular dystrophy. Similarly, Hayashi et al. (99) administered 0.1 g/kg creatine per day for 12 weeks to patients with childhood systemic lupus erythematosus and detected no deleterious changes in laboratory parameters of inﬂammation, hematology, skeletal muscle enzymes, or kidney and liver function. REFERENCES 11. Kim HJ, Kim CK, Carpentier A, Poortmans JR. Studies on the safety of creatine supplementation. Amino acids (2011) 40:1409–18. doi: 10.1007/s00726-011-0878-2 1. Kerksick CM, Wilborn CD, Roberts MD, Smith-Ryan A, Kleiner SM, Jager R, et al. ISSN exercise & sports nutrition review update: research & recommendations. J Int Soc Sports Nutr. (2018) 15:38. doi: 10.1186/s12970-018-0242-y 12. Kreider RB, Melton C, Rasmussen CJ, Greenwood M, Lancaster S, Cantler EC, et al. Long-term creatine supplementation does not signiﬁcantly aﬀect clinical markers of health in athletes. Mol Cell Biochem. (2003) 244:95–104. doi: 10.1023/A:1022469320296 2. Kreider RB, Kalman DS, Antonio J, Ziegenfuss TN, Wildman R, Collins R, et al. International society of sports nutrition position stand: safety and eﬃcacy of creatine supplementation in exercise, sport, and medicine. J Int Soc Sports Nutr. (2017) 14:18. doi: 10.1186/s12970-017- 0173-z 2. Kreider RB, Kalman DS, Antonio J, Ziegenfuss TN, Wildman R, Collins R, et al. International society of sports nutrition position stand: safety and eﬃcacy of creatine supplementation in exercise, sport, and medicine. J Int Soc Sports Nutr. (2017) 14:18. doi: 10.1186/s12970-017- 0173-z 13. Poortmans JR, Francaux M. Long-term oral creatine supplementation does not impair renal function in healthy athletes. Med Sci Sports Exer. (1999) 31:1108–10 14. Cancela P, Ohanian C, Cuitino E, Hackney A. Creatine supplementation does not aﬀect clinical health markers in football players. Br J Sports Med. (2008) 429:731–5. doi: 10.1136/bjsm.2007.030700 3. Harris RC, Soderlund K, Hultman E. Elevation of creatine in resting and exercised muscle of normal subjects by creatine supplementation. Clin Sci. (1992) 83:367–74 15. Greenwood M, Kreider RB, Greenwood L, Byars A. Cramping and injury incidence in collegiate football players are reduced by creatine supplementation. J Athlet Train. (2003) 38:216–9. 4. Jager R, Purpura M, Shao A, Inoue T, Kreider RB. Analysis of the eﬃcacy, safety, and regulatory status of novel forms of creatine. Amino Acids (2011) 40:1369–83. doi: 10.1007/s00726-011-0874-6 16. Greenwood M, Kreider RB, Melton C, Rasmussen C, Lancaster S, Cantler E, et al. Creatine supplementation during college football training does not increase the incidence of cramping or injury. Mol Cell Biochem. (2003) 244:83–8. doi: 10.1023/A:1022413202549 5. Gastin PB. Energy system interaction and relative contribution during maximal exercise. Sports Med. (2001) 31:725–41. doi: 10.2165/00007256-200131100-00003 6. Allen DG, Lamb GD, Westerblad H. Skeletal muscle fatigue: cellular mechanisms. Physiol Rev. (2008) 88:287–332. doi: 10.1152/physrev.00015.2007 17. Mayhew DL, Mayhew JL, Ware JS. A Balanced Perspective Despite the overwhelming body of evidence supporting the safety and eﬃcacy of creatine supplementation in adult athletes, there still exists a plethora of misconceptions and concerns regarding the use of creatine in adolescents that are not well- substantiated within the literature. Commonly, conversations that start with discussions of proper fueling and hydration can transition into unsubstantiated statements surrounding the purported safety of creatine use in a younger population, often associating creatine with illegal performance-enhancing drugs. For example, Greydanus and Patel (100) assessed the incidence of sports doping in adolescents and included creatine as an “anabolic-like agent” when describing anabolic steroid use. As another example, Ranby et al. (101) included creatine with anabolic steroids in the same category when describing changes in a survey designed to determine the intention or November 2018 \| Volume 5 \| Article 115 9 Creatine Use and Adolescents Jagim et al. It is our sincere hope that this article will serve as a guide for other researchers, Institutional Review Boards (IRBs), clinicians, professional organizations, and governing bodies to use when determining the safety and eﬃcacy of creatine use in youth and adolescent populations. In this respect and in completing this review, we have identiﬁed areas where scientiﬁcally controlled, high-quality studies are needed to help build and progress this body of literature. Most importantly, short-term (<7 days) and long-term (weeks to months) studies that employ well-powered, randomized, double-blind, placebo- controlled study designs are desperately needed to examine the impact of creatine supplementation on traditional markers of clinical safety (hemodynamic changes, urine parameters, complete blood counts, and comprehensive metabolic and lipid panels) after acute and prolonged creatine supplementation regimens in adolescent populations. Therefore, this review may lend itself as a call to action for future work in this area by providing a comprehensive summary of the relevant literature and identifying the need to assess the clinical safety of creatine supplementation within this population. Such work is of paramount importance, as it will begin to demonstrate the safety of creatine use in adolescent populations under scientiﬁcally controlled conditions. From there, studies that examine the minimum eﬀective dose of creatine or any prudent modiﬁcations to the regimens of creatine supplementation that are discussed in this article and elsewhere should be undertaken. A Balanced Perspective In this respect, creatine turnover in the adult is known (and was brieﬂy discussed), but when one considers the eating patterns of children and how creatine turnover may diﬀer in this population, the need to explore relevant dosing amounts and patterns is also important. AUTHOR CONTRIBUTIONS AJ and CK conceptualized the topic of this review. AJ, RS, PH, JE, and CK researched and analyzed the literature and assisted in manuscript preparation. JE provided scholarly guidance on the topic, assisted in manuscript preparation, and revised the manuscript critically. PH and CK performed the systematic review. All authors contributed to manuscript revision, have read, and approve of the ﬁnal version of the manuscript. ACKNOWLEDGMENTS We would like to thank and recognize Mark Tarnopolsky, MD, Ph.D., of McMaster University for critically reviewing and oﬀering valuable insight regarding the direction and interpretation of this manuscript. REFERENCES Acta Physiol Hung. (2009) 96:325–36. doi: 10.1556/APhysiol.96.2009.3.6 28. Solis MY, Hayashi AP, Artioli GG, Roschel H, Sapienza MT, Otaduy MC, et al. Eﬃcacy and safety of creatine supplementation in juvenile dermatomyositis: a randomized, double-blind, placebo-controlled crossover trial. Muscle Nerve (2016) 53:58–66. doi: 10.1002/mus.24681. Nerve (2016) 53:58–66. doi: 10.1002/mus.24681. 48. Silva AJ, Machado Reis V, Guidetti L, Bessone Alves F, Mota P, Freitas J, et al. Eﬀect of creatine on swimming velocity, body composition and hydrodynamic variables. J Sports Med Phys Fitness (2007) 47:58–64. 29. Vannas-Sulonen K, Sipila I, Vannas A, Simell O, Rapola J. Gyrate atrophy of the choroid and retina. A ﬁve-year follow-up of creatine supplementation. Ophthalmology (1985) 92:1719–27. 49. Kreider RB, Ferreira M, Wilson M, GrindstaﬀP, Plisk S, Reinardy J, et al. Eﬀects of creatine supplementation on body composition, strength, and sprint performance. Med Sci Sports Exer. (1998) 30:73–82. 30. Pereira RTdS, Dörr FA, Pinto E, Solis MY, Artioli GG, Fernandes AL, et al. Can creatine supplementation form carcinogenic heterocyclic amines in humans? J Physiol. (2015) 593:3959–71. doi: 10.1113/JP2 70861 50. Stone MH, Sanborn K, Smith LL, O’Bryant HS, Hoke T, Utter AC, et al. Eﬀects of in-season (5 weeks) creatine and pyruvate supplementation on anaerobic performance and body composition in American football players. Int J Sport Nutr. (1999) 9:146–65. 31. Armentano MJ, Brenner AK, Hedman TL, Solomon ZT, Chavez J, Kemper GB, et al. The eﬀect and safety of short-term creatine supplementation on performance of push-ups. Military Med. (2007) 172:312–7. doi: 10.7205/MILMED.172.3.312 51. Bemben MG, Bemben DA, Loftiss DD, Knehans AW. Creatine supplementation during resistance training in college football athletes. Med Sci Sports Exer. (2001) 33:1667–73. doi: 10.1097/00005768-200110000- 00009 32. Galvan E, Walker DK, Simbo SY, Dalton R, Levers K, O’Connor A, et al. Acute and chronic safety and eﬃcacy of dose dependent creatine nitrate supplementation and exercise performance. J Int Soc Sports Nutr. (2016) 13:12. doi: 10.1186/s12970-016-0124-0 52. Hoﬀman J, Ratamess N, Kang J, Mangine G, Faigenbaum A, Stout J. Eﬀect of creatine and beta-alanine supplementation on performance and endocrine responses in strength/power athletes. Int J Sport Nutr Exerc Metab. (2006) 16:430–46. doi: 10.1123/ijsnem.16.4.430 33. Lugaresi R, Leme M, de Salles Painelli V, Murai IH, Roschel H, Sapienza MT, et al. Does long-term creatine supplementation impair kidney function in resistance-trained individuals consuming a high-protein diet? J Int Soc Sports Nutr. (2013) 10:26. doi: 10.1186/1550-2783-10-26 53. Chilibeck PD, Magnus C, Anderson M. REFERENCES Creatine supplementation with speciﬁc view to exercise/sports performance: an update. J Int Soc Sports Nutr. (2012) 9:33. doi: 10.1186/1550-2783-9-33. 41. Branch JD. Eﬀect of creatine supplementation on body composition and performance: a meta-analysis. Int J Sport Nutr Exerc Metab. (2003) 132:198– 226. doi: 10.1123/ijsnem.13.2.198 22. Neves M, Jr, Gualano B, Roschel H, Lima FR, Lucia de Sa-Pinto A, Seguro AC, et al. Eﬀect of creatine supplementation on measured glomerular ﬁltration rate in postmenopausal women. Appl Physiol Nutr Metab. (2011) 36:419–22. doi: 10.1139/h11-014 42. Rawson ES, Volek JS. Eﬀects of creatine supplementation and resistance training on muscle strength and weightlifting performance. J Strength Cond Res. (2003) 17:822–31. doi: 10.1519/00124278-200311000-00031 23. Ropero-Miller JD, Paget-Wilkes H, Doering PL, Goldberger BA. Eﬀect of oral creatine supplementation on random urine creatinine, pH, and speciﬁc gravity measurements. Clin Chem. (2000) 46:295–7. 43. Volek JS, Duncan ND, Mazzetti SA, Staron RS, Putukian M, Gomez AL, et al. Performance and muscle ﬁber adaptations to creatine supplementation and heavy resistance training. Med Sci Sports Exer. (1999) 31:1147–56. 24. Persky AM, Rawson ES. Safety of creatine supplementation. Sub-Cell Biochem. (2007) 46:275–89. doi: 10.1007/978-1-4020-6486-9_14 44. Cornish SM, Chilibeck PD, Burke DG. The eﬀect of creatine monohydrate supplementation on sprint skating in ice-hockey players. J Sports Med Phys Fitness (2006) 46:90–8. 25. Poortmans JR, Auquier H, Renaut V, Durussel A, Saugy M, Brisson GR. Eﬀect of short-term creatine supplementation on renal responses in men. Eur J Appl Physiol Occup Physiol. (1997) 76:566–7. 26. Bender A, Samtleben W, Elstner M, Klopstock T. Long-term creatine supplementation is safe in aged patients with Parkinson disease. Nutr Res. (2008) 28:172–8. doi: 10.1016/j.nutres.2008.01.001 45. Dawson B, Vladich T, Blanksby BA. Eﬀects of 4 weeks of creatine supplementation in junior swimmers on freestyle sprint and swim bench performance. J Strength Cond Res. (2002) 16:485–90. doi: 10.1519/00124278-200211000-00001 27. Gualano B, de Salles Painelli V, Roschel H, Lugaresi R, Dorea E, Artioli GG, et al. Creatine supplementation does not impair kidney function in type 2 diabetic patients: a randomized, double-blind, placebo-controlled, clinical trial. Eur J Appl Physiol. (2011) 111:749–56. doi: 10.1007/s00421-010- 1676-3 46. GrindstaﬀPD, Kreider R, Bishop R, Wilson M, Wood L, Alexander C, et al. Eﬀects of creatine supplementation on repetitive sprint performance and body composition in competitive swimmers. Int J Sport Nutr. (1997) 7:330–46. 47. Juhasz I, Gyore I, Csende Z, Racz L, Tihanyi J. Creatine supplementation improves the anaerobic performance of elite junior ﬁn swimmers. REFERENCES Eﬀects of long-term creatine supplementation on liver and kidney functions in American college football players. Int J Sport Nutr Exerc Metab. (2002) 12:453-60. doi: 10.1123/ijsnem.12.4.453 7. Hultman E, Soderlund K, Timmons JA, Cederblad G, GreenhaﬀPL. Muscle creatine loading in men. J Appl Physiol. (1996) 81:232–7. 7. Hultman E, Soderlund K, Timmons JA, Cederblad G, GreenhaﬀPL. Muscle creatine loading in men. J Appl Physiol. (1996) 81:232–7. 18. Gualano B, Ugrinowitsch C, Novaes RB, Artioli GG, Shimizu MH, Seguro AC, et al. Eﬀects of creatine supplementation on renal function: a randomized, double-blind, placebo-controlled clinical trial. Eur J Appl Physiol. (2008) 103:33–40. doi: 10.1007/s00421-007-0669-3 8. Mesa JL, Ruiz JR, Gonzalez-Gross MM, Gutierrez Sainz A, Castillo Garzon MJ. Oral creatine supplementation and skeletal muscle metabolism in physical exercise. Sports Med. (2002) 32:903–44. doi: 10.2165/00007256-200232140-00003 8. Mesa JL, Ruiz JR, Gonzalez-Gross MM, Gutierrez Sainz A, Castillo Garzon MJ. Oral creatine supplementation and skeletal muscle metabolism in physical exercise. Sports Med. (2002) 32:903–44. doi: 10.2165/00007256-200232140-00003 19. Joy JM, Lowery RP, Falcone PH, Mosman MM, Vogel RM, Carson LR, et al. 28 days of creatine nitrate supplementation is apparently safe in healthy individuals. J Int Soc Sports Nutr. (2014) 11:60. doi: 10.1186/s12970-014-0060-9 9. GreenhaﬀPL, Casey A, Short AH, Harris R, Soderlund K, Hultman E. Inﬂuence of oral creatine supplementation of muscle torque during repeated bouts of maximal voluntary exercise in man. Clin Sci. (1993) 84:565–71. 20. Lobo DM, Tritto AC, da Silva LR, de Oliveira PB, Benatti FB, Roschel H, et al. Eﬀects of long-term low-dose dietary creatine supplementation in older women. Exp Gerontol. (2015) 70:97–104. doi: 10.1016/j.exger.2015.07.012 10. Vandenberghe K, Goris M, Van Hecke P, Van Leemputte M, Vangerven L, Hespel P. Long-term creatine intake is beneﬁcial to muscle performance during resistance training. J Appl Physiol. (1997) 83:2055–63. 10. Vandenberghe K, Goris M, Van Hecke P, Van Leemputte M, Vangerven L, Hespel P. Long-term creatine intake is beneﬁcial to muscle performance during resistance training. J Appl Physiol. (1997) 83:2055–63. November 2018 \| Volume 5 \| Article 115 Frontiers in Nutrition \| www.frontiersin.org 10 Jagim et al. Creatine Use and Adolescents 21. Mihic S, MacDonald JR, McKenzie S, Tarnopolsky MA. Acute creatine loading increases fat-free mass, but does not aﬀect blood pressure, plasma creatinine, or CK activity in men and women. Med Sci Sports Exer. (2000) 32:291–6. doi: 10.1097/00005768-200002000-00007 40. Cooper R, Naclerio F, Allgrove J, Jimenez A. REFERENCES Knapik JJ, Steelman RA, Hoedebecke SS, Austin KG, Farina EK, Lieberman HR. Prevalence of dietary supplement use by athletes: systematic review and meta-analysis. Sports Med. (2016) 46:103–23. doi: 10.1007/s40279-015-0387-7. 59. Volek JS, Ratamess NA, Rubin MR, Gomez AL, French DN, McGuigan MM, et al. The eﬀects of creatine supplementation on muscular performance and body composition responses to short-term resistance training overreaching. Eur J Appl Physiol. (2004) 91:628–37. doi: 10.1007/s00421-003-1031-z. 81. McGuine TA, Sullivan JC, Bernhardt DA. Creatine supplementation in Wisconsin high school athletes. WMJ (2002) 101:25–30. 60. Buford TW, Kreider RB, Stout JR, Greenwood M, Campbell B, Spano M, et al. International Society of Sports Nutrition position stand: creatine supplementation and exercise. J Int Soc Sports Nutr. (2007) 4:6. doi: 10.1186/1550-2783-4-6 82. Ziegler PJ, Nelson JA, Jonnalagadda SS. Use of dietary supplements by elite ﬁgure skaters. Int J Sport Nutr Exerc Metab. (2003) 13:266–76. doi: 10.1123/ijsnem.13.3.266 83. Braun H, Koehler K, Geyer H, Kleiner J, Mester J, Schanzer W. Dietary supplement use among elite young German athletes. Int J Sport Nutr Exerc Metab. (2009) 19:97–109. doi: 10.1123/ijsnem.19.1.97 61. Tarnopolsky MA. Potential beneﬁts of creatine monohydrate supplementation in the elderly. Curr Opin Clin Nutr Metab Care (2000) 3:497–502. doi: 10.1097/00075197-200011000-00013 84. Mohebbi H, Rahnama N, Moghadassi M, Ranjbar K. Eﬀect of creatine supplementation on sprint and skill performance in Young Soccer Players. Middle-East J Sci Res. (2012) 12:397–401. 62. Rawson ES, Venezia AC. Use of creatine in the elderly and evidence for eﬀects on cognitive function in young and old. Amino acids (2011) 40:1349–62. doi: 10.1007/s00726-011-0855-9 85. Ostojic SM. Creatine supplementation in young soccer players. Int J Sport Nutr Exerc Metab. (2004) 14:95–103. doi: 10.1123/ijsnem.14.1.95 63. Kreider RB, Wilborn CD, Taylor L, Campbell B, Almada AL, Collins R, et al. ISSN exercise & sport nutrition review: research & recommendations. J Int Soc Sports Nutr. (2010) 7:7. doi: 10.1186/1550-2783-7-7 86. Yanez-Silva A, Buzzachera CF, Picarro IDC, Januario RSB, Ferreira LHB, McAnulty SR, et al. Eﬀect of low dose, short-term creatine supplementation on muscle power output in elite youth soccer players. J Int Soc Sports Nutr. (2017) 14:5. doi: 10.1186/s12970-017-0162-2 64. Devries MC, Phillips SM. Creatine supplementation during resistance training in older adults-a meta-analysis. Med Sci Sports Exer. (2014) 46:1194– 203. doi: 10.1249/mss.0000000000000220 87. Theodorou AS, Cooke CB, King RFGJ, Hood C, Denison T, Wainwright BG, et al. The eﬀect of longer-term creatine supplementation on elite swimming performance after an acute creatine loading. REFERENCES Eﬀect of in-season creatine supplementation on body composition and performance in rugby union football players. Appl Physiol Nutr Metab. (2007) 32:1052–7. doi: 10.1139/h07-072 34. Robinson TM, Sewell DA, Casey A, Steenge G, GreenhaﬀPL. Dietary creatine supplementation does not aﬀect some haematological indices, or indices of muscle damage and hepatic and renal function. Br J Sports Med. (2000) 34:284–8. doi: 10.1136/bjsm.34.4.284 54. Claudino JG, Mezencio B, Amaral S, Zanetti V, Benatti F, Roschel H, et al. Creatine monohydrate supplementation on lower-limb muscle power in Brazilian elite soccer players. J Int Soc Sports Nutr. (2014) 11:32. doi: 10.1186/1550-2783-11-32 35. Groeneveld GJ, Beijer C, Veldink JH, Kalmijn S, Wokke JH, van den Berg LH. Few adverse eﬀects of long-term creatine supplementation in a placebo-controlled trial. Int J Sports Med. (2005) 26:307–13. doi: 10.1055/s-2004-817917 55. Kerksick CM, Rasmussen C, Lancaster S, Starks M, Smith P, Melton C, et al. Impact of diﬀering protein sources and a creatine containing nutritional formula after 12 weeks of resistance training. Nutrition (2007) 23:647–56. doi: 10.1016/j.nut.2007.06.015. 36. Bemben MG, Lamont HS. Creatine supplementation and exercise performance: recent ﬁndings. Sports Med. (2005) 352:107–25. doi: 10.2165/00007256-200535020-00002 j 56. Kerksick CM, Wilborn CD, Campbell WI, Harvey TM, Marcello BM, Roberts MD, et al. The eﬀects of creatine monohydrate supplementation with and without D-pinitol on resistance training adaptations. J strength Cond Res. (2009) 23:2673–82. doi: 10.1519/JSC.0b013e3181b 3e0de 37. Kreider RB. Eﬀects of creatine supplementation on performance and training adaptations. Mol Cell Biochem. (2003) 244:89–94. doi: 10.1023/A:1022469320296 38. Lanhers C, Pereira B, Naughton G, Trousselard M, Lesage FX, Dutheil F. Creatine Supplementation and lower limb strength performance: a systematic review and meta-analyses. Sports Med. (2015) 45:1285–94. doi: 10.1007/s40279-015-0337-4 57. Volek JS, Kraemer WJ, Bush JA, Boetes M, Incledon T, Clark KL, et al. Creatine supplementation enhances muscular performance during high-intensity resistance exercise. J Am Diet Assoc. (1997) 97:765–70. doi: 10.1016/s0002-8223(97)00189-2 39. Butts J, Jacobs B, Silvis M. Creatine Use in Sports. Sports Health (2017) 101:31–4. doi: 10.1177/1941738117737248 November 2018 \| Volume 5 \| Article 115 Frontiers in Nutrition \| www.frontiersin.org 11 Jagim et al. Creatine Use and Adolescents the high-performance athlete. Br J Sports Med. (2018) 52:439–55. doi: 10.1136/bjsports-2018-099027 the high-performance athlete. Br J Sports Med. (2018) 52:439–55. doi: 10.1136/bjsports-2018-099027 58. Volek JS, Mazzetti SA, Farquhar WB, Barnes BR, Gomez AL, Kraemer WJ. Physiological responses to short-term exercise in the heat after creatine loading. Med. Sci Sports Exer. (2001) 33:1101–8. doi: 10.1097/00005768-200107000-00006 80. REFERENCES J Sports Sci. (1999) 17:853–9. doi: 10.1080/026404199365416. 65. Wiroth JB, Bermon S, Andrei S, Dalloz E, Hebuterne X, Dolisi C. Eﬀects of oral creatine supplementation on maximal pedalling performance in older adults. Eur J Appl Physiol. (2001) 84:533–9. doi: 10.1007/s004210000370 88. Unnithan VB, Veehof SH, Vella CA, Kern M. Is there a physiologic basis for creatine use in children and adolescents? J Strength Cond Res. (2001) 15:524–8. doi: 10.1519/00124278-200111000-00021 66. McMorris T, Mielcarz G, Harris RC, Swain JP, Howard A. Creatine supplementation and cognitive performance in elderly individuals. Neuropsychol Dev Cogn B Aging Neuropsychol Cogn. (2007) 14:517–28. doi: 10.1080/13825580600788100 89. Merege-Filho CA, Otaduy MC, de Sa-Pinto AL, de Oliveira MO, de Souza Goncalves L, Hayashi AP, et al. Does brain creatine content rely on exogenous creatine in healthy youth? a proof-of-principle study. Appl Physiol Nutr Metab. (2017) 42:128–34. doi: 10.1139/apnm-2016-0406 67. Rawson ES, Clarkson PM. Acute creatine supplementation in older men. Int J Sports Med. (2000) 21:71–5. doi: 10.1055/s-2000-8859 68. Aguiar AF, Januario RS, Junior RP, Gerage AM, Pina FL, do Nascimento MA, et al. Long-term creatine supplementation improves muscular performance during resistance training in older women. Eur J Appl Physiol. (2013) 113:987–96. doi: 10.1007/s00421-012-2514-6 90. Solis MY, Artioli GG, Otaduy MCG, Leite CDC, Arruda W, Veiga RR, et al. Eﬀect of age, diet, and tissue type on PCr response to creatine supplementation. J Appl Physiol. (2017) 123:407–14. doi: 10.1152/japplphysiol.00248.2017 69. Smith J, Dahm DL. Creatine use among a select population of high school athletes. Mayo Clinic (2000) 75:1257–63. doi: 10.4065/75.12.1257. 91. Braissant O, Henry H, Beard E, Uldry J. Creatine deﬁciency syndromes and the importance of creatine synthesis in the brain. Amino acids (2011) 40:1315–24. doi: 10.1007/s00726-011-0852-z 70. Metzl JD, Small E, Levine SR, Gershel JC. Creatine use among young athletes. Pediatrics (2001) 108:421–5. doi: 10.1542/peds.108.2.421 92. Sykut-Cegielska J, Gradowska W, Mercimek-Mahmutoglu S, Stockler- Ipsiroglu S. Biochemical and clinical characteristics of creatine deﬁciency syndromes. Acta Biochim Pol. (2004) 514:875–82. 71. Kayton S, Cullen RW, Memken JA, Rutter R. Supplement and ergogenic aid use by competitive male and female high school athletes. Med Sci Sports Exerc. (2002) 34:S193. doi: 10.1097/00005768-200205001-01087 72. O’Dea JA. Consumption of nutritional supplements among adolescents: usage and perceived beneﬁts. Health Edu Res. (2003) 18:98–107. doi: 10.1093/her/18.1.98 93. Stockler-Ipsiroglu S, van Karnebeek CD. Cerebral creatine deﬁciencies: a group of treatable intellectual developmental disorders. Semin Neurol. (2014) 34:350–6. doi: 10.1055/s-0034-1386772. 73. REFERENCES Bell A, Dorsch KD, McCreary DR, Hovey R. A look at nutritional supplement use in adolescents. J Adolescent Health (2004) 34:508–16. doi: 10.1016/j.jadohealth.2003.07.024 94. Battini R, Alessandri MG, Leuzzi V, Moro F, Tosetti M, Bianchi MC, et al. Arginine:glycine amidinotransferase (AGAT) deﬁciency in a newborn: early treatment can prevent phenotypic expression of the disease. J Pediatr. (2006) 1486:828–30. doi: 10.1016/j.jpeds.2006.01.043 74. Hoﬀman JR, Faigenbaum AD, Ratamess NA, Ross R, Kang J, Tenenbaum G. Nutritional supplementation and anabolic steroid use in adolescents. Med Sci Sports Exer. (2008) 40:15–24. doi: 10.1249/mss.0b013e31815a5181 95. Stockler-Ipsiroglu S, van Karnebeek C, Longo N, Korenke GC, Mercimek- Mahmutoglu S, Marquart I, et al. Guanidinoacetate methyltransferase (GAMT) deﬁciency: outcomes in 48 individuals and recommendations for diagnosis, treatment and monitoring. Mol Genet Metab. (2014) 111:16–25. doi: 10.1016/j.ymgme.2013.10.018 75. Petroczi A, Naughton DP. The age-gender-status proﬁle of high performing athletes in the UK taking nutritional supplements: lessons for the future. J Int Soc Sports Nutr. (2008) 5:2. doi: 10.1186/1550-2783-5-2 76. Diehl K, Thiel A, Zipfel S, Mayer J, Schnell A, Schneider S. Elite adolescent athletes’ use of dietary supplements: characteristics, opinions, and sources of supply and information. Int J Sport Nutr Exerc Metab. (2012) 22:165–74. doi: 10.1123/ijsnem.22.3.165 96. Tarnopolsky MA, Mahoney DJ, Vajsar J, Rodriguez C, Doherty TJ, Roy BD, et al. Creatine monohydrate enhances strength and body composition in Duchenne muscular dystrophy. Neurology (2004) 62:1771–7. doi: 10.1212/01.WNL.0000125178.18862.9D 97. Louis M, Lebacq J, Poortmans JR, Belpaire-Dethiou MC, Devogelaer JP, Van Hecke P, et al. Beneﬁcial eﬀects of creatine supplementation in dystrophic patients. Muscle Nerve (2003) 27:604–10. doi: 10.1002/mus.10355 77. Evans MW, Jr, Ndetan H, Perko M, Williams R, Walker C. Dietary supplement use by children and adolescents in the United States to enhance sport performance: results of the National Health Interview Survey. J Primary Prevent. (2012) 33:3–12. doi: 10.1007/s10935-012-0261-4 98. Escolar DM, Buyse G, Henricson E, Leshner R, Florence J, Mayhew J, et al. CINRG randomized controlled trial of creatine and glutamine in Duchenne muscular dystrophy. Ann Neurol. (2005) 58:151–5. doi: 10.1002/ana.20523 78. Kerksick CM, Fox E. Sports Nutrition Needs for Child and Adolescent Athletes. Boca Raton, FL: CRC Press (2016). 99. Hayashi AP, Solis MY, Sapienza MT, Otaduy MC, de Sa Pinto AL, Silva CA, et al. Eﬃcacy and safety of creatine supplementation in 79. Maughan RJ, Burke LM, Dvorak J, Larson-Meyer DE, Peeling P, Phillips SM, et al. Frontiers in Nutrition \| www.frontiersin.org childhood-onset systemic lupus erythematosus: a randomized, double- blind, placebo-controlled, crossover trial. Lupus (2014) 23:1500–11. doi: 10.1177/0961203314546017 November 2018 \| Volume 5 \| Article 115 REFERENCES IOC consensus statement: dietary supplements and November 2018 \| Volume 5 \| Article 115 Frontiers in Nutrition \| www.frontiersin.org 12 Jagim et al. Creatine Use and Adolescents childhood-onset systemic lupus erythematosus: a randomized, double- blind, placebo-controlled, crossover trial. Lupus (2014) 23:1500–11. doi: 10.1177/0961203314546017 Conﬂict of Interest Statement: The authors declare that the research was conducted in the absence of any commercial or ﬁnancial relationships that could be construed as a potential conﬂict of interest. 100. Greydanus DE, Patel DR. Sports doping in the adolescent: the Faustian conundrum of Hors de Combat. Pediatr Clin North Am. (2010) 57:729–50. doi: 10.1016/j.pcl.2010.02.008 Copyright © 2018 Jagim, Stecker, Harty, Erickson and Kerksick. This is an open- access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. 101. Ranby KW, Aiken LS, MacKinnon DP, Elliot DL, Moe EL, McGinnis W, et al. A Mediation analysis of the ATHENA intervention for female athletes: prevention of athletic-enhancing substance use and unhealthy weight loss behaviors. J Pediatr Psychol. (2009) 34:1069–83. doi: 10.1093/jpepsy/ jsp025 November 2018 \| Volume 5 \| Article 115 Frontiers in Nutrition \| www.frontiersin.org 13
https://openalex.org/W3011158255	https://www.degruyter.com/document/doi/10.1515/jtm-2019-0003/pdf	English	null	Transition of adolescents with congenital heart disease from pediatric to adult congenital cardiac care: lessons from a retrospective cohort study	Journal of transition medicine	2,020	cc-by	7,445	Abstract Methods: This is a retrospective cohort study in which all patients eligible for transfer from pediatric to adult con- genital cardiac care from January 2006 to December 2012 were identified from the McMaster Children’s Hospital database. Successful transfer was defined as attendance at the ACCC within 2 years of discharge from Pediatric Cardiology. Patient and context-related variables include gender, severity of the CHD diagnosis, years since pedi- atric follow-up, and distance from the patient’s home to MUMC. The relationship between patient- and context- related variables available at baseline and unsuccessful transfer was assessed by univariate analysis. Background: The transfer from pediatric to adult care is a key milestone for adolescents living with chronic health conditions. Over the past few decades, pediatric cardiac care has witnessed outstanding advancements leading to a dramatic increase in the number of children with con- genital heart disease (CHD) surviving into adulthood. Suc- cessful transfer from pediatric to adult congenital cardiac care is critical because many adults with CHD require reg- ular long-term cardiac care for optimal health outcomes. Objectives: This study aims to (1) determine the rate of suc- cessful transfer of adolescents with CHD from pediatric to adult congenital cardiac care at the McMaster University Medical Centre (MUMC), a tertiary care level centre, and (2) to explore available patient- and context-related fac- tors associated with unsuccessful transfer. MUMC includes both the McMaster Children’s Hospital, which offers Background: The transfer from pediatric to adult care is a key milestone for adolescents living with chronic health conditions. Over the past few decades, pediatric cardiac care has witnessed outstanding advancements leading to a dramatic increase in the number of children with con- genital heart disease (CHD) surviving into adulthood. Suc- cessful transfer from pediatric to adult congenital cardiac care is critical because many adults with CHD require reg- ular long-term cardiac care for optimal health outcomes. Objectives: This study aims to (1) determine the rate of suc- cessful transfer of adolescents with CHD from pediatric to adult congenital cardiac care at the McMaster University Medical Centre (MUMC), a tertiary care level centre, and (2) to explore available patient- and context-related fac- tors associated with unsuccessful transfer. MUMC includes both the McMaster Children’s Hospital, which offers Results: A total of 279 patients were identified, of which, 269 patients (96.4%) were successfully transferred to adult congenital cardiac care. Tapas K. Mondal, Varun Muddaluru, Ahmad Jaafar, Shikha Gupta-Bhatnagar, Lucas J. Greville and Jan Willem Gorter https://doi.org/10.1515/jtm-2019-0003 Received October 18, 2019; accepted January 27, 2020 Pediatric Cardiology services, and Adult Outpatient Ser- vices, which offers the Adult Congenital Cardiac Clinic (ACCC). J Transition Med 2020; aop Open Access. © 2020 Tapas K. Mondal, Varun Muddaluru, Ahmad Jaafar, Shikha Gupta-Bhatnagar, Lucas J. Greville and Jan Willem Gorter, published by De Gruyter. This work is licensed under the Creative Commons Attribution 4.0 International License. Abstract Out of the 10 patients (3.6%) who were lost to follow-up, 8 had mild, 1 had moderate, and 1 had severe CHD. Based on the point estimates expressed as odds ratio (OR), factors that are potentially associated with a higher risk for loss to follow-up were: male gender (OR 1.8, 95% CI 0.5–7.3) and travel distance greater than 200 km to MUMC (OR 7.7, 95% CI 0.7–81.5), while moder- ate and severe CHD could potentially be a protective factor against loss to follow up when compared to mild CHD (OR 0.2, 95% CI 0–1.1). Corresponding author: Tapas K. Mondal, Faculty of Health Sciences, McMaster University, Hamilton, Ontario, Canada; and Department of Pediatrics, McMaster Children’s Hospital, Hamilton, Ontario, Canada, E-mail: mondalt@mcmaster.ca. https://orcid.org/ 0000-0003-2077-1710 Discussion: The medical and administrative practices that may be contributing to the high transfer rate of 96.4% include early and developmentally appropriate discus- sions, engaging patients and their families in cardiac care, proximity of the pediatric and adult congenital cardiac clinics, and an information pamphlet regarding the tran- sition process, amongst others. Learning from our retro- spective study we now work with the patients identified as potential high risk for loss to follow-up to understand and eliminate barriers and to implement sustainable methods that will ensure a successful transition to adult health care for all patients with CHD. Varun Muddaluru: Graduate Entry Medicine, Royal College of Surgeons in Ireland, Dublin, Ireland, E-mail: varunmuddaluru@rcsi.com Ahmad Jaafar: Department of Pediatrics, McMaster Children’s Hospital, Hamilton, Ontario, Canada, E-mail: ahmad.jaafar@medportal.ca Shikha Gupta-Bhatnagar: Department of Pediatrics, McMaster Children’s Hospital, Hamilton, Ontario, Canada, E-mail: guptas37@mcmaster.ca Lucas J. Greville: McMaster University, Hamilton, Ontario, Canada, E-mail: grevillj@mcmaster.ca Jan Willem Gorter: Department of Pediatrics, McMaster Children’s Hospital, Hamilton, Ontario, Canada; and McMaster University, Hamilton, Ontario, Canada, E-mail: gorter@mcmaster.ca Varun Muddaluru: Graduate Entry Medicine, Royal College of Surgeons in Ireland, Dublin, Ireland, E-mail: varunmuddaluru@rcsi.com Ahmad Jaafar: Department of Pediatrics, McMaster Children’s Hospital, Hamilton, Ontario, Canada, E-mail: ahmad.jaafar@medportal.ca Shikha Gupta-Bhatnagar: Department of Pediatrics, McMaster Children’s Hospital, Hamilton, Ontario, Canada, E-mail: guptas37@mcmaster.ca Lucas J. Abstract Greville: McMaster University, Hamilton, Ontario, Canada, E-mail: grevillj@mcmaster.ca Jan Willem Gorter: Department of Pediatrics, McMaster Children’s Hospital, Hamilton, Ontario, Canada; and McMaster University, Hamilton, Ontario, Canada, E-mail: gorter@mcmaster.ca Varun Muddaluru: Graduate Entry Medicine, Royal College of Surgeons in Ireland, Dublin, Ireland, E-mail: varunmuddaluru@rcsi.com Ahmad Jaafar: Department of Pediatrics, McMaster Children’s Hospital, Hamilton, Ontario, Canada, E-mail: ahmad.jaafar@medportal.ca Shikha Gupta-Bhatnagar: Department of Pediatrics, McMaster Children’s Hospital, Hamilton, Ontario, Canada, E-mail: guptas37@mcmaster.ca Lucas J. Greville: McMaster University, Hamilton, Ontario, Canada, E-mail: grevillj@mcmaster.ca Jan Willem Gorter: Department of Pediatrics, McMaster Children’s Hospital, Hamilton, Ontario, Canada; and McMaster University, Hamilton, Ontario, Canada, E-mail: gorter@mcmaster.ca 2 \| Mondal et al.: Lessons learned from the transition of patients from pediatric to adult congenital cardiac clinic 2 \| Mondal et al.: Lessons learned from the transition of patients from pediatric to adult congenital cardiac clinic 2 Keywords: adolescents; adults; congenital heart disease; retrospective cohort study; transfer; transition. Ontario, Canada [11]. According to Mackie et al., only 39% of patients between 18 and 22 years old were still being seen in outpatient cardiology in Quebec, Canada [12]. Fail- ure to maintain continuous cardiac care into adulthood can delay recognition of evolving and new cardiac prob- lems, thus increasing the risk of negative health complica- tions and premature mortality [14–16]. Complications that adults with CHD may develop include arrhythmias leading to sudden cardiac death, progressive heart failure, [16, 17] and the risk of reoperation [18, 19]. For these reasons, the successful transfer of patients with CHD from pediatric to adult congenital cardiac care is of particular importance. Significance of the continuity of cardiac care The transition from adolescence to adulthood is challeng- ing for young individuals with chronic health conditions. These individuals become responsible for their health management while facing additional life challenges such as higher education, starting a career, and moving out of their parents’ home [1]. A key element of this transition to adulthood is the transfer of individuals from pediatric services to adult health care settings. While the divide in the pediatric and adult health care system often lim- its the ability to track the outcomes after transfer, clini- cians in pediatric care are encouraged to evaluate their success rates of transitioning young people. This study assesses the transfer of adolescents with congenital heart disease (CHD) from pediatric to adult congenital cardiac care at the McMaster University Medical Centre (MUMC), a tertiary care level centre in Hamilton, Ontario, Canada. MUMC includes both the McMaster Children’s Hospital, which offers Pediatric Cardiology services, and Adult Out- patient Services, which offers the Adult Congenital Cardiac Clinic (ACCC). Transfer vs. transition of care The terms ‘transfer’ and ‘transition’ of care are related but not synonymous. Transfer of care is an event that occurs at some point during pediatric care in which the patient is transferred to adult care [20]. In the pediatric cardiac care setting, it implies that the patient has been discharged from the pediatric cardiac care program with a referral to follow-up with the adult congenital cardiac care program. On the other hand, transition of care is a continuous process that occurs over years, and can start even before reaching adolescence. It works at enabling the patient to take charge and responsibility of their life and health [21]. It aims at preparing the patient to be followed-up by the adult care program, and it often requires a collaboration between the patient, their family, and the pediatric and adult health care providers [22]. The American Academy of Pediatrics states that “the goal of transition in health care for young adults with special health care needs is to maximize life-long functioning and potential through the provision of high-quality, developmentally appropri- ate health care services that continue uninterrupted as the individual moves from adolescence to adulthood” [23]. The importance of cardiac care transition was raised in the five task-force reports obtained from the Bethesda conference on the care of the adult with CHD [4, 7, 24–26]. CHD is considered the most common birth defect with an estimated incidence of 0.8% in neonates [2]. Recent advancements in pediatric cardiac care including diag- nostic modalities, medical therapies, and surgical and catheter-based interventions have dramatically increased the number of children with CHD surviving into adult- hood [3]. Currently, between 80% and 85% of children with CHD survive into adulthood with an estimated 100,000 Canadian and 1 million American adults living with CHD [4, 5]. After a catheter-based or surgical cardiac interven- tion, many CHD patients, particularly those with complex lesions, require life-long care. This highlights the impor- tance for cardiac care to extend into adulthood [6]. Accord- ing to the Canadian medical guidelines, more than 50% of adults living with CHD are recommended to regularly follow-up with an adult cardiologist specializing in CHD every 12–24 months [4, 7–10]. Predictors of unsuccessful transfer Baseline data were obtained from patient charts to poten- tially identify predictors of unsuccessful transfer. Such data include (1) gender, (2) severity of CHD, and (3) dis- tance from MUMC. The severity of the cardiac diagnoses was classified into mild, moderate or severe using the Adult Congenital Heart Disease Consensus Statement [30]. MUMC is located on the same site as the McMaster Chil- dren’s Hospital, and includes the McMaster Children’s Hospital, Women’s Health Centre, and the Adult Outpa- tient Services, where the ACCC is located. The linear dis- tance to MUMC, which contains both the pediatric and adult congenital cardiac clinics was calculated based on the postal code of the patient’s residence. In addition, the patient population was stratified based on the time (in years) since their last pediatric follow up. This was to assess any difference in transfer rates over the study period of January 2006 – December 2012. For example, 2 years since pediatric follow up means 2 years prior to 2012. Addi- tional data reviewed in the study include attendance at the pediatric cardiology clinic appointments as well as docu- mentation of the recommended follow-up with adult con- genital cardiac care. Any missing referrals to ACCC were reviewed and included in the study as well. Setting The study evaluated our services at McMaster Children’s Hospital, a tertiary care level pediatric academic hospi- tal located in Hamilton, Ontario, Canada. It serves health- care to children and youth up to 18 years of age in a wide range of pediatric specialities, including pediatric cardiology. Our transition service is integrated in routine clinical practice, and delivered by the Pediatric Cardio- logists from the Department of Pediatrics. A transition pamphlet is provided prior to the actual transfer process (Supplemental Figure 1). Recently, in collaboration with the CanChild Centre for Childhood Disability Research at McMaster University, we have implemented the MyTran- sition mobile phone application available on the Apple App Store and Google Play Store [27, 28]. The app contains the Transition-Q questionnaire which tracks the develop- ment of skillset required in health management [29]. The Pediatric and ACCC are located at the same site in Hamil- ton. This enables that all tests including ECG, holter, and exercise tests are done in the same place. If patients miss their appointment with the Pediatric or Adult service, they are contacted by the respective departments via telephone call. The ACCC has a dedicated nurse practitioner who contacts the patients. When patients miss their appoint- ment, a mailing letter is sent and if we are unable to con- tact the patient via telephone more than twice, their GP is contacted. Ethical considerations This is a retrospective cohort study that included all patients who were eligible for transfer from pediatric to adult congenital cardiac care from January 2006 to Decem- ber 2012, with ages ranging from 17 to 26 years, based on the Canadian Cardiovascular Society (CCS) consensus cri- teria [8–10]. No exclusion criteria were set in the study. The data were obtained from the McMaster Children’s Hospital database. Ethical board approval was obtained from the Hamilton Integrated Research Ethics Board (HiREB) before conduct- ing the study. To secure patient confidentiality, no per- sonal information from any participant was utilized or shared in the process of conducting the study. Each par- ticipant was assigned a random number when collecting and analyzing the data, without referring to any personal information such as name or chart number. Methods time frame recommended by most pediatric cardiac clin- ics based on resources and demands placed on the clin- ics. However, adult congenital cardiac clinics try to pro- vide appointments to patients within 1 year from their last appointment with the pediatric cardiac care program. Objectives This study aims to (1) determine the rate of success- ful transfer of adolescents with CHD from pediatric to adult congenital cardiac care at a single tertiary care level centre – McMaster University Medical Centre located in Hamilton, Ontario, and (2) to evaluate available patient- and context-related factors associated with unsuccessful transfer. Studies have shown that less than 50% of children with CHD are successfully transferred to adult congeni- tal cardiac care [11–13]. Reid et al. reported a 47% (95% CI: 42–52) transfer rate for patients aged 19–21 years with complex CHD between the Hospital for Sick Children and the Toronto Congenital Cardiac Centre for Adults in Mondal et al.: Lessons learned from the transition of patients from pediatric to adult congenital cardiac clinic Mondal et al.: Lessons learned from the transition of patients from pediatric to adult congenital cardiac clinic 3 Data analysis The primary outcome in the study was the successful transfer of children with CHD from Pediatric Cardiology to the Adult Congenital Cardiac Clinic at MUMC. Successful transfer was defined as attendance (original or resched- uled appointment) at the ACCC within 2 years of dis- charge from Pediatric Cardiology. This definition fits the Descriptive statistical analysis was utilized for variables available at study entry (i.e. baseline). The relationship between factors available at baseline and our outcome of interest, i.e. successful versus unsuccessful transfer was assessed by univariate analyses, and expressed as an odds ratio with 95% confidence interval. 4 \| Mondal et al.: Lessons learned from the transition of patients from pediatric to adult congenital cardiac clinic Results Also, patients seen in earlier years (>3 years prior to the study period) were potentially at a higher risk for loss to follow-up when compared to those seen in recent years prior to the study (OR 1.4, CI 0.4–5.2). Table 1: Participants’ demographic characteristics. All participants (n = 279) Successful transfer (n = 269) Unsuccessful transfer (n = 10) Age (years) 17–22, n (%) 180 (67) 8 (80) >22, n (%) 65 (24) 2 (20) Not available on chart, n (%) 24 (9) 0 (0) Gender Females, n (%) 119 (44) 3 (30) Males, n (%) 150 (56) 7 (70) Severity based on ACHD rating Mild, n (%) 129 (48) 8 (80) Moderate, n (%) 103 (38) 1 (10) Severe, n (%) 37 (14) 1 (10) Distance from patient’s home to MUMCa <50 km, n (%) 153 (57) 5 (50) 50–200 km, n (%) 75 (28) 1 (10) >200 km, n (%) 4 (1) 1 (10) Not available in chart, n (%) 37 (14) 3 (30) Years since pediatric follow-up 0, n (%) 9 (3) 0 (0) 1, n (%) 34 (13) 0 (0) 2, n (%) 38 (14) 2 (20) 3, n (%) 44 (16) 4 (40) 4, n (%) 25 (9) 0 (0) 5, n (%) 30 (11) 2 (20) 6, n (%) 32 (12) 1 (10) 7, n (%) 32 (12) 1 (10) 8, n (%) 1 (0) 0 (0) Not available on chart, n (%) 24 (9) 0 (0) aMUMC – McMaster University Medical Centre. the low number of patients who were lost to follow-up (n = 10) statistically significant associations could not be established as expressed in the wide 95% confidence intervals. However, the point estimates did demonstrate an increased risk for loss to follow-up for male patients (OR 1.8, 95% CI 0.5–7.3) and those travelling greater than 200 km to the hospital (OR 7.7, 95% CI 0.7–81.5). Next, mod- erate and severe CHD could potentially be a protective fac- tor against loss to follow up when compared to mild CHD (OR 0.2, 95% CI 0–1.1). Also, patients seen in earlier years (>3 years prior to the study period) were potentially at a higher risk for loss to follow-up when compared to those seen in recent years prior to the study (OR 1.4, CI 0.4–5.2). Results severe CHD. Most participants (56.6%) were living within 50 kilometers (km) from McMaster Children’s Hospital, 27% of patients were living >50 km and 2% of patients were living >200 km. A total of 279 patients were identified and included in the study based on the inclusion criteria. Of these patients, 157 (56%) were male and 122 (44%) were female (Table 1). The years since pediatric follow up ranging from 0 to 8 years is shown in Table 1. A total of 269 (96.4%) patients were suc- cessfully transferred from Pediatric Cardiology to the ACCC at MUMC and 10 patients (3.6%) were lost to follow-up between January 2006 and December 2012. Out of the 10 patients who were lost to follow-up with the ACCC, 1 patient did not attend their last pedi- atric cardiology follow-up, 1 patient did not have a docu- mented recommendation to follow-up with the ACCC and 2 patients were missing referrals to the ACCC (data not shown). Table 2 contains additional detailed information regarding the 10 patients who were lost to follow-up with the ACCC including their cardiac diagnoses, associated co- morbidities, past surgical or catheter-based interventions, and any medications the patients were prescribed, divided into cardiac and non-cardiac medications. We evaluated the demographics of our population according to patient- and context-related factors (Table 1). Approximately half of the reviewed patients (49.1%) had mild CHD, with the remaining patients having either mod- erate (37.3%) or severe (13.6%) CHD. Of the 10 patients who were lost to follow-up, 8 had mild, 1 moderate, and 1 had Factors that are potentially influencing the risk of unsuccessful transfer are presented in Table 3. Due to the low number of patients who were lost to follow-up (n = 10) statistically significant associations could not be established as expressed in the wide 95% confidence intervals. However, the point estimates did demonstrate an increased risk for loss to follow-up for male patients (OR 1.8, 95% CI 0.5–7.3) and those travelling greater than 200 km to the hospital (OR 7.7, 95% CI 0.7–81.5). Next, mod- erate and severe CHD could potentially be a protective fac- tor against loss to follow up when compared to mild CHD (OR 0.2, 95% CI 0–1.1). aMUMC – McMaster University Medical Centre. Results Bicuspid aortic valve (mild ARb, mild ASc) None Moderate 1998 – balloon dilatation and stent placement 2005 – diagnostic angiogram Bisoprolol 5 mg once daily August 2010 2. Coarctation of aorta 3. Hypertension 20 Male 1. Corrected TGAd with supero-inferiorly related ventricles None Severe 1992 – VSD closure 1994 – pacemaker insertion None None 2. VSDe 3. Right-sided AVf valve regurgitation 4. Second degree AV block 20 Female 1. PDAg (very small left-to-right shunt) Developmental delay Mild 2010 – PDA repair Non-cardiac June 2012 19 Male 1. Perimembranous VSD (no leak) Prematurity, ADHD, cleft palate, hypospadias, velopharyngeal repair Mild VSD repair None September 2010 2. Resection of RVh muscular bundles 22 Male 1. Small perimembranous VSD None Mild None None May 2009 2. Small-to-moderate ASDi 3. Premature ventricular contractions 23 Female 1. Bicuspid aortic valve (no AR or AS) Chromosomal disorder Mild None None January 2007 2. Subaortic ridge (jpg: 49 mmHg) 3. Mild LVk outflow tract gradient 22 Male 1. Perimembranous VSD (pg: 119 mmHg) None Mild None None May 2007 24 Male 1. Perimembranous VSD (pg: 100 mmHg) Chromosomal disorder, leukopenia Mild None None May 2006 2. Small PFOl 20 Male 1. Mild pulmonary stenosis (pg: 25 mmHg) None Mild None None February 2009 2. Moderate pulmonary insuflciency 0 20 Female 1. Bicuspid aortic valve (mild AR) Depression Mild None Non-cardiac December 2009 No. – number; bAR – aortic regurgitation; cAS – aortic stenosis; dTGA – transposition of great arteries; eVSD – ventricular septal defect; fAV – atrioventricular; gPDA – patent ductus arteriosus; RV – right ventricle; iASD – atrial septal defect; jpg – peak gradient; kLV – left ventricle; lPFO – patent foramen ovale. Table 2: Demographics and relevant medical histories of patients lost to follow-up. Case No.a Age Gender Cardiac diagnosis Co-morbidities ACHD class Surgery/intervention Medication Last visit 1 19 Male 1. Bicuspid aortic valve (mild ARb, mild ASc) None Moderate 1998 – balloon dilatation and stent placement 2005 – diagnostic angiogram Bisoprolol 5 mg once daily August 2010 2. Coarctation of aorta 3. Hypertension 2 20 Male 1. Corrected TGAd with supero-inferiorly related ventricles None Severe 1992 – VSD closure 1994 – pacemaker insertion None None 2. VSDe 3. Right-sided AVf valve regurgitation 4. Second degree AV block 3 20 Female 1. PDAg (very small left-to-right shunt) Developmental delay Mild 2010 – PDA repair Non-cardiac June 2012 4 19 Male 1. Results Discussion McMaster Children’s Hospital & the rate of successful cardiac transfer The Pediatric Cardiology program at McMaster Children’s Hospital recognizes the importance of successful transfer from pediatric to adult congenital cardiac care. Although the hospital does not have a formal transition program, we are excited to demonstrate in the cardiac clinic with an integrated approach to transition a high rate of success- ful transfer, 96.4%, exceeding many other figures reported in the literature. For instance, Reid et al. reported that 47% of adolescents with CHD underwent successful trans- fer of cardiac care in Toronto, Canada [11]. Their defini- tion of successful transfer was attendance to at least 1 cardiac appointment of any type (e.g. clinic, echocardio- gram, cardiac catheterization, or surgical), without speci- fying a timeframe, while in the current study, successful Table 1: Participants’ demographic characteristics. All participants (n = 279) Successful transfer (n = 269) Unsuccessful transfer (n = 10) Age (years) 17–22, n (%) 180 (67) 8 (80) >22, n (%) 65 (24) 2 (20) Not available on chart, n (%) 24 (9) 0 (0) Gender Females, n (%) 119 (44) 3 (30) Males, n (%) 150 (56) 7 (70) Severity based on ACHD rating Mild, n (%) 129 (48) 8 (80) Moderate, n (%) 103 (38) 1 (10) Severe, n (%) 37 (14) 1 (10) Distance from patient’s home to MUMCa <50 km, n (%) 153 (57) 5 (50) 50–200 km, n (%) 75 (28) 1 (10) >200 km, n (%) 4 (1) 1 (10) Not available in chart, n (%) 37 (14) 3 (30) Years since pediatric follow-up 0, n (%) 9 (3) 0 (0) 1, n (%) 34 (13) 0 (0) 2, n (%) 38 (14) 2 (20) 3, n (%) 44 (16) 4 (40) 4, n (%) 25 (9) 0 (0) 5, n (%) 30 (11) 2 (20) 6, n (%) 32 (12) 1 (10) 7, n (%) 32 (12) 1 (10) 8, n (%) 1 (0) 0 (0) Not available on chart, n (%) 24 (9) 0 (0) aMUMC – McMaster University Medical Centre. Table 1: Participants’ demographic characteristics. Table 1: Participants’ demographic characteristics. Mondal et al.: Lessons learned from the transition of patients from pediatric to adult congenital cardiac clinic \| 5 ble 2: Demographics and relevant medical histories of patients lost to follow-up. ase No.a Age Gender Cardiac diagnosis Co-morbidities ACHD class Surgery/intervention Medication Last visit 19 Male 1. b. Engaging the patient in their cardiac care transfer is defined as patients seen within 2 years since discharge from the pediatric clinic. The former study only included patients with complex CHD while the current study includes mild, moderate and severe CHD patients. Mackie et al. found that individuals with simple CHD were at a higher risk for loss to follow up when compared to patients with complex CHD. The high transfer rate (96.4%) at McMaster, which includes this high-risk population of mild and moderate CHD patients, indicates the success of McMaster in transferring patients from pediatric cardiol- ogy to adult cardiac care. Of note are the limitations of our study. The current study covered a specific period in time, which is between January 2006 and December 2012. This may not reflect our current figures of successful trans- fer, however, no major changes in the clinical processes of transfer took place over the past few years, making a sig- nificant decrease in the figure less likely. It is worth men- tioning that this was our first evaluation of transition out- comes conducted at McMaster Children’s Hospital using a retrospective chart review, and further, prospective stud- ies are now anticipated to monitor the outcomes of our transitional care over time. Encouraging patients to take a proactive role in their cardiac care is an important strategy implemented by the Pediatric Cardiology service. Foremost, the patient is educated on their condition using simple drawings and easy-to-follow pictures, and simple tablet educational applications. Next, patients are encouraged to share how they cope with CHD with their family members and peers. This is in hope that educating the public will decrease any stigmatization or labels placed on individuals with CHD. Moreover, patients are actively involved in their cardiac health management. For instance, patients are engaged in discussions regarding why a specific treatment method is preferred over the other, why an investigation is required within a given timeframe, and why long-term follow-up is required. An interesting finding is that during the 3 years prior to the study in 2013 (2010–2013), the follow-up outcomes of patients seem to be more favourable when compared to those who were transferred greater than 3 years prior to the study (2006–2009) (OR 1.4; 95% CI: 0.4–5.2). This could be due to changes introduced in 2009 that increased awareness about the importance of transition in pediatric healthcare, lead by the CanChild research centre which is affiliated with McMaster Children’s Hospital. Results Perimembranous VSD (no leak) Prematurity, ADHD, cleft palate, hypospadias, velopharyngeal repair Mild VSD repair None September 2010 2. Resection of RVh muscular bundles 5 22 Male 1. Small perimembranous VSD None Mild None None May 2009 2. Small-to-moderate ASDi 3. Premature ventricular contractions 6 23 Female 1. Bicuspid aortic valve (no AR or AS) Chromosomal disorder Mild None None January 2007 2. Subaortic ridge (jpg: 49 mmHg) 3. Mild LVk outflow tract gradient 7 22 Male 1. Perimembranous VSD (pg: 119 mmHg) None Mild None None May 2007 8 24 Male 1. Perimembranous VSD (pg: 100 mmHg) Chromosomal disorder, leukopenia Mild None None May 2006 2. Small PFOl 9 20 Male 1. Mild pulmonary stenosis (pg: 25 mmHg) None Mild None None February 2009 2. Moderate pulmonary insuflciency 10 20 Female 1. Bicuspid aortic valve (mild AR) Depression Mild None Non-cardiac December 2009 aNo. – number; bAR – aortic regurgitation; cAS – aortic stenosis; dTGA – transposition of great arteries; eVSD – ventricular septal defect; fAV – atrioventricular; gPDA – patent ductus arteriosus; hRV – right ventricle; iASD – atrial septal defect; jpg – peak gradient; kLV – left ventricle; lPFO – patent foramen ovale. 6 \| Mondal et al.: Lessons learned from the transition of patien Table 3: Risk implications of patient- and contextual related factors available at study entry (i.e. baseline variables) for unsuccessful transfer. Patient-related Factors Unsuccessful transfer (n = 10) Odds ratio 95% CI Gender Male 1.8 0.5–7.3 ACHD severity Moderate vs. Mild 0.2 0–1.2 Severe vs. Mild 0.4 0–3.6 Moderate + Severe vs. Mild 0.2 0–1.1 Contextual-related factors Years since pediatric follow-up >3 years 1.4 0.4–5.2 Distance to MUMC (km) >50 0.4 0–3.6 >50 + >200 0.8 0.2–4.1 >200 7.7 0.7–81.5 6 \| Mondal et al.: Lessons learned from the transition of patients from pediatric to adult congenital cardiac clinic a. Early and developmentally appropriate discussions Early and developmentally appropriate discussions aim to provide the patient and parents with a better under- standing of their specific CHD, future expectations, and the importance of continuing with the adult congenital cardiac clinic visits. These discussions may begin as early as 10–12 years of age, as recommended by the American College of Cardiology Task Force and the 2008 ACC/AHA guidelines on the Management of Adults with CHD [31]. It is important to highlight that patients with developmen- tal delays/syndromes did not have a particular problem in the transition process as the parents and caregivers were heavily involved in the patient’s care and understood the complexity of the patient’s heart condition, along with the prognosis in adult life. The parents and caregivers con- tinue to bring them to the ACCC as in their pediatric age. d. Transition pamphlet At the last pediatric cardiology clinic appointment, patients are given a transition pamphlet regarding their transition to the ACCC. This pamphlet (Supplemental Figure 1) includes important instructions and information about the adult program, its importance, and how to reach and communicate with the clinic. Please note that the contact information of the ACCC staff has been removed from Supplemental Figure 1. In addition, a transition pam- phlet is also sent to the patients by the ACCC immediately after receipt of the referral so that the patient is not lost to follow-up. Similar resources provided by the pediatric and adult congenital cardiac services ensures that there is no gap in knowledge and that the patient is comfortable during the transfer process. Furthermore, the proximity of the pediatric and adult congenital cardiac clinics to each other may be another correlate of the successful transfer seen at McMaster Chil- dren’s Hospital, as both clinics are located at the MUMC. This makes it easier for the patient to follow-up with the ACCC, as they would be familiar with the building environment, parking, it’s facilities, and medical diag- nostic units, which are the same for many pediatric and adult procedures such as X-ray, EKG, echocardiogram, etc. f. Proximity of the adult congenital cardiac clinic visit in which this recommendation takes place. This is in keeping with Mackie et al. who discussed that docu- menting the need to follow-up has a protective role against the loss of follow-up [13]. However, our current study found 9 of the 10 patients who were lost to follow-up had documented recommendation to do so. Therefore, it is important to apply other means to reassure success- ful transition even when the need to follow-up has been documented. The proximity of the adult congenital cardiac clinic to the patient’s residence may be a critical variable in the suc- cessful transfer of care. As evident in this study, more than half of the patients (56.6%) who were under the care of the pediatric cardiology program lived within 50 kilome- ters of MUMC. Similarly, most patients (56.9%) who’s care was successfully transferred were living within the same distance. Overall, there was potentially a 7.7-fold increased risk for those who live >200 km from the hospital (OR 7.7; 95% CI 0.7–81.5) to be lost to follow-up, indicating that dis- tance could be a barrier for patients to attend the ACCC though it did not reach statistical significance due to the wide CI. c. Medical record documentation for the need of follow-up Several factors may have played a role in the high rate of successful cardiac care transfer at McMaster Children’s Hospital. These may include the following: Documenting the recommendation to follow-up with an adult congenital cardiac clinic is important at every clinic Mondal et al.: Lessons learned from the transition of patients from pediatric to adult congenital cardiac clinic \| 7 7 References 1. Menrath I, Ernst G, Szczepanski R, Lange K, Bomba F, Staab D, et al. Effectiveness of a generic transition-oriented patient education program in a multicenter, prospective and con- trolled study. J Transition Med [Internet]. 2018;1 [cited 2019 Aug 20];1. Available from: https://www.degruyter.com/view/j/ jtm.2019.1.issue-1/jtm-2018-0001/jtm-2018-0001.xml?format= INT. doi: https://doi.org/10.1515/jtm-2018-0001.f Recommendations 2. Hoffman JI, Kaplan S. The incidence of congenital heart disease. J Am Coll Cardiol 2002;39:1890–900. A. Have regular, early, and developmentally appropriate discussions about the importance of continuity of car- diac care with patients and parents, if patients are eligible for transfer to adult congenital cardiac care. 3. Marelli AJ, Mackie AS, Ionescu-Ittu R, Rahme E, Pilote L. Con- genital heart disease in the general population: changing prevalence and age distribution. Circulation 2007;115:163–72.f B. Improve the patient and parent’s understanding of their form of CHD and encourage the patient to take a proactive role in their cardiac care. 4. Warnes CA, Liberthson R, Danielson GK, Dore A, Harris L, Hoff- man JI, et al. Task force 1: the changing profile of congenital heart disease in adult life. J Am Coll Cardiol 2001;37:1170–5. 5. Reid GJ, Irvine MJ, McCrindle BW, Sananes R, Ritvo PG, Siu SC, et al. Prevalence and Correlates of Successful Transfer From Pediatric to Adult Health Care Among a Cohort of Young Adults With Complex Congenital Heart Defects. Pediatrics 2004;113:e197–e205. 5. Reid GJ, Irvine MJ, McCrindle BW, Sananes R, Ritvo PG, Siu SC, et al. Prevalence and Correlates of Successful Transfer From Pediatric to Adult Health Care Among a Cohort of Young Adults With Complex Congenital Heart Defects. Pediatrics 2004;113:e197–e205. C. Ensure that the transfer of cardiac care occurs in a smooth, sensitive manner. D. Standardize the referring process to avoid missing referrals. 6. Wacker A, Kaemmerer H, Hollweck R, Hauser M, Deutsh MA, Brodherr-Heberlein S, et al. Outcome of operated and unoper- ated adults with congenital cardiac disease lost to follow-up for more than five years. Am J Cardiol 2005;95:776–9. E. Consider a pre-transfer visit to the adult clinic (if not in the same facility) to familiarize patients and their fam- ilies with the new environment. This might include meeting the nurse practitioner in the adult clinic (if available) and a brief tour of the clinic and diagnos- tic areas to ease the transition process and facilitate successful transfer. i 7. Landzberg MJ, Murphy DJ, Davidson WR, Jarcho JA, Krumholz HM, Mayer JE, et al. Task force 4: organization of delivery sys- tems for adults with congenital heart disease. J Am Coll Cardiol 2001;37:1187–93. 8. Therrien J, Dore A, Gersony W, Iserin L, Liberthson L, Meijbom F, et al. CCS consensus conference 2001 update: Recommen- dations for the management of adults with congenital heart disease. e. Early adult congenital cardiac clinic follow-up and reminder call 8 \| Mondal et al.: Lessons learned from the transition of patients from pediatric to adult congenital cardiac clinic 8 \| Mondal et al.: Lessons learned from the transition of patients from pediatric to adult congenital cardiac clinic 8 Research funding: None declared. Research funding: None declared. Author contributions: All authors have accepted respon- sibility for the entire content of this manuscript and approved its submission. Competing interests: Authors state no conflict of interest. Informed consent: Due to the nature of this retrospective study, the HiREB waived informed consent. Ethical approval: Research involving human subjects com- plied with all relevant national regulations, institutional policies and is in accordance with the tenets of the Helsinki Declaration (as revised in 2013). Ethical board approval was obtained from Hamilton Integrated Research Ethics Board (HiREB) before conducting the study. Next steps in improving transitional care Acknowledgements: We acknowledge Laura Lee Walter, Retired Nurse Practitioner and Dr. Elaine Gordon, Retired Adult Congenital Heart Disease Specialist in the Adult Congenital Cardiac Clinic (ACCC), McMaster University Medical Centre, for providing patient data, support and help. Acknowledgements: We acknowledge Laura Lee Walter, Retired Nurse Practitioner and Dr. Elaine Gordon, Retired Adult Congenital Heart Disease Specialist in the Adult Congenital Cardiac Clinic (ACCC), McMaster University Medical Centre, for providing patient data, support and help. To promote smooth transition and a successful transfer, we need to obtain comprehensive insights about patients’ transition readiness, or the ability to self-manage their own health. Since our retrospective chart review, we have implemented the TRANSITION-Q [31] and MyTransition app [27] which were developed by our partner, CanChild research institute. Further studies comparing the trans- fer processes and outcomes between sites, cities, and countries for congenital heart disease, along with other chronic health conditions are recommended. Recently, Mackie et al.’s review focused on the state of transfer and transition in CHD in Canada and proposed a transition curriculum aimed at CHD education and self-management and self-advocacy skills development [33]. Their findings along with future studies will help to identify best prac- tices and unique patient- and context-related factors influ- encing successful transfer between pediatric and adult health care settings. Subsequently, this will enable clini- cians to develop strategies to achieve improved and sus- tainable transition outcomes for the patient and their family. e. Early adult congenital cardiac clinic follow-up and reminder call In the current study, although significant conclusions could not be established, male patients (OR 1.8, 95% CI 0.5–7.3) and patients travelling greater than 200 km to the hospital (OR 7.7, 95% CI 0.7–81.5) were potentially at a higher risk for loss to follow-up. Goossens et al. also found that male gender (OR: 1.80; 95% CI: 1.02–3.17) was a pre- dictor of patients who were lost to follow-up with adult congenital cardiac care [32]. Our study found that moder- ate and severe CHD could potentially be a protective factor against loss to follow up when compared to mild CHD (OR 0.2; 95% CI: 0–1.1). This is in line with findings reported by Mackie et al., where they found that patients with a simple shunt lesion have a greater risk for unsuccessful follow-up when compared to patients with complex/severe lesions. It is important to note that patients with mild CHD will be more likely to seek cardiac care from a community cardi- ologist rather than an adult congenital cardiac clinic [32]. This maybe the case with the patient population in this study as 8 out of the 10 patients who were lost to follow up had mild CHD. The first adult congenital cardiac clinic appointment at MUMC is usually fixed to be within 1 year of discharge from pediatric cardiac care, regardless of the severity of the lesion. Although some patients are followed less fre- quently with the pediatric cardiac clinics, they are given an earlier appointment with the ACCC in order to facili- tate successful transfer and to avoid losing patients due to change of address, leaving home for school, etc. Patients are given a reminder call before their first ACCC appointment. If no answer was obtained a second call is made at a later time. This is to ensure that the patient and/or their parents are aware of the date, time and loca- tion of the appointment. It is worth noting that the ACCC at McMaster University Medical Centre has a dedicated nurse practitioner who corresponds with new patients to ensure successful follow-up. When patients miss their appoint- ment, a mailing letter is sent and if we are unable to con- tact the patient via telephone more than twice, their GP is contacted. Recommendations Part I. Can J Cardiol 2001;17:940–59. F. Conduct further studies to better understand the pre- dictors of unsuccessful transfer to adult congenital cardiac care. Mondal et al.: Lessons learned from the transition of patients from pediatric to adult congenital cardiac clinic \| 9 Mondal et al.: Lessons learned from the transition of patients from pediatric to adult congenital cardiac clinic \| 9 9 23. American Academy of Pediatrics, American Academy of Family Physicians, and American College of Physicians-American Soci- ety of Internal Medicine. A consensus statement on health care transitions for young adults with special health care needs. Pediatrics 2002;110(6 Pt 2):1304–6. 9. Therrien J, Gatzoulis M, Graham T, Bink-Boelkens M, Connelly M, Niwa K, et al. Canadian cardiovascular society consensus conference 2001 update: recommendations for the manage- ment of adults with congenital heart disease. Part II. Can J Cardiol 2001;17:1029–50. 10. Therrien J, Warnes C, Daliento L, Hess J, Hoffmann A, Marelli A, et al. Canadian Cardiovascular Society Consensus Confer- ence 2001 update: recommendations for the management of adults with congenital heart disease. Part III. Can J Cardiol 2001;17:1135–58. 24. Foster E, Graham TP, Driscoll DJ, Reid GJ, Russell IA, Sermer M, et al. Task force 2: special health care needs of adults with congenital heart disease. J Am Coll Cardiol 2001;37: 1176–83. 25. Child JS, Collins-Nakai RL, Alpert JS, Deanfield JE, Harris L, McLaughlin P, et al. Task force 3: workforce description and educational requirements for the care of adults with congenital heart disease. J Am Coll Cardiol 2001;37:1183–7. 11. Reid GJ, Irvine MJ, McCrindle BW, Sananes R, Ritvo PG, Siu SC, et al. Prevalence and correlates of successful transfer from pediatric to adult health care among a cohort of young adults with complex congenital heart defects. Pediatrics 2004;113(3 Pt 1):e197–205. 26. Skorton DJ, Garson A, Allen HD, Fox JM, Truesdell SC, Webb GD, et al. Task force 5: adults with congenital heart disease: access to care. J Am Coll Cardiol 2001;37:1193–8. 12. Mackie AS, Lonescu-Lttu R, Therrien J, Pilote L, Abrahamow- icz M, Marelli AJ. Children and adults with congenital heart disease lost to follow-up. Circulation 2009;120:302–9. 27. Apple App Store. Hamilton. CanChildMcmaster; [17 January 2019]. MyTransition. Available from: https://apps.apple. com/us/app/mytransition-app/id1327036414. 13. Mackie AS, Rempel GR, Rankin KN, Nicholas D, Magill- Evans J. Risk factors to loss of follow-up among children and young adults with congenital heart disease. Cardiol Young 2012;22:307–15. 28. Google Play. Hamilton. Recommendations CanChildMcmaster; [17 January 2019]. MyTransition. Available from: https://play.google.com/store/ apps/details?id=com.iDeaWorks.myTransition&hl=en_IE. 14. Warnes CA. The adult with congenital heart disease: born to be bad? J Am Coll Cardiol 2005;46:1–8.f 29. Klassen AF, Grant C, Barr R, Brill H, Kraus de Camargo O, Ronen GM, et al. Development and validation of a generic scale for use in transition programmes to measure self-management skills in adolescents with chronic health conditions: the TRANSITION-Q. Child Care Health Dev 2015;41:547–58. 15. Perloff JK, Warnes CA. Challenges posed by adults with repaired congenital heart disease. Circulation 2001;103: 2637–43. 30. Webb GD, Williams RG. 32nd Bethesda conference: “care of the adult with congenital heart disease”. J Am Coll Cardiol 2001;37:1161–98. 16. Oechslin EN, Harrison DA, Connelly MS, Webb GD, Siu SC. Mode of death in adults with congenital heart disease. Am J Cardiol 2000;86:1111–6. 31. Warnes CA, William RG, Bashore TM, Childs JS, Connolly HM, Dearani JA, et al. ACC/AHA 2008 guidelines for the manage- ment of adults with congenital heart disease: a report of the American College of Cardiology/American Heart Associa- tion Task Force on Practice Guidelines (writing committee to develop guidelines on the management of adults with congenital heart disease). Circulation 2008;118:2395–451. 17. Harrison DA, Connelly M, Harris L, Luk C, Webb GD, McLaughlin PR. Sudden cardiac death in the adult with congenital heart disease. Can J Cardiol 1996;12:1161–3. 18. Webb GD, McLaughlin PR, Gow RM, Liu PP, Williams WG. Transposition complexes. Cardiol Clin 1993;11:651–64. 19. Oechslin EN, Harrison DA, Harris L, Downar E, Webb GD, Siu SS, et al. Reoperation in adults with repair of tetralogy of Fallot: indications and outcomes. J Thorac Cardiovasc Surg 1999;118:245–51. 32. Goossens E, Stephani I, Hilderson D, Gewilling M, Budts W, Van Deyk K, et al. Transfer of adolescents with congenital heart disease from pediatric cardiology to adult health care: an analysis of transfer destinations. J Am Coll Cardiol 2011;57: 2368–74. 20. Sable C, Foster E, Uzark K, Bjornsen K, Canobbio MM, Connolly HM, et al. Best practices in managing transition to adulthood form adolescents with congenital heart disease: the transi- tion process and medical and psychosocial issues. Circulation 2011;123:1454–85. 33. Mackie AS, Fournier A, Swan L, Marelli A, Kovacs A. Transition and transfer from pediatric to adult congenital heart disease care in Canada: call for strategic implementation. Can J Cardiol [Internet] 2019;35:1640–51. [cited 2019 Sep 4]; Available from: https://doi.org/10.1016/j.cjca.2019.08.014. 21. Knauth MA, Verstappen A, Reiss J, Webb GD. Supplementary Material: The online version of this article offers supplementary material (https://doi.org/10.1515/jtm-2019-0003). Recommendations Transition and transfer from pediatric to adult care of the young adult with complex congenital heart disease. Cardiol Clin 2006;24: 619–29. 22. Kovacs AH, Cullen-Dean G, Aiello S, Wasyliw C, Harrison JL, Li Q, et al. The Toronto congenital heart disease transition task force. Prog Pediatr Cardiol 2012;34:21–6.
https://openalex.org/W2065251709	https://www.ajol.info/index.php/bcse/article/download/80280/70540	English	null	Two Zn and Hg bromide salts based on 1-ethyl-3-methyl imidazolium ionic liquid: Ionothermal synthesis, structures and supramolecular organization	Bulletin of the Chemical Society of Ethiopia	2,012	cc-by	3,788	Bull. Chem. Soc. Ethiop. 2012, 26(3), 407-414. Printed in Ethiopia DOI: http://dx.doi.org/10.4314/bcse.v26i3.9 Bull. Chem. Soc. Ethiop. 2012, 26(3), 407-414. Printed in Ethiopia DOI: http://dx.doi.org/10.4314/bcse.v26i3.9 Bull. Chem. Soc. Ethiop. 2012, 26(3), 407-414. Printed in Ethiopia ISSN 1011-3924  2012 Chemical Society of Ethiopia DOI: http://dx.doi.org/10.4314/bcse.v26i3.9 KEY WORDS: Crystal structure, 1-Ethyl-3-methyl imidazolium bromide, Ionothermal synthesis, Supramolecular organization KEY WORDS: Crystal structure, 1-Ethyl-3-methyl imidazolium bromide, Ionothermal synthesis, Supramolecular organization Xiu-Cheng Zhang1* and Bing Liu2* Xiu-Cheng Zhang1* and Bing Liu2* 1College of Science, Northeast Forestry University, 150040 Harbin, Heilongjiang, P. R. China 2Department of Chemistry, University of Aveiro, CICECO, 3810-193 Aveiro, Portugal __________ *Corresponding author. E-mail: bliu_1203@yahoo.com.cn (Received October 14, 2011; revised July 16, 2012) ABSTRACT. Two Zn(II) and Hg(II) bromide salts, [EMI]2[ZnBr4] (1) and [EMI][HgBr3] (2), have been synthesized under ionothermal conditions using 1-ethyl-3-methyl imidazolium bromide ([EMI]Br) as solvents. 1 consists of tetrahedral anion [ZnBr4]2− and 2 consists of 1D double chain locating in the cavities surrounded by [EMI]+ cations. Both compounds exhibit 3D supramolecular architectures organized by the C-H···Br hydrogen bondings and alkyl-alkyl interactions. EXPERIMENTAL Materials and physical measurements Materials and physical measurements The reagents and solvents were used directly as supplied commercially without further purification except [EMI]Br. [EMI]Br was synthesized from the reaction of ethylbromide with 1-methylimidazole according to literature processes [22-23]. The IR spectra were recorded on a Nicolet Magna 750 FT-IR spectrometer with KBr pellet in the range 4000-400 cm-1. Elemental analyses of C, H and N were carried out on a Vario EL III elemental analyzer. The melting point of [EMI]Br was measured at X-4 microscopic melting point determinator. Synthesis of 1-ethyl-3-methylimidazolium bromide Synthesis of 1-ethyl-3-methylimidazolium bromide Under inert nitrogen atmosphere conditions, 100 mL degassed bromoethane (146.0 g, 1.34 mol) was added to 35.7 mL redistilled 1-methylimidazole (37.0 g, 0.45 mol) with constant stirring. The mixture was refluxed at 40 °C for 3 h and then cooled to room temperature. A pale yellow oil was produced. 100 mL ethyl acetate was added, and the product crashed out of solution. The product was washed with ethyl acetate, and dried under a vacuum at 25 °C for one day to give 1-ethyl-3-methylimidazolium bromide as a white solid (m.p. = 80-83 oC, yield: 85%). Synthesis of [EMI]2[ZnBr4] (1). 0.1 mmol, 29.6 mg Zn(NO3)2·6H2O and 0.1 mmol, 12.8 mg 3- amino-1,2,4-triazole-5-carboxylic aid were placed in a 25 mL Teflon-lined stainless-steel autoclave mixed with [EMI]Br (ca. 1.0 g). The mixture was kept in a furnace at 160 °C for 5 days, and then cooled at 2 ºC/h to room temperature to get colorless crystals of 1 suitable for X- ray diffraction. The crystals were manually selected under a microscope and used for all characterizations. Elemental analysis found (calcd) for C12H22Br4N4Zn: C, 23.69 (23.73); H, 3.73 (3.65); N, 9.11 (9.23). IR data (in KBr, cm-1) for 1: 3434(w), 3127(s), 2981(m), 1700(w), 1594(s), 1533(s), 1468(w), 1444(m), 1360(w), 1292(m), 1228(s), 1114(s), 1033(m), 950(m), 860(m), 752(m), 668(m), 624(w). Synthesis of [EMI][HgBr3] (2). The reaction process of 2 is similar to that of 1 using Hg(NO3)2·H2O (0.1 mmol, 34.2 mg) replacing Zn(NO3)2·6H2O. Elemental analysis found (calcd) for C6H11Br3HgN2: C, 12.89 (13.07); H, 2.11 (2.01); N, 5.11 (5.08). IR data (in KBr, cm-1) for 2: 3430(w), 3118(s), 2938(m), 1700(w), 1594(s), 1525(s), 1468(w), 1360(w), 1249(m), 1108(s), 963(m), 852(m), 738(m), 651(m), 624(w). Synthesis of [EMI][HgBr3] (2). The reaction process of 2 is similar to that of 1 using Hg(NO3)2·H2O (0.1 mmol, 34.2 mg) replacing Zn(NO3)2·6H2O. INTRODUCTION In 2002, ionic liquid [BMIM][BF4] (BMIM = 1-butyl-3-methylimidozolium) was used as reaction media for the first time in the synthesis of metal-organic frameworks [1]. Ionothermal synthesis is now flourishing in many fields, such as nanomaterials [2-4], framework materials [5-8], organic synthesis, electrochemistry and catalysis [9-11]. The principal feature of ionothermal synthesis is that the ionic liquids (ILs) act as both the “designed” green solvents and the structure-template providers, which have demonstrated their potential in novel material discovery. ILs consist of a bulky organic cation and an inorganic/organic anion [12]. Bulky organic cations usually are quaternary ammonium, phosphonium, pyridinium, and imidazolium [13-15]. Thus ILs are somewhat similar to polar solvents, suitable for dissolving the inorganic components required for the synthesis process [16-17]. ILs have many unique properties: low melting point (< 100 oC), extremely low volatility, chemical and thermal stability, nonflammability, high ionic conductivity, high heat capacity, high thermal conductivity, and wide electrochemical potential window [18-19]. For example, negligible vapor pressure will eliminate the safety concerns associated with high self-engendered pressures and lead to the integration with microwave synthesis [20]. Compared with hydro(solvo)thermal synthesis, no other solvents added to the reaction system as space fillers, no competition in ionothermal synthesis between template-framework and solvent-framework interactions will offer significantly different reaction environment [21]. Despite the difficulties in crystallization, we used 1-ethyl-3-methyl imidazolium bromide ([EMI]Br) as starting material reacting with Zn(NO3)2/Hg(NO3)2 to obtain two new d10 bromide salts, [EMI]2[ZnBr4] (1) and [EMI][HgBr3] (2). The structural features and supramolecular organizations in both bromide salts are described. 408 Xiu-Cheng Zhang and Bing Liu EXPERIMENTAL EXPERIMENTAL Elemental analysis found (calcd) for C6H11Br3HgN2: C, 12.89 (13.07); H, 2.11 (2.01); N, 5.11 (5.08). IR data (in KBr, cm-1) for 2: 3430(w), 3118(s), 2938(m), 1700(w), 1594(s), 1525(s), 1468(w), 1360(w), 1249(m), 1108(s), 963(m), 852(m), 738(m), 651(m), 624(w). Synthesis of [EMI][HgBr3] (2). The reaction process of 2 is similar to that of 1 using Hg(NO3)2·H2O (0.1 mmol, 34.2 mg) replacing Zn(NO3)2·6H2O. Elemental analysis found (calcd) for C6H11Br3HgN2: C, 12.89 (13.07); H, 2.11 (2.01); N, 5.11 (5.08). IR data (in KBr, cm-1) for 2: 3430(w), 3118(s), 2938(m), 1700(w), 1594(s), 1525(s), 1468(w), 1360(w), 1249(m), 1108(s), 963(m), 852(m), 738(m), 651(m), 624(w). Structural determination and refinement Symmetry codes: 1) A = -x+1, -y+3/2, z; B = -y+5/4, x+1/4, -z+1/4; D -y+1/4, x+1/4, -z+1/4; E = y-1/4, - x+1/4, -z+1/4; F -x, -y+1/2, z; 2) A = x+1/2, -y-3/2, -z-1; B = x-1/2, -y-3/2, -z-1. RESULTS AND DISCUSSION Synthesis The imidazolium Zn and Hg bromide salts 1 and 2 were synthesized under ionothermal conditions at 160 °C for 5 days under autogenous pressure. Unexpectedly, 3-amino-1,2,4- triazole-5-carboxylic acid did not participate in the final structures. Our other series experiments 1 Zn1-Br1 2.3731(5) Zn2-Br2 2.3740(5) Br1-Zn1-Br1A 111.16(3) Br2-Zn2-Br2E 110.45(1) Br1-Zn1-Br1B 108.63(1) Br2-Zn2-Br2F 107.52(3) 2 Hg1-Br2 2.8210(8) Hg1-Br1B 3.220(1) Hg1-Br1 2.9017(8) Hg1-Br3A 3.1396(9) Hg1-Br3 3.0077(9) Br1-Hg1-Br3 89.59(3) Br1-Hg1-Br1B 169.41(2) Br2-Hg1-Br3A 92.93(3) Br3-Hg1-Br3A 174.70(2) Br1-Hg1-Br3A 85.26(2) Br2-Hg1-Br1 91.77(3) Br3-Hg1-Br3A 174.70(2) Br2-Hg1-Br3 88.48(3) Hg1-Br3-Hg1B 83.79(3) Table 1. Crystal data and structure refinement parameters for 1 and 2. Table 1. Crystal data and structure refinement parameters for 1 and 2. Parameter 1 2 Empirical formula C12H22Br4N4Zn C6H11Br3HgN2 Color and habit Colorless block Colorless prism Crystal size (mm) 0.26×0.24×0.24 0.22×0.17×0.06 Crystal system Tetragonal Orthorhombic Space group I41/a P212121 a (Ǻ) 14.443(1) 8.207(2) b (Ǻ) 14.443(1) 9.824(2) c (Ǻ) 19.959(4) 15.542(4) V(Ǻ3) 4163.5(9) 1253.1(5) Z 8 4 Fw 607.35 551.49 Dcalcd (Mgm-3) 1.938 2.923 µ (mm-1) 8.856 21.807 F(000) 2336 984 Reflections measured 18530 11590 Independent reflections 2583 (Rint = 0.0487) 3074 (Rint = 0.0425) Observed reflection [I>2σ(I)] 1459 2566 Final R1 a, wR2 b [I>2σ(I)] 0.0589, 0.1530 0.0286, 0.0568 R1 a, wR2 b indices (all) 0.1093, 0.1817 0.0417, 0.0604 GOF on F2 1.037 1.059 (∆/σ)max/min 0.001, 0.000 0.018, 0.001 Largest difference peak (eǺ-3) 1.666, -0.793 0.769, -1.526 R1 a = (Σ\|\|Fo\| - \|Fc \|\| / Σ \|Fo\|). wR2 b = [Σ (w(Fo 2 - Fc 2)2) / Σ (w \|Fo 2\|2)]1/2. Table 2. Selected bond distances (Ǻ) and angles (°) for 1 and 2. Table 2. Selected bond distances (Ǻ) and angles ( ) for 1 and 2. 1 Zn1-Br1 2.3731(5) Zn2-Br2 2.3740(5) Br1-Zn1-Br1A 111.16(3) Br2-Zn2-Br2E 110.45(1) Br1-Zn1-Br1B 108.63(1) Br2-Zn2-Br2F 107.52(3) 2 Hg1-Br2 2.8210(8) Hg1-Br1B 3.220(1) Hg1-Br1 2.9017(8) Hg1-Br3A 3.1396(9) Hg1-Br3 3.0077(9) Br1-Hg1-Br3 89.59(3) Br1-Hg1-Br1B 169.41(2) Br2-Hg1-Br3A 92.93(3) Br3-Hg1-Br3A 174.70(2) Br1-Hg1-Br3A 85.26(2) Br2-Hg1-Br1 91.77(3) Br3-Hg1-Br3A 174.70(2) Br2-Hg1-Br3 88.48(3) Hg1-Br3-Hg1B 83.79(3) Symmetry codes: 1) A = -x+1, -y+3/2, z; B = -y+5/4, x+1/4, -z+1/4; D -y+1/4, x+1/4, -z+1/4; E = y-1/4, - x+1/4, -z+1/4; F -x, -y+1/2, z; 2) A = x+1/2, -y-3/2, -z-1; B = x-1/2, -y-3/2, -z-1. Bull. Chem. Soc. Ethiop. 2012, 26(3) Structural determination and refinement Data collections of compounds 1 and 2 were performed on Rigaku Mercury CCD diffractometer equipped with graphite-monochromated MoKα radiation (λ = 0.71073 Å). Intensity data were collected by the narrow frame method at 293 K and corrected for Lorentz and polarization effects as well as for absorption by the ω scan technique and were reduced using CrystalClear program [24]. The structures were solved by direct methods using SHELXTLTM package of crystallographic software and refined by full-matrix least-squares technique on F2 [25]. All non- hydrogen atoms were refined with anisotropic thermal parameters. Hydrogen atoms attached to C atoms were located at geometrically calculated positions and refined with isotropic thermal parameters included in the final stage of the refinement on calculated positions bonded to their carrier atoms. Because of bad crystal quality of 1, the not so good crystal diffraction data results in a large final wR2 (0.1530). A summary of the structural determinations and refinements for 1 and 2 is listed in Table 1. Selected bond distances and angles of 1 and 2 are shown in Table 2. Bull. Chem. Soc. Ethiop. 2012, 26(3) Zn and Hg bromide salts based on 1-ethyl-3-methyl imidazolium ionic liquid 409 Zn and Hg bromide salts based on 1-ethyl-3-methyl imidazolium ionic liquid 409 Table 1. Crystal data and structure refinement parameters for 1 and 2. Parameter 1 2 Empirical formula C12H22Br4N4Zn C6H11Br3HgN2 Color and habit Colorless block Colorless prism Crystal size (mm) 0.26×0.24×0.24 0.22×0.17×0.06 Crystal system Tetragonal Orthorhombic Space group I41/a P212121 a (Ǻ) 14.443(1) 8.207(2) b (Ǻ) 14.443(1) 9.824(2) c (Ǻ) 19.959(4) 15.542(4) V(Ǻ3) 4163.5(9) 1253.1(5) Z 8 4 Fw 607.35 551.49 Dcalcd (Mgm-3) 1.938 2.923 µ (mm-1) 8.856 21.807 F(000) 2336 984 Reflections measured 18530 11590 Independent reflections 2583 (Rint = 0.0487) 3074 (Rint = 0.0425) Observed reflection [I>2σ(I)] 1459 2566 Final R1 a, wR2 b [I>2σ(I)] 0.0589, 0.1530 0.0286, 0.0568 R1 a, wR2 b indices (all) 0.1093, 0.1817 0.0417, 0.0604 GOF on F2 1.037 1.059 (∆/σ)max/min 0.001, 0.000 0.018, 0.001 Largest difference peak (eǺ-3) 1.666, -0.793 0.769, -1.526 R1 a = (Σ\|\|Fo\| - \|Fc \|\| / Σ \|Fo\|). wR2 b = [Σ (w(Fo 2 - Fc 2)2) / Σ (w \|Fo 2\|2)]1/2. Table 2. Selected bond distances (Ǻ) and angles (°) for 1 and 2. Symmetry codes: 1) A = -x+1, -y+3/2, z; B = -y+5/4, x+1/4, -z+1/4; D -y+1/4, x+1/4, -z+1/4; E = y-1/4, - x+1/4, -z+1/4; F -x, -y+1/2, z; 2) A = x+1/2, -y-3/2, -z-1; B = x-1/2, -y-3/2, -z-1. Synthesis The imidazolium Zn and Hg bromide salts 1 and 2 were synthesized under ionothermal conditions at 160 °C for 5 days under autogenous pressure. Unexpectedly, 3-amino-1,2,4- triazole-5-carboxylic acid did not participate in the final structures. Our other series experiments Bull. Chem. Soc. Ethiop. 2012, 26(3) Xiu-Cheng Zhang and Bing Liu 410 proved that 3-amino-1,2,4-triazole-5-carboxylic acid would decompose beyond 130 °C. The contrast experiment of omitting 3-amino-1,2,4-triazole-5-carboxylic acid did not yield any crystals but white powders. In our estimation, possibly, the existence and/or decomposing of 3- amino-1,2,4-triazole-5-carboxylic acid can offer a suitable pH range, which affects the reaction and/or crystallization environments of the systems. This similar phenomena also appeared in the synthetic processes of imidazolium bromoplumbates [EMI]PbBr3 and [BMI]2PbBr4 (BMI = 1- buthyl-3-methylimidazolium cation) [26]. Structural description of [EMI]2[ZnBr4] (1) Structural description of [EMI]2[ZnBr4] (1) Compound 1 crystallizes in the tetragonal space group I41/a, whose asymmetric unit comprises of two distinct Zn(II) atoms of a quarter occupancy, two Br- anions and a [EMI]+ cation (Figure 1a). Each Zn(II) atom bonds to four symmetry-related Br- anions to form a negative [ZnBr4]2- tetrahedron with Zn1-Br1 = 2.3731(5), Zn2-Br2 = 2.3740(5) Ǻ and Br-Zn-Br angles in the ranges of 108.63(1)-111.16(3) and 107.52(3)-110.45(1)° respectively (Table 2). The positive charge on [EMI]+ cation is balanced by two [ZnBr4]2- units of a quarter occupancy. [EMI]+ cation exhibits weak C-H···Br hydrogen bondings (C12-H12A···Br1, C14-H14A···Br2) through the C atoms in the imidazolium ring, with two separated Br- anions (Figure 1b), and alkyl-alkyl interactions (methyl-methyl: C1-H1C···H3B-C3; C3-H3B···H1A-C1; methyl-aromatic alkyl: C1- H1A···H14A-C14; C2-H2A···H15A-C15) (Figure 1b, Table 3). Each [ZnBr4]2- unit is surrounded by four [EMI]+ cations. Through the C-H···Br and and C1-H1C···H3B-C3 interaction, a 2D supramolecular sheet is shaped, and further connected by the rest interactions into a 3D supramolecular architecture (Figure 2a). The aforementioned alkyl-alkyl interactions lead to the formation of the channels of 6.08×5.72 Ǻ2 with [EMI]+ as walls, which are fully occupied by [ZnBr4]2- tetrahedra (Figure 2b). Table 3. Hydrogen bondings and alkyl-alkyl interactions in 1 and 2. D-H H···A D···A <(DHA) Interactions 1 0.93 2.88 3.749(5) 156.5 C14-H14A···Br2 0.93 2.74 3.646(4) 164.1 C12-H12A···Br1 0.96 3.38 3.829(11) 110.6 C1-H1C···H3B-C3 0.96 3.05 3.829(11) 139.2 C3-H3B···H1A-C1 0.96 3.35 3.809(8) 111.3 C1-H1A···H14A-C14 0.97 2.96 3.838(9) 151.7 C2-H2A···H15A-C15 2 0.96 3.04 3.843(8) 141.7 C1-H1B···Br1 0.96 3.18 3.830(8) 126.2 C1-H1A···Br2 0.97 3.23 3.857(9) 123.6 C2-H2B···Br3 0.96 2.99 3.778(8) 140.5 C3-H3A···Br1 0.97 3.06 3.994(11) 163.4 C2-H2B···H1B-C1 0.96 3.34 3.917(12) 120.9 C3-H3B···H1B-C1 0.96 3.10 3.866(11) 138.2 C3-H3A···H14A-C14 0.96 3.11 3.665(12) 118.3 C3-H3A···H15A-C15 Table 3. Hydrogen bondings and alkyl-alkyl interactions in 1 and 2. Bull. Chem. Soc. Ethiop. 2012, 26(3) Zn and Hg bromide salts based on 1-ethyl-3-methyl imidazolium ionic liquid 411 Figure 1. (a) The structural motif of 1; (b) The C-H···Br hydrogen bondings and alkyl-alkyl interactions in 1. Figure 1. (a) The structural motif of 1; (b) The C-H···Br hydrogen bondings and alkyl-alkyl interactions in 1. Figure 2. (a) The 3D supramolecular architecture in 1; (b) the channels along the a-direction in the 3D supramolecular architecture. Figure 2. (a) The 3D supramolecular architecture in 1; (b) the channels along the a-direction in the 3D supramolecular architecture. Structural description of [EMI][HgBr3] (2) Structural description of [EMI][HgBr3] (2) The structural analysis indicates that compound 2 features a 1D inorganic chain anchoring in the channels made up of imidazolium cations. Compound 2 crystallizes in orthorhombic chiral space group P212121, whose structural flack factor is -0.0445 with 1291 Friedel pairs, indicating the accuracy of its absolute structure. In its asymmetry unit, there are one Hg(II), three distinct Br- anions and one [EMI]+ cation. Three positive charges on Hg(II) and [EMI]+ cations are balanced by three Br- anions. The Hg(II) center is five-coordinated by five bromine atoms in a slightly distorted HgBr5 square pyramid with Br1, Br3 and their symmetry-related Br1B and Br3A atoms (A = 0.5+x, -1.5-y, -1-z; B = x-0.5, -1.5-y, -1-z) shaping the basal plane and Br2 occupying the apex. The maximum deviation of Hg1 is 0.0939 Ǻ from the best least-square basal plane built by Br1, Br1B, Br3 and Br3A atoms. The HgBr5 square pyramid displays τ = 0.0882 [τ = (β-α)/60, where α and β are the two biggest bond angles around Hg(II) center; τ = 0 for ideal square pyramid; and τ = 1 for ideal trigonal bipyramid] [27]. All Hg-Br distances range in 2.8210(8)-3.220(1) Å, which are within the van der Waals contact limits (Table 2). Br1 and Bull. Chem. Soc. Ethiop. 2012, 26(3) Xiu-Cheng Zhang and Bing Liu 412 Br3 adopt µ2-mode linking neighboring Hg(II) centers into a double zigzag chain, in which a HgBr5 square pyramid shares two edges with neighboring two HgBr5 square pyramids (Figure 3a). Similarly, there exist C-H···Br hydrogen bondings (C1-H1B···Br1; C1-H1A···Br2; C2- H2B···Br3; C3-H3A···Br1) between alkyl and three distinct Br- anions (Figure 3a). Through alkyl-alkyl interactions (C2-H2B···H1B-C1; C3-H3B···H1B-C1; C3-H3A···H14A-C14; C3- H3A···H15A-C15) between neighboring [EMI]+ cations (Figure 3b, Table 3), the 3D supramolecular architecture is formed and further stabilized (Figure 4a). The 1D double chains are surrounded by [EMI]+ cations, thus to form channels of 5.90×5.75Ǻ2 with [EMI]+ cations as walls, which are fully occupied by the 1D chains (Figure 4b). Figure 3. (a) 1D [HgBr5] chain and its hydrogen bondings between EMI+ and Br‒ in 2; (b) The C-H···Br hydrogen bondings and alkyl-alkyl interactions in 2. Figure 3. (a) 1D [HgBr5] chain and its hydrogen bondings between EMI+ and Br‒ in 2; (b) The C-H···Br hydrogen bondings and alkyl-alkyl interactions in 2. Figure 4. (a) The 3D supramolecular architecture in 2; (b) the channels along the a-direction in the 3D supramolecular architecture. Figure 4. Structural description of [EMI][HgBr3] (2) (a) The 3D supramolecular architecture in 2; (b) the channels along the a-direction in the 3D supramolecular architecture. Compounds 1 and 2 were obtained from similar ionothermal processes and derived from the same [EMI]Br ionic liquid. Compound 1 is to some degree like the reported [bpyr][AlCl4] (bpyr = 1-butylpyridinium) [26] and Compound 2 has the same structure as [EMI]PbBr3 [26], but contains richer interactions. Observing this type of the compounds [26, 28, 29-32], we can find a phenomenon that [organic cation]X (X = halide or halide-containing anions) reacts with metal salts, specially the metal halides, X- tends to change into larger metal-containing polyatomic anion. This can be demonstrated in the compounds in this work and those reported in the literatures [26, 31, 32]. For example, [Im]Cl with FeCl3 can produce [Im][FeCl4] (Im = imidazolium) [32], [bpyr][AlCl4] with V2O5 to be [bpyr]4[V4O4Cl12] [28] and [EMI][AlCl4] with Bull. Chem. Soc. Ethiop. 2012, 26(3) 413 Zn and Hg bromide salts based on 1-ethyl-3-methyl imidazolium ionic liquid KZr6CCl15 to be [EMI]4[Zr6CCl18] [33]. The larger anions should be possible to promote the structural stability with the presence of the large organic cations. The metal-containing anions play a key role in the formations of the final 3D supramolecular architectures of 1, 2 and [EMI]PbBr3 [26]. The anions result in the difference in the hydrogen bondings between the inorganic blocks and [EMI]+ cations. Compound 1 has only two types of C-H···Br hydrogen bondings per [EMI]+, both through the alkyl group in the imidazolium ring, none from the alkyl group of side chains. While in 2, more C-H···Br hydrogen bondings exist around [EMI]+; contrarily, all four types of C-H···Br hydrogen bondings appear through the alkyl side chains, none from imidazolium ring. The alkyl-alkyl interactions in both compounds are very similar, showing little effect of anions on the [EMI]+ itself. More interactions also induce the difference in sizes of cavities. The cavities in 2 are supposed to be bigger than those in 1, due to the bigger inorganic anions. Actually, the cavities in 1 are a little larger than those in 2, which reasonably relates with the interactions. SUPPLEMENTARY Crystallographic data for the structure reported here has been deposited with the Cambridge Crystallographic Data Centre (Deposition No. CCDC: 824348-824349). That data can be obtained free of charge via http://www.ccdc.cam.ac.uk/perl/catreq.cgi (or from the CCDC, 12 Union Road, Cambridge CB2 1EZ, UK; fax: +44 1223 336033; e-mail: deposit@ccdc.cam.ac.uk). CONCLUSIONS Our attempt in the investigation of alkyl-imidazolium ILs produced two d10 dialkylimidazolium bromide salts, [EMI]2[ZnBr4] (1) and [EMI][HgBr3] (2), from ionothermal reactions with 1- ethyl-3-methyl imidazolium bromide as solvents. 1 consists of tetrahedral anion [ZnBr4]2− and 2 consists of 1D double chain, both locating in the cavities surrounded by [EMI]+ cations. Compounds 1 and 2 are 3D supramolecular architectures based on the connections of C-H···Br and alkyl-alkyl interactions. The structural analysis in C-H···Br and alkyl-alkyl interactions, and cavity sizes suggests the anions in 1 and 2 play an important role in their 3D supramolecular structures. ACKNOWLEDGEMENTS The authors gratefully acknowledge the financial support of the NSF of Heilongjiang Province (B2000714) and the Dissertation Funds for the Graduate Students of Northeast Forestry University (GRAM09). Bull. Chem. Soc. Ethiop. 2012, 26(3) REFERENCES 1. Jin, K.; Huang, X.; Pang, L.; Li, J.; Appel, A.; Wherland, S. Chem. Commun. 2002, 2872. 1. Jin, K.; Huang, X.; Pang, L.; Li, J.; Appel, A.; Wherland, S. Chem. Commun. 2002, 2872. 2. Antonietti, M.; Kuang, D.; Smarsly, B.; Zhou, Y. Angew. Chem., Int. Ed. 2004, 43, 4988. 2. Antonietti, M.; Kuang, D.; Smarsly, B.; Zhou, Y. Angew. Chem., Int. Ed. 2004, 43, 4988 3. Biswas, K.; Rao, C.N.R. Chem. Eur. J. 2007, 13, 6123. 3. Biswas, K.; Rao, C.N.R. Chem. Eur. J. 2007, 13, 6123. 4. Itoh, H.; Naka, K.; Chujo, Y. J. Am. Chem. Soc. 2004, 126, 3026. 4. Itoh, H.; Naka, K.; Chujo, Y. J. Am. Chem. Soc. 2004, 126, 3026. 5. Cooper, E.R.; Andrews, C.D.; Wheatley, P.S.; Webb, P.B.; Wormald, P.; Morris, R.E. Nature 2004, 30, 1012. 5. Cooper, E.R.; Andrews, C.D.; Wheatley, P.S.; Webb, P.B.; Wormald, P.; Morris, R.E. Natu 2004, 30, 1012. 6. Parnham, E.R.; Morris, R.E. Acc. Chem. Res. 2007, 40, 1005. 6. Parnham, E.R.; Morris, R.E. Acc. Chem. Res. 2007, 40, 1005. 7. Xu, L.; Yan, S.H.; Choi, E.Y.; Lee, J.Y.; Kwon. Y.U. Chem. Commun. 2009, 3431. 8. Xu, L.; Choi, E.Y.; Kwon. Y.U. Inorg. Chem. 2008, 47, 1907. 9. Ohno, H. Electrochemical Aspects of Ionic Liquids, Wiley-Interscience: New York; 2005. Ohno, H. Electrochemical Aspects of Ionic Liquids, Wi Bull. Chem. Soc. Ethiop. 2012, 26(3) 414 Xiu-Cheng Zhang and Bing Liu 10. Lancaster, M. Green Chemistry: An Introductory Text, The Royal Society of Chemistry: Cambridge, UK; 2002. g 11. Pârvulescu, V.I.; Hardacre, C. Chem. Rev. 2007, 107, 2615. 12. Wasserscheid, P.; Welton, T. Ionic Liquids in Synthesis, Wiley-VCH: Weinehim, Germany; 2003. 13. Visser, A.E.; Swatloski, R.P.; Reichert, W.M.; Griffin, S.Y.; Rogers, R.D. Ind, Eng. Chem. Res. 2000, 39, 3596. 14. Hussey, G. L. Pure Appl. Chem. 1998, 60, 1763. 15. Ito, Y.; Nohira, T. Electrochim. Acta. 2000, 45, 2611. 15. Ito, Y.; Nohira, T. Electrochim. Acta. 2000, 45, 261 16. Huddleston, J.G.; Visser, A.E.; Reichert, W.M.; Willauer, H.D.; Broker, G.A.; Rogers, R.D. Green Chem. 2001, 3, 156. 17. Welton, T. Coord. Chem. Rev. 2004, 248, 2459. 18. Seddon, K.R. J. Chem. Technol. Biotechnol. 1997, 68, 351. 19. Seddon, K.R. Chem. Eng. 2002, 730, 33. 20. Lin, Z.; Wragg, D.S.; Morris, R.E. Chem. Commun. 2006, 2021. 21. Cooper, E.R.; Andrews, C.D.; Wheatley, P.S.; Webb, P.B.; Wormald, P.; Morris, R.E. Stud. Surf. Sci. Catal., A 2005, 158, 247. 22. Bull. Chem. Soc. Ethiop. 2012, 26(3) REFERENCES Parnham, E.R.; Morris, R.E. Chem. Mater. 2006, 18, 4882. 23. Bonhote, P.; Dias, A.P.; Papageorgiou, N.; Kalyanasundaram, K.; Cratzel, M. Inorg. Chem. 1996, 35, 1168. 24. Rigaku, CrystalClear 1.3.6, Software User's Guide for the Rigaku. R-Axis, Mercury and Jupiter CCD Automated X-ray Imaging System, Rigaku Molecular Structure Corporation: Utah, USA; 2002. 25. Siemens, SHELXTLTM Version 5 Reference Manual, Siemens Engergy & Automation Inc.: Madison, Wisconsin, USA; 1994. 26. Thirumurugan, C.; Rao, C.N.R. Cryst. Growth Des. 2008, 8, 1640. 27. Addision, A.W.; Rao, T.N.; Reedijk, J.; Riju, J.V.; Verschoor, G.C. J. Chem. Soc., Dalton Trans. 1984, 1349. 28. Mahjoor, P.; Latturner, S.E.; Cryst. Growth Des. 2009, 9, 1385. 29. Ding, J.; Wu, J.; MacFarlane, D.R.; Price, W.E.; Wallace, G. Phys. Chem. Chem. Phys. 2008, 10, 5863. 30. Endres, F.; El Abedin, S.Z.; Saad, A.Y.; Moustafa, E.M.; Borissenko, N.; Price, W.E.; Wallace, G.G.; MacFarlane, D.R.; Newman, P.J.; Bund, A. Phys. Chem. Chem. Phys. 2008, 10, 2189. 31. Babai, A.; Mudring, A.V.Z. Anorg. Allg. Chem. 2006, 632, 1956. 32. Lin, I.J.B.; Vasam, C.S. J. Organomet. Chem. 2005, 690, 3498. 33. Sun, D.; Hughbanks, T. Inorg. Chem. 2000, 39, 1964. Bull. Chem. Soc. Ethiop. 2012, 26(3)
https://openalex.org/W3111171304	https://www.researchsquare.com/article/rs-126673/latest.pdf	English	null	Cardiometabolic risks of SARS-CoV-2 hospitalization using Mendelian Randomization	medRxiv (Cold Spring Harbor Laboratory)	2,021	cc-by	4,437	Cardiometabolic Risks of SARS-CoV-2 Hospitalization Using Mendelian Randomization Noah Lorincz-Comi Case Western Reserve University Xiaofeng Zhu (  xxz10@case.edu ) Case Western Reserve University Noah Lorincz-Comi Case Western Reserve University Xiaofeng Zhu (  xxz10@case.edu ) Case Western Reserve University Cardiometabolic risks of SARS-CoV-2 hospitalization using Mendelian Randomization Cardiometabolic risks of SARS-CoV-2 hospitalization using Mendelian Randomization Research Article Version of Record: A version of this preprint was published at Scienti¦c Reports on April 12th, 2021. See the published version at https://doi.org/10.1038/s41598-021-86757-3. Intro Many cardiometabolic conditions have demonstrated associative evidence with COVID-19 hospitalization risk. However, the observational designs of the studies in which these associations are observed preclude causal inferences of hospitalization risk. Mendelian Randomization (MR) is an alternative risk estimation method more robust to these limitations that allows for causal inferences. Results Although no significant causal effect evidence was observed, our data suggested a trend of increasing hospitalization risk for Type II diabetes (IMRP OR, 95% CI: 1.67, 0.96-2.92) and pulse pressure (OR, 95% CI: 1.27, 0.97-1.66) in the multi-ethnic sample. Methods & materials We applied four MR methods (MRMix, IMRP, IVW, MREgger) to publicly available GWAS summary statistics from European (COVID-19 GWAS n=2,956) and multi-ethnic populations (COVID-19 GWAS n=10,808) to better understand extant causal associations between Type II Diabetes (GWAS n=659,316), BMI (n=681,275), diastolic and systolic blood pressure, and pulse pressure (n=757,601 for each) and COVID-19 hospitalization risk across populations. Noah Lorincz-Comi1, Xiaofeng Zhu1 Noah Lorincz-Comi1, Xiaofeng Zhu1 Department of Population and Quantitative Health Sciences, Case Western Reserve University, Cleveland, OH, USA. Corresponding author Xiaofeng Zhu, PhD Department of Population and Quantitative Health Sciences, Case Western Reserve University Wolstein Research Building Room 1317, 2103 Cornell Rd., Cleveland, OH 44106. E-mail: xxz10@case.edu Corresponding author Xiaofeng Zhu, PhD Department of Population and Quantitative Health Sciences, Case Western Reserve University Wolstein Research Building Room 1317, 2103 Cornell Rd., Cleveland, OH 44106. E-mail: xxz10@case.edu Introduction SARS-CoV2-2, a novel coronavirus (COVID-19), has infected millions of individuals globally since it first emerged in December 2019. There is tremendous effort to learn more about why some individuals progress from a relatively stable illness to a more serious one requiring hospitalization. Observational studies, while providing some evidence for the conferral of hospitalisation risk by a number of risk factors, have sometimes produced conflicting results[1]. Equally, it would be inappropriate to infer from these studies alone causal associations between a given exposure and COVID-19 hospitalization. The explanations for this vary but may include unresolved bias, confounding, and reverse causation[2,3]. One alternative to observational studies is the randomised controlled trial (RCT) that can effectively eliminate confounding and reverse causation effects via the randomisation process. However, an RCT to identify causal risks for COVID-19 hospitalization would be expensive and unfeasible because it would be impossible to randomize patients based on a hypothesized risk factor in practice. Mendelian randomisation (MR) is an alternative to the RCT that retains the randomisation component but is both feasible and inexpensive[4]. MR capitalizes on the randomisation of an individual’s genetic information during meiosis. This means that the effects of each genetic variant on a given outcome should be less subject to the effects of residual confounding, such as reverse causation since genes are at conception fixed. This makes MR extremely powerful, which could explain its recent popularity in biomedical research. SARS-CoV2-2, a novel coronavirus (COVID-19), has infected millions of individuals globally since it first emerged in December 2019. There is tremendous effort to learn more about why some individuals progress from a relatively stable illness to a more serious one requiring hospitalization. Observational studies, while providing some evidence for the conferral of hospitalisation risk by a number of risk factors, have sometimes produced conflicting results[1]. Equally, it would be inappropriate to infer from these studies alone causal associations between a given exposure and COVID-19 hospitalization. The explanations for this vary but may include unresolved bias, confounding, and reverse causation[2,3]. One alternative to observational studies is the randomised controlled trial (RCT) that can effectively eliminate confounding and reverse causation effects via the randomisation process. However, an RCT to identify causal risks for COVID-19 Observational epidemiological studies have identified Type II Diabetes, hypertension, and body mass index (BMI) as potential risks for COVID-19 hospitalization (vs non- hospitalization). Conclusions Type II diabetes and Pulse pressure demonstrates a potential causal association with COVID- 19 hospitalization risk, the proper treatment of which may work to reduce the risk of a severe COVID-19 illness requiring hospitalization. However, GWAS of COVID-19 with large sample size is warranted to confirm the causality. Introduction Pooled estimated odds ratios are 2.75 (95% CI: 2.09-3.62) for Type II diabetes[5], 2.30 (1.76-3.00) for hypertension[6] (binary), and 2.67 (1.52-3.82) for BMI[7]. Not only did all these pooled estimates display heterogeneity in the meta-analysis from which they came, the observational designs of the included studies preclude making any causal inferences. We attempt here to further investigate some of these associations using MR, namely the associations between Type II Diabetes, body mass, diastolic and systolic blood pressure, and pulse pressure. We test for the causal relationships between each of these health conditions and COVID-19 severity, as indicated by hospitalization risk, using publicly available summary statistics from genome-wide association studies (GWAS). These exposures were selected for their hypothesized associations with COVID-19 hospitalization. Data COVID-19 summary statistics were retrieved from the COVID-19 Host Genetics Initiative[8] (covid19hg.rg/results) for 928 European cases and 2,028 controls and 2,430 cases and 8,378 controls of European, African, Hispanic, and Middle Eastern ethnicity. Cases were those hospitalised COVID-19 patients and controls were non-hospitalised COVID-19 patients. GWAS estimates, from three research groups[9-11] (UK Biobank, deCODE Genetics, FinnGen), of European individuals were meta-analyzed to yield the complete set of summary statistics for Europeans. GWAS estimates from 10 additional research groups[12-19] (BoSCO, SPGRX, GNH, PMBB, QGP, MVP, Ancestry, BQC19; full descriptions of these groups are found at the respective links in the References section) were then analysed with the original three European cohorts to form the expanded set of estimates from more ethnically diverse individuals. GWAS summary statistics from three European descent studies were meta-analysed using inverse variance-weighting approach to yield a single set of summary statistics for 12,029,423 and 9,503,351 SNPs for European and variable ethnicity populations, respectively. Only alleles with frequency >0.1% were included in the meta-analysis. Summary statistics for all five exposures, namely diastolic and systolic blood pressure (DBP, SBP), pulse pressure (PP), body mass index (BMI), and Type II diabetes (T2D), were available from public GWAS repositories. We used PLINK v1.90b6.11[20] to select independent GWAS significant SNPs from the available millions of SNPs for each exposure. These SNPs had in their GWAS a p-value less than 5e-8 and a linkage disequilibrium coefficient less than 0.1. For BMI (n=681,275)[21], DBP, SBP, PP (n=757,601 for each)[22], and Type II Diabetes (n=659,316)[23] 1,494, 1,202, 1,134, 926, and 174 SNPs meeting these criteria were identified, respectively. The GWAS data repositories for each exposure are listed in the respective referenced research. For each exposure, we merged the COVID-19 summary statistics to each exposure data set by matching genetic variant and effect alleles between the exposure and COVID-19 set of summary statistics. No SNPs in the original COVID-19 GWAS had between-study heterogenous effect estimates significant at a Bonferroni-adjusted p-value threshold. In the set of summary statistics from only European cohorts, 129, 1,186, 947, 884, 723 selected SNPs (using selection procedures described above) were included in the analyses for Type II Diabetes, BMI, DBP, SBP, and PP, respectively; in cohorts of individuals of variable ethnicity, 129, 1,188, 953, 887, and 726 SNPs were selected for Type II Diabetes, BMI, DBP, SBP, and PP, respectively. Data Selected SNP counts different between samples because the COVID-19 GWAS SNP counts differed between samples from the beginning. Mendelian Randomization Analysis To estimate the causal effects of each exposure on the probability of COVID-19 hospitalization status, we used Mendelian Randomization (MR). This method intends to eliminate potential confounding effects by capitalizing on the random assignment of genetic information at conception. The causal effect of each variant 𝑖 is generally estimated as θ̂i = Γ̂i/γ̂i where Γ̂𝑖 and γ̂i are the standardized effect estimates of SNP 𝑖 on the outcome and exposure, respectively. Standardized effect estimates are calculated as γ̂𝑖= β̂𝐸,𝑖/ (𝑠𝑒β̂𝐸,𝑖𝑛𝐸,𝑖 1/2) (1) (1) for SNP 𝑖 of the exposure (𝐸) and Results Table 1 presents summary statistics for the effect sizes of each exposure and for the correlations between GWAS effect estimate t-statistics for each exposure and COVID-19. These summary statistics indicate that, for each exposure in European only or multiple ethnic sample, the mean exposure effect sizes of only the selected SNPs are all <0.001. In the European-only sample, the correlations of effect sizes for the respective exposure and COVID-19 hospitalization [i.e., 𝜌̂(𝛤̂𝑖, 𝛾̂𝑖)] are all negative. In the multiple ethnicity sample, the correlations are more robust. These associations are reflected in Figure 1. Using the summary statistics from only European individuals, we did not observe significant causal effects of the exposures to COVID-19 hospitalization after adjusting for multiple tests, although a nominal evidence for BMI was observed in the IMRP analysis (OR=0.66, 95% CI: 0.46-0.96). No uniformly significant causal effect evidence was observed in the multiple ethnicity sample although all the exposures demonstrated increasing risk of COVID-19 hospitalization (Table 2). In the sample of individuals of multiple ethnicities, Type II Diabetes was positively associated with COVID-19 hospitalization using IMRP (1.672, 0.956-2.92). Positive associations are also detected by MRMix, IVW, and MREgger models, though with attenuated effect estimates and wider confidence intervals. Using IMRP, six of the 129 variants in the Type II Diabetes analyses displayed evidence of pleiotropic effects (P<0.05) which are shown in Figure 2, although none remain after Bonferroni correction. The intercept term in the MREgger model was non-significant (p=0.784), indicated an absence of pleiotropic effects or low statistical power for detecting them. A positive causal association was detected for pulse pressure in the multiple ethnicity sample. The IMRP model produced an estimated odds ratio of 1.27 (95% CI: 0.97-1.66). Positive associations are also detected by IVW (1.39, 1.07-1.81) and MREgger (3.05, 1.20-7.77), and marginally so for MRMix (1.36, 0.92-2.02). Thirty-two SNPs displayed evidence of pleiotropic effects (P<0.05) as indicated by IMRP (SNPs included in Figure 2, detailed in Supplementary Table S1), none of which remained after a Bonferroni adjustment (MREgger intercept p-value: 0.088). To better understand why for diastolic and systolic BP in the mixed ethnicity sample MRMix produced a causal effect estimate in the direction opposite of that produced by IMRP, IVW, and MREgger, we also plotted the causal effect estimate (𝜃̂) against its likelihood function (see Figure 3). For both variables a clear but flat peak is present, suggesting large standard error. for SNP 𝑖 of the exposure (𝐸) and for SNP 𝑖 of the exposure (𝐸) and Γ̂𝑖= β̂𝑌,𝑖[2𝑀𝐴𝐹𝑖(1 −𝑀𝐴𝐹𝑖)]1/2 (2) (2) for the outcome 𝑌 where 𝑀𝐴𝐹𝑖 is the minor allele frequency. The standard errors of the effect sizes are also adjusted from the original standard errors in each GWAS as: for the outcome 𝑌 where 𝑀𝐴𝐹𝑖 is the minor allele frequency. The standard errors of the effect sizes are also adjusted from the original standard errors in each GWAS as: 𝑠𝑒γ̂i = 𝑛𝐸,𝑖−1/2, 𝑠𝑒Γ̂𝑖= 𝑠𝑒β̂𝑌,𝑖[2𝑀𝐴𝐹𝑖(1 −𝑀𝐴𝐹𝑖)]1/2. (3,4) (3,4) Many methods for estimating the causal effect exist, of which we chose four that are: MRMix[24], IVW[25], MR-Egger[26] and IMRP[27]. Comparing the results of four similar but distinct analytical procedures provided a more complete picture of any existing causal relationships between each exposure and COVID-19 hospitalization. All analyses are more fully described elsewhere[28] and were done using the MRMix and IMRP packages in R[29]. Results A possible reason is the small sample size of COVID-19. Table 1: Summary statistics for exposure effect sizes Effect size Europeansa Multiple Ethnicitya Exposure Mean SD Min. Max. 𝝆̂ (exposure, COVID-19)b (SE) Mean SD Min. Max. 𝝆̂ (exposure, COVID-19)b (SE) Type II Diabetes <0.001 0.010 -0.021 0.020 -0.033 (0.089) <0.001 0.010 -0.021 0.020 0.153 (0.084) BMI <0.001 0.009 -0.022 0.043 -0.024 (0.029) <0.001 0.009 -0.022 0.043 0.013 (0.029) DBP <0.001 0.009 -0.025 0.023 -0.035 (0.033) <0.001 0.009 -0.025 0.023 0.026 (0.032) SBP <0.001 0.009 -0.026 0.023 -0.027 (0.034) <0.001 0.009 -0.026 0.023 0.036 (0.033) PP <0.001 0.009 -0.022 0.027 -0.021 (0.037) <0.001 0.009 -0.022 0.027 0.089 (0.037) a: Estimates from the European set are from 2,956 European individuals; estimates from the variable ethnicity set are from 10,808 individuals of either of European, African, Hispanic, or Middle Eastern descent. b: the correlation between the effect sizes for each SNP’s estimated effect on the probability of the respective exposure or COVID-19 hospitalization after all mentioned exclusions (SNP independence, p<5e-8, Pearson’s r<0.1 with index SNP) were applied. a: Estimates from the European set are from 2,956 European individuals; estimates from the variable ethnicity set are from 10,808 individuals of either of European, African, Hispanic, or Middle Eastern descent. b: the correlation between the effect sizes for each SNP’s estimated effect on the probability of the respective exposure or COVID-19 hospitalization after all mentioned exclusions (SNP independence, p<5e-8, Pearson’s r<0.1 with index SNP) were applied. Figure 1: Exposure, COVID-19 Effect Size Associations – Mixed Ethnicity Sample These figures display the effect size associations between each exposure (Type II Diabetes, BMI, diastolic and systolic blood pressure, and pressure) and COVID-19 hospitalization risk. Overlaid on the scatterplots are univariate linear regression fitted values and their associated 9 confidence intervals. Figure 1: Exposure, COVID-19 Effect Size Associations – Mixed Ethnicity Sample These figures display the effect size associations between each exposure (Type II Diabetes, BMI, diastolic and systolic blood pressure, and pulse pressure) and COVID-19 hospitalization risk. Overlaid on the scatterplots are univariate linear regression fitted values and their associated 95% confidence intervals. he effect size associations between each exposure (Type II Diabetes, BMI, diastolic and systolic blood pressure, and pulse 19 hospitalization risk. Results Overlaid on the scatterplots are univariate linear regression fitted values and their associated 95% Table 2: Causal effect estimates MRMix IMRP IVW MREgger Exposure/sample OR SElog P-value OR SElog P-value OR SElog P-value OR SElog P-value European only Type II Diabetes 1.000 0.000 1.000 0.554 0.514 0.250 0.848 0.537 0.758 3.153 1.947 0.556 BMI 1.916 2.150 0.762 0.662 0.188 0.028 0.894 0.183 0.541 0.507 0.621 0.282 Diastolic BP 0.779 0.253 0.322 1.036 0.205 0.863 0.814 0.200 0.304 0.751 0.681 0.675 Systolic BP 0.966 0.245 0.887 1.065 0.213 0.767 0.847 0.208 0.423 1.904 0.717 0.369 Pulse Pressure 0.923 0.278 0.773 0.863 0.240 0.539 0.881 0.234 0.588 1.845 0.829 0.460 Mixed ethnicity Type II Diabetes 1.105 19.500 0.996 1.672 0.285 0.071 1.487 0.278 0.155 1.148 0.985 0.889 BMI 1.323 0.199 0.159 1.082 0.108 0.469 1.117 0.108 0.308 1.742 0.362 0.125 Diastolic BP 0.803 0.676 0.745 1.028 0.119 0.820 1.047 0.119 0.701 1.430 0.402 0.374 Systolic BP 0.670 14.000 0.977 1.029 0.124 0.820 1.072 0.121 0.567 1.711 0.417 0.197 Pulse Pressure 1.363 0.201 0.124 1.265 0.138 0.087 1.393 0.135 0.014 3.047 0.478 0.020 Note: Odds ratios (OR) are exponentiated causal effect estimates. Standard errors (SE) correspond to the causal effect estimates themselves (not the ORs), which are in natural log scale. Figure 2: IMRP Pleitropy Evidence – Mixed Ethnicity Sample Displayed are those SNPs demonstrating evidence of pleiotropic effects for pulse pressure and Type II Diabetes in the multi-ethnicity sample as estimated by IMRP. The false discovery rate-adjusted threshold is determined by a Bonferroni-adjusted Type I Error rate of p=0.05. Figure 2: IMRP Pleitropy Evidence – Mixed Ethnicity Sample Displayed are those SNPs demonstrating evidence of pleiotropic effects for pulse pressure and Type II Diabetes in the multi-ethnicity sample as estimated by IMRP. The false discovery rate-adjusted threshold is determined by a Bonferroni-adjusted Type I Error rate of p=0.05. Displayed are those SNPs demonstrating evidence of pleiotropic effects for pulse pressure and Type II Diabetes in the multi-ethnicity sample as estimated by IMRP. The false discovery rate-adjusted threshold is determined by a Bonferroni-adjusted Type I Error rate of p=0.05. Figure 3: MRMix Estimation Performance – Mixed Ethnicity Sample These two plots display the estimated causal effect (theta, θ̂) during maximum likelihood estimation for diastolic and systolic blood pressure by MRMix in the multi-ethnicity sample. A clear, sharp peak indicates stable performance in the estimation of theta. Results The dotted red lines indicate the most likely estimates of theta (θ) reported in the Results section, respectively for each exposure. Figure 3: MRMix Estimation Performance – Mixed Ethnicity Sample These two plots display the estimated causal effect (theta, θ̂) during maximum likelihood estimation for diastolic and systolic blood pressure by MRMix in the multi-ethnicity sample. A clear, sharp peak indicates stable performance in the estimation of theta. The dotted red lines indicate the most likely estimates of theta (θ) reported in the Results section, respectively for each exposure. These two plots display the estimated causal effect (theta, θ̂) during maximum likelihood estimation for diastolic and systolic blood pressure by MRMix in the multi-ethnicity sample. A clear, sharp peak indicates stable performance in the estimation of theta. The dotted red lines indicate the most likely estimates of theta (θ) reported in the Results section, respectively for each exposure. Discussion Using summary statistics from European-only samples, we failed to detect any compelling evidence supporting any causal associations between hospitalization from COVID-19 and Type II Diabetes, body mass index, diastolic and systolic blood pressure, and pulse pressure. Conversely, we provide evidence of existing positive causal associations between Type II Diabetes and pulse pressure and the probability of hospitalization from COVID-19. A positive causal association between pulse pressure and COVID-19 hospitalization is biologically plausible as pulse pressure is moderately correlated with lung function (r=-0.37)[30], the failure of which is a primary reason for COVID-19 hospitalization. Similarly, although Type II diabetes is associated with a number of risks also hypothesized to be causally associated with COVID-19 severity (e.g., obesity, hypertension, cardiovascular disease), risk for hospitalization may also be conferred because of known inflammatory immune responses in patients with Type II Diabetes[31,32]. Nonetheless, our study is limited by many factors, not least of which is the relatively small sample sizes from which the COVID-19 hospitalization summary statistics were produced (n=2,956 for only Europeans, 10,808 for variable ethnicities). While we cannot for certain say that this alone can explain some of the null findings, using COVID-19 summary statistics from even larger GWAS samples would have allowed for more accurate estimates of any extant causal associations between each exposure and hospitalization from COVID-19. As more GWAS using COVID-19 patients are completed, more accurate summary statistics produced from larger samples will become publicly available and thus future work will work to reduce the probability of making Type II errors. Conversely, our study is strengthened by the very large sizes of the samples from which the exposure summary statistics were drawn. Each of these GWAS included >650,000 participants. Equally, our study is strengthened by the availability of GWAS summary statistics from different populations of individuals. This of course may allow us to generalize our findings to more people globally. Lastly, the comparison of results across multiple causal effect estimation methods should provide greater confidence in the findings for each exposure. We also benefit from a relatively new method (IMRP) which can detect SNPs displaying evidence of pleiotropic effects. Future work should build on this research to better understand COVID-19 disease burden within and across many populations so as to minimize national and global hospitalization risk. References [1] Emami, A., Javanmardi, F., Pirbonyeh, N., & Akbari, A. (2020). Prevalence of underlying diseases in hospitalized patients with COVID-19: a systematic review and meta- analysis. AAEM, 8(1). [2] Sheehan, N.A., Didelez, V., Burton, P.R., & Tobin, M.D. (2008). Mendelian randomisation and causal inference in observational epidemiology. PLoS Med., 5(8), e177. [3] Flegal, K.M., Graubard, B.I., Williamson, D.F., & Cooper, R.S. (2011). Reverse causation and illness-related weight loss in observational studies of body weight and mortality. Am J Epidemiol, 173(1), 1-9. [4] Smith, G. D. (2010). Mendelian randomization for strengthening causal inference in observational studies: application to gene× environment interactions. Perspectives on Psychol. Sci., 5(5), 527-545. [5] Kumar, A. et al. (2020). Is diabetes mellitus associated with mortality and severity of COVID-19? A meta-analysis. Diabetes & Metab. Syndr.: Clinical Research & Reviews. [6] Chen, Y., Gong, X., Wang, L., & Guo, J. (2020). Effects of hypertension, diabetes and coronary heart disease on COVID-19 diseases severity: a systematic review and meta- analysis. MedRxiv. [7] Yang, J., Hu, J., & Zhu, C. (2020). Obesity aggravates COVID‐19: a systematic review and meta‐analysis. J Med Virol. [8] The COVID-19 Host Genetics Initiative. The COVID-19 Host Genetics Initiative, a global initiative to elucidate the role of host genetic factors in susceptibility and severity of the SARS-CoV-2 virus pandemic. Eur. J. Hum. Genet., 28, 715–718 (2020). https://doi.org/10.1038/s41431-020-0636-6. [9] UK Biobank Limited (2020). COVID-19 hub. URL: https://www.ukbiobank.ac.uk/ [10] Sulem, P., Sveinbjornsson, G., & Stefansson, K. (2020). deCODE Genetics. URL: https://www.decode.com/ [11] Ganna, A. & Daly, M. (2020). FinnGen. URL: https://www.finngen.fi/en [12] BoSCO: Bonn Study of COVID19 Genetics. URL: https://bosco-studie.de/ [13] SPGRX: Determining the Molecular Pathways and Genetic Predisposition of the Acute Inflammatory Process Caused by SARS-CoV-2. GENYO-Center for Genomics and Oncological Research, URL: https://www.genyo.es/?lang=en [14] GNH: Genes & Health. Queen Mary University of London, URL: http://www.genesandhealth.org/ [15] PMBB: Penn Medicine Biobank (2020). Perelman School of Medicine, URL: https://www.itmat.upenn.edu/biobank/ [16] QGP: Qatar Genome Program (2020). Qatar Genome, URL: https://qatargenome.org.qa/node/5 [17] MVP: Million Veterans Program (2020). Department of Veterans Affairs, URL: https://www.mvp.va.gov/ [18] Ancestry (2020). Ancestry: https://www.ancestry.com/ [19] BQC19: Biobanque Quebec COVID19 (2020). URL: https://bqc19.ca/ [20] Chang, C.C. et al. (2015) Second-generation PLINK: rising to the challenge of larger and richer datasets. GigaScience, 4. URL: www.cog-genomics.org/plink/1.9/) [21] Yengo, L., et al. (2018). Meta-analysis of genome-wide association studies for height and body mass index in ∼700000 individuals of European ancestry. Hum. Mol. Genet., 27(20), 3641-3649. [22] Evangelou, E. et al. Disclosures There are no conflicts of interest. References (2018). Genetic analysis of over 1 million people identifies 535 new loci associated with blood pressure traits. Nat. Genet., 50(10), 1412-1425. [23] Xue, A., et al. (2018). Genome-wide association analyses identify 143 risk variants and putative regulatory mechanisms for type 2 diabetes. Nat. Commun., 9(1), 1-14. [24] Qi, G., & Chatterjee, N. (2019). Mendelian randomization analysis using mixture models for robust and efficient estimation of causal effects. Nat. Commun., 10(1), 1-10. [25] Burgess, S. & Bowden, J. (2015). Integrating summarized data from multiple genetic variants in Mendelian randomization: bias and coverage properties of inverse- variance weighted methods. arXiv eprint arXiv: 1512.04486 [26] Egger, M., Smith, G.D., Schneider, M., & Minder, C. (1997). Bias in meta-analysis detected by a simple, graphical test. BMJ, 315(7109), 629-634. [27] Zhu, X., Li, X., Xu, R., & Wang, T. (2020). An iterative approach to detect pleiotropy and perform Mendelian Randomization analysis using GWAS summary statistics, In Press, Bioinformatics. Software: https://github.com/xiaofengzhucase/IMRP [28] Zhu, X. (2020). Mendelian randomization and pleiotropy analysis. Quant. Biol., 1-11. [29] R Core Team (2020). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL: https://www.R- project.org/ [30] Jankowich, M.D., Taveira, T., & Wu, W.C. (2010). Decreased lung function is associated with increased arterial stiffness as measured by peripheral pulse pressure: data from NHANES III. Am. J. Hypertens., 23(6), 614-619. [31] Apicella, M. et al. (2020). COVID-19 in people with diabetes: understanding the reasons for worse outcomes. Lancet Diabetes Endocrinol.. [32] Guo, W. et al. (2020). Diabetes is a risk factor for the progression and prognosis of COVID‐19. Diabetes Metab. Res. Rev., e3319. Figure 2: IMRP Pleitropy Evidence – Mixed Ethnicity Sample Displayed are those SNPs demonstrating evidence of pleiotropic effects for pulse pressure and Type II Diabetes in the multi-ethnicity sample as estimated by IMRP. The false discovery rate-adjusted threshold is determined by a Bonferroni-adjusted Type I Error rate of p=0.05. Figure legends Figure 1: Exposure, COVID-19 Effect Size Associations – Mixed Ethnicity Sample These figures display the effect size associations between each exposure (Type II Diabetes, BMI, diastolic and systolic blood pressure, and pulse pressure) and COVID-19 hospitalization risk. Overlaid on the scatterplots are univariate linear regression fitted values and their associated 95% confidence intervals. y p p p ospitalization risk. Overlaid on the scatterplots are univariate linear regression fitted valu nd their associated 95% confidence intervals. Acknowledgements This work was supported by grant HG011052 (to X.Z.) from the National Human Genome Research Institute (NHGRI). N. L-C. was supported by grants T32 HL007567 from the National Heart, Lung and Blood Institute (NHLBI). Figure 3: MRMix Estimation Performance – Mixed Ethnicity Sample These two plots display the estimated causal effect (theta, θ̂) during maximum likelihood estimation for diastolic and systolic blood pressure by MRMix in the multi-ethnicity sample. A clear, sharp peak indicates stable performance in the estimation of theta. The dotted red lines indicate the most likely estimates of theta (θ) reported in the Results section, respectively for each exposure. Figures Figure 1 Exposure, COVID-19 Effect Size Associations – Mixed Ethnicity Sample These ¦gures display the effect size associations between each exposure (Type II Diabetes, BMI, diastolic and systolic blood pressure, and pulse pressure) and COVID-19 hospitalization risk. Overlaid on the scatterplots are univariate linear regression ¦tted values and their associated 95% con¦dence intervals. Figures Figure 1 E COVID 19 Eff Si A i i Mi d E h i i S l Th ¦ di l h ff Figure 1 Exposure, COVID-19 Effect Size Associations – Mixed Ethnicity Sample These ¦gures display the effect size associations between each exposure (Type II Diabetes, BMI, diastolic and systolic blood pressure, and pulse pressure) and COVID-19 hospitalization risk. Overlaid on the scatterplots are univariate linear regression ¦tted values and their associated 95% con¦dence intervals. Figure 2 IMRP Pleitropy Evidence – Mixed Ethnicity Sample Displayed are those SNPs demonstrating evidence of pleiotropic effects for pulse pressure and Type II Diabetes in the multi-ethnicity sample as estimated by IMRP. The false discovery rate-adjusted threshold is determined by a Bonferroni-adjusted Type I Error rate of p=0.05. Figure 2 Figure 3 MRMix Estimation Performance – Mixed Ethnicity Sample These two plots display the estimated causal effect (theta, θ ฀) during maximum likelihood estimation for diastolic and systolic blood pressure by MRMix in the multi-ethnicity sample. A clear, sharp peak indicates stable performance in the estimation of theta. The dotted red lines indicate the most likely estimates of theta (θ) reported in the Results section, respectively for each exposure. Figure 2 IMRP Pleitropy Evidence – Mixed Ethnicity Sample Displayed are those SNPs demonstrating evidence of pleiotropic effects for pulse pressure and Type II Diabetes in the multi-ethnicity sample as estimated by IMRP. The false discovery rate-adjusted threshold is determined by a Bonferroni-adjusted Type I Error rate of p=0.05. p Supplement.pdf Supplementary Files This is a list of supplementary ¦les associated with this preprint. Click to download. Supplement.pdf
https://openalex.org/W3180073716	https://ieeexplore.ieee.org/ielx7/7333/9363468/09477597.pdf	English	null	Measuring Phase-Amplitude Coupling Based on the Jensen-Shannon Divergence and Correlation Matrix	IEEE transactions on neural systems and rehabilitation engineering	2,021	cc-by	10,492	Measuring Phase-Amplitude Coupling Based on the Jensen-Shannon Divergence and Correlation Matrix Zhaohui Li , Xiaochen Bai, Rui Hu , and Xiaoli Li Abstract— Phase-amplitude coupling (PAC) measures the relationshipbetween the phase of low-frequency oscilla- tions (LFO) and the amplitude of high-frequencyoscillations (HFO). It plays an important functional role in neural infor- mation processing and cognition. Thus, we propose a novel method based on the Jensen-Shannon (JS) divergence and correlation matrix. The method takes the amplitude distribu- tions of the HFO located in the corresponding phase bins of the LFO as multichannel inputs to construct a correlation matrix, where the elements are calculated based on the JS divergence between pairs of amplitude distributions. Then, the omega complexity extracted from the correlation matrix is used to estimate the PAC strength. The simu- lation results demonstrate that the proposed method can effectively reﬂect the PAC strength and slightly vary with the data length. Moreover, it outperforms ﬁve frequently used algorithms in the performance against additive white Gaussian noise and spike noise and the ability of detecting PAC in wide frequency ranges. To validate our proposed method with real data, it was applied to analyze the local ﬁeld potential recorded from the dorsomedial striatum in a male Sprague-Dawley rat. It can replicate previous results obtained with other PAC metrics. In conclusion, these results suggest that our proposed method is a powerful tool for measuring the PAC between neural oscillations. ﬁrst observed in 1928 [7]. Other characteristic brain activities have also been successively investigated, including the delta (0.5-4 Hz), theta (4-8 Hz), beta (12-30 Hz), and gamma (above 30 Hz) frequency bands [8]–[10]. Moreover, it was found that the neural oscillations in different frequency bands interact mutually, which is called cross-frequency coupling (CFC) in the literature [11]–[13]. Several theoretical analyses and exper- imental ﬁndings have demonstrated that CFC is fundamental for neural communication and encoding, which bridges the gap between neural oscillations and brain functions [14]–[20]. Additionally, it has been proven that CFC is a pathologi- cal pattern in various conditions correlating with symptom severity [21]–[23]. Most previous studies focused on the following three types of CFCs: PAC [24]–[26], phase-phase coupling (PPC) [27]–[29] and amplitude-amplitude coupling (AAC) [30], [31]. Here, we intended to make contributions to measuring the PAC because of its signiﬁcant role in neural information processing [16], [32]–[34]. Generally, PAC reﬂects the degree of synchroniza- tion between the amplitude of HFO and the phase of LFO [26], [35]. 1375 1375 IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 29, 2021 This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ Measuring Phase-Amplitude Coupling Based on the Jensen-Shannon Divergence and Correlation Matrix It has been used to investigate many brain functions [16], [21], [22], [36]–[41]. For example, the PAC between low-frequency and high-frequency brain oscilla- tions subserved sensory information detection [40]; the PAC between oscillations in theta and gamma bands could help to read out information from long-term memory in the regions involved in working memory and sensory integration [16]; and the delta/theta phase and high gamma amplitude coupling play a crucial role in the control of visuospatial attention [41]. In addition, the variation in PAC under different pathological conditions has been utilized to explain the generation and evolution of many diseases, such as epilepsy [18], [24], [42], schizophrenia [43], multiple sclerosis [44], Parkinson’s dis- ease [22], [45], and Alzheimer’s disease [46]. Index Terms— Neural oscillation, phase-amplitude cou- pling, correlation matrix, Jensen-Shannondivergence, local ﬁeld potential. I. INTRODUCTION I. INTRODUCTION N N EURAL oscillations play a fundamental role in many brain functions, such as learning, memory, and percep- tion [1]–[6]. Alpha (8-12 Hz) rhythms in the human brain were Manuscript received November 19, 2020; revised April 8, 2021 and June 16, 2021; accepted July 1, 2021. Date of publication July 8, 2021; date of current version July 26, 2021. This work was supported by the National Natural Science Foundation of China under Grant 61971374, Grant 61827811, and Grant 61603327. (Corresponding author: Xiaoli Li.) ) Zhaohui Li is with the School of Information Science and Engi- neering, Yanshan University, Qinhuangdao 066004, China, and also with the Hebei Key Laboratory of Information Transmission and Signal Processing, Yanshan University, Qinhuangdao 066004, China (e-mail: lizhaohui@ysu.edu.cn). Several approaches have been introduced to quantify PAC; nevertheless, there is still room for improvement in robustness and accuracy. The traditional and frequently used methods to measure the PAC include PLV [47], MVL [35], KLmi [26], and GLM [48]. Recently, some new methods have also been developed in the literature, such as driven autoregressive (DAR) models [49], bispectrum [50], Gaussian-copula [51], mutual information [25], and adaptive decomposition [52]. Considering the popularity of these methods, in this study, y ) Xiaochen Bai and Rui Hu are with the School of Information Sci- ence and Engineering, Yanshan University, Qinhuangdao 066004, China (e-mail: baixiaochen@stumail.ysu.edu.cn; hurui@stumail.ysu.edu.cn). Xiaoli Li is with the State Key Laboratory of Cognitive Neuro- science and Learning, Beijing Normal University, Beijing 100875, China (e-mail: xiaoli@bnu.edu.cn). Xiaochen Bai and Rui Hu are with the School of Information Sci- ence and Engineering, Yanshan University, Qinhuangdao 066004, China (e-mail: baixiaochen@stumail.ysu.edu.cn; hurui@stumail.ysu.edu.cn). Xiaoli Li is with the State Key Laboratory of Cognitive Neuro- science and Learning, Beijing Normal University, Beijing 100875, China (e-mail: xiaoli@bnu.edu.cn). ( ) Digital Object Identiﬁer 10.1109/TNSRE.2021.3095510 er a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/b IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 29, 2021 1376 Fig. 2. Schematic diagram for the calculation procedure of the proposed JSCM method. (A) Two seconds of simulated PAC data. (B) The ﬁltered HFO (60-70 Hz, blue line) and LFO (4-8 Hz, green line). (C) The amplitude (blue line) of HFO and the phase (green line) of LFO. (D) The amplitude distribution (blue bar and green bar) for two equal intervals in each phase bin ([−π, π] is separated by dotted lines). I. INTRODUCTION study were performed by utilizing eegﬁlt.m in the EEGLAB toolbox [53]. Then, the Hilbert transform [54], [55] is used to Fig. 1. An example where the KLmi fails to detect the coupling but the JSCM does not fail. Fig. 1. An example where the KLmi fails to detect the coupling but the JSCM does not fail. the PLV, MVL, KLmi, GLM and DAR are used for com- parison to illustrate the superiority of our proposed method. Speciﬁcally, the idea behind KLmi is to discretize the phase of LFO into N bins and calculate the average amplitude of HFO in each phase bin to obtain an amplitude distribution, subsequently deﬁning the Kullback-Leibler distance between this amplitude distribution and the uniform distribution as KLmi [26]. However, in this way, the N segments are only represented by an average of the amplitudes. Thus, some waveform details of the time series falling into each phase bin are completely ignored. An example to illustrate this point is shown in Fig. 1. The average of the amplitudes in each bin is identical. Thus, the KLmi is zero, which means that it fails to detect the signiﬁcant PAC originally existing in the neural oscillations. Regardless, KLmi is of great importance in the ﬁeld of PAC quantiﬁcation. Drawing lessons from the idea of separating the phase into bins, in this study, we proposed a new method based on JS divergence and correlation matrix to measure the PAC and therefore named it JSCM. It can pick up some global waveform details of the amplitudes in each phase bin. Again, considering the amplitude envelope in Fig. 1, the JSCM can give an effective reﬂection of the PAC. Moreover, the performance of the approach was assessed by simulated data, and the validity was tested on the local ﬁeld potential (LFP) recorded from the rat dorsomedial striatum. We hope it will help to characterize the PAC in neural oscillations and explore the principle of brain functions. Fig. 2. Schematic diagram for the calculation procedure of the proposed JSCM method. (A) Two seconds of simulated PAC data. (B) The ﬁltered HFO (60-70 Hz, blue line) and LFO (4-8 Hz, green line). (C) The amplitude (blue line) of HFO and the phase (green line) of LFO. (D) The amplitude distribution (blue bar and green bar) for two equal intervals in each phase bin ([−π, π] is separated by dotted lines). I. INTRODUCTION (E) The constructed correlation matrix based on the JS divergence. Fig. 1. An example where the KLmi fails to detect the coupling but the JSCM does not fail. the PLV, MVL, KLmi, GLM and DAR are used for com- parison to illustrate the superiority of our proposed method. Speciﬁcally, the idea behind KLmi is to discretize the phase of LFO into N bins and calculate the average amplitude of HFO in each phase bin to obtain an amplitude distribution, subsequently deﬁning the Kullback-Leibler distance between this amplitude distribution and the uniform distribution as KLmi [26]. However, in this way, the N segments are only represented by an average of the amplitudes. Thus, some waveform details of the time series falling into each phase bin are completely ignored. An example to illustrate this point is shown in Fig. 1. The average of the amplitudes in each bin is identical. Thus, the KLmi is zero, which means that it fails to detect the signiﬁcant PAC originally existing in the neural oscillations. Regardless, KLmi is of great importance in the ﬁeld of PAC quantiﬁcation. Drawing lessons from the idea of separating the phase into bins, in this study, we proposed a new method based on JS divergence and correlation matrix to measure the PAC and therefore named it JSCM. It can pick up some global waveform details of the amplitudes in each phase bin. Again, considering the amplitude envelope in Fig. 1, the JSCM can give an effective reﬂection of the PAC. Moreover, the performance of the approach was assessed by simulated data, and the validity was tested on the local ﬁeld potential (LFP) recorded from the rat dorsomedial striatum. We hope it will help to characterize the PAC in neural oscillations and explore the principle of brain functions. II. MATERIALS AND METHODS A. Modulation Index Based on JS Divergence and Fig. 2. Schematic diagram for the calculation procedure of the proposed JSCM method. (A) Two seconds of simulated PAC data. (B) The ﬁltered HFO (60-70 Hz, blue line) and LFO (4-8 Hz, green line). (C) The amplitude (blue line) of HFO and the phase (green line) of LFO. (D) The amplitude distribution (blue bar and green bar) for two equal intervals in each phase bin ([−π, π] is separated by dotted lines). (E) The constructed correlation matrix based on the JS divergence. I. INTRODUCTION (E) The constructed correlation matrix based on the JS divergence. II. MATERIALS AND METHODS C. Mean Vector Length The mean vector length (MVL) modulation index is another widely used method that constructs a complex time series by multiplying the instantaneous phase of the slow oscillation by the instantaneous amplitude of the fast oscillation in a direct way to estimate the PAC [35]. Mathematically, MVL can be denoted by: MV L = 1 m m t=1 ateiϕt , (8) (8) where m is the total number of time points, at is the instan- taneous amplitude of the modulated signal, and ϕt is the instantaneous phase of the modulating signal at time point t. II. MATERIALS AND METHODS A. Modulation Index Based on JS Divergence and Correlation Matrix (6) (6) DJ S (P, Q)= 1 2 DK L P, P+Q 2 + 1 2 DK L Q, P+Q 2 , ( (1) Obviously, the JSCM ranges from 1 to N. A higher JSCM value indicates less synchronization between the amplitude distribution of different phase bins, which corresponds to a stronger PAC. Conversely, a lower JSCM value means greater similarity between these distributions, which implies weaker PAC strength. The matlab code is fully available: https://www. mathworks.com/matlabcentral/ﬁleexchange/95388-jscm. where P and Q denote the amplitude distribution in two phase bins and DK L indicates the KL distance between these two distributions, which is deﬁned as [57]: DK L (P, Q) = P log P Q . (2) (2) The JS divergence between P and Q ranges from 0 to 1. The JS divergence between P and Q ranges from 0 to 1. B. Phase Locking Value To assess these relative contributions, the eigenvalues are normalized to unit sum, i.e., where λn are the eigenvalues with λ1 ≤λ2 ≤· · · ≤ λN and un are the eigenvectors corresponding toλn. The eigenvalues have the following properties: 1) Because C is a real symmetric matrix, all eigenvalues are real numbers, and the sum of the eigenvalues equals the number of phase bins N. 2) If the amplitude distributions are fully uncorrelated, C becomes an identity matrix, and all the eigenvalues are one. 3) Once all the amplitude distributions are completely corre- lated, the maximum eigenvalue is equal to the number of phase bins N, and other eigenvalues fall to zero. Thus, eigenvalues can provide information about the correlation between the amplitude distributions [58]–[60]. Finally, omega complexity is employed to characterize the global correlation of multiple amplitude distributions. The computation of omega complexity is based on decomposition of the data into spatial principal components. The eigenvalues represent the contributions of the respective components to the total variance [61], [62]. To assess these relative contributions, the eigenvalues are normalized to unit sum, i.e., II. MATERIALS AND METHODS A. Modulation Index Based on JS Divergence and Correlation Matrix study were performed by utilizing eegﬁlt.m in the EEGLAB toolbox [53]. Then, the Hilbert transform [54], [55] is used to obtain the instantaneous phase of X L (t) and the instantaneous amplitude of X H (t), which are denoted as ϕL (t) and AH (t), respectively. The range [−π, π] is discretized into 18 equal phase bins. The amplitudes that fell into each phase bin are denoted by an amplitude sequence ⟨AH (t)⟩(n), n = 1, 2, . . . 18. In fact, the 18 amplitude sequences usually have different lengths. Additionally, the correlation matrix should be symmetric. Therefore, the JS divergence [56] is used to measure the similarity between these amplitude sequences. More concretely, according to the maximum and minimum value of all the amplitudes (denoted by Amax and Amin respectively), M intervals are equally separated in the range from Amin to Amax. M is set to 2 in this study, and the distinction between the results for different M is shown in the results section. Then, in each phase bin, the amplitude A. Modulation Index Based on JS Divergence and Correlation Matrix The main idea of the proposed method is to take the amplitude distributions of HFO corresponding to N phase bins of the LFO as multichannel signals and calculate the index of PAC by constructing a correlation matrix, where each correlation coefﬁcient is obtained by subtracting the JS divergence between pairs of amplitude distributions from 1. Speciﬁcally, N = 18 (i.e., 18 phase bins) is adopted in our method, which is consistent with previous studies [21], [23]. Additionally, the results for different numbers of bins are demonstrated in the results section. A diagrammatic sketch of the computing process is illustrated in Fig. 2, and more details of the algorithm are described in the following sections. First, it is necessary to ﬁlter the raw data X (t) to extract narrow-band oscillations X L (t) and X H (t). All ﬁlters in this LI et al.: MEASURING PAC 1377 Therefore, the novel modulation index of PAC based on the JS divergence and correlation matrix can be deﬁned as: distribution can be obtained by calculating the proportion of the number of amplitudes within each interval compared to the total number of amplitudes. The JS divergence is computed as: J SCM = exp − N n=1 λ′ n log λ′ n . B. Phase Locking Value DJ S = 0 means that the two distributions are identical, while The phase-locking value (PLV) indicates the degree to which the amplitude envelope of the higher components oscil- lates with the lower frequency components, which can be calculated from two series of phase angles. Speciﬁcally, it can be deﬁned as [47]: DJ S = 1 implies that they are completely different. Next, it is feasible to construct the correlation matrix C whose elements can be calculated based on the JS divergence, i.e., cij = 1 −DJ S [P (i) , P ( j)], i, j = 1, 2, 3, · · · N, (3) PLV = 1 v v t=1 ei(ϕp,t −ϕA,t) , (7) (7) where N is the number of phase bins. Obviously, C is a real symmetric matrix, and all diagonal elements are equal to 1 [58]. Then, the eigenvalue decomposition of C is given by where N is the number of phase bins. Obviously, C is a real symmetric matrix, and all diagonal elements are equal to 1 [58]. Then, the eigenvalue decomposition of C is given by where v is the total number of time points, ϕp,t is the phase of the low-frequency signal, and ϕA,t is the phase of the amplitude envelope of the high-frequency signal. Any consistent relationship between the two signals results in a PLV greater than 0. In particular, PLV = 1 means that there is perfect coupling. (4) Cun = λnun, (4) where λn are the eigenvalues with λ1 ≤λ2 ≤· · · ≤ λN and un are the eigenvectors corresponding toλn. The eigenvalues have the following properties: 1) Because C is a real symmetric matrix, all eigenvalues are real numbers, and the sum of the eigenvalues equals the number of phase bins N. 2) If the amplitude distributions are fully uncorrelated, C becomes an identity matrix, and all the eigenvalues are one. 3) Once all the amplitude distributions are completely corre- lated, the maximum eigenvalue is equal to the number of phase bins N, and other eigenvalues fall to zero. Thus, eigenvalues can provide information about the correlation between the amplitude distributions [58]–[60]. Finally, omega complexity is employed to characterize the global correlation of multiple amplitude distributions. The computation of omega complexity is based on decomposition of the data into spatial principal components. The eigenvalues represent the contributions of the respective components to the total variance [61], [62]. D. Kullback-Leibler Modulation Index λ′ n = λn N n=1 λn = λn tr (C), (5) λ′ n = λn N n=1 λn = λn tr (C), (5) The Kullback-Leibler modulation index (KLmi) is con- ceptually intuitive and assesses the PAC by characterizing the deviation of the amplitude distribution from the uniform distribution in a phase-amplitude plot [26]. The calculation procedures of the method are brieﬂy summarized as follows: it ﬁrst discretizes the phase of LFO into R bins and calculates the distribution for the average amplitudes of HFO in individual phase bins, subsequently quantifying the PAC by the KL distance between this amplitude distribution (denoted by S) (5) where tr (C) is the trace of the correlation matrix C. Here, tr (C) is equal to N since C is an N × N matrix and its diag- onal elements are 1. The entropy of the normalized eigenvalues is deﬁned as the logarithm of the omega complexity [61], [62], which is used to quantify the strength of coupling in this study. 378 IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 29, 2021 IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 29, 2021 1378 Fig. 3. Construction of the simulation data. (A) Superimposed waveforms in the high-frequency band and low-frequency band, indicated by blue line and green line, respectively. (B) Normalized HFO and LFO, indicated by blue line and green line, respectively. (C) The modulated HFO generated via the Von-Mises coupling method. (D) The simulation data constructed by combining the normalized LFO and the modulated HFO. and the uniform distribution (denoted by U). The amplitude distribution S is calculated by: and the uniform distribution (denoted by U). The amplitude distribution S is calculated by: S (r) = ⟨AH⟩(r) R r=1 ⟨AH⟩(r) , (9) (9) where ⟨AH⟩(r) is the mean amplitude in each phase bin. Then, the KLmi can be deﬁned as [26]: where ⟨AH⟩(r) is the mean amplitude in each phase bin. Then, the KLmi can be deﬁned as [26]: K Lmi = DK L (S, U) log (R) = R r=1 S (r) log S (r) log (R) + 1. (10) E. General Linear Model Fig. 3. Construction of the simulation data. (A) Superimposed waveforms in the high-frequency band and low-frequency band, indicated by blue line and green line, respectively. (B) Normalized HFO and LFO, indicated by blue line and green line, respectively. (C) The modulated HFO generated via the Von-Mises coupling method. (D) The simulation data constructed by combining the normalized LFO and the modulated HFO. The main idea of the general linear model (GLM) is that the amplitudes are speciﬁcally high at certain phases if the PAC is available, while the amplitudes are rather similar across all possible phases when the PAC does not exist [48]. Similar to other approaches, the phase is extracted from slower oscillations and the amplitude from faster oscillations. Phase and amplitude values are depicted in a scatter plot. In the ﬁrst case, a curve (called the spline model) that follows the amplitude pattern would best model the data. In the second case, the data would be best modeled by a horizontal line (called the null model), and the phase value would have no predictive power. hand, the frequencies of the HFO ranged from 60 to 70 Hz (gamma band) with a step of 1 Hz, and the amplitudes and phases were determined in the same manner as for LFO. Both oscillations were normalized to the range from −1 to 1, aiming to remove the ﬂuctuation of their amplitudes and retain the variation of their instantaneous frequencies [67]. Next, the amplitude envelope of the HFO was produced by the Von-Mises coupling method [68]: The more the spline model differs from the null model, the stronger the PAC in the data. A more detailed description and the main MATLAB code of the GLM method can be found in the original article by Kramer and Eden [48]. Ahigh (t) = c exp λ exp [λ × low (t)], (11) (11) where c controls the maximum amplitude of HFO and λ is a concentration parameter. A larger λ generates a larger amplitude of HFO around the preferred phase, while λ = 0 produces HFO with equal amplitudes at all phases. In this study, the parameters were set as c = 1 and λ = 1. Therefore, the raw data with PAC can be generated as follows: G. Acquisition of Simulation It Data First, by adding multiple sine waves with different fre- quencies, amplitudes, and phases [63], [64], LFO and HFO with a sampling frequency of 1000 Hz were generated and denoted by low (t) and high (t), respectively. The frequencies of the LFO ranged from 4 to 8 Hz (theta band) with a step of 1 Hz, the amplitudes of the components were inversely proportional to their frequencies [63], [65], [66], and the phases were randomly selected from [−π, π]. On the other F. Driven Autoregressive Model The driven autoregressive (DAR) model is a natural exten- sion of the well-known AR model. It utilizes nonlinear autoregressive models to depict the time-varying spectral content of the signals and subsequently estimates the mod- ulation of a signal (HFO) conditional on a driving signal (LFO) [49]. By evaluating a score of the “goodness of ﬁt” via the likelihood, the probabilistic DAR model enables easy model selection and parameter comparison. It can measure the PAC independently of the driving signal phase where the high-frequency component is the strongest. The DAR model builds the entire spectrum simultaneously; therefore, it can avoid the issues caused by incorrect ﬁltering or the Hilbert transform applied to wide-band signals (for further details see Tom Dupré la Tour et al., 2017). Raw = low (t) + Ahigh (t) × high (t). (12) (12) An illustration of the process for generating the simulation data is shown in Fig. 3. Then, to adjust the PAC strength, the raw data were mixed with an interferential signal (IFS), which was generated and normalized in the same manner as the simulated LFO. Thus, the raw data in Eq. (12) can be reconstructed as: Raw = k × low (t) + (1 −k) × I FS (t) + Ahigh (t) × high (t). (13) (13) where k indicates the PAC strength with a range from 0 to 1. Four examples of the simulation data with different coupling strengths are plotted in Fig. 4. The simulation data are utilized to evaluate the performance of the proposed JSCM method in the following sections. A. Selection of Hyperparameters for JSCM It should be noted that two hyperparameters must be deter- mined before using the JSCM, i.e., the number of phase bins (N) to construct the correlation matrix and the number of statistical intervals (M) to calculate the amplitude distribution. In fact, we have recommended appropriate values for these two parameters, i.e., N = 18 and M = 2. In this section, we inves- tigate the effect of different N and M on the performance of JSCM. H. Acquisition of Experimental Data The experimental dataset was collected from a male Sprague-Dawley rat after behavioral training. The data were LI et al.: MEASURING PAC 1379 Fig. 4. Four examples of the simulation data with different coupling strengths. (A) k = 0.1. (B) k = 0.4. (C) k = 0.7. (D) k = 1. Fig. 6. Effect of the number of intervals on the performance of the JSCM method. There were 100 realizations for each coupling strength. Fig. 6. Effect of the number of intervals on the performance of the JSCM method. There were 100 realizations for each coupling strength. Fig. 4. Four examples of the simulation data with different coupling strengths. (A) k = 0.1. (B) k = 0.4. (C) k = 0.7. (D) k = 1. On the other hand, to measure the similarity of the ampli- tude series falling into the phase bins, we use several equal intervals to obtain the amplitude distribution and subsequently calculate the JS divergence between the distributions. Then, it is necessary to investigate the effect of the number of inter- vals on the algorithm’s output. As shown in Fig. 6, the JSCM values for different numbers of intervals vary consistently with each other. The distinction can be explained as follows: the more intervals there are, the smaller the values in the correlation matrix, which leads to a larger JSCM value. Thus, we can choose any number of intervals. However, considering the robustness to resist noise, a small number of intervals is appropriate, such as 2. Fig. 5. Effect of the number of phase bins on the performance of the JSCM method. There were 100 realizations for each coupling strength. Fig. 5. Effect of the number of phase bins on the performance of the JSCM method. There were 100 realizations for each coupling strength. III. RESULTS III. RESULTS B. Extraction of PAC From Correlation Matrix acquired from Charles River Laboratories, France. More details of the experiments can be found in [77]. The elec- trode for the LFP recording was located in the dorsomedial striatum. The raw LFP was ampliﬁed 100 times, ﬁltered by 0.3 Hz-1 kHz bandpass ﬁlter and digitalized at 10 kHz. The recording time was 1800 s, and the data were downsampled to 350 Hz in this study. In addition to hyperparameter selection, another issue should be considered, i.e., how to extract a PAC index from the correlation matrix. In our proposed JSCM method, omega complexity is adopted to quantify the PAC. In fact, other methods can also be used to characterize the correlation matrix, such as the S-estimator [69]. Here, we evaluate the robustness of these two methods by adding white Gaussian noise into the simulated data. The signal-to-noise ratio (SNR) ranged from −10 dB to 10 dB with a step of 1 dB. The data length was 10 s, and the PAC strength was set to 0.3, 0.5, and 0.8. We found that the PAC index based on the omega complexity is more robust to noise than the S-estimator, as shown in Fig. 7. Therefore, it is appropriate to use the omega complexity to deﬁne the PAC index in the JSCM method. IV. DISCUSSION Neural oscillations reﬂect rhythmic ﬂuctuations in the syn- chronization of neuronal populations and play a critical role in neural processing [70]. The coupling between neural oscil- lations with different frequencies, i.e., CFC lays a foundation for studies of brain functions [15], [24], [71]. PAC is one of the commonly used CFC measures. Considering that there is still room for improvement of robustness and accuracy, we proposed a novel method based on the JS divergence and correlation matrix (JSCM). The simulation results demonstrate that it can effectively measure the PAC strength and slightly vary with the data length. Although two hyperparameters should be determined before using this method, they do not inﬂuence the performance signiﬁcantly, and we have provided recommended values for them. In fact, there are many tradi- tional and frequently used approaches to estimate the PAC, and we have introduced some typical approaches in the methods section, including the PLV, KLmi, MVL, GLM, and DAR. Here, by using the simulated data, we compared our method with these methods to indicate its superiority and limitation. In addition, to further illustrate its usability in real data, we applied the JSCM method to the LFP recorded from a male Sprague-Dawley rat. Fig. 7. Effect of white Gaussian noise on the PAC index based on the Omega-complexity and S-estimator. The coupling strength was set to (A) 0.3, (B) 0.5, and (C) 0.8. There were 100 realizations for each strength. Fig. 8. Effect of data length on the JSCM method at different coupling strengths (100 realizations for each strength). Fig. 8. Effect of data length on the JSCM method at different coupling strengths (100 realizations for each strength). almost no inﬂuence on the performance of the JSCM method. Thus, we can choose an appropriate data length according to the speciﬁc application. This property is of crucial importance for making comparisons between experimental recordings with different lengths. It is also found that the JSCM values do not equal the simulated coupling strengths, which is caused by the principle of JSCM calculation. However, the method can still differentiate these different couplings, which is further illustrated in the next subsection. almost no inﬂuence on the performance of the JSCM method. Thus, we can choose an appropriate data length according to the speciﬁc application. This property is of crucial importance for making comparisons between experimental recordings with different lengths. A. Robustness of JSCM Against Noise A. Robustness of JSCM Against Noise Inevitably, there is some noise mixed in the experimental recordings. Then, the robustness against noise is essential for evaluating the performance of PAC measures. In the comparison of these methods, the robustness was evaluated by adding white Gaussian noise into the simulated data. The signal-to-noise ratio (SNR) ranged from −10 dB to 10 dB with a step of 1dB. Additionally, the data length was 10 s, and the PAC strength was set to 0.3, 0.5, and 0.8. Then, the PAC IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 29, 2021 IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 29, 2021 1380 Fig. 7. Effect of white Gaussian noise on the PAC index based on the Omega-complexity and S-estimator. The coupling strength was set to (A) 0.3, (B) 0.5, and (C) 0.8. There were 100 realizations for each strength. Fig. 9. Variation of the JSCM with coupling strength (100 realizations). Fig. 9. Variation of the JSCM with coupling strength (100 realizations). Given a data length of 10 s, Fig. 9 plots the PAC measured by the JSCM approach while the coupling strength varies from 0 to 1 by an increased step of 0.05. Obviously, the JSCM estimation increases monotonically with increasing coupling strength. In particular, when the coupling strength is relatively small (k < 0.2) or large (k > 0.8), the JSCM estimation increases slower. The reason is that the output waveforms of the simulation model present similar PAC patterns in either weak or strong couplings. In contrast, when the coupling strength takes median values, the output waveforms exhibit PAC patterns with obvious differences. IV. DISCUSSION It is also found that the JSCM values do not equal the simulated coupling strengths, which is caused by the principle of JSCM calculation. However, the method can still differentiate these different couplings, which is further illustrated in the next subsection. C. Variation of JSCM With Data Length The PAC estimation in this study depends on the statistical calculation of the amplitude distribution; therefore, it is nec- essary to assess the inﬂuence of data length on the proposed JSCM algorithm. Given the duration of the simulation data ranging from 5 to 30s (5,000 to 30,000 sample points) with a step of 1 s, Fig. 8 plots the PAC index estimated by the JSCM at coupling strengths of 0.3, 0.5, and 0.8. Obviously, the impact of the data length on the algorithm’s output is extremely small for the selected coupling strengths. In other words, when the PAC strength is constant in a period, the data length exerts On the one hand, given the coupling strength ranging from 0 to 1 with a step of 0.05 and a data length of 10 s, the results for 9, 12, 15, 18, 24, and 30 phase bins are shown in Fig. 5. Clearly, the inﬂuence of phase bins on the performance of the algorithm is very small. That is, when using a number of bins between 9 and 30, the PAC estimation is stable. We suggest that the choice of the default 18 bins is a safe option and is also in line with previous works [24], [26]. Speciﬁcally, the phase of −π to π is discretized into 18 intervals of π/9. D. Variation of JSCM With Coupling Strength Phase-amplitude comodulograms of the six methods when white Gaussian noise is mixed in the simulated data (100 realizations). The x-axis represents the frequency of the modulation phase, the y-axis represents the frequency of the modulated amplitude. The center fre- quencies (LFO: 3-10 Hz with a step of 0.5 Hz; HFO: 30-100 Hz with a step of 2 Hz) of the bandpass ﬁlter correspond to the coordinates in each comodulogram. There is no high-frequency ﬁltering used in panel E. (A) PLV. (B) MVL. (C) KLmi. (D) GLM. (E) DAR. (F) JSCM. Fig. 11. Phase-amplitude comodulograms of the six methods when the JSCM is more localized at the involved frequency bands (i.e., 4-8 Hz and 60-70 Hz) with a relatively high noise level. Speciﬁcally, the GLM has some spurious results below 4 Hz. In other words, the JSCM is a more reliable measure when the recordings are mixed with large additive noise. B. Detection of Frequency Pairs With PAC In the analysis of experimental data, we usually need to determine the pairs of frequency bands that generate the PAC. Therefore, the ability of detecting the frequency band pairs presenting the PAC is of signiﬁcance to assess the PAC measures. The phase-amplitude comodulogram is an effective tool to investigate the PAC distribution in broad frequency ranges. Speciﬁcally, the comodulogram was obtained by scan- ning frequency band pairs and estimating the PAC on each of them, which could report the PAC among multiple bands simultaneously [26]. In this subsection, the overall frequency band of the simulation data was expanded to 1-100 Hz. However, PAC existed only between the frequency bands of 4-8 Hz and 60-70 Hz. The PAC strength was set to 0.8, and the data length was 10 s. Moreover, white Gaussian noise with SNR = 5 was added to the simulation data. In the calculation of PAC, there were 15 frequency bands with a bandwidth of 4 Hz uniformly distributed in the 3-10 Hz range, which was used as the scanning LFO. Additionally, 36 frequency bands with a bandwidth of 10 Hz were uniformly distributed in the 30-100 Hz range, which was used as the scanning HFO. The comodulograms of the ﬁve previously published PAC measures and our improved method are plotted in Fig. 11. Clearly, compared with the other ﬁve methods, D. Variation of JSCM With Coupling Strength Generally, the PAC measure is used to observe the dis- tinction under different conditions; thus, there is a need to examine the variation of the JSCM with coupling strength. LI et al.: MEASURING PAC 1381 Fig. 10. Effect of white Gaussian noise on the simulated data for different coupling strengths (100 realizations for each strength). (A) k = 0.3. (B) k = 0.5. (C) k = 0.8. estimation at the different noise intensities was calculated. Moreover, to facilitate the comparison across different meth- ods, the results of noisy data for each method were divided by its PAC estimation of the clean data. As demonstrated in Fig. 11. Phase-amplitude comodulograms of the six methods when white Gaussian noise is mixed in the simulated data (100 realizations). The x-axis represents the frequency of the modulation phase, the y-axis represents the frequency of the modulated amplitude. The center fre- quencies (LFO: 3-10 Hz with a step of 0 5 Hz; HFO: 30-100 Hz with Fig. 10. Effect of white Gaussian noise on the simulated data for different coupling strengths (100 realizations for each strength). (A) k = 0.3. (B) k = 0.5. (C) k = 0.8. Fig. 11. Phase-amplitude comodulograms of the six methods when white Gaussian noise is mixed in the simulated data (100 realizations). The x-axis represents the frequency of the modulation phase, the y-axis represents the frequency of the modulated amplitude. The center fre- quencies (LFO: 3-10 Hz with a step of 0.5 Hz; HFO: 30-100 Hz with a step of 2 Hz) of the bandpass ﬁlter correspond to the coordinates in each comodulogram. There is no high-frequency ﬁltering used in panel E. (A) PLV. (B) MVL. (C) KLmi. (D) GLM. (E) DAR. (F) JSCM. Fig. 10. Effect of white Gaussian noise on the simulated data for different coupling strengths (100 realizations for each strength). (A) k = 0.3. (B) k = 0.5. (C) k = 0.8. estimation at the different noise intensities was calculated. Moreover, to facilitate the comparison across different meth- ods, the results of noisy data for each method were divided by its PAC estimation of the clean data. As demonstrated in Fig. 10, the PAC index estimated by these methods decreases with the amplitude of added white Gaussian noise. More importantly, the novel JSCM method is superior to the others at characterizing the PAC in simulated data with noise, especially for low coupling strengths. Fig. 11. C. Sensibility to Spurious it Coupling In the previous simulation analysis, we only considered additive white Gaussian noise. In fact, the noise content may be much more complex and stronger in real EEG recordings, such as the sharp waveform of spikes and large artifacts [51], [72]–[75]. Speciﬁcally, movement arti- facts are inevitable and unexpected during prolonged record- ings (∼24 hours) for closed loop applications [23], [76]. Sharp-edged transients in the time domain correspond to broadband harmonic components in the frequency domain. They inﬂuence both the phase of low-frequency compo- nents and the amplitude of high-frequency components. Thus, it is possible to generate spurious CFC (includ- ing PAC) even though a true physiological interaction is absent [72], [74], [75]. The methods based on the instanta- neous estimation of the phase and the amplitude are poten- tially affected by spurious PAC, such as MVL and KLmi. Recently, some new measures have been proposed without using phase and amplitude estimation, including DAR, bis- pectral approaches [50], and adaptive decompositions [52]. IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 29, 2021 IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 29, 2021 1382 Fig. 12. An example of the simulated signal with spike noise in time domain. (A) Three randomly distributed spikes. (B) Simulated signal without spikes. (C) Simulated signal with spikes superimposed on the background activity. Fig. 13. Phase-amplitude comodulograms of the six methods when a Fig. 12. An example of the simulated signal with spike noise in time domain. (A) Three randomly distributed spikes. (B) Simulated signal without spikes. (C) Simulated signal with spikes superimposed on the background activity. Fig. 13. Phase-amplitude comodulograms of the six methods when a train of spikes are superimposed on the background signal (100 real- izations). The x-axis represents the frequency for the modulation phase and the y-axis represents the frequency for the modulated amplitude. The center frequencies (LFO: 3-10 Hz with a step of 0.5 Hz; HFO: 30-100 Hz with a step of 2 Hz) of the bandpass ﬁlter corresponding to the coordinates in each comodulogram. There is no high- frequency ﬁltering used in panel E. (A) PLV. (B) MVL. (C) KLmi. (D) GLM. (E) DAR. (F) JSCM. Fig. 12. An example of the simulated signal with spike noise in time domain (A) Three randomly distributed spikes (B) Simulated signal Fig. 12. An example of the simulated signal with spike noise in time domain. (A) Three randomly distributed spikes. C. Sensibility to Spurious it Coupling Some issues have also been pointed out in the application of PAC algorithms, mainly focusing on the spurious PAC [37], [72], [75]. More importantly, Aru et al. listed 9 practical recommendations to minimize technical pitfalls and overinterpretation of PAC measures and ensure that the detected coupling is truly contained in the signals [72]. Based on these concerns, we chose a dataset recorded from the dor- somedial striatum for the validation of our JSCM method [77]. Moreover, this dataset was used for PAC analysis in a previous study, where the DAR model was proposed [49]. Considering that the data length was very long (1800 s), in this study, we used a sliding time window (30 s with an overlap of 20 s) to calculate the PAC. Additionally, the ﬁrst 20 s (70 k points) of the signal were removed because of the large amplitude artifacts. It has been reported that to observe any amplitude modulation, the bandwidth of the HFO has to be at least twice the highest frequency of the LFO [37], [49], [78]. Thus, in the PLV, MVL, KLmi, GLM, and JSCM methods, high frequencies were extracted with a bandwidth of 20 Hz. In DAR models, the hyperparameters used are the cross-validated values, which were described in [49]. For all methods, the low frequencies were extracted with the optimal bandwidth of 3.2 Hz [49]. Between the comodulograms in the windows obtained by a speciﬁc method, no signiﬁcant difference was observed. Therefore, only the average comodulogram of the C. Sensibility to Spurious it Coupling (B) Simulated signal without spikes. (C) Simulated signal with spikes superimposed on the background activity. Fig. 13. Phase-amplitude comodulograms of the six methods when a train of spikes are superimposed on the background signal (100 real- izations). The x-axis represents the frequency for the modulation phase and the y-axis represents the frequency for the modulated amplitude. The center frequencies (LFO: 3-10 Hz with a step of 0.5 Hz; HFO: 30-100 Hz with a step of 2 Hz) of the bandpass ﬁlter corresponding to the coordinates in each comodulogram. There is no high- frequency ﬁltering used in panel E. (A) PLV. (B) MVL. (C) KLmi. (D) GLM. (E) DAR. (F) JSCM. In our proposed JSCM method, we used the Hilbert transform to extract the phase and amplitude. Therefore, inevitably, its performance is affected by spurious PAC. Here, considering that the detection of spurious coupling with no interactive oscillations within the signal has been investigated [49], we evaluated the inﬂuence of spikes on the six methods in the case of strong coupling that occurs between the HFO of 60-70 Hz and the LFO of 4-8 Hz. Speciﬁcally, the simulated coupling strength was set as 0.8, the data length was 10 s, and three randomly distributed spikes with a 10 ms duration and a 5-10 standard deviation height (selected randomly for each realization) were superimposed on the background signal. An example of the simulated signal with spike noise is plotted in Fig. 12. Fig. 13 shows the comodulograms computed on this simulated spurious PAC dataset. An ideal characterization of the simulated coupling would be visible by a compact blob in the comodulogram. Clearly, the spikes exert great inﬂuence on all six methods. In each panel, the blob corresponding to the coupling frequency pairs is blurred or even shifted. Additionally, coupling at unexpected frequency pairs can be observed. However, we can also ﬁnd that the JSCM measure performs slightly better than the MVL but much better than PLV, KLmi, GLM, and DAR. On the other hand, because no ﬁltering at high frequencies is used in the DAR, there are no unexpected frequency pairs carrying the PAC presented in panel E. Therefore, although the JSCM method is not able to completely solve the issue of spurious PAC, it is still an effective approach to measure PAC when there are spikes in the signal. diseases [21], [22], [36]–[38], [40]–[43], [45], [46]. REFERENCES [1] J. E. Pina, M. Bodner, and B. Ermentrout, “Oscillations in working memory and neural binding: A mechanism for multiple memories and their interactions,” PLoS Comp. Biol., vol. 14, no. 11, Nov. 2018, Art. no. e1006517. [2] C. C. Liu, S. G. Hajra, G. Pawlowski, S. D. Fickling, X. Song, and R. C. N. D’Arcy, “Differential neural processing of spontaneous blinking under visual and auditory sensory environments: An EEG inves- tigation of blink-related oscillations,” NeuroImage, vol. 218, Sep. 2020, Art. no. 116879. pp We also note that there is a contradiction between the results of simulated data and real data. Speciﬁcally, PLV and MVL show similar comodulograms in Figs. 11 and 13, suggesting that they are also robust against noise to some extent. At the same time, GLM fails to detect the coupling when the simulated data are mixed with noise, as shown in Figs. 11 and 13. However, in Fig. 14, the PLV and MVL methods cannot characterize the PAC, but the GLM can. This can be explained as follows. In fact, the simulated signals are periodic, while the experimental data are of great complexity, which is an inherent characteristic of EEG signals. On the other hand, the noise used to examine the robustness of these measures is white Gaussian noise or spike noise. It is possible that none of these noise experiments can replicate the challenge in real data. From Figs. 11 and 13, it can be found that although PLV and MVL are able to resist some noise, they cannot reﬂect the coupling exactly. Moreover, as shown in Fig. 14, they fail to detect the signiﬁcant PAC in analyzing the real data because of their simple algorithm based on phase extraction. In addition, GLM is inferior to the other approaches due to its sensitivity to large white Gaussian noise and spikes. However, it performs well in the analysis of real data because of the absence of large noise, especially spikes. [3] N. Kopell, G. B. Ermentrout, M. A. Whittington, and R. D. Traub, “Gamma rhythms and beta rhythms have different synchronization properties,” Proc. Nat. Acad. Sci. USA, vol. 97, no. 4, pp. 1867–1872, Feb. 2000. [4] C. E. Schroeder and P. Lakatos, “Low-frequency neuronal oscillations as instruments of sensory selection,” Trends Neurosci., vol. 32, no. 1, pp. 9–18, Jan. 2009. pp [5] F. L. da Silva, “EEG and MEG: Relevance to neuroscience,” Neuron, vol. 80, no. V. CONCLUSION V. CONCLUSION V. CONCLUSION In this study, we proposed a novel measure based on JS divergence and correlation matrix to estimate the PAC in con- tinuous electrophysiological signals, such as EEG and iEEG. We explained that the JSCM method can detect coupling that may be missed by the KLmi approach. With the help of simulated data, we ﬁrst provided recommended values for the hyperparameters of the JSCM algorithm. Then, we illustrated that it can effectively characterize the coupling strength and slightly vary with data length. Moreover, we compared the performance of six PAC measures, including JSCM, PLV, MVL, KLmi, GLM and DAR. The results demonstrated that the proposed JSCM method exhibited superior performance, including robustness to resist additive white Gaussian noise, the ability to detect the contributing coupling frequency pairs in a wide frequency range of a signal with a high noise level, and the capability to depict the comodulogram of a signal containing spike noise. Finally, we applied the novel method to analyze the experimental dataset collected from the rodent striatum and observed a result consistent with previous studies. To conclude, we think the proposed JSCM method is an alternative tool to estimate the PAC available in neural oscillations. Fig. 14. Phase-amplitude comodulograms of the six methods applied to experimental data. LFO: 2-10 Hz with a step of 0.5 Hz; HFO: 20-120 Hz with a step of 2 Hz. There is no high- frequency ﬁltering used in panel E. (A) PLV. (B) MVL. (C) KLmi. (D) GLM. (E) DAR. (F) JSCM. ACKNOWLEDGMENT The authors would like to thank Tom Dupré la Tour for pro- viding them with the experimental data, which were recorded at the Charles River Laboratories, France. six methods was plotted, as shown in Fig. 14. It is demon- strated that the JSCM method is able to replicate previous results obtained with other metrics [49]. Although our method does not present signiﬁcant superiority over the KLmi, GLM and DAR in this data analysis, at least it can achieve the same performance. Moreover, considering that the JSCM method can resist noise better, we believe that it is an effective tool to characterize the PAC in real applications. D. Application to Experimental Data As we mentioned in the introduction section, PAC has attracted great attention in studying brain functions and LI et al.: MEASURING PAC 1383 Fig. 14. Phase-amplitude comodulograms of the six methods applied to experimental data. LFO: 2-10 Hz with a step of 0.5 Hz; HFO: 20-120 Hz with a step of 2 Hz. There is no high- frequency ﬁltering used in panel E. (A) PLV. (B) MVL. (C) KLmi. (D) GLM. (E) DAR. (F) JSCM. IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 29, 2021 IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 29, 2021 1384 [12] P. Sauseng, C. Peylo, A. L. Biel, E. V. C. Friedrich, and C. Romberg-Taylor, “Does cross-frequency phase coupling of oscilla- tory brain activity contribute to a better understanding of visual working memory?” Brit. J. Psychol., vol. 110, no. 2, pp. 245–255, May 2019. [35] R. T. Canolty et al., “High gamma power is phase-locked to theta oscilla- tions in human neocortex,” Science, vol. 313, no. 5793, pp. 1626–1628, Sep. 2006. [36] A. B. L. Tort et al., “Dynamic cross-frequency couplings of local ﬁeld potential oscillations in rat striatum and hippocampus during performance of a T-maze task,” Proc. Natl. Acad. Sci. USA, vol. 105, no. 51, pp. 20517–20522, Dec. 2008. [13] T. Popov, O. Jensen, and J.-M. Schoffelen, “Dorsal and ventral cortices are coupled by cross-frequency interactions during working memory,” NeuroImage, vol. 178, pp. 277–286, Sep. 2018. [37] D. Dvorak and A. A. Fenton, “Toward a proper estimation of phase– amplitude coupling in neural oscillations,” J. Neurosci. Methods, vol. 225, pp. 42–56, Mar. 2014. [14] O. Jensen and L. L. Colgin, “Cross-frequency coupling between neu- ronal oscillations,” Trends Cognit. Sci., vol. 11, no. 7, pp. 267–269, Jul. 2007. [38] L. Gagliano, E. B. Assi, D. K. Nguyen, S. Rihana, and M. Sawan, “Bilateral preictal signature of phase-amplitude coupling in canine epilepsy,” Epilepsy Res., vol. 139, pp. 123–128, Jan. 2018. [15] A. B. L. Tort, R. W. Komorowski, J. R. Manns, N. J. Kopell, and H. Eichenbaum, “Theta-gamma coupling increases during the learning of item-context associations,” Proc. Nat. Acad. Sci. USA, vol. 106, no. 49, pp. 20942–20947, Dec. 2009. [39] M. Esghaei, M. R. Daliri, and S. Treue, “Attention decreases phase- amplitude coupling, enhancing stimulus discriminability in cortical area MT,” Frontiers Neural Circuits, vol. 9, Dec. 2015, Art. no. 82. [16] N. Axmacher et al., “Cross-frequency coupling supports multi-item working memory in the human hippocampus,” Proc. Nat. Acad. Sci. USA, vol. 107, no. 7, pp. 3228–3233, Feb. 2010. [40] B. Handel and T. Haarmeier, “Cross-frequency coupling of brain oscilla- tions indicates the success in visual motion discrimination,” NeuroImage, vol. 45, no. 3, pp. 1040–1046, Apr. 2009. [17] O. M. Velarde et al., “Bifurcation structure determines different phase- amplitude coupling patterns in the activity of biologically plausible neural networks,” NeuroImage, vol. 202, Nov. 2019, Art. no. 116031. [41] S. M. IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 29, 2021 Szczepanski et al., “Dynamic changes in phase-amplitude coupling facilitate spatial attention control in fronto-parietal cortex,” Plos Biol., vol. 12, no. 8, Aug. 2014, Art. no. e1001936. [18] D. Dellavale et al., “Two types of ictal phase-amplitude couplings in epilepsy patients revealed by spectral harmonicity of intracerebral EEG recordings,” Clin. Neurophysiol., vol. 131, no. 8, pp. 1866–1885, May 2020. [42] M. Bandarabadi et al., “Assessing epileptogenicity using phase-locked high frequency oscillations: A systematic comparison of methods,” Frontiers Neurol., vol. 10, Oct. 2019, Art. no. 1132. [19] Z. Li et al., “Characterizing the orientation selectivity in V1 and V4 of macaques by quadratic phase coupling,” J. Neural Eng., vol. 17, no. 3, Jun. 2020, Art. no. 036028. [43] E. A. Allen et al., “Components of cross-frequency modulation in health and disease,” Frontiers Syst. Neurosci., vol. 5, Jul. 2011, Art. no. 59. [20] S. Jafakesh, F. Z. Jahromy, and M. R. Daliri, “Decoding of object categories from brain signals using cross frequency coupling methods,” Biomed. Signal Process. Control, vol. 27, pp. 60–67, May 2016. [44] A. Ahmadi, S. Davoudi, and M. R. Daliri, “Computer aided diagnosis system for multiple sclerosis disease based on phase to amplitude cou- pling in covert visual attention,” Comput. Methods Programs Biomed., vol. 169, pp. 9–18, Feb. 2019. [21] J. Lopez-Azcarate et al., “Coupling between beta and high-frequency activity in the human subthalamic nucleus may be a pathophysiological mechanism in Parkinson’s disease,” J. Neurosci., vol. 30, no. 19, pp. 6667–6677, May 2010. [45] Y. Salimpour and W. S. Anderson, “Cross-frequency coupling based neu- romodulation for treating neurological disorders,” Frontiers Neurosci., vol. 13, Feb. 2019, Art. no. 125. [22] C. de Hemptinne et al., “Exaggerated phase-amplitude coupling in the primary motor cortex in parkinson disease,” Proc. Nat. Acad. Sci. USA, vol. 110, no. 12, pp. 4780–4785, Mar. 2013. [46] R. Goutagny et al., “Alterations in hippocampal network oscillations and theta–gamma coupling arise before Aβ overproduction in a mouse model of Alzheimer’s disease,” Eur. J. Neurosci., vol. 37, no. 12, pp. 1896–1902, Jun. 2013. [23] M. Ozturk, A. Abosch, D. Francis, J. Wu, J. Jimenez-Shahed, and N. F. Ince, “Distinct subthalamic coupling in the ON state describes motor performance in Parkinson’s disease,” Movement Disorders, vol. 35, no. 1, pp. 91–100, Jan. 2020. [47] F. Mormann et al., “Phase/amplitude reset and theta–gamma interaction in the human medial temporal lobe during a continuous word recognition memory task,” Hippocampus, vol. 15, no. 7, pp. IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 29, 2021 890–900, 2005. [24] R. Zhang et al., “Temporal-spatial characteristics of phase-amplitude coupling in electrocorticogram for human temporal lobe epilepsy,” Clin. Neurophysiol., vol. 128, no. 9, pp. 1707–1718, Sep. 2017. [48] M. A. Kramer and U. T. Eden, “Assessment of cross-frequency coupling with conﬁdence using generalized linear models,” J. Neurosci. Methods, vol. 220, no. 1, pp. 64–74, Oct. 2013. [25] R. Martinez-Cancino et al., “Measuring transient phase-amplitude coupling using local mutual information,” NeuroImage, vol. 185, pp. 361–378, Jan. 2019. [49] T. D. la Tour et al., “Non-linear auto-regressive models for cross- frequency coupling in neural time series,” PLoS Comput. Biol., vol. 13, no. 12, Dec. 2017, Art. no. e1005893. [26] A. B. L. Tort, R. Komorowski, H. Eichenbaum, and N. Kopell, “Measur- ing phase-amplitude coupling between neuronal oscillations of different frequencies,” J. Neurophysiol., vol. 104, no. 2, pp. 1195–1210, 2010. [50] C. K. Kovach, H. Oya, and H. Kawasaki, “The bispectrum and its relationship to phase-amplitude coupling,” NeuroImage, vol. 173, pp. 518–539, Jun. 2018. [27] M. Belluscio et al., “Cross-frequency phase-phase coupling between theta and gamma oscillations in the hippocampus,” J. Neurosci., vol. 32, no. 2, pp. 423–435, Jan. 2012. [51] E. Combrisson et al., “Tensorpac: An open-source Python toolbox for tensor-based phase-amplitude coupling measurement in electrophysio- logical brain signals,” PLoS Comput. Biol., vol. 16, no. 10, Oct. 2020, Art. no. e1008302. [28] R. Scheffer-Teixeira and A. B. L. Tort, “On cross-frequency phase-phase coupling between theta and gamma oscillations in the hippocampus,” Elife, vol. 5, Dec. 2016, Art. no. e20515. [52] B. Pittman-Polletta, W.-H. Hsieh, S. Kaur, M.-T. Lo, and K. Hu, “Detect- ing phase-amplitude coupling with high frequency resolution using adaptive decompositions,” J. Neurosci. Methods, vol. 226, pp. 15–32, Apr. 2014. [29] T. Rings et al., “No evidence for spontaneous cross-frequency phase– phase coupling in the human hippocampus,” Eur. J. Neurosci., vol. 51, no. 8, pp. 1735–1742, Apr. 2020. [53] A. Delorme and S. Makeig, “EEGLAB: An open source toolbox for analysis of single-trial EEG dynamics including independent com- ponent analysis,” J. Neurosci. Methods, vol. 134, no. 1, pp. 9–21, Mar. 2004. [30] C. Yeh, M. Lo, and K. Hu, “Spurious cross-frequency amplitude– amplitude coupling in nonstationary, nonlinear signals,” Phys. A, Stat. Mech. Appl., vol. 454, pp. 143–150, Jul. 2016. [54] M. G. Rosenblum, A. S. Pikovsky, and J. Kurths, “Phase synchronization of chaotic oscillators,” Phys. Rev. Lett., vol. 76, no. 11, pp. 1804–1807, Mar. 1996. [31] M. REFERENCES 5, pp. 1112–1128, Dec. 2013. [6] S. Hanslmayr, J. Matuschek, and M.-C. Fellner, “Entrainment of prefrontal beta oscillations induces an endogenous echo and impairs memory formation,” Current Biol., vol. 24, no. 8, pp. 904–909, Apr. 2014. p [7] H. I. Berger, “Über das elektrenkephalogramm des menschen,” Eur. Arch. Psychiatry Clin. Neurosci., vol. 87, pp. 527–570, Dec. 1928. [8] L. L. Colgin, “Mechanisms and functions of theta rhythms,” Annu. Rev. Neurosci., vol. 36, no. 1, pp. 295–312, Jul. 2013. [9] A. M. M. Fransen, F. van Ede, and E. Maris, “Identifying neuronal oscillations using rhythmicity,” NeuroImage, vol. 118, pp. 256–267, Sep. 2015. [10] W. Klimesch, “Alpha-band oscillations, attention, and controlled access to stored information,” Trends Cogn. Sci., vol. 16, no. 12, pp. 606–617, Dec. 2012. [11] R. T. Canolty and R. T. Knight, “The functional role of cross- frequency coupling,” Trends Cognit. Sci., vol. 14, no. 11, pp. 506–515, Nov. 2010. IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 29, 2021 Siems and M. Siegel, “Dissociated neuronal phase- and amplitude- coupling patterns in the human brain,” NeuroImage, vol. 209, Apr. 2020, Art. no. 116538. [55] G. M. Ibrahim et al., “Phase-amplitude coupling within the anterior thalamic nuclei during seizures,” J. Neurophysiol., vol. 119, no. 4, pp. 1497–1505, Apr. 2018. [32] M. Parham and S. Sepideh, “Oscillation-based connectivity architecture is dominated by an intrinsic spatial organization, not cognitive state or frequency,” J. Neurosci., vol. 41, no. 1, pp. 179–192, Jan. 2021. [56] J. Lin, “Divergence measures based on the Shannon entropy,” IEEE Trans. Inf. Theory, vol. 37, no. 1, pp. 145–151, Jan. 1991. [33] R. M. G. Reinhart and J. A. Nguyen, “Working memory revived in older adults by synchronizing rhythmic brain circuits,” Nat. Neurosci., vol. 22, no. 4, pp. 820–827, Apr. 2019. [57] S. Kullback, “The Kullback-Leibler distance,” Amer. Statistician, vol. 41, no. 4, pp. 340–341, 1987. [34] B. Voytek et al., “Oscillatory dynamics coordinating human frontal networks in support of goal maintenance,” Nature Neurosci., vol. 18, no. 9, pp. 1318–1324, Jun. 2015. [58] X. Li et al., “Synchronization measurement of multiple neuronal popu- lations,” J. Neurophysiol., vol. 98, no. 6, pp. 3341–3348, Dec. 2007. LI et al.: MEASURING PAC 1385 [59] Z. Li, D. Cui, and X. Li, “Unbiased and robust quantiﬁcation of synchronization between spikes and local ﬁeld potential,” J. Neurosci. Methods, vol. 269, pp. 33–38, Aug. 2016. [69] C. Carmeli, M. G. Knyazeva, G. M. Innocenti, and O. D. Feo, “Assessment of EEG synchronization based on state-space analysis,” NeuroImage, vol. 25, no. 2, pp. 339–354, 2005. [60] D. Cui, X. Liu, Y. Wan, and X. Li, “Estimation of genuine and random synchronization in multivariate neural series,” Neural Netw., vol. 23, no. 6, pp. 698–704, Aug. 2010. [70] G. Buzsáki and A. Draguhn, “Neuronal oscillations in cortical networks,” Science, vol. 304, no. 5679, pp. 1926–1929, Jun. 2004. pp [71] S. Samiee and S. Baillet, “Time-resolved phase-amplitude cou- pling in neural oscillations,” NeuroImage, vol. 159, pp. 270–279, Oct. 2017. [61] M. Yoshimura et al., “Global approach to multichannel electroen- cephalogram analysis for diagnosis and clinical evaluation in mild Alzheimer’s disease,” Neuropsychobiology, vol. 49, no. 3, pp. 163–166, 2004. [72] J. Aru et al., “Untangling cross-frequency coupling in neuroscience,” Curr. Opin. Neurobiol., vol. 31, pp. 51–61, Apr. 2015. [62] B. Czigler et al., “Quantitative EEG in early Alzheimer’s disease patients—Power spectrum and complexity features,” Int. J. Psychophys- iol., vol. 68, no. IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 29, 2021 1, pp. 75–80, Apr. 2008. [73] S. R. Cole et al., “Nonsinusoidal beta oscillations reﬂect cortical pathophysiology in Parkinson’s disease,” J. Neurosci., vol. 37, no. 18, pp. 4830–4840, May 2017. pp p [63] U. Rutishauser et al., “Human memory strength is predicted by theta- frequency phase-locking of single neurons,” Nature, vol. 464, no. 7290, pp. 903–907, Apr. 2010. [74] D. Lozano-Soldevilla, N. ter Huurne, and R. Oostenveld, “Neuronal oscillations with non-sinusoidal morphology produce spurious phase- to-amplitude coupling and directionality,” Frontiers Comput. Neurosci., vol. 10, Aug. 2016, Art. no. 87. pp p [64] P. Fries et al., “Oscillatory neuronal synchronization in primary visual cortex as a correlate of stimulus selection,” J. Neurosci., vol. 22, no. 9, pp. 3739–3754, May 2002. [75] E. M. Gerber et al., “Non-sinusoidal activity can produce cross- frequency coupling in cortical signals in the absence of functional inter- action between neural sources,” PLoS ONE, vol. 11, no. 12, Dec. 2016, Art. no. e1008302. pp y [65] G. Buzsaki, N. Logothetis, and W. Singer, “Scaling brain size, keeping timing: Evolutionary preservation of brain rhythms,” Neuron, vol. 80, no. 3, pp. 751–764, Oct. 2013. [76] M. Arlotti et al., “Monitoring subthalamic oscillations for 24 hours in a freely moving Parkinson’s disease patient,” Mov. Disord., vol. 34, no. 5, pp. 757–759, May 2019. [66] A. J. Ibanez-Molina, M. F. Soriano, and S. Iglesias-Parro, “Mutual information of multiple rhythms for EEG signals,” Frontiers Neurosci., vol. 14, Dec. 2020, Art. no. 574796. pp y [77] G. Dallerac et al., “Updating temporal expectancy of an aversive event engages striatal plasticity under amygdala control,” Nature Commun., vol. 8, Jan. 2017, Art. no. e1008302. [67] N. Cheng et al., “Permutation mutual information: A novel approach for measuring neuronal phase-amplitude coupling,” Brain Topogr., vol. 31, no. 2, pp. 186–201, Mar. 2017. [78] J. I. Berman et al., “Variable bandwidth ﬁltering for improved sensitivity of cross-frequency coupling metrics,” Brain Connect., vol. 2, no. 3, pp. 155–163, Jun. 2012. pp [68] A. Papoulis, Probability, Random Variables and Stochastic Processes, 3th ed. New York, NY, USA: McGraw-Hill, 1991.
https://openalex.org/W2048062896	https://europepmc.org/articles/pmc3846734?pdf=render	English	null	MagFRET: The First Genetically Encoded Fluorescent Mg2+ Sensor	PloS one	2,013	cc-by	10,847	Introduction Magnesium is the most abundant intracellular divalent cation and is involved in numerous essential cellular processes including replication, transcription, translation and energy metabolism. In addition to its omnipresent role as an essential enzymatic cofactor, Mg2+ is also important for chromatin stability and regulates specific ion channels [1,2]. The total cellular Mg2+ concentration ranges between 14–20 mM, but the concentration of free Mg2+ in the cytosol has been estimated to be between 0.1 and 1.5 mM [3– 7]. Given its importance to so many different cellular processes, the intracellular Mg2+ concentration is generally believed to be strongly buffered and tightly regulated by the combined action of magnesium binding (macro)molecules (proteins, ribonucleic acids, ATP, etc.), storage in organelles and the action of Mg2+ channels [8–10]. Hereditary disorders related to Mg2+ homeostasis have been shown to result in diminished kidney functioning and in severe cases to renal failure, muscle spasms and seizures [11]. Magnesium deficiency has also been shown to accelerate cellular senescence [12], providing a potential link between low dietary magnesium intake and the early onset of aging diseases such as diabetes [13], cardiovascular diseases [14] and osteoporosis [15]. Recent studies suggested that T cell activation following antigen receptor stimulation was dependent on a transient influx of Mg2+ Despite the abundance and importance of Mg2+, the intracel- lular regulation of Mg2+ homeostasis and the putative role of Mg2+ in intracellular signal transduction are not well understood. In part this is because of a lack of convenient molecular tools to image the intracellular Mg2+ concentration in single living cells in real time [17]. Magnesium-selective microelectrodes have been used to determine cytosolic Mg2+ levels in different muscle cells, revealing concentrations between 0.7 and 0.9 mM [4]. However, these microelectrodes are highly invasive and do not provide spatial information. Another method to probe the intracellular concen- tration of Mg2+ is the measurement of the ratio of Mg2+-bound and Mg2+-free ATP using 31P NMR [18]. While non-invasive, 31P NMR measures Mg2+ indirectly and averaged over a large collection of cells [19,20]. The currently most commonly applied approach uses synthetic dyes that alter their fluorescent properties upon binding of Mg2+ [21–27]. However, many of the available dyes show limited specificity for Mg2+ and often bind Ca2+ with low micromolar affinity [22,25,26], which has been shown to interfere in an intracellular setting [28]. Abstract nding: This work was supported by the Netherlands Organization of Scientific Research NWO (VIDI grant 700.56.428) and by an ERC e funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. ¤ Current address: Molecular Diagnostics, Philips Research, Eindhoven, The Netherlands ¤ Current address: Molecular Diagnostics, Philips Research, Eindhoven, The Netherlands ¤ Current address: Molecular Diagnostics, Philips Research, Eindhoven, The Netherlands . These authors contributed equally to this work. in the cytosol, implicating a novel role for Mg2+ as second messenger in intracellular signal transduction [16]. MagFRET: The First Genetically Encoded Fluorescent Mg2+ Sensor Laurens H. Lindenburg., Jan L. Vinkenborg.¤, Jorn Oortwijn, Stijn J. A. Aper, Maarten Merkx* Laboratory of Chemical Biology, Department of Biomedical Engineering, Eindhoven University of Technology, Eindhoven, The Netherlands Abstract Magnesium has important structural, catalytic and signaling roles in cells, yet few tools exist to image this metal ion in real time and at subcellular resolution. Here we report the first genetically encoded sensor for Mg2+, MagFRET-1. This sensor is based on the high-affinity Mg2+ binding domain of human centrin 3 (HsCen3), which undergoes a transition from a molten- globular apo form to a compactly-folded Mg2+-bound state. Fusion of Cerulean and Citrine fluorescent domains to the ends of HsCen3, yielded MagFRET-1, which combines a physiologically relevant Mg2+ affinity (Kd = 148 mM) with a 50% increase in emission ratio upon Mg2+ binding due to a change in FRET efficiency between Cerulean and Citrine. Mutations in the metal binding sites yielded MagFRET variants whose Mg2+ affinities were attenuated 2- to 100-fold relative to MagFRET-1, thus covering a broad range of Mg2+ concentrations. In situ experiments in HEK293 cells showed that MagFRET-1 can be targeted to the cytosol and the nucleus. Clear responses to changes in extracellular Mg2+ concentration were observed for MagFRET- 1-expressing HEK293 cells when they were permeabilized with digitonin, whereas similar changes were not observed for intact cells. Although MagFRET-1 is also sensitive to Ca2+, this affinity is sufficiently attenuated (Kd of 10 mM) to make the sensor insensitive to known Ca2+ stimuli in HEK293 cells. While the potential and limitations of the MagFRET sensors for intracellular Mg2+ imaging need to be further established, we expect that these genetically encoded and ratiometric fluorescent Mg2+ sensors could prove very useful in understanding intracellular Mg2+ homeostasis and signaling. Editor: Claudio M. Soares, Instituto de Tecnologica Quı´mica e Biolo´gica, UNL, Portugal Received September 23, 2013; Accepted October 29, 2013; Published December 2, 2013 Received September 23, 2013; Accepted October 29, 2013; Published December 2, 2013 Copyright: 2013 Lindenburg et al. This is an open-access article distributed under the terms of the Creative Comm unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Copyright: 2013 Lindenburg et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Citation: Lindenburg LH, Vinkenborg JL, Oortwijn J, Aper SJA, Merkx M (2013) MagFRET: The First Genetically Encoded Fluorescent Mg2+ Sensor. PLoS ONE 8(12): e82009. doi:10.1371/journal.pone.0082009 Protein expression and purification Protein expression and purification E. coli BL21(DE3) cells were used for protein expression. A single colony was used to inoculate 5 mL LB medium (10 g/L NaCl, 10 g/L peptone, 5 g/L yeast extract) supplemented with 30 mg/mL kanamycin which was grown overnight at 225 rpm at 37uC. Overnight cultures were diluted in 500 mL LB medium containing kanamycin (30 mg/mL) and grown until an optical density of 0.6–0.8 was reached at 600 nm wavelength. Protein expression was induced by the addition of 0.1 mM isopropyl b-D- 1-thiogalactopyranoside (IPTG, Sigma). Bacteria were cultured overnight at 225 rpm at 25uC and harvested by centrifugation at 10,000 g for 10 minutes at 4uC. The cell pellets were lysed using Bugbuster reagent (Novagen) according to the manufacturer’s instructions. The resulting soluble protein fraction was used for further purification. The expressed MagFRET proteins contain an N-terminal hexahistidine-tag. Ni2+-NTA resin (His-bind, Nova- gen) was used for affinity chromatography following the manu- facturer’s instructions. After elution of the protein using 0.5 M imidazole, the protein was dialyzed overnight against 100 volumes of 20 mM Tris-HCl (pH 8.4), 150 mM NaCl and 2.5 mM CaCl2 using a 12–14 kDa Molecular Weight Cut-Off (MWCO) dialysis membrane (Spectropore) at 4uC. The hexahistidine-tag was subsequently removed by the addition of 0.3 U thrombin protease (Novagen) per mg protein at a 0.2 mg/mL protein concentration and incubated for 24 hours at 4uC. His-tags and uncleaved proteins were removed using Ni2+ affinity chromatography. The flow-through was further purified using size exclusion chromatog- raphy (SEC) on a Sephacryl S-200 High resolution column (GE Healthcare). Fractions containing pure proteins were pooled, concentrated, frozen in liquid nitrogen and stored in aliquots at 280uC. The purity and correct molecular weight of the obtained proteins was confirmed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE, with 12% acrylamide) analysis. The protein concentration was determined using the absorption at 515 nm (ND-1000 Nanodrop) and a molar extinction coefficient of 77,000 M21cm21 for Citrine [44]. Surprisingly, no genetically encoded sensors have thus far been reported for Mg2+. One of the specific challenges in this case is metal binding specificity. Mg2+ and Ca2+ show similar coordina- tion chemistry and often bind to the same metal binding proteins, with Ca2+ typically showing stronger binding. Protein expression and purification Here we report the first genetically encoded fluorescent sensor (MagFRET-1) for Mg2+ by taking advantage of the particular metal binding properties of the N-terminal part of the HsCen3 protein, which binds both Ca2+ and Mg2+ with high affinity [42]. We show that Mg2+ binding to MagFRET-1 induces folding from a molten- globule state that results in an increase in FRET. Mutagenesis of metal binding site residues allowed further tuning of the metal binding properties, yielding MagFRET variants with Kd values for Mg2+ binding ranging between 0.15 and 15 mM. While also responsive to Ca2+ in vitro, we show that the Ca2+ affinities of the MagFRET sensors are sufficiently attenuated that they are not responsive to normal Ca2+ fluctuations in situ. Cloning of expression plasmids DNA encoding the N-terminal fragment of HsCen3 (residues 23 to 98 [42]) was obtained as a synthetic pUC57 construct (GenScript, USA). Restriction of this construct with restriction enzymes NheI and NcoI yielded an insert fragment that was compatible with a pET28a acceptor vector encoding for His6- Cerulean-(GGS)18-Citrine [43] that had been treated with restriction enzymes SpeI (creating an NheI-compatible cohesive overhang) and NcoI. A ligation was carried out at equimolar vector-to-insert ratio using T4 DNA ligase (TaKaRA Mighty Mix, Takara, USA) at 16uC for 1 hour following the manufacturer’s instructions, resulting in pET28a-MagFRET-1 (Figure S1). The SpeI restriction site in the pET28a-His6-Cerulean-(GGS)18-Citrine acceptor vector was located 8 residues upstream of the Cerulean C-terminus, such that in the final MagFRET construct, the native flexible C-terminus of Cerulean was deleted, resulting in tighter allosteric coupling between changes in HsCen3 conformation and changes in the fluorescent domains’ interchromophore distance. A FRET Sensor for Intracellular Mg2+ Imaging A FRET Sensor for Intracellular Mg2+ Imaging to physiological changes in cytosolic Ca2+ concentration [24,29,30]. Recently a variant of this dye, KMG-103 was reported that showed preferred accumulation in mitochondria [27]. Like most synthetic Mg2+ dyes, the KMG dyes are intensiometric, making Mg2+ quantification challenging and sensitive to changes in sensor concentration. A few ratiometric Mg2+ fluorescent dyes (e.g. Mag-Fura and Mag-Indo) exist, yet these have the disadvantage that they require potentially cytotoxic UV excitation [17]. The mammalian expression vector for MagFRET-1 was obtained by digesting pET28a-MagFRET-1 using restriction enzymes AgeI and NotI. Ligation into a peCALWY-1 vector [36] that was digested with the same restriction enzymes resulted in pCMV- MagFRET-1 (Figure S2). Mutations in metal binding loop I and II of MagFRET-1 were introduced using site-directed mutagenesis (QuikChange Multi Site-Directed Mutagenesis Kit for mutations in loop I and QuikChange Site-Directed Mutagenesis Kit for mutations in loop II), following the kit manufacturer’s (Qiagen) instructions. Primers used to introduce these mutations are listed in Table S1. To obtain the mammalian expression vector encoding for a nuclear-targeted MagFRET-1 (MagFRET-1-NLS), a pUC57 vector containing a synthetic gene encoding for the final part of Citrine together with three PKKKRKV repeats was digested using restriction enzymes HindIII and NotI, followed by ligation into a pCMV-MagFRET-1 plasmid that was treated with the same restriction enzymes (Figure S3). The correct open reading frame of each sensor was confirmed by Sanger dideoxy sequencing (Baseclear, Leiden, The Netherlands). Genetically encoded fluorescent sensor proteins provide an attractive alternative to small-molecule fluorescent sensors, because they do not require cell-invasive procedures, their concentration can be tightly controlled and they can be targeted to different locations in the cell [31]. Many of these sensors consist of metal binding domain(s) fused to a donor and an acceptor fluorescent domain capable of Fo¨rster Resonance Energy Transfer (FRET). Modulation of the distance and/or orientation of the fluorescent domains following metal binding affects the FRET efficiency, which can be detected as change in the emission ratio, an output signal that is independent of sensor concentration. In addition, the use of natural metal binding protein domains often ensures a physiologically relevant metal binding affinity and specificity. The wealth of genetically encoded sensors that have been developed for Ca2+ [32–35], and more recently also for Zn2+ [36–39] and Cu+ [40,41], have made important contributions to the understanding of intracellular metal homeostasis and signaling. Introduction A notable exception is KMG-104 and related dyes developed by Kuzuki and coworkers, whose affinity for Mg2+ is higher than for Ca2+ (Kd = 2.1 and 7.5 mM, respectively), rendering these dyes completely insensitive December 2013 \| Volume 8 \| Issue 12 \| e82009 1 PLOS ONE \| www.plosone.org Imaging of response to changes in intracellular Mg2+ 2 Imaging of response to changes in intracellular Mg Prior to addition of Mg2+ or EDTA, cells were permeabilized by a 6 minute incubation of HEK293 cells in 400 mL intracellular buffer (IB) containing 10 mg/ml digitonin (Sigma). IB comprised 20 mM Hepes (pH 7.05), 140 mM KCl, 10 mM KH2PO4, 100 mM ATP, 2 mM Na+ succinate and 5.5 mM glucose. After 6 minutes, recordings were started and buffers containing increasing concentrations of EDTA or MgCl2 were added as stated in the main text. When adding MgCl2, KCl concentrations were reduced accordingly to maintain the Cl2 concentration at 140 mM. Imaging frequency was 0.1 Hz. Prior to addition of Mg2+ or EDTA, cells were permeabilized by a 6 minute incubation of HEK293 cells in 400 mL intracellular buffer (IB) containing 10 mg/ml digitonin (Sigma). IB comprised 20 mM Hepes (pH 7.05), 140 mM KCl, 10 mM KH2PO4, 100 mM ATP, 2 mM Na+ succinate and 5.5 mM glucose. After 6 minutes, recordings were started and buffers containing increasing concentrations of EDTA or MgCl2 were added as stated in the main text. When adding MgCl2, KCl concentrations were reduced accordingly to maintain the Cl2 concentration at 140 mM. Imaging frequency was 0.1 Hz. R~Rsz DR1: Ca2z Kd1z Ca2z ½ z DR2: Ca2z Kd2z Ca2z ½ ð2Þ ð2Þ where RS is the starting emission ratio in absence of Ca2+, DR1 is the difference between the emission ratio in the Ca2+-free state and the state in which a single Ca2+ ion has bound to the first metal binding loop of HsCen3, DR2 the difference between the latter state and the state in which a second Ca2+ ion has bound to EF- hand II. For the metal specificity measurements, either BaCl2, NiSO4, CuSO4, ZnCl2 or FeCl3 was added from 10006 concentration stock solutions to a final concentration of 10 mM, followed by addition of 1 mM MgCl2. For titrations (with MgCl2, NaCl and ammonium acetate) testing the effect of ionic strength on the sensor, a low salt buffer was used, consisting of 20 mM Hepes (pH 7.1), 10 mM NaCl and 10% (v/v) glycerol. To test for pH sensitivity of MagFRET-1, MgCl2 and CaCl2 titrations were carried out in buffers where Hepes was replaced with either MES (2-(N-morpholino)ethanesulfonic acid) or Tris (tris(hydroxymethy- l)aminomethane) in the standard measurement buffer. Cell culturing and transfection HEK293 cells were grown in Dulbecco’s modified Eagle medium (DMEM, Sigma) containing 10% (vol/vol) fetal bovine serum (FBS, Life Technologies), 3 mM glucose, 2 mM glutamine, 100 units mL21 penicillin and 100 mg mL21 streptomycin at 37uC and 5% CO2. Cells were plated on poly-L-lysine (Sigma) treated glass coverslips and transfected with 1.5 mg of plasmid DNA and 5 mg polyethyleneimine (PEI). Cells were imaged for transient expression 2 days after transfection. In addition to the pCMV- MagFRET-1 and pCMV-MagFRET-1-NLS constructs described above, cells were also transfected with peZinCh-NB [36], which encodes for the Cerulean-linker-Citrine protein that was used as a negative control. Details of the Western blot procedure are provided in Method S1. Imaging of response to changes in intracellular Mg2+ 2 The MES- containing buffer was prepared at pH 6 while the Tris-containing buffer was prepared at pH 8. where RS is the starting emission ratio in absence of Ca2+, DR1 is the difference between the emission ratio in the Ca2+-free state and the state in which a single Ca2+ ion has bound to the first metal binding loop of HsCen3, DR2 the difference between the latter state and the state in which a second Ca2+ ion has bound to EF- hand II. For the metal specificity measurements, either BaCl2, NiSO4, CuSO4, ZnCl2 or FeCl3 was added from 10006 concentration stock solutions to a final concentration of 10 mM, followed by addition of 1 mM MgCl2. For titrations (with MgCl2, NaCl and ammonium acetate) testing the effect of ionic strength on the sensor, a low salt buffer was used, consisting of 20 mM Hepes (pH 7.1), 10 mM NaCl and 10% (v/v) glycerol. To test for pH sensitivity of MagFRET-1, MgCl2 and CaCl2 titrations were carried out in buffers where Hepes was replaced with either MES (2-(N-morpholino)ethanesulfonic acid) or Tris (tris(hydroxymethy- l)aminomethane) in the standard measurement buffer. The MES- containing buffer was prepared at pH 6 while the Tris-containing buffer was prepared at pH 8. Intracellular Ca2+ specificity Intracellular calcium specificity measurements were performed on HEK293 cells transfected as described above. Modified Krebs- bicarbonate buffer was used, consisting of 10 mM Hepes (pH 7.4), 140 mM NaCl, 3.6 mM KCl, 0.5 mM NaH2PO4, 1.5 mM CaCl2, 25 mM NaHCO3 and 3 mM glucose. Where indicated, PAR-1 agonist peptide (sequence SFLLRN, Genscript, USA) or ATP (Sigma) were added to a final concentration of 50 mM. Control experiments were performed using non-transfected HEK293 cells that were loaded with 10 mM Oregon Green- BAPTA-AM (Life Technologies, Netherlands) in phosphate buffered saline (PBS) with 0.01% (w/v) Pluronic F-127 (Life Technologies) for 30 minutes. At the end of each experiment in which Oregon Green-BAPTA-AM was used, 20 mM of calcium ionophore A23187 (Sigma) was added. Imaging frequency was 0.2 Hz. Sensor design The construction of a FRET sensor for Mg2+ requires the availability of a metal binding domain that undergoes a large conformational change and displays a relatively high affinity for Mg2+ compared to Ca2+. Cox and coworkers previously reported that a truncated version of HsCen3 containing the first two of its four native EF-hand metal binding sites, undergoes a dramatic change in conformation upon metal binding from a molten globular (MG) state to a compact, natively-folded state [42]. Unlike most other EF hand-like proteins, which typically bind Ca2+ orders of magnitude more strongly than Mg2+, HsCen3’s first EF hand is a high-affinity mixed Mg2+/Ca2+ binding site, with a reported Kd for Mg2+ of 10–28 mM and a Kd for Ca2+ of 1.5– 8 mM. The second metal binding site was reported to bind only Ca2+, but with a much weaker affinity (Kd = 140 mM). HsCen3 is one of the four isoforms of human Centrin, a family of proteins that is involved in centriole duplication. We based our design on the structure of HsCen2, which shows high homology to HsCen3 and is the only isoform for which an X-ray structure has been determined (Figure 1A). The 11 kDa N-terminal fragment studied by Cox and coworkers contained the complete a-helix A FRET Sensor for Intracellular Mg2+ Imaging A FRET Sensor for Intracellular Mg2+ Imaging 420 nm. MgCl2 and CaCl2 (both from Sigma) were added at increasing concentrations from a concentrated stock solution in water. To determine the MagFRET-1 dissociation constant (Kd) for Mg2+, the emission ratio (R) as a function of MgCl2 concentration ([Mg2+]) was fit to equation 1, for FRET measurements was performed on an Axio observer D.1 (Zeiss) equipped with an Axiocam MRm monochrome digital camera (Zeiss) using Axiovision 4.7 software. Samples were excited using a HXP 120 Mercury lamp (Zeiss) and Cerulean and Citrine emission was recorded sequentially using filter set 47 (excitation BP 436/20, dichroic 455, emission BP 480/40) and 48 (excitation BP 436/20, dichroic 455, emission BP 535/30) (Zeiss) in a motorized filter turret. Emission of Oregon Green-BAPTA was recorded using filterset 38 HE (excitation 470/40, dichroic 495, emission BP 525/50) (Zeiss). Images were acquired using an apochromat 406objective, with an exposure time of 200–300 ms. for FRET measurements was performed on an Axio observer D.1 (Zeiss) equipped with an Axiocam MRm monochrome digital camera (Zeiss) using Axiovision 4.7 software. Samples were excited using a HXP 120 Mercury lamp (Zeiss) and Cerulean and Citrine emission was recorded sequentially using filter set 47 (excitation BP 436/20, dichroic 455, emission BP 480/40) and 48 (excitation BP 436/20, dichroic 455, emission BP 535/30) (Zeiss) in a motorized filter turret. Emission of Oregon Green-BAPTA was recorded using filterset 38 HE (excitation 470/40, dichroic 495, emission BP 525/50) (Zeiss). Images were acquired using an apochromat 406objective, with an exposure time of 200–300 ms. R~Rsz DR: Mg2z Kdz Mg2z ½ ð1Þ ð1Þ where RS is the starting emission ratio in absence of Mg2+ and DR the difference in emission ratio between the Mg2+-free and Mg2+- saturated form of MagFRET-1. To determine the MagFRET Kd1 and Kd2 values associated with the first and second Ca2+-binding events respectively, the emission ratio (R) as a function of CaCl2 concentration ([Ca2+]) was fit to a double binding event using equation 2, Fluorescence spectroscopy Unless otherwise mentioned, magnesium and calcium titrations were performed in 150 mM Hepes (4-(2-hydroxyethyl)-1-piper- azineethanesulfonic acid) (pH 7.1), 100 mM NaCl, 10% (v/v) glycerol. Fluorescence emission spectra were recorded between 450 and 600 nm at a 0.2 mM protein concentration on a Varian Cary Eclipse fluorometer with an excitation wavelength of December 2013 \| Volume 8 \| Issue 12 \| e82009 PLOS ONE \| www.plosone.org 2 Fluorescence microscopy To demonstrate correct localization and subcellular targeting of MagFRET-1, confocal microscopy (Leica TCS SP5 X) was used to image the sensor with high spatial resolution. Samples were excited using a 405 nm laser and emission was detected using a hybrid APD/PMT detector (HyD, Leica). Spectral emission windows were set to 460–490 nm for the Cerulean channel and 510–550 nm for the Citrine channel, using an acousto optical beam splitter (AOBS, Leica). Widefield fluorescence microscopy December 2013 \| Volume 8 \| Issue 12 \| e82009 PLOS ONE \| www.plosone.org 3 A FRET Sensor for Intracellular Mg2+ Imaging was fitted to a 2:1 binding model (Figure 2C). Binding of Ca2+ to the high affinity site showed a Kd of 1064 mM and resulted in a 19% increase in emission ratio. An additional 20% increase in emission ratio was observed at Ca2+ concentrations above 1 mM, but the low affinity for this site precluded accurate determination of its Kd. While the absolute affinity of MagFRET-1 for Ca2+ is higher than for Mg2+, the sensor would not be expected to be sensitive to normal fluctuations in bulk cytosolic Ca2+ concentra- tions, which range between 0.1 and 1 mM [45]. No increase in emission ratio was observed upon addition of 10 mM Ba2+, Ni2+, Cu2+ or Fe3+, while only a very small increase was seen for 10 mM Zn2+ (Figure 2D), a concentration that is 10,000-fold higher than the free Zn2+ concentration found in the cytosol [36]. Another important aspect of sensor performance is pH sensitivity. Ca2+ and Mg2+ titrations performed at pH 6 and pH 8 showed that metal binding affinities were unaffected within this pH range (Figure S4A–D). As expected, the absolute emission ratios where somewhat lower at pH 6, due to the pH sensitivity of Citrine, which has a pKa of 5.7 [44]. Finally, we noticed that the emission ratio of the apo form of the sensor is dependent on the ionic strength of the buffer (Figure S5A). When the Mg2+ titration was repeated in a low ionic strength buffer, the emission ratio of the apo form decreased to 2.2, whereas the emission ratio in the Mg2+- bound state was the same (Figure S5B) and the Mg2+ affinity remained mostly unaffected (Kd = 231610 mM). This ionic strength dependence most likely reflects the influence of ionic strength on the compactness of the molten globule structure of HsCen3. In vitro characterization of MagFRET-1 To allow characterization of MagFRET-1 in vitro, a His-tagged sensor construct was expressed in good yield in E. coli BL21(DE3) and purified using Ni2+-affinity and size exclusion chromatogra- phy. A relatively high ratio of Citrine to Cerulean emission of 3.6 was observed in the absence of Mg2+, indicating that the molten globule state of the metal binding domain is relatively compact bringing the fluorescent domains close together (Figure 2A). As expected, a further increase in emission ratio of 50% was observed upon addition of Mg2+, which is consistent with the formation of a more compact metal-bound, native state. The increase in emission ratio could be fitted using a 1:1 binding model, yielding a Kd for Mg2+ of 148623 mM (Figure 2B). Fortunately, this affinity is in the (lower) range of the cytosolic [Mg2+]free reported by previous methods, and 10-fold weaker than that reported by Cox et al. for their N-terminal variant of HsCen3 [42]. Since HsCen3 was reported to not only bind Mg2+ but also contain two Ca2+ binding sites [42], the Ca2+ response of MagFRET-1 was also tested. Addition of Ca2+ led to a biphasic increase in emission ratio, which Figure 1. Design of the genetically encoded magnesium FRET sensor MagFRET. (A) Crystal structure (PDB code 2GGM) of HsCen2 in the calcium-bound, compact state. The typical helix-loop-helix structure can be observed, with EF-hands indicated by Roman numerals. The dotted lines indicate the N-terminal truncated part of the domain used in the sensor. In HsCen3, the high-affinity Mg2+/Ca2+ binding site is in loop I, and a much weaker Ca2+-binding site is found in loop II. (B) Schematic representation of MagFRET, where the N-terminal truncation of HsCen3 is flanked by Cerulean and Citrine. In absence of Mg2+, the HsCen3 domain is in a molten globule-like state, with little tertiary structure and a relatively large average distance between the fluorescent domains. Mg2+-binding induces a compact, well-defined tertiary structure, resulting in increased energy transfer between Cerulean and Citrine. doi:10 1371/journal pone 0082009 g001 Fluorescence microscopy Although the effect is less pronounced at physiologically relevant salt concentrations, it does mean that large changes in ionic strength should be avoided when applying MagFRET-1 in situ. connecting the 2nd and 3rd EF hand sites. To decrease the distance between the N- and C-termini of the receptor part in the Mg2+- bound state, we decided to truncate this helix to approximately half its size (aa 23–98) and fuse it to the fluorescent proteins Cerulean and Citrine (Figure 1B). To ensure that a conforma- tional change of the HsCen3 domain in MagFRET-1 was translated to a maximal change in relative orientation of the fluorescent domains, the final 8 amino acids from the flexible C- terminus of Cerulean were removed. Tuning metal binding affinities g g To test whether we could further tune the metal affinity and specificity of the MagFRET sensor we explored several mutations in both metal binding sites. Targeting key residues in the 1st EF hand (D1A, D3E, A7D and D5E/A7E) resulted in a reduction of both the Mg2+ and Ca2+ affinity to the millimolar regime, indicating that these residues are indeed directly involved in high affinity metal binding (Table 1, Figure 3C–F). Only a single Ca2+ binding event was observed for these mutants, suggesting that the two EF hands in MagFRET-3-6 have a similar Ca2+ affinity, making the two binding events indistinguishable. Inter- estingly, an E6D substitution (MagFRET-2) did not alter the affinity for Mg2+ or Ca2+ (Figure 3B, Table 1), showing that the presence of a glutamic acid at this position is not essential for high affinity metal binding. Although the change in emission ratio for binding Mg2+ is attenuated to 33% in this variant, the response to Ca2+ binding is almost absent for the high affinity site (3%), rendering this variant effectively Ca2+ insensitive. In an effort to abolish Ca2+ binding to the weakly Ca2+-binding EF hand II, we replaced aspartic acid 1 (MagFRET-7) and glycine 6 (MagFRET- 8) at that site by positively charged lysine residues. Surprisingly, upon titration of Ca2+, both sensor variants still displayed the same biphasic response as seen with MagFRET-1 (Figure 3A, G, H), showing that neither of these residues is essential for the low affinity Ca2+ binding event in EF hand II. Interestingly, both the D1K and the G6K mutation subtly attenuated the high affinity mixed Ca2+/Mg2+ site in EF-hand I, leading to a 6- and 5-fold decrease of the Mg2+ affinity and a 6- and 4-fold decrease in Ca2+ affinity, respectively (Figure 3G, H). The somewhat weaker affinities for both Mg2+ and Ca2+ observed for MagFRET-7 and MagFRET-8 could prove beneficial for imaging Mg2+ homeostasis Figure 1. Design of the genetically encoded magnesium FRET sensor MagFRET. (A) Crystal structure (PDB code 2GGM) of HsCen2 in the calcium-bound, compact state. The typical helix-loop-helix structure can be observed, with EF-hands indicated by Roman numerals. The dotted lines indicate the N-terminal truncated part of the domain used in the sensor. In HsCen3, the high-affinity Mg2+/Ca2+ binding site is in loop I, and a much weaker Ca2+-binding site is found in loop II. Tuning metal binding affinities (B, C) Emission ratio (Citrine to Cerulean) of MagFRET-1 as a function of the Mg2+ (B) or Ca2+ (C) concentration. Solid lines indicate a fit to a single (B) or a double (C) binding event, yielding a Kd of 0.1560.02 mM for Mg2+ and Kd’s of 1064 mM and ,35 mM for Ca2+, respectively. (D) Emission ratios of MagFRET-1 in absence of metal, in the presence of 10 mM Ba2+, Ni2+, Cu2+, Zn2+ or Fe3+, and in the presence of the same metals and 1 mM Mg2+. Measurements were performed in triplicate, error bars indicate SEM. All measurements were performed in 150 mM Hepes (pH 7.1), 100 mM NaCl and 10% (v/v) glycerol with 0.2 mM sensor protein. doi:10.1371/journal.pone.0082009.g002 under conditions where the intracellular Mg2+ concentrations are higher. (Figure S7A), while cells loaded with Oregon Green-BAPTA showed a clear response to the same treatment (Figure S7B). (Figure S7A), while cells loaded with Oregon Green-BAPTA showed a clear response to the same treatment (Figure S7B). p ( g ) Having established that MagFRET-1 could be targeted and that it was insensitive to normal cytosolic Ca2+ fluctuations, we next characterized MagFRET-1’s ability to report on changes in cytosolic [Mg2+]free. Surprisingly, attempts to perturb the free concentration of Mg2+ in the cytosol of intact HEK293 cells using previously reported procedures did not induce a significant ratiometric response in HEK293 cells transfected with Mag- FRET-1 (not shown). These protocols included incubation of the cells in 50 mM MgCl2 [29], exposure to the Mg2+ competitor Li+ [53] and varying of the sodium concentration to affect the Mg2+/ Na+ exchanger [54]. To verify that the transiently expressed sensor is still responsive, we permeabilized HEK293 cells transfected with MagFRET-1 using 10 mg/mL digitonin and exposed them to buffers with different Mg2+ or EDTA concentrations (Figure 4G– J). Addition of 2 mM Mg2+ resulted in an increase in emission ratio for MagFRET-1, indicating metal binding to this sensor (Figure 4G). Subsequent addition of increasing Mg2+ concentra- tions did not result in a further increase in emission ratio up to 10 mM, while rounding of cells was observed at concentrations of 15 and 30 mM Mg2+ (not shown). To verify that the sensor could also monitor a decrease in cytosolic Mg2+ levels, the metal chelator EDTA was added to permeabilized cells expressing MagFRET-1. Tuning metal binding affinities (B) Schematic representation of MagFRET, where the N-terminal truncation of HsCen3 is flanked by Cerulean and Citrine. In absence of Mg2+, the HsCen3 domain is in a molten globule-like state, with little tertiary structure and a relatively large average distance between the fluorescent domains. Mg2+-binding induces a compact, well-defined tertiary structure, resulting in increased energy transfer between Cerulean and Citrine. Figure 1. Design of the genetically encoded magnesium FRET sensor MagFRET. (A) Crystal structure (PDB code 2GGM) of HsCen2 in the calcium-bound, compact state. The typical helix-loop-helix structure can be observed, with EF-hands indicated by Roman numerals. The dotted lines indicate the N-terminal truncated part of the domain used in the sensor. In HsCen3, the high-affinity Mg2+/Ca2+ binding site is in loop I, and a much weaker Ca2+-binding site is found in loop II. (B) Schematic representation of MagFRET, where the N-terminal truncation of HsCen3 is flanked by Cerulean and Citrine. In absence of Mg2+, the HsCen3 domain is in a molten globule-like state, with little tertiary structure and a relatively large average distance between the fluorescent domains. Mg2+-binding induces a compact, well-defined tertiary structure, resulting in increased energy transfer between Cerulean and Citrine. doi:10.1371/journal.pone.0082009.g001 December 2013 \| Volume 8 \| Issue 12 \| e82009 PLOS ONE \| www.plosone.org 4 A FRET Sensor for Intracellular Mg2+ Imaging Figure 2. Metal binding properties of MagFRET-1. (A) Normalized fluorescence emission spectra of MagFRET-1 at 0 and at 16 mM Mg2+ after excitation at 420 nm. (B, C) Emission ratio (Citrine to Cerulean) of MagFRET-1 as a function of the Mg2+ (B) or Ca2+ (C) concentration. Solid lines indicate a fit to a single (B) or a double (C) binding event, yielding a Kd of 0.1560.02 mM for Mg2+ and Kd’s of 1064 mM and ,35 mM for Ca2+, respectively. (D) Emission ratios of MagFRET-1 in absence of metal, in the presence of 10 mM Ba2+, Ni2+, Cu2+, Zn2+ or Fe3+, and in the presence of the same metals and 1 mM Mg2+. Measurements were performed in triplicate, error bars indicate SEM. All measurements were performed in 150 mM Hepes (pH 7.1), 100 mM NaCl and 10% (v/v) glycerol with 0.2 mM sensor protein. doi:10.1371/journal.pone.0082009.g002 Figure 2. Metal binding properties of MagFRET-1. (A) Normalized fluorescence emission spectra of MagFRET-1 at 0 and at 16 mM Mg2+ after excitation at 420 nm. Tuning metal binding affinities Upon addition of EDTA, cells expressing MagFRET-1 showed a decrease in emission ratio that is consistent with a decrease in cytosolic Mg2+ levels (Figure 4H). Importantly, no changes in emission ratio were observed upon addition of Mg2+ or EDTA to digitonin-treated cells expressing a negative control construct consisting of Cerulean, a flexible linker and Citrine but lacking any metal binding sites (Figure 4I, J). These results exclude the possibility that changes in emission ratio observed in MagFRET-1- expressing cells may have resulted from changes in the fluorescent Discussion To the best of our knowledge, this work represents the first report of a genetically encoded fluorescent sensor for Mg2+. The new sensor principle of metal-induced folding of an EF-hand protein was used to create a FRET-based sensor protein that combines a physiologically relevant Mg2+ affinity with a 50% increase in emission ratio upon Mg2+ binding. Mutations introduced in the metal binding domains yielded sensor variants with different degrees of attenuation in Mg2+ affinity, generating a toolbox of MagFRET variants for different applications. Unlike most synthetic fluorescent Mg2+ probes reported so far, Mag- FRET-1 allows emission ratiometric detection of Mg2+ and is thus less sensitive to fluctuations in sensor concentration or background fluorescence. A general advantage of genetically encoded sensors is that their subcellular localization can be easily controlled, as we demonstrated by targeting MagFRET-1 to the cytosol and nucleus of HEK293 cells. Importantly, while MagFRET-1 is also sensitive to Ca2+, its Ca2+ affinity is sufficiently attenuated to make the sensor effectively unresponsive to the Ca2+ levels reached during signaling. Although MagFRET-1 was clearly responsive to changes in Mg2+ in permeabilized HEK293 cells, we did not observe similar changes in emission ratio for intact cells. A lack of selective Mg2+ ionophores and chelators is a fundamental problem in the Mg2+ imaging field [4] and prevented us from calibrating the sensor’s resting emission ratio by depleting and saturating the sensor in situ, as is commonly done for genetically encoded Ca2+ and Zn2+ sensors. g y g y g g The Mg2+ and Ca2+ affinity of MagFRET-1 were found to be attenuated compared to the affinities previously reported for the N-terminal domain of HsCen3 (Kd = 10–28 mM for Mg2+; Kd = 1.5–8 and 140 mM for Ca2+). The difference in metal binding affinity might be explained by the fact that the central a-helix that connects the 2nd and 3rd EF hands in HsCen3 was reduced to half its length in MagFRET-1, possibly further destabilizing the N- terminal domain, resulting in a net decrease in Mg2+ affinity. Fortunately, in this case the attenuation yielded a sensor that is sensitive to physiologically relevant Mg2+ concentrations, and insensitive to normal cytosolic Ca2+ concentrations. The limited number of metal binding domain mutations that were explored in this study revealed that mutations in the first EF hand typically result in strongly attenuated metal binding affinities. Discussion A more subtle, 4–6 fold attenuation of metal binding affinity was obtained after introduction of positively charged amino acids in the 2nd EF hand. These effects could be due to direct allosteric coupling between the two EF hands in the metal-bound state, but alternatively could also result from further stabilization of the molten globule state. An important goal is to develop sensor variants that are less sensitive to Ca2+ yet retain affinity for Mg2+, as this would allow targeting to organelles known to have much higher resting levels of Ca2+. The similar coordination chemistries of Ca2+ and Mg2+ make rational design of such variants challenging, although mutations in an EF-hand-like protein have been reported that decreased the Ca2+ affinity 100-fold while simultaneously doubling the Mg2+ affinity [58]. Further optimi- zation of metal binding affinity and specificity, but also the sensor’s dynamic range, may benefit from directed evolution approaches Despite the intrinsically large conformational change associated with protein folding, surprisingly few examples of FRET sensors exist where ligand-induced folding of a partially unfolded receptor domain is employed in FRET sensor design. Most FRET sensors developed so far either use ligand binding domains that are known to undergo significant conformational changes upon ligand binding (e.g. the periplasmic binding proteins [55]), or receptor domains that undergo a ligand-induced interaction with another peptide/protein domain (e.g. Cameleons [32]). Ligand-induced folding is believed to occur for intrinsically-disordered proteins, which may account for 35–51% of all eukaryotic protein domains Table 1. Sensor properties of the different MagFRET variants. Table 1. Sensor properties of the different MagFRET variants. Table 1. Sensor properties of the different MagFRET variants. Variant 1st EF-h and sequence1 2nd EF-hand sequence1 Kd Mg2+/mM ± SE2 D.R. Mg2+ binding event3 Kd,1 Ca2+/mM ± SE2 D.R. 1st Ca2+ binding event3 MagFRET-1 DTDKDEAIDYHE DREATGKITFED 0.1560.02 49% 1063.7 19% MagFRET-2 DTDKDDAIDYHE DREATGKITFED 0.3560.03 33% 1569.8 3.1% MagFRET-3 ATDKDEAIDYHE DREATGKITFED 9.260.7 58% 45006243 66% MagFRET-4 DTEKDEAIDYHE DREATGKITFED 8.560.5 62% 45006311 64% MagFRET-5 DTDKDEDIDYHE DREATGKITFED 7.460.5 74% 16006116 49% MagFRET-6 DTDKEEEIDYHE DREATGKITFED 1560.8 50% 79006786 55% MagFRET-7 DTDKDEAIDYHE KREATGKITFED 0.7860.04 38% 5765 23% MagFRET-8 DTDKDEAIDYHE DREATKKITFED 0.8960.06 56% 3665 25% 1Mutations introduced in the first or second 12-residue metal binding loops of HsCen3 are indicated in bold and are underlined. 2The dissociation constant (Kd) for each variant’s Mg2+ and first Ca2+ binding event is indicated, together with the standard error (SE). A FRET Sensor for Intracellular Mg2+ Imaging domains’ properties due to their sensitivity to pH or [Cl2] and confirm that MagFRET-1 is capable of responding to changes in intracellular Mg2+ levels. [56]. In addition, ligand binding domains can be intentionally destabilized to turn conformationally silent ligand binding domains into attractive input domains for FRET sensor design [57]. This work revealed that ligand-induced folding of intrinsi- cally-disordered proteins is an attractive mechanism for FRET sensor design. However, it also identified a potential disadvantage as we observed that the conformation and thus the amount of energy transfer of the ligand-free state is sensitive to ionic strength. Although this effect is most apparent below physiologically relevant salt concentrations, it is important to be aware of this phenomenon and use these sensors under conditions of constant ionic strength or use appropriate control sensors that have a strongly attenuated Mg2+ affinity, such as e.g. MagFRET-6. In situ characterization of MagFRET-1 in HEK293 cells To assess the sensor properties of MagFRET-1 in situ, CMV vectors were constructed to allow transient expression of MagFRET-1 in HEK293 cells. Fluorescence microscopy images revealed homogeneous expression of the sensor in the cytosol (Figure 4A, B) and Western blot analysis showed a single band corresponding to the full-length protein (Figure S6). In addition, transfection of cells with a construct containing three repeats of the nuclear localization sequence PKKKRKV [46,47] at the C- terminus of the sensor protein (MagFRET-1-NLS), resulted in a clear cyan and yellow emission in the nucleus (Figure 4C, D), which demonstrates the ability to target MagFRET-1 to a specific location in the cell. Although the Kd of MagFRET-1 for Ca2+ determined in vitro is an order of magnitude higher than the maximum Ca2+ concentration that is typically observed in the cytosol during signaling, it was still important to verify that the MagFRET-1 sensor does not respond to stimuli that are known to transiently induce increases in cytosolic Ca2+ concentrations. Ca2+ signaling in HEK293 cells was activated via addition of 50 mM of the protease activated receptor-1 (PAR-1) agonist peptide [48,49]. No changes in the emission ratio of MagFRET-1 were observed after addition of this stimulant (Figure 4E). Cells loaded with the synthetic Ca2+ dye Oregon Green-BAPTA did show a transient increase in fluorescence upon addition of PAR-1 agonist peptide, confirming that the expected increase in intracellular Ca2+ concentration was induced under these conditions (Figure 4F). Similar results were obtained with ATP, another commonly used stimulant for Ca2+ signaling [50–52]. Addition of 50 mM ATP did not affect the emission ratio of MagFRET-1 in HEK293 cells December 2013 \| Volume 8 \| Issue 12 \| e82009 PLOS ONE \| www.plosone.org 5 A FRET Sensor for Intracellular Mg2+ Imaging stant (Kd) for each variant’s Mg2+ and first Ca2+ binding event is indicated, together with the standard error (SE). i (D R ) i d fi d h diff i i i i b h b d d f ll l b d f di 3A binding event’s dynamic range (D.R.) is defined as the difference in emission ratio between the unbound and fully metal bound form divided by the emission ratio in the unbound form, multiplied by 100%. doi:10 1371/journal pone 0082009 t001 ns introduced in the first or second 12-residue metal binding loops of HsCen3 are indicated in bold and are underlined. i ti t t (K ) f h i t’ M2+ d fi t C2+ bi di t i i di t d t th ith th t d d (SE) Discussion 3A binding event’s dynamic range (D.R.) is defined as the difference in emission ratio between the unbound and fully metal bound form divided by the emission ratio in the unbound form, multiplied by 100%. doi:10.1371/journal.pone.0082009.t001 2The dissociation constant (Kd) for each variant’s Mg2+ and first Ca2+ binding event is indicated, together with the standard error (SE). 3A binding event’s dynamic range (D.R.) is defined as the difference in emission ratio between the unbound and fully metal bound form divided by the emission ratio in the unbound form, multiplied by 100%. doi:10.1371/journal.pone.0082009.t001 December 2013 \| Volume 8 \| Issue 12 \| e82009 PLOS ONE \| www.plosone.org PLOS ONE \| www.plosone.org 6 A FRET Sensor for Intracellular Mg2+ Imaging Figure 3. Mg2+/Ca2+ titrations of MagFRET variants with mutations in metal binding site MagFRET variants as a function of the Mg2+ (black circles) or Ca2+ (grey triangles) concentration. So event, while grey traces indicate a fit to either double Ca2+-binding events (A, B, G, H) or a single Ca2+- summarized in Table 1. Measurements were performed in 150 mM Hepes (pH 7.1), 100 mM NaCl an protein. doi:10.1371/journal.pone.0082009.g003 Figure 3. Mg2+/Ca2+ titrations of MagFRET variants with mutations in metal binding sites. (A–H) Emission ratio (Citrine to Cerulean) of MagFRET variants as a function of the Mg2+ (black circles) or Ca2+ (grey triangles) concentration. Solid black traces indicate a fit to a Mg2+-binding event, while grey traces indicate a fit to either double Ca2+-binding events (A, B, G, H) or a single Ca2+-binding event (C–F). Results of the titrations are summarized in Table 1. Measurements were performed in 150 mM Hepes (pH 7.1), 100 mM NaCl and 10% (v/v) glycerol and 0.2 (A) or 1 (B–H) mM protein. p doi:10.1371/journal.pone.0082009.g003 doi:10.1371/journal.pone.0082009.g003 similar to the ones that were recently applied to develop new color variants of the Ca2+ sensor GECO [34,59]. HEK 293 cells. This result was expected based on the affinity of MagFRET-1 that was determined in vitro (Kd = 10 mM) and previous reports that show that bulk cytosolic Ca2+ concentrations typically reach a maximum of 1 mM during signaling [45]. In contrast, synthetic Mg2+ dyes with similar affinity to Ca2+ as MagFRET-1 have been reported to respond to Ca2+. Discussion This may be explained by the fact that MagFRET-1 is exclusively localized in In situ characterization of MagFRET-1 in HEK293 cells revealed that the sensor is readily expressed in the cytosol and can be targeted to the nucleus. Two ligands that are known to induce Ca2+ signaling in cells, PAR-1 agonist peptide and ATP, did not affect the emission ratio of the MagFRET-1 sensor in December 2013 \| Volume 8 \| Issue 12 \| e82009 PLOS ONE \| www.plosone.org 7 A FRET Sensor for Intracellular Mg2+ Imaging Figure 4. In situ characterization of MagFRET-1 in HEK293 cells. (A–D) Confocal fluorescence microscopy images showing HEK293 cells expressing MagFRET-1 (A, B) and MagFRET-1-NLS (C, D) showing Cerulean (A,C) or Citrine emission (B, D). (E, F) Investigation of MagFRET-1’s in situ Ca2+ sensitivity. (E) Emission ratio over time of intact HEK293 cells expressing MagFRET-1 measured by widefield fluorescence microscopy. At t = 120 s, 50 mM of PAR-1 agonist peptide was added to activate Ca2+ signaling. (F) To confirm Ca2+ signaling took place in stimulated cells, the fluorescence intensity of intact HEK293 cells loaded with Ca2+-dye Oregon Green–BAPTA was followed. At t = 120 s, 50 mM of PAR-1 agonist peptide was added to activate Ca2+ signaling, and at t = 240 s, 20 mM A23187 was added. In E and F, each trace represents the response of an individual cell, with ratio (E) or intensity (F) normalized to the value at t = 0 s. (G, H) Response of MagFRET-1 expressed in permeabilized HEK293 cells to changes in [Mg2+]. MagFRET-1 emission ratio was followed over time as the concentration of MgCl2 (G) or EDTA (H) was increased, as indicated on the panels. (I, J) Response of negative control construct Cerulean-linker-Citrine expressed in permeabilized HEK293 cells to changes in [Mg2+]. To maintain an isotonic solution, the increase in Cl2 concentration due to addition of MgCl2 was compensated for by reducing the KCl concentration in the buffer. Prior to imaging, cells were permeabilized using 10 mg/mL digitonin. Traces in G to J represent averages of at least 9 cells, error bars indicate SEM, ratios were normalized to the emission ratio at t = 0. doi:10.1371/journal.pone.0082009.g004 Figure 4. In situ characterization of MagFRET-1 in HEK293 cells. (A–D) Confocal fluorescence microscopy images showing HEK293 cells expressing MagFRET-1 (A, B) and MagFRET-1-NLS (C, D) showing Cerulean (A,C) or Citrine emission (B, D). Discussion (E, F) Investigation of MagFRET-1’s in situ Ca2+ sensitivity. (E) Emission ratio over time of intact HEK293 cells expressing MagFRET-1 measured by widefield fluorescence microscopy. At t = 120 s, 50 mM of PAR-1 agonist peptide was added to activate Ca2+ signaling. (F) To confirm Ca2+ signaling took place in stimulated cells, the fluorescence intensity of intact HEK293 cells loaded with Ca2+-dye Oregon Green–BAPTA was followed. At t = 120 s, 50 mM of PAR-1 agonist peptide was added to activate Ca2+ signaling, and at t = 240 s, 20 mM A23187 was added. In E and F, each trace represents the response of an individual cell, with ratio (E) or intensity (F) normalized to the value at t = 0 s. (G, H) Response of MagFRET-1 expressed in permeabilized HEK293 cells to changes in [Mg2+]. MagFRET-1 emission ratio was followed over time as the concentration of MgCl2 (G) or EDTA (H) was increased, as indicated on the panels. (I, J) Response of negative control construct Cerulean-linker-Citrine expressed in permeabilized HEK293 cells to changes in [Mg2+]. To maintain an isotonic solution, the increase in Cl2 concentration due to addition of MgCl2 was compensated for by reducing the KCl concentration in the buffer. Prior to imaging, cells were permeabilized using 10 mg/mL digitonin. Traces in G to J represent averages of at least 9 cells, error bars indicate SEM, ratios were normalized to the emission ratio at t = 0. doi:10.1371/journal.pone.0082009.g004 substantial and physiologically relevant fluctuations in Mg2+ concentration could still occur locally, e.g. at the plasma membrane near Mg2+-specific ion channels. Although Mag- FRET-1 responded to changes in Mg2+ concentration in the order of seconds both in vitro and in cells, protein-based sensors often display slower kinetics than small molecule sensors, so that MagFRET-1 might fail to respond to extremely fast Mg2+ transients, should they occur. In addition, overall Mg2+ levels could change over longer periods of time, e.g. as a function of the cell cycle. These changes may be more reliably monitored using lifetime imaging, which might also be the preferred method to allow quantification of intracellular Mg2+ concentrations. Al- though we confirmed that MagFRET-1 is responsive to changes in Mg2+ concentration in permeabilized cells, we cannot completely the cytosol, whereas synthetic dyes sometimes partially mislocalize to Ca2+-rich organelles such as the ER or even leak into the external buffer [60]. Method S1 Western blotting. (PDF) Table S1 Primers used for mutagenesis of HsCen3. (PDF) Acknowledgments The authors are grateful for the use of the fluorescence confocal microscope in the Laboratory of Soft Tissue Biomechanics & Engineering (Eindhoven University of Technology) and appreciate the technical assistance provided by Dr. Ir. M. van Turnhout. The authors would like to thank Dr. Kees Jalink (Netherlands Cancer Institute, Amsterdam), Dr. Stan van de Graaf (Academic Medical Centre), Dr. Sjoerd Verkaart, Dr. Jenny van der Wijst, Prof. Joost Hoenderop and Prof. Rene´ Bindels (Radboud University Nijmegen) for fruitful discussions. Figure S5 Effect of ionic strength on MagFRET-1. (A) Emission ratio of MagFRET-1 at increasing concentrations of ammonium acetate or NaCl in a buffer with low ionic strength. (B) Emission ratio of MagFRET-1 as a function of Mg2+ concentra- tion in a buffer with low ionic strength. The low ionic strength buffer used in (A, B) was 20 mM Hepes (pH 7.1), 10 mM NaCl, Table S1 Primers used for mutagenesis of HsCen3. (PDF) Figure S4 Effect of pH on MagFRET-1. To check for pH sensitivity, the MagFRET-1 emission ratio was followed as a function of Mg2+ (A, B) and Ca2+ (C, D) concentration, at pH 6 (A, C) and pH 8 (B, D). Fitting of the data revealed a MagFRET-1 Kd for Mg2+ of 230635 mM at pH 6 and 99618 mM at pH 8. The sensor’s Kd for Ca2+ (first binding event) at pH = 6 was found to be 5.661.7 mM, while at pH 8 it was 5.961.9 mM. Buffers used were 150 mM MES (pH 6), 100 mM NaCl and 10% glycerol for pH 6 and 150 mM Tris (pH 8), 100 mM NaCl and 10% glycerol for pH 8. (TIF) Discussion Surprisingly, HEK293 cells expressing MagFRET-1 did not respond to procedures that were previously reported to affect the intracellular Mg2+ concentration. A possible explanation is that the free concentration of Mg2+ in the cytosol is tightly buffered and controlled and not easily changed by external stimuli. The free concentration of Mg2+ in the cytosol is at least 1000-fold higher than that of Ca2+ and 106-fold higher than that of Zn2+. For this reason and because Mg2+ is essential to such a wide variety of biological processes, it would not be surprising that manipulation of intracellular free Mg2+ is much more difficult than that of other metals. Nonetheless, it is conceivable that while the overall free Mg2+ concentration in cells is relatively constant, December 2013 \| Volume 8 \| Issue 12 \| e82009 PLOS ONE \| www.plosone.org 8 A FRET Sensor for Intracellular Mg2+ Imaging 10% (v/v) glycerol. Fitting of the data using a single binding event revealed a Kd for Mg2+ of 231610 mM. (TIF) rule out that for some unknown reason MagFRET-1 is less responsive in intact cells. The potential and limitations of the MagFRET sensors for intracellular Mg2+ imaging therefore remain to be further established. rule out that for some unknown reason MagFRET-1 is less responsive in intact cells. The potential and limitations of the MagFRET sensors for intracellular Mg2+ imaging therefore remain to be further established. Figure S6 Western blot analysis of MagFRET-1 express- ing HEK293 cells. A molecular weight marker (Precision Plus Protein Standards, Bio-Rad) was loaded in the left-hand lane. Lane 1 displays the lysate of HEK293 cells transfected with a vector encoding for MagFRET-1 under control of a CMV promoter. The blotting membrane was incubated with mouse anti- GFP (Ab3277, Abcam), followed by HRP-functionalized goat anti- mouse antibody (Dako). The calculated molecular weight for MagFRET-1 is 62 kDa. (TIF) Author Contributions Conceived and designed the experiments: LHL JLV MM. Performed the experiments: LHL JLV JO SJAA. Analyzed the data: LHL JLV JO SJAA MM. Wrote the paper: LHL JLV MM. Conceived and designed the experiments: LHL JLV MM. Performed the experiments: LHL JLV JO SJAA. Analyzed the data: LHL JLV JO SJAA MM. Wrote the paper: LHL JLV MM. 13. Chaudhary DP, Sharma R, Bansal DD (2010) Implications of magnesium deficiency in type 2 diabetes: a review. Biol Trace Elem Res 134: 119–129. 12. Killilea DW, Ames BN (2008) Magnesium deficiency accelerates cellular senescence in cultured human fibroblasts. Proc Natl Acad Sci U S A 105: 5768– 5773. 6. Gunther T (2006) Concentration, compartmentation and metabolic function of intracellular free Mg2+. Magnes Res 19: 225–236. Supporting Information Figure S1 Nucleotide sequence of bacterial expression vector pET28a-MagFRET-1 ORF. The DNA sequence is shown in lowercase, with the single letter amino acid code shown beneath each codon in uppercase. The His-tag is highlighted in bright green, the thrombin cleavage site in pink, Cerulean in turquoise, HsCen3 in red and Citrine in yellow. The two EF-hand motifs are underlined in white. (PDF) Figure S7 Investigation of MagFRET-1 response to elevated cytosolic Ca2+ induced by ATP. (A) Emission ratio over time of intact HEK293 cells expressing MagFRET-1 measured by widefield fluorescence microscopy. At t = 104 s, 50 mM ATP was added to activate Ca2+ signaling. (B) To confirm Ca2+ signaling took place in stimulated cells, the fluorescence intensity of intact HEK293 cells loaded with Ca2+-dye Oregon Green–BAPTA was followed. At t = 104 s, 50 mM ATP was added to activate Ca2+ signaling, and at t = 226 s, 20 mM of the Ca2+ ionophore A23187 was added. In A and B, each trace represents the response of an individual cell, with ratio (A) or intensity (B) normalized to the value at t = 0 s. Figure S2 Nucleotide sequence of mammalian expres- sion vector pCMV-MagFRET-1 ORF. The DNA sequence is shown in lowercase, with the single letter amino acid code shown beneath each codon in uppercase. Cerulean is highlighted in turquoise, HsCen3 in red and Citrine in yellow. The two EF-hand motifs are underlined in white. (PDF) Figure S3 Nucleotide sequence of mammalian expres- sion vector pCMV-MagFRET-1-NLS ORF. The DNA sequence is shown in lowercase, with the single letter amino acid code shown beneath each codon in uppercase. Cerulean is highlighted in turquoise, HsCen3 in red, Citrine in yellow and the three PKKKRKV repeats in grey. The two EF-hand motifs are underlined in white. (PDF) A FRET Sensor for Intracellular Mg2+ Imaging PLoS One 6: e23684. 51. Nagai T, Yamada S, Tominaga T, Ichikawa M, Miyawaki A (2004) Expanded dynamic range of fluorescent indicators for Ca2+ by circularly permuted yellow fluorescent proteins. Proc Natl Acad Sci U S A 101: 10554–10559. 28. Hurley TW, Ryan MP, Brinck RW (1992) Changes of cytosolic Ca2+ interfere with measurements of cytosolic Mg2+ using mag-fura-2. Am J Physiol Cell Physiol 263: C300–307. 52. Qi Z, Murase K, Obata S, Sokabe M (2000) Extracellular ATP-dependent activation of plasma membrane Ca2+ pump in HEK-293 cells. Br J Pharmacol 131: 370–374. 29. Zhou H, Clapham DE (2009) Mammalian MagT1 and TUSC3 are required for cellular magnesium uptake and vertebrate embryonic development. Proc Natl Acad Sci U S A 106: 15750–15755. 53. Fonseca CP, Montezinho LP, Baltazar G, Layden B, Freitas DM, et al. (2000) Li+ influx and binding, and Li+/Mg2+ competition in bovine chromaffin cell suspensions as studied by 7Li NMR and fluorescence spectroscopy. Met Based Drugs 7: 357–364. 30. Xie J, Sun B, Du J, Yang W, Chen HC, et al. (2011) Phosphatidylinositol 4,5- bisphosphate (PIP(2)) controls magnesium gatekeeper TRPM6 activity. Sci Rep 1: 146. 54. Schweigel M, Park HS, Etschmann B, Martens H (2006) Characterization of the Na+-dependent Mg2+ transport in sheep ruminal epithelial cells. Am J Physiol Gastrointest Liver Physiol 290: G56–65. 31. Palmer AE, Qin Y, Park JG, McCombs JE (2011) Design and application of genetically encoded biosensors. Trends Biotechnol 29: 144–152. 32. Miyawaki A, Llopis J, Heim R, Michael McCaffery J, Adams JA, et al. (1997) Fluorescent indicators for Ca2+ based on green fluorescent proteins and calmodulin. Nature 388: 882–887. y 55. Okumoto S, Looger LL, Micheva KD, Reimer RJ, Smith SJ, et al. (2005) Detection of glutamate release from neurons by genetically encoded surface- displayed FRET nanosensors. Proc Natl Acad Sci U S A 102: 8740–8745. 33. Palmer AE, Giacomello M, Kortemme T, Hires SA, Lev-Ram V, et al. (2006) Ca2+ indicators based on computationally redesigned calmodulin-peptide pairs. Chem Biol 13: 521–530. p y 56. Dunker AK, Brown CJ, Lawson JD, Iakoucheva LM, Obradovic Z (2002) Intrinsic disorder and protein function. Biochemistry 41: 6573–6582. p y 57. Kohn JE, Plaxco KW (2005) Engineering a signal transduction mechanism for protein-based biosensors. Proc Natl Acad Sci U S A 102: 10841–10845. 34. Zhao YX, Araki S, Jiahui WH, Teramoto T, Chang YF, et al. (2011) An expanded palette of genetically encoded Ca2+ indicators. A FRET Sensor for Intracellular Mg2+ Imaging ChemBioChem 11: 2264–2267. 21. Kim HM, Yang PR, Seo MS, Yi J-S, Hong JH, et al. (2007) Magnesium ion selective two-photon fluorescent probe based on a benzo[h]chromene derivative for in vivo imaging. J Org Chem 72: 2088–2096. 44. Griesbeck O, Baird GS, Campbell RE, Zacharias DA, Tsien RY (2001) Reducing the environmental sensitivity of yellow fluorescent protein. J Biol Chem 276: 29188–29194. 22. Hwan Myung K, Cheol J, Bo Ra K, Soon-Young J, Jin Hee H, et al. (2007) Environment-sensitive two-photon probe for intracellular free magnesium ions in live tissue. Angew Chem Int Ed Engl 46: 3460–3463. 45. Bootman MD (2012) Calcium signaling. Cold Spring Harb Perspect Biol 4: a011171. in live tissue. Angew Chem Int Ed Engl 46: 3460–3463. 23. Komatsu H, Miki T, Citterio D, Kubota T, Shindo Y, et al. (2005) Single molecular multianalyte (Ca2+, Mg2+) fluorescent probe and applications to bioimaging. J Am Chem Soc 127: 10798–10799. 46. Kalderon D, Roberts BL, Richardson WD, Smith AE (1984) A short amino acid sequence able to specify nuclear location. Cell 39: 499–509. 47. Fischer-Fantuzzi L, Vesco C (1988) Cell-dependent efficiency of reiterated nuclear signals in a mutant simian virus 40 oncoprotein targeted to the nucleus. Mol Cell Biol 8: 5495–5503. 24. Komatsu H, Iwasawa N, Citterio D, Suzuki Y, Kubota T, et al. (2004) Design and synthesis of highly sensitive and selective fluorescein-derived magnesium fluorescent probes and application to intracellular 3D Mg2+ imaging. J Am Chem Soc 126: 16353–16360. 48. Jiang T, Danilo P Jr, Steinberg SF (1998) The thrombin receptor elevates intracellular calcium in adult rat ventricular myocytes. J Mol Cell Cardiol 30: 2193–2199. 25. Raju B, Murphy E, Levy LA, Hall RD, London RE (1989) A fluorescent indicator for measuring cytosolic free magnesium. Am J Physiol Cell Physiol 256: C540–548. 49. Hui KY, Jakubowski JA, Wyss VL, Angleton EL (1992) Minimal sequence requirement of thrombin receptor agonist peptide. Biochem Biophys Res Commun 184: 790–796. 26. Paredes RM, Etzler JC, Watts LT, Zheng W, Lechleiter JD (2008) Chemical calcium indicators. Methods 46: 143–151. 50. Isshiki M, Ando J, Korenaga R, Kogo H, Fujimoto T, et al. (1998) Endothelial Ca2+ waves preferentially originate at specific loci in caveolin-rich cell edges. Proc Natl Acad Sci U S A 95: 5009–5014. 27. Shindo Y, Fujii T, Komatsu H, Citterio D, Hotta K, et al. (2011) Newly developed Mg2+-selective fluorescent probe enables visualization of Mg2+ dynamics in mitochondria. A FRET Sensor for Intracellular Mg2+ Imaging 14. Bo S, Pisu E (2008) Role of dietary magnesium in cardiovascular disease prevention, insulin sensitivity and diabetes. Curr Opin Lipidol 19: 50–56. 37. Dittmer PJ, Miranda JG, Gorski JA, Palmer AE (2009) Genetically encoded sensors to elucidate spatial distribution of cellular zinc. J Biol Chem 284: 16289– 16297. prevention, insulin sensitivity and diabetes. Curr Opin Lipidol 19 p y p p 15. Rude RK, Gruber HE (2004) Magnesium deficiency and osteoporosis: animal and human observations. J Nutr Biochem 15: 710–716. 38. Miranda JG, Weaver AL, Qin Y, Park JG, Stoddard CI, et al. (2012) New alternately colored FRET sensors for simultaneous monitoring of Zn2+ in multiple cellular locations. PLoS One 7: e49371. and human observations. J Nutr Biochem 15: 710–7 16. Li FY, Chaigne-Delalande B, Kanellopoulou C, Davis JC, Matthews HF, et al. (2011) Second messenger role for Mg2+ revealed by human T-cell immunode- ficiency. Nature 475: 471–476. 39. Qiao W, Mooney M, Bird AJ, Winge DR, Eide DJ (2006) Zinc binding to a regulatory zinc-sensing domain monitored in vivo by using FRET. Proc Natl Acad Sci U S A 103: 8674–8679. 17. Trapani V, Farruggia G, Marraccini C, Iotti S, Cittadini A, et al. (2010) Intracellular magnesium detection: imaging a brighter future. Analyst 135: 1855–1866. 40. Wegner SV, Arslan H, Sunbul M, Yin J, He C (2010) Dynamic copper(I) imaging in mammalian cells with a genetically encoded fluorescent copper(I) sensor. J Am Chem Soc 132: 2567–2569. 18. Cohen SM, Burt CT (1977) 31P nuclear magnetic relaxation studies of phosphocreatine in intact muscle: determination of intracellular free magnesium. Proc Natl Acad Sci U S A 74: 4271–4275. 41. Koay MS, Janssen BM, Merkx M (2013) Tuning the metal binding site specificity of a fluorescent sensor protein: from copper to zinc and back. Dalton Trans 42: 3230–3232. 19. Resnick LM, Gupta RK, Laragh JH (1984) Intracellular free magnesium in erythrocytes of essential hypertension: relation to blood pressure and serum divalent cations. Proc Natl Acad Sci U S A 81: 6511–6515. 42. Cox JA, Tirone F, Durussel I, Firanescu C, Blouquit Y, et al. (2005) Calcium and magnesium binding to human centrin 3 and interaction with target peptides. Biochemistry 44: 840–850. 20. Gunther T (2007) Total and free Mg2+ contents in erythrocytes: a simple but still undisclosed cell model. Magnes Res 20: 161–167. y 43. Golynskiy MV, Rurup WF, Merkx M (2010) Antibody detection by using a FRET-based protein conformational switch. References 1. Shi J, Krishnamoorthy G, Yang Y, Hu L, Chaturvedi N, et al. (2002) Mechanism of magnesium activation of calcium-activated potassium channels. Nature 418: 876–880. 7. Corkey BE, Duszynski J, Rich TL, Matschinsky B, Williamson JR (1986) Regulation of free and bound magnesium in rat hepatocytes and isolated mitochondria. J Biol Chem 261: 2567–2574. J 8. Maguire ME (2006) The structure of CorA: a Mg2+-selective channel. Curr Opin Struct Biol 16: 432–438. 2. Nadler MJ, Hermosura MC, Inabe K, Perraud AL, Zhu Q, et al. (2001) LTRPC7 is a Mg.ATP-regulated divalent cation channel required for cell viability. Nature 411: 590–595. 9. Schlingmann KP, Waldegger S, Konrad M, Chubanov V, Gudermann T (2007) TRPM6 and TRPM7-Gatekeepers of human magnesium metabolism. Biochim Biophys Acta, Molecular Basis of Disease 1772: 813–821. 3. Romani AM (2007) Magnesium homeostasis in mammalian cells. Front Biosci 12: 308–331. 10. Hattori M, Tanaka Y, Fukai S, Ishitani R, Nureki O (2007) Crystal structure of the MgtE Mg2+ transporter. Nature 448: 1072–1075. 4. Grubbs RD (2002) Intracellular magnesium and magnesium buffering. Biometals 15: 251–259. 11. Konrad M, Weber S (2003) Recent advances in molecular genetics of hereditary magnesium-losing disorders. J Am Soc Nephrol 14: 249–260. 5. Rutter GA, Osbaldeston NJ, McCormack JG, Denton RM (1990) Measurement of matrix free Mg2+ concentration in rat heart mitochondria by using entrapped fluorescent probes. Biochem J 271: 627–634. 12. Killilea DW, Ames BN (2008) Magnesium deficiency accelerates cellular senescence in cultured human fibroblasts. Proc Natl Acad Sci U S A 105: 5768– 5773. 6. Gunther T (2006) Concentration, compartmentation and metabolic function of intracellular free Mg2+. Magnes Res 19: 225–236. 13. Chaudhary DP, Sharma R, Bansal DD (2010) Implications of magnesium deficiency in type 2 diabetes: a review. Biol Trace Elem Res 134: 119–129. PLOS ONE \| www.plosone.org 9 December 2013 \| Volume 8 \| Issue 12 \| e82009 A FRET Sensor for Intracellular Mg2+ Imaging A FRET Sensor for Intracellular Mg2+ Imaging Science 333: 1888– 1891. protein-based biosensors. Proc Natl Acad Sci U S A 102: 10841–1084 58. Wang W, Barnabei MS, Asp ML, Heinis FI, Arden E, et al. (2013) Noncanonical EF-hand motif strategically delays Ca2+ buffering to enhance cardiac performance. Nat Med 19: 305–312. 35. Tian L, Hires SA, Mao T, Huber D, Chiappe ME, et al. (2009) Imaging neural activity in worms, flies and mice with improved GCaMP calcium indicators. Nat Methods 6: 875–U113. p 59. Lindenburg L, Merkx M (2012) Colorful calcium sensors. ChemBioChem 13: 349–351. 36. Vinkenborg JL, Nicolson TJ, Bellomo EA, Koay MS, Rutter GA, et al. (2009) Genetically encoded FRET sensors to monitor intracellular Zn2+ homeostasis. Nat Methods 6: 737–740. 60. Trapani V, Schweigel-Rontgen M, Cittadini A, Wolf FI (2012) Intracellular magnesium detection by fluorescent indicators. Methods Enzymol 505: 421– 444. December 2013 \| Volume 8 \| Issue 12 \| e82009 PLOS ONE \| www.plosone.org 10
https://openalex.org/W132102926	https://journals.openedition.org/pontourbe/pdf/1102	Portuguese	null	Foguetes e fitas coloridas: encontros nos movimentos pela cidade	Ponto Urbe	2,012	cc-by	3,183	11 \| 2012 Ponto Urbe 11 Edição impressa ç p Data de publição: 1 dezembro 2012 Foguetes e fitas coloridas: encontros nos movimentos pela cidade Martina Ahlert Editora Núcleo de Antropologia Urbana da Universidade de São Paulo Refêrencia eletrónica Refêrencia eletrónica Martina Ahlert, « Foguetes e fitas coloridas: encontros nos movimentos pela cidade », Ponto Urbe [Online], 11 \| 2012, posto online no dia 01 dezembro 2012, consultado o 15 setembro 2020. URL : http://journals.openedition.org/pontourbe/1102 Este documento foi criado de forma automática no dia 15 setembro 2020. This work is licensed under a Creative Commons Attribution 4.0 International License. Foguetes e fitas coloridas: encontros nos movimentos pela cidade 1 Foguetes e fitas coloridas: encontros nos movimentos pela cidade Martina Ahlert 1 Escrevo este relato etnográfico a partir do encontro de uma procissão em homenagem a Todos os Santos e um carro de som que festejava a vitória de Dilma Rousseff nas eleições presidenciais, em primeiro de novembro de 2009. Este evento aconteceu nos primeiros meses da pesquisa de campo do meu doutorado, em uma das maiores cidades do interior do Maranhão, conhecida como Codó. Localizada na rota entre as cidades de São Luís (MA) e Teresina (PI), Codó possui cerca de 118 mil habitantes e é afamada por seus grandes feiticeiros, seus muitos pais e mães de santo e como cidade berço do Terecô, religião afro-brasileira de provável matriz banto, com elementos jeje e nagô, também referenciada por Tambor da Mata ou Brincadeira de Santa Bárbara (MACHADO, 1999; FERRETTI, M. 1998, 2000, 2001; BARROS, 2000, entre outros). 2 O evento de que trato é o Festejo para Todos os Santos, realizado por Dona Maria Alice, na cidade de Codó. Antes de falar especificamente sobre aquele primeiro de novembro, é importante dizer que, na cidade, são inúmeras as festas realizadas para santos, para encantados1 e para orixás, no âmbito do catolicismo popular e também nas tendas de Terecô, Umbanda e Candomblé. As festas têm variados períodos de duração, podendo ser realizadas em uma, três ou nove noites. Ainda no âmbito desta breve introdução um pouco mais genérica, cabe dizer que as festas incluem elementos semelhantes: a reza do Terço, ladainhas em português ou em latim, procissões, comes e bebes, serestas animadas por grupos musicais e ainda, no caso de acontecerem em tendas, noites inteiras de tambor. Neste último caso, o tambor é tocado (no ritmo intenso do “toque da mata”) e dançado pelos “brincantes” (como são chamadas as pessoas das religiões afro- brasileiras) e pelas entidades, até o amanhecer. 2 O evento de que trato é o Festejo para Todos os Santos, realizado por Dona Maria Alice, na cidade de Codó. Antes de falar especificamente sobre aquele primeiro de novembro, é importante dizer que, na cidade, são inúmeras as festas realizadas para santos, para encantados1 e para orixás, no âmbito do catolicismo popular e também nas tendas de Terecô, Umbanda e Candomblé. As festas têm variados períodos de duração, podendo ser realizadas em uma, três ou nove noites. Foguetes e fitas coloridas: encontros nos movimentos pela cidade Foguetes e fitas coloridas: encontros nos movimentos pela cidade Ainda no âmbito desta breve introdução um pouco mais genérica, cabe dizer que as festas incluem elementos semelhantes: a reza do Terço, ladainhas em português ou em latim, procissões, comes e bebes, serestas animadas por grupos musicais e ainda, no caso de acontecerem em tendas, noites inteiras de tambor. Neste último caso, o tambor é tocado (no ritmo intenso do “toque da mata”) e dançado pelos “brincantes” (como são chamadas as pessoas das religiões afro- brasileiras) e pelas entidades, até o amanhecer. 3 A festa de Dona Maria Alice tem nove noites, encerrando na véspera de Todos os Santos, em primeiro de novembro, dia do evento que aqui relato. Naquela noite, sai da minha casa a caminho da casa de Seu Ribinha, que promovia um Tambor de Crioula em homenagem a São Luiz Gonzaga. No meu deslocamento, quando caminhava ao lado da Igreja de São Sebastião - um dos padroeiros de Codó - encontrei uma pequena Ponto Urbe, 11 \| 2012 Foguetes e fitas coloridas: encontros nos movimentos pela cidade 2 procissão. Cerca de cinquenta pessoas seguiam, a pé, atrás de um andor carregado por dois homens. Em princípio tive dificuldade em identificar o santo sob a estrutura de madeira coberta de pano branco, só no caminhar junto à procissão pude perceber que eram diferentes santos, amarrados uns aos outros por fitas de cetim, tão coloridas como as próprias imagens – eram, evidentemente, “todos os santos” homenageados. As pessoas seguiam rezando Pai Nosso, Ave Maria e, por vezes, cantando algum hino religioso, amparadas por uma pequena banda de instrumentos de sopro e percussão. O som das rezas e das músicas era acompanhado pelos foguetes. Eu me juntei à procissão, conhecendo e conversando com algumas pessoas que participavam do evento. Aqueles, entretanto, não eram os únicos foguetes que festejavam a noite na cidade. Quando descemos da rua da Igreja, no caminho do bairro onde morava a festeira Dona Maria Alice, a procissão encontrou um carro de som de onde também provinham foguetes. Os hinos cantados pelos devotos se misturaram ao som do carro/trio elétrico, que tocava, alegremente, uma música que muito tínhamos ouvido na cidade naqueles dias: o “Arrastão do Treze”. Naquela mesma noite, em que Dona Maria Alice festejava seus santos, a cidade comemorava a vitória de Dilma Rousseff no segundo turno das eleições federais. Foguetes e fitas coloridas: encontros nos movimentos pela cidade O “Arrastão do Treze” era uma variante do “Arrastão do Quinze”, dedicado à campanha de Roseana Sarney ao governo do estado e, posteriormente, à eleição à presidência, já que os partidos das duas candidatas estavam coligados no Maranhão e, em Codó, eram apoiados pelo prefeito. Por breves instantes, tivemos uma mistura entre os sons, as pessoas e os foguetes, que anunciavam e afirmavam, cada qual na sua perspectiva, a alegria e a comemoração daquela noite. O carro de som seguiu na direção do centro da cidade. A procissão se deslocou na direção oposta, entrando no bairro Codó Novo, até a pequena travessa onde se localizava a casa da festeira. O bairro cresceu a partir dos anos oitenta, com a vinda de muitos moradores da zona rural para o local. A casa simples de Dona Maria Alice fora tomada pela festa, assim como a própria rua, que recebeu mesas, cadeiras e um palco para a seresta. A separação entre a casa – desde o pátio dos fundos, onde a comida era preparada – e a rua, divisão celebrada como metáfora para pensar o Brasil (ver, por exemplo, DA MATTA, 1997) ganhava um novo tom no festejo de Dona Maria Alice, onde suas fronteiras eram transpostas e contaminadas pela circulação dos santos e das diversas pessoas presentes. 6 Entre os devotos com os quais conversei na procissão estavam familiares de Dona Maria Alice. Com eles e com a própria festeira, fiquei sabendo que o Festejo para Todos os Santos é realizado por anos suficientes para se esquecer da conta. Disseram-me que nasceu como resposta a uma graça alcançada por meio de uma promessa. Naquele contexto era feito pela avó de Maria Alice, sendo herdado por uma das suas filhas (a mãe de Maria Alice) e assim alcançando sua atual festeira. Ninguém pareceu se importar com minha pergunta sobre a continuidade da festa, como se não fosse algo que merecesse efetiva preocupação. Promessas, festas, santos (estátuas e imagens) e encantados (para os “brincantes”) são recebidos, também em diversos outros casos, como herança familiar. Desta forma, eventos e entidades perpassam gerações. Talvez por este caráter os festejos ainda sejam momentos importantes do calendário dos parentes, que vem de outras cidades e estados para acompanhar estes eventos. 7 Quando a procissão chegou a casa de Dona Maria Alice, o andor foi depositado sobre uma mesa ao lado do palco da seresta. Foguetes e fitas coloridas: encontros nos movimentos pela cidade Ao final das músicas, Dona Maria Alice disse, com uma potencia impressionante diante do seu pequeno tamanho, “Eu sou neta de Dona Casimira e filha de Teresinha!” – remetendo às mulheres que a antecederam na organização da festa. O festejo é um encontro de família, os que moram na cidade, os que vivem distante e se deslocam para a cidade, os vivos e os antepassados. 9 Animadamente a “Turma dos Descalços” comemorou quando as mulheres, acompanhadas pela banda, cantaram a música que antecedia o início da comensalidade. O andor foi carregado para um primeiro cômodo da casa e depositado sobre uma mesa que servia de altar. Nas paredes, pintadas de colorido com “tinta em pó”, havia alguns pôsteres do Flamengo, um dos times mais populares entre os codoenses. Depositado o andor, as músicas religiosas deram lugar às profanas e alguns meninos da “Turma dos Descalços” se tornaram par de dança das mulheres – bem mais velhas que eles – que estavam cantando alegremente. Ao final das músicas, Dona Maria Alice disse, com uma potencia impressionante diante do seu pequeno tamanho, “Eu sou neta de Dona Casimira e filha de Teresinha!” – remetendo às mulheres que a antecederam na organização da festa. O festejo é um encontro de família, os que moram na cidade, os que vivem distante e se deslocam para a cidade, os vivos e os antepassados. 10 Em seguida as pessoas se dispersaram, muitas foram para a cozinha. Sobre a mesa, diversos copos estavam espalhados e uma grande panela guardava o chocolate que era servido. Entrando em uma fila, recebíamos os pedaços de bolo das mãos de uma das filhas de Maria Alice. A “Turma dos Descalços” desenvolvia diferentes estratégias para pegar bolo mais de uma vez: entravam na fila novamente, iam com ou sem boné para disfarçar. Quando a filha da festeira percebia suas tentativas, os espantava com um pedaço de papelão que antes fora parte de um foguete. Os adolescentes se divertiam, correndo por toda a casa. 10 Em seguida as pessoas se dispersaram, muitas foram para a cozinha. Sobre a mesa, diversos copos estavam espalhados e uma grande panela guardava o chocolate que era servido. Entrando em uma fila, recebíamos os pedaços de bolo das mãos de uma das filhas de Maria Alice. Foguetes e fitas coloridas: encontros nos movimentos pela cidade As pessoas que conheci caminhando me levaram para ver os bolos e o chocolate (quente, preparado com leite de coco babaçu, uma Ponto Urbe, 11 \| 2012 Foguetes e fitas coloridas: encontros nos movimentos pela cidade 3 iguaria das festas), que seria servido depois da reza. Voltando para a rua em frente a casa, eu me sentei em uma das mesas, esperando o início do Terço e da Ladainha. Logo fui cercada por um grupo de adolescentes (entre doze e dezesseis anos) que sentou à mesa comigo, me fez diversas perguntas e contou histórias. Aqueles adolescentes vinham todos os anos ao festejo, porque moravam nas redondezas do local. Vinham, segundo me contaram, para comer bolo, tomar refrigerante e chocolate. Por causa de seu objetivo claro, conheciam a ordem das rezas e das músicas cantadas até chegar o momento do bolo. 8 A partir daí, a programação ‘oficial’ do evento ficou, pra mim, em segundo plano. Com a presença dos adolescentes na minha mesa, se tornava evidente a multiplicidade de pessoas, de perspectivas e de interesses que compõe um mesmo ritual. “Espalhando lixo”, “Catando gato”, “Peru” e seus amigos tomaram minha atenção, contando sobre motos, sobre os empregos que conseguiam, as escolas onde estudavam, a falta de expectativa em relação aos estudos e o ingresso em alguma faculdade, a dificuldade em conseguir dinheiro para levar as garotas pra tomar sorvete. Perguntaram ainda se eu estava na cidade “fazendo pesquisa que observa as pessoas” – conhecimento que tinham adquirido em uma aula com a professora de história. Contaram-me que o bairro não era violento ( “a não ser que você esteja devendo pra alguém”) e que em vários locais da cidade existem grupos de rapazes como o deles e que são conhecidos como “turmas”. Eles eram a “Turma dos Descalços”. 9 Animadamente a “Turma dos Descalços” comemorou quando as mulheres, acompanhadas pela banda, cantaram a música que antecedia o início da comensalidade. O andor foi carregado para um primeiro cômodo da casa e depositado sobre uma mesa que servia de altar. Nas paredes, pintadas de colorido com “tinta em pó”, havia alguns pôsteres do Flamengo, um dos times mais populares entre os codoenses. Depositado o andor, as músicas religiosas deram lugar às profanas e alguns meninos da “Turma dos Descalços” se tornaram par de dança das mulheres – bem mais velhas que eles – que estavam cantando alegremente. Foguetes e fitas coloridas: encontros nos movimentos pela cidade Conta ainda que dança com ele desde que era nova, ainda quando a família dos dois morava no interior do município. Tanto ela, com seu festejo para Todos os Santos, quanto seu Ribinha, cuja família festeja São Luiz Gonzaga desde 1930, trouxeram suas festas de povoados do interior para a cidade, quando, por diferentes motivos, tiveram que deixar o campo2. Antigamente, naquela mesma noite, as mulheres da festa de Dona Maria Alice, ao início da seresta, passavam na casa de Seu Ribinha, para dançar algumas toadas de Tambor. Depois da despedida, um dos sobrinhos da festeira se dispôs a me ajudar a encontrar uma “moto” (ou um mototaxista) para meu deslocamento. Embora andasse a pé durante todo o tempo, não era considerado seguro que mulheres andassem sozinhas à noite pela cidade. Minha carona foi com um dos adolescentes da “Turma dos Descalços”, na moto que seu padrasto tinha “tirado” para ele, já que ainda não tinha idade para dirigir. A seresta terminaria só ao amanhecer, assim como o Tambor de Crioula na casa de Seu Ribinha. 12 Eventos e rituais têm uma característica importante no trabalho de campo em antropologia, eles condensam muitas informações sobre os grupos que os mantém. Neste sentido, a relação entre sagrado e profano, as festas que são tanto públicas quanto domésticas, os santos tomando espaço nas casas, a presença intergeracional, elementos que me chamaram atenção no festejo para Todos os Santos, foram recorrentes durante toda a pesquisa de campo que realizo em Codó. No caminhar das pessoas e dos santos em procissão, na vinda dos parentes para o festejo, nos antepassados relembrados, os movimentos se dão tanto no espaço quanto no tempo (INGOLD, 2005). Entre uma mulher que ocupa, pela primeira vez, um importante posto político no país, e as mulheres ancestrais de Dona Maria Alice, há um compartilhar do espaço-tempo, nos diferentes festejos daquele primeiro de novembro. Foguetes e fitas coloridas: encontros nos movimentos pela cidade A “Turma dos Descalços” desenvolvia diferentes estratégias para pegar bolo mais de uma vez: entravam na fila novamente, iam com ou sem boné para disfarçar. Quando a filha da festeira percebia suas tentativas, os espantava com um pedaço de papelão que antes fora parte de um foguete. Os adolescentes se divertiam, correndo por toda a casa. 11 As pessoas voltaram para a rua onde sentaram novamente nas mesas e cadeiras disponíveis. Como a reza acabara, a venda de bebida alcoólica estava liberada e em pouco tempo começaria a seresta. A banda da seresta era composta por um tecladista e um cantor, mas, não raro, as bandas são acompanhadas de dançarinos e dançarinas. Recebi convite para ficar na seresta, mas já tinha combinado de encontrar alguns amigos no Tambor de Crioula de Seu Ribinha. Quando fui me despedir, usando desta informação, Dona Maria Alice me contou que ela mesma dança na casa de Seu Ribinha. Ponto Urbe, 11 \| 2012 Foguetes e fitas coloridas: encontros nos movimentos pela cidade BIBLIOGRAFIA AHLERT, Martina. Casa sagrada, casa doméstica: uma etnografia da relação entre parentes, humanos e encantados em Codó – MA. Trabalho apresentado na 36ª Reunião da ANPOCS. Águas de Lindóia, 2012. BARROS, Sulivan Charles. Encantaria de Bárbara Soeira: a construção do imaginário do medo em Codó/MA. 163f. Dissertação (Mestrado em Sociologia) – Universidade de Brasília. Brasília, 2000. DA MATTA, Roberto. A casa e a rua: espaço, cidadania, mulher e morte no Brasil. Rio de Janeiro: Rocco, 1997. FERRETTI, Mundicarmo. Terecô: a linha de Codó. VIII Jornadas sobre Alternativas Religiosas na América Latina. São Paulo, 22 a 25 de setembro, 1998. __________. Maranhão encantado: encantaria maranhense e outras histórias. São Luis: UEMA Editora, 2000. __________. Encantaria de Barba Soeira: Codó, Capital da magia negra? São Paulo: Siciliano, 2001. Ponto Urbe 11 \| 2012 AHLERT, Martina. Casa sagrada, casa doméstica: uma etnografia da relação entre parentes, h d d b lh d d Á AHLERT, Martina. Casa sagrada, casa doméstica: uma etnografia da relação entre parentes, humanos e encantados em Codó – MA. Trabalho apresentado na 36ª Reunião da ANPOCS. Águas de Lindóia, 2012. humanos e encantados em Codó – MA. Trabalho apresentado na 36ª Reunião da ANPOCS. Águas de Lindóia, 2012. BARROS, Sulivan Charles. Encantaria de Bárbara Soeira: a construção do imaginário do medo em Codó/MA. 163f. Dissertação (Mestrado em Sociologia) – Universidade de Brasília. Brasília, 2000. DA MATTA, Roberto. A casa e a rua: espaço, cidadania, mulher e morte no Brasil. Rio de Janeiro: Rocco, 1997. FERRETTI, Mundicarmo. Terecô: a linha de Codó. VIII Jornadas sobre Alternativas Religiosas na América Latina. São Paulo, 22 a 25 de setembro, 1998. __________. Maranhão encantado: encantaria maranhense e outras histórias. São Luis: UEMA Editora, 2000. __________. Encantaria de Barba Soeira: Codó, Capital da magia negra? São Paulo: Siciliano, 2001. Ponto Urbe, 11 \| 2012 Foguetes e fitas coloridas: encontros nos movimentos pela cidade 5 INGOLD, Tim. Jornada ao longo de um caminho de vida – mapas, descobridor-caminho e navegação. In: Revista Religião e Sociedade, volume 25, número 1, p. 76-110, julho 2005. MACHADO, João Batista. Codó, histórias do fundo do baú. São Luís, FACT/UEMA, 1999. MOURA, Flávia de Almeida. Escravos da precisão. Economia familiar e estratégias de sobrevivência de trabalhadores rurais em Codó (MA). São Luís: EDUFMA, 2009. PEIRANO, Mariza. Rituais ontem e hoje. Rio de Janeiro: Zahar, 2003. NOTAS 1. “No Maranhão o termo encantado é encontrado nos terreiros de Mina, tanto nos fundados por africanos, quanto nos mais novos e sincréticos, e nos salões de curadores e pajés. Refere-se a uma categoria de seres espirituais recebidos em transe mediúnico, que não podem ser observados diretamente ou que se acredita poderem ser vistos, ouvidos e sentidos em sonho, ou por pessoas dotadas de vidência, mediunidade ou de percepção extrassensorial, como alguns preferem denominar. Os encantados, apesar de totalmente invisíveis para a maioria das pessoas, tornam-se “visíveis” quando os médiuns em quem incorporam manifestam alterações de consciência e assumem outra identidade. Apresentam-se à comunidade religiosa como alguém que teve vida terrena há muitos anos e que desapareceu misteriosamente ou tornou-se invisível, encantou-se” (M. Ferretti, 2000, p. 15). 2. A partir da década de 1970, a vinda de pessoas do campo para a cidade aumentou consideravelmente no município, especialmente em virtude da expulsão dos trabalhadores de suas terras e do processo de grilagem de terras. Sobre estes movimentos, ver Moura, 2009. MARTINA AHLERT Doutoranda em Antropologia Social na Universidade de Brasília (PPGAS/UnB), professora substituta do Departamento de Antropologia da Universidade Federal do Paraná (DEAN/UFPR). Email: mah_poa@yahoo.com.br Ponto Urbe, 11 \| 2012
https://openalex.org/W4380724041	https://zenodo.org/records/8043862/files/18.%20Aparecida%20Zuin%20287-302.pdf	Portuguese	null	O direito ao alimento na perspectiva da Teoria Sistêmica e por uma ecologia dos direitos humanos	Zenodo (CERN European Organization for Nuclear Research)	2,023	cc-by	12,135	© Cadernos de Dereito Actual Nº 21. Núm. Ordinario (2023), pp. 287-302 ·ISSN 2340-860X - ·ISSNe 2386-5229 © Cadernos de Dereito Actual Nº 21. Núm. Ordinario (2023), pp. 287-302 ·ISSN 2340-860X - ·ISSNe 2386-5229 © Cadernos de Dereito Actual Nº 21. Núm. Ordinario (2023), pp. 287-302 ·ISSN 2340-860X - ·ISSNe 2386-5229 O direito ao alimento na perspectiva da Teoria Sistêmica e por uma ecologia dos direitos humanos The right to food in the perspective of Systemic Theory and for an ecology of human rights Raffaele De Giorgi1 Università del Salento (Lecce-Itália) Aparecida Luzia Alzira Zuin2 Universidade Federal de Rondônia (UNIR) Aparecida Luzia Alzira Zuin2 Universidade Federal de Rondônia (UNIR) Sumário: 1. Introdução. 2. O tratamento sobre o direito ao alimento como direito humano. 3. Da produção excedente de comida à fome como estrutura de poder. 4. Considerações finais. 5. Referências. Resumo: Este trabalho apresenta o direito ao alimento como direito humano, com enfoque na Teoria Sistêmica de Niklas Luhmann e por uma ecologia dos direitos humanos. O objetivo da investigação é desenvolver a perspectiva sociossistêmica sobre os direitos humanos, discutindo como se desenvolvem e têm funções específicas na sociedade. Apresenta o problema da fome nos seguintes contextos: a produção de alimento como commodities na Amazônia brasileira e a produção e/ou projeto da fome como estrutura de poder. Conclui sobre a paradoxalidade dessa sociedade, cujos eventos operam nas produções de riscos: mais produção de alimentos, mais fome; mais comunicação e educação, maior desinformação; mais saber, maior o não saber. Palavras-chave: Direito ao alimento; Ecologia dos direitos humanos; Teoria Sistêmica; Risco. Abstract: This paper presents the right to food as a human right, focusing on Niklas Luhmann's Systemic Theory and an ecology of human rights. The objective of the investigation is to develop a socio-systemic perspective on human rights, discussing how they develop and have specific functions in society. It examines the issue of hunger in the following contexts: the production of food as commodities in the Brazilian Amazon, and the production and/or project of hunger as a power structure. It concludes on the paradox of this society, whose events operate in the production 1Doutor em Filosofia pela Universidade de Roma, La Sapienza. Professor Titular de Sociologia do Direito, Teoria Geral do Direito e Filosofia do Direito (Università del Salento). É diretor do Centro de Estudos do Risco, fundado em 1990, com Niklas Luhmann (Lecce – Italia). Escreveu com Niklas Luhmann a obra Teoria da Sociedade. Atualmente dedica-se, dentre várias linhas de pesquisa: descrição da complexidade da sociedade contemporânea; as formas modernas de escravidão; as formas de inclusão da exclusão na sociedade contemporânea; a memória do direito e da transição da Democracia no Brasil; Futuro da Teoria da Sociedade. ORCID: https://orcid.org/0000-0002-7548-560X E-mail: raffaeledegiorgi@gmail.com 2Doutora em Comunicação e Semiótica (PUC-SP). Docente nos Programas de Pós-Graduação: Doutorado em Educação na Amazônia, Mestrado Profissional Interdisciplinar em Direitos Humanos e Desenvolvimento da Justiça (DHJUS), Mestrado Acadêmico em Educação (PPGE). 1Doutor em Filosofia pela Universidade de Roma, La Sapienza. Professor Titular de Sociologia do Direito, Teoria Geral do Direito e Filosofia do Direito (Università del Salento). É diretor do Centro de Estudos do Risco, fundado em 1990, com Niklas Luhmann (Lecce – Italia). Escreveu com Niklas Luhmann a obra Teoria da Sociedade. Atualmente dedica-se, dentre várias linhas de pesquisa: descrição da complexidade da sociedade contemporânea; as formas modernas de escravidão; as formas de inclusão da exclusão na sociedade contemporânea; a memória do direito e da transição da Democracia no Brasil; Futuro da Teoria da Sociedade. ORCID: https://orcid.org/0000-0002-7548-560X E-mail: raffaeledegiorgi@gmail.com 2Doutora em Comunicação e Semiótica (PUC-SP). Docente nos Programas de Pós-Graduação: Doutorado em Educação na Amazônia, Mestrado Profissional Interdisciplinar em Direitos Humanos e Desenvolvimento da Justiça (DHJUS), Mestrado Acadêmico em Educação (PPGE). Lotada no Departamento de Ciências Jurídicas. Realizou estágio de pós-doutorado na Università del Salento, sob supervisão do prof. Dr. Raffaele De Giorgi, no Centro de Estudos do Risco, com o tema: direito alimentar, segurança e insegurança alimentar na sociedade do risco. ORCID: https://orcid.org/0000-0002-5838-2123 E-mail: alazuin@gmail.com 3BORGES, J.L. Ficções. Companhia das Letras, São Paulo, 2007. 4LUHMANN, N. Introdução à teoria dos sistemas. Vozes, Petrópolis, 2011. 5DÜRRENMATT, F. Der Auftrag: oder Vom Beobachten des Beobachters der Beobachter. Novelle in vierundzwanzig Sätzen. Diogenes Verlag, Zurique, 1988. 6“Per complessità del mondo Luhmann intende la totalità degli eventi possibili. Il mondo è estremamente complesso laddove il margine di attenzione della nostra esperienza intenzionale e del nostro agire è estremamente ridotto: la sovrabbondanza del possibile supera sempre ciò che noi siamo capaci di elaborare attraverso l’azione o l’esperienza”. LUHMANN, N. e DE GIORGI, R. Teoria della società. Franco Angeli, Milano, 2013, p. 219. 7LUHMANN, N. Sociologia do Direito I. Edições Tempo Brasileiro, Rio de Janeiro, 1983. 1. INTRODUÇÃO Na América Latina, no Brasil e na Argentina em particular, a Teoria dos Sistemas foi tratada como o pensamento inimigo de classe, um pensamento reacionário a ser combatido e atacado. No Brasil, combinaram a Teoria dos Sistemas ao Marxismo, à Teoria Crítica, à Teoria dos Valores e ao Sociologismo de origem norte-americana. Na Teoria Jurídica havia uma velha tradição kelseniana misturada com um kelsenismo importado e uma espécie de Teologia da Libertação; na Argentina, o espaço foi ocupado pelo Direito Natural Neotomista, Pensamento Crítico e teorias de origem oxoniense. Todos tiveram que citar Niklas Luhmann, escrever sobre a Teoria dos Sistemas e criticá-la. Por conseguinte, tratar a Teoria dos Sistemas como pensamento reacionário era ser crítico, e ser crítico era um título de honra. Agora, ao contrário, uma Teoria Crítica dos Sistemas está “na moda”, lê-se Luhmann citando passagens de Karl Marx. Daí os motivos pelos quais temos que compreender o porquê Luhmann minou a forma da racionalidade ocidental, quebrou a linearidade de suas construções, simplesmente transformando o que era óbvio para todos em um problema de investigação. Logo, as certezas foram sendo gradualmente quebradas; o conhecimento passou a ser revelado como uma técnica da ocultação; os conceitos nos quais o conhecimento se baseava tornaram-se infundados. A individualidade do indivíduo, assim como a subjetividade do sujeito ou a racionalidade da razão, começou a arder como as ruínas circulares do famoso conto de Jorge Luis Borges3. A isso se deve, a partir da pergunta que Niklas Luhmann4 tomou como basilar para os estudos da Teoria Sistêmica: “como isso é possível?”. Com esta pergunta, Luhmann simplesmente se coloca do outro lado: ele passa a observar como os outros do pensamento tradicional observam, ou seja, o que eles mesmos constroem como se observassem um mundo objetivo. Porém, Luhmann não apenas observou os modos como este fato acontecia, mas passou a conferir sentido “aquele mundo do outro lado, de outra maneira”, tal como a atitude descrita no subtítulo do maravilhoso romance de Friedrich Dürrenmatt, Der Auftrag, oder, Vom Beobachten des Beobachters der Beobachter5 (“A ordem: ou da observação dos observadores do observador”). Não é possível entender Luhmann6 e os problemas do direito se não nos colocarmos do outro lado. Assim, não pode ser lido Das Recht der Gesellschaft7 se continuarmos a pensar que o direito é uma estrutura de regras que se aplicam e têm validade. Raffaele De Giorgi; Aparecida Luzia Alzira Zuin. O direito ao alimento (…) O direito ao alimento (…) of risks: more food production, more hunger; more communication and education, more misinformation; more to know, more to not know. of risks: more food production, more hunger; more communication and education, more misinformation; more to know, more to not know. Keywords: Right to food; Ecology of human rights; Systemic Theory; Risk. , g , , , p UHMANN, N. Sociologia do Direito I. Edições Tempo Brasileiro, Rio de Janeiro, 1983. O direito ao alimento na perspectiva da Teoria Sistêmica e por uma ecologia dos direitos humanos The right to food in the perspective of Systemic Theory and for an ecology of human rights Lotada no Departamento de Ciências Jurídicas. Realizou estágio de pós-doutorado na Università del Salento, sob supervisão do prof. Dr. Raffaele De Giorgi, no Centro de Estudos do Risco, com o tema: direito alimentar, segurança e insegurança alimentar na sociedade do risco. ORCID: https://orcid.org/0000-0002-5838-2123 E-mail: alazuin@gmail.com 1Doutor em Filosofia pela Universidade de Roma, La Sapienza. Professor Titular de Sociologia do Direito, Teoria Geral do Direito e Filosofia do Direito (Università del Salento). É diretor do Centro de Estudos do Risco, fundado em 1990, com Niklas Luhmann (Lecce – Italia). Escreveu com Niklas Luhmann a obra Teoria da Sociedade. Atualmente dedica-se, dentre várias linhas de pesquisa: descrição da complexidade da sociedade contemporânea; as formas modernas de escravidão; as formas de inclusão da exclusão na sociedade contemporânea; a memória do direito e da transição da Democracia no Brasil; Futuro da Teoria da Sociedade. ORCID: https://orcid.org/0000-0002-7548-560X E-mail: raffaeledegiorgi@gmail.com 2 ã ó ó ã ORCID: https://orcid.org/0000 0002 7548 560X E mail: raffaeledegiorgi@gmail.com 2Doutora em Comunicação e Semiótica (PUC-SP). Docente nos Programas de Pós-Graduação: Doutorado em Educação na Amazônia, Mestrado Profissional Interdisciplinar em Direitos Humanos e Desenvolvimento da Justiça (DHJUS), Mestrado Acadêmico em Educação (PPGE). Lotada no Departamento de Ciências Jurídicas. Realizou estágio de pós-doutorado na Università del Salento, sob supervisão do prof. Dr. Raffaele De Giorgi, no Centro de Estudos do Risco, com o tema: direito alimentar, segurança e insegurança alimentar na sociedade do risco. ORCID: https://orcid.org/0000-0002-5838-2123 E-mail: alazuin@gmail.com Recibido: 05/05/2023 Aceptado: 15/06/2023 DOI: 10.5281/zenodo.8043862 Raffaele De Giorgi; Aparecida Luzia Alzira Zuin. Cadernos de Dereito Actual Nº 21. Núm. Ordinario, (2023) Cadernos de Dereito Actual Nº 21. Núm. Ordinario, (2023) 289 Saramago — Ensaio sobre a cegueira8, Ensaio sobre a lucidez9 e As intermitências da morte10. Luhmann observa a estrutura dos conceitos fundamentais do pensamento tradicional e as construções da realidade em que foram criadas a partir de sua semântica: a sociedade, por exemplo, ou a finalidade desta; ou ainda o conceito de direito, ou dos direitos fundamentais; contudo, acerca dos conceitos ainda resta a pergunta orientadora para explicá-los: “como é possível ser assim?”. Esta questão é a que o coloca do outro lado. O que faz toda a diferença, porque se coloca entre o que se move, quem fica estático e quem observa o movimento por outra perspectiva11. Saramago — Ensaio sobre a cegueira8, Ensaio sobre a lucidez9 e As intermitências da morte10. Luhmann observa a estrutura dos conceitos fundamentais do pensamento tradicional e as construções da realidade em que foram criadas a partir de sua semântica: a sociedade, por exemplo, ou a finalidade desta; ou ainda o conceito de direito, ou dos direitos fundamentais; contudo, acerca dos conceitos ainda resta a pergunta orientadora para explicá-los: “como é possível ser assim?”. Esta questão é a que o coloca do outro lado. O que faz toda a diferença, porque se coloca entre o que se move, quem fica estático e quem observa o movimento por outra perspectiva11. p p Desse modo, Luhmann observa que o que é tomado como substância é, na verdade, uma construção artificial. Nas suas obras, a semântica dos conceitos permite ver a função do trabalho teórico que consiste na necessidade de ocultar o paradoxo da artificialidade que se manifesta, quando uma artificialidade é assumida como substância12. Esse trabalho é feito pelo pensamento autorreflexivo da teoria e da Sociologia. Nessas primeiras obras ainda existe toda a linguagem clássica da Sociologia, da Teoria das Organizações, do pensamento jurídico tradicional e do pensamento filosófico sensível à Fenomenologia. Tem Max Weber, Talcott Parsons, George Herbert Mead e Hebert Simon. No entanto, na cultura alemã dos anos 1970, fortificações foram erguidas para se defender dos perigos dessas obras e para criticar, com linguagem às vezes verdadeiramente brutal, o que os críticos-não-leitores afirmavam que as obras de Luhmann significavam. Todavia, o pensamento de Luhmann se voltava a outra realidade, a outra necessidade. Agora é compreensível quando se pode ler Soziologische Aufklärung, Soziale Systeme, Vertrauen e Das Recht der Gesellschaft. 8SARAMAGO, J. Ensaio sobre a cegueira. Companhia das Letras, São Paulo, 1995. 9SARAMAGO, J. Ensaio sobre a lucidez. Companhia das Letras, São Paulo, 2004. 10SARAMAGO, J. As intermitências da morte. Companhia das Letras, São Paulo, 2005. 11DE GIORGI, R. Direito, democracia e risco. Vínculos com o futuro. Fabris, Porto Alegre, 1998, p. 81-97. ç p 15DE GIORGI, R. Direito, democracia e risco. Vínculos com o futuro. Fabris, Porto Alegre, 199 LUHMANN, N. Introdução à teoria dos sistemas. Vozes, Petrópolis, 2 14BORGES, J.L. Ficções. Companhia das Letras, São Paulo, 2007. 1. INTRODUÇÃO Se olharmos do outro lado, por outras vertentes, vê-se que as regras são fatos. Por outro lado, pode-se conferir que esta razão nada tem de racional. , p q Nas primeiras obras de Luhmann é possível conferir como a linguagem flui por si mesma, a construção se organiza tal como nos romances da trilogia de José 288 10SARAMAGO, J. As intermitências da morte. Companhia das Letras, São Paulo, 2005. 11DE GIORGI, R. Direito, democracia e risco. Vínculos com o futuro. Fabris, Porto Alegre, 199 p. 81-97. 8SARAMAGO, J. Ensaio sobre a cegueira. Companhia das Letras, São Paulo, 1995. 9SARAMAGO, J. Ensaio sobre a lucidez. Companhia das Letras, São Paulo, 2004. 10SARAMAGO J A i t itê i d t C hi d L t Sã P l 2005 8SARAMAGO, J. Ensaio sobre a cegueira. Companhia das Letras, São Paulo, 1995. 9SARAMAGO J E i b l id C hi d L Sã P l 2004 13LUHMANN, N. Introdução à teoria dos sistemas. Vozes, Petrópolis, 2011. 14BORGES, J.L. Ficções. Companhia das Letras, São Paulo, 2007. 8SARAMAGO, J. Ensaio sobre a cegueira. Companhia das Letras, São Paulo, 1995. 9SARAMAGO, J. Ensaio sobre a lucidez. Companhia das Letras, São Paulo, 2004. 10SARAMAGO, J. As intermitências da morte. Companhia das Letras, São Paulo, 2005. 11DE GIORGI, R. Direito, democracia e risco. Vínculos com o futuro. Fabris, Porto Alegre, 1998, p. 81-97. 12DE GIORGI, R. Op.Cit., p. 83. 13LUHMANN, N. Introdução à teoria dos sistemas. Vozes, Petrópolis, 2011. 14BORGES, J.L. Ficções. Companhia das Letras, São Paulo, 2007. 15DE GIORGI, R. Direito, democracia e risco. Vínculos com o futuro. Fabris, Porto Alegre, 1998. 12DE GIORGI, R. Op.Cit., p. 83. 13LUHMANN N Introdução à teoria dos sistemas Vozes Petrópolis 2011 DE GIORGI, R. Op.Cit., p. 83. LUHMANN, N. Introdução à teoria dos sistemas. Vozes, Petrópolis, 2011. , g p , , ARAMAGO, J. Ensaio sobre a lucidez. Companhia das Letras, São Paulo, 2004. SARAMAGO J A i t itê i d t C hi d L t Sã P l 2005 12DE GIORGI, R. Op.Cit., p. 83. Cadernos de Dereito Actual Nº 21. Núm. Ordinario, (2023) Sabe-se, porém, que as possíveis representações contidas nestas diferenciações podem esconder a violência e a repressão, e, desse modo, convida o observador a retirar o manto que cobre a superfície da água para ser visto o que está depositado no seu fundo. Nas condições de seu normal funcionamento, a racionalidade imanente a essa forma da diferenciação social, que é a racionalidade da inclusão universal de todos, produz continuamente exclusão, amplia pequenas diferenças originárias e, sem garantias externas, produz um contínuo excesso de alteridade que se acumula no seu interior. A alteridade que foi produzida no início da modernidade e que estava localizada é agora substituída por uma produção deslocalizada, flutuante, desmaterializada de alteridade produzida pelo funcionamento racional da sociedade moderna, pelo normal funcionamento de sua forma de diferenciação17. Nesse sentido, o construtivismo de Luhmann é revolucionário. O pensamento tradicional representava para si a comédia fatal da razão moderna; esse pensamento escreveria a dramaturgia dessa razão. Para o lugar onde se situava aquele pensamento, Luhmann18 recoloca o espaço aberto em que se representa a circularidade da inclusão, oportunizando a invenção de si através da representação do outro. As correntes “enferrujadas da comédia da razão” são substituídas pelo espaço aberto da construção possível de outros mundos. Com Luhmann, substitui-se o espaço cognitivo, tal qual é o espaço da criação na poesia, não como o espaço na comédia. Assim, não saber sobre a alteridade, inclusão e diferença faz com que a responsabilidade aumente, principalmente porque a responsabilidade dita que não tem cor moral, é laica, tem como sanção a redução do espaço da ação, da comunicação, do ser sem ontologia. ç g A responsabilidade pela perspectiva de Luhmann se tornou abrangente por meio de uma duplicação reflexiva do não saber: por meio da autoironia. Essa autoironia transformou sua prática teórica e a urgência do fazer em um jogo, e fez de si uma construção da “inocência infantil”. Mas seu mundo também era uma construção dessa ingenuidade ou simplesmente simplicidade. Luhmann tinha uma atitude cognitiva que lhe permitia quebrar a estabilidade, as certezas, as determinações conceituais com as quais o pensamento construía suas representações do mundo, porque, ao mesmo tempo que complexo, também fora analisado como óbvio, haja vista a simplicidade contida na pergunta: “como isso é possível?”. Cadernos de Dereito Actual Nº 21. Núm. Ordinario, (2023) Neste aporte, é possível afirmar: Niklas Luhmann era sua teoria. Ele pensou e agiu como parte do mundo que construiu. Sabia que era responsável por aquele mundo porque fazia parte dele. E também sabia que a alteridade do outro era o reflexo de sua alteridade e que a identidade nada mais é do que o resultado da manutenção da diferença da alteridade recíproca. Surgiu, desse modo, o caráter ético e político de seu pensamento. Uma ética que não precisava de regras, uma teoria da moral que surgiu a partir de sua teoria da comunicação e de seu construtivismo epistemológico, uma teoria da política construída no Iluminismo Sociológico13. g Desde 1968, o Iluminismo Sociológico foi um manifesto político e, depois, o programa de uma teoria de observação do presente. A coerência cognitiva daquela eticidade era avassaladora: o observador, o fazedor, como dizia Borges14; aquele que faz o mundo sabe que não vê, que não vê o que não vê, sabe que vê apenas o resultado de sua construção e desse resultado é ele mesmo, e que ele mesmo se torna possível como alteridade de outra alteridade. Da consciência da circularidade da construção surge a linearidade das concatenações semânticas do ser. Uma ética sem normas, uma teoria do ser sem ontologias, uma teoria da identidade que nasce da função constitutiva da alteridade na construção da ação dotada de sentido15. Tudo soa impressionante, porque à linearidade da representação da ação, a linearidade que se encontra em todas as teorias da ação social, opõe-se aqui a circularidade da produção de sentido por meio da comunicação. Nessas teorias ainda se encontram as distinções que caracterizaram antigas formas de diferenciação social, a distinção: 1) sujeito-objeto; 2) identidade-alteridade; 3) norma-fato; 4) Raffaele De Giorgi; Aparecida Luzia Alzira Zuin. O direito ao alimento (…) prescrição-descrição; 5) liberdade-subjugação; 6) produção de sentido-transferência de sentido; 7) ator-receptor16. prescrição-descrição; 5) liberdade-subjugação; 6) produção de sentido-transferência de sentido; 7) ator-receptor16. ) p Às análises destas diferenciações ficam a indagação luhmanniana sobre a casuística de todas elas. Essas diferenciações impregnaram a legibilidade do mundo moderno, justificaram a centralidade da razão moderna, possibilitaram a representação de sua comédia como desenvolvimento, progresso, democracia, liberdade. 16DE GIORGI, R. Op. Cit., p. 84-85. 17DE GIORGI, Raffaele. Periferias da modernidade. Revista Direito Mackenzie, 2017, p. 44. 18LUHMANN, N. Die Soziologie und der Mensch. Westdeutscher Verlag, Opladen, 1995. 19BRÖKER, F. M.; FOERSTER, H. von. Teil der Welt: Fraktale einer Ethik - oder Heinz von Foersters Tanz mit der Welt, Carl-Auer, Heidelberg, 2014. 17DE GIORGI, Raffaele. Periferias da modernidade. Revista Direito Mackenzie, 2017, p. 44. 18LUHMANN, N. Die Soziologie und der Mensch. Westdeutscher Verlag, Opladen, 1995. 19BRÖKER, F. M.; FOERSTER, H. von. Teil der Welt: Fraktale einer Ethik - oder Heinz vo Foersters Tanz mit der Welt, Carl-Auer, Heidelberg, 2014. Cadernos de Dereito Actual Nº 21. Núm. Ordinario, (2023) Cadernos de Dereito Actual Nº 21. Núm. Ordinario, (2023) 291 constrói. Ele não está fora, está dentro. Não só isso: ele não pode dizer de si mesmo que é objeto de observação sem reconhecer ao mesmo tempo que ele próprio é objeto de observação. Até o assunto é diferente; para que se atue como sujeito, deve ser observado, percebido, tratado como outra coisa. constrói. Ele não está fora, está dentro. Não só isso: ele não pode dizer de si mesmo que é objeto de observação sem reconhecer ao mesmo tempo que ele próprio é objeto de observação. Até o assunto é diferente; para que se atue como sujeito, deve ser observado, percebido, tratado como outra coisa. Dessa forma, porém, a alteridade não é mais a outra parte de uma distinção, mas a diferença que faz a diferença. O observador não usa uma distinção para indicar um lado ou outro que ele observa. O observador exime-se da diferença e ao mesmo tempo inclui-se na distinção. Aqui o observador é simultaneamente o todo e uma parte dele: precisamente, a sociedade e nós, a humanidade e a nossa cultura. Ele trata o outro como se fosse algo fora da sociedade e por isso pode ameaçá-la. Assim, ele justifica a expectativa de que o outro permaneça no ambiente ao qual naturalmente pertence. Como foi feito com a monstruosidade do monstro, com a pobreza e com a doença mental, quando foram inventados como outras coisas. E com a desumanidade dos nativos, quando também eles foram inventados como outros em suas terras. Nesse contexto, a Teoria dos Sistemas de Luhmann20, apesar de toda a ênfase no conceito de uma sociedade (mundial) única, não é indiferente aos problemas que, como consequência do desenvolvimento assimétrico dessa sociedade, emergem em diversas regiões do globo. Neves21, em seu trabalho, corrobora neste estudo, pois, o que está em jogo aqui é, também, dentre outras coisas, como o Estado Democrático de Direito trata o direito22. Em sua contribuição, Ökologie des Nochtwissens (“Ecologia do não saber”), Luhmann escreveu que toda a base desse argumento é, sobretudo, influenciada pelo fato de que a fonte primária de insegurança na sociedade não é mais o outro indivíduo, mas o ambiente ecológico e o contexto em que a sociedade evolui23. 20“É por fundamentos políticos que se persiste na segmentação regional do sistema político da sociedade mundial em Estados, apesar de permanente perigo de guerra; e são fundamentos econômicos que forçam a diferenciação da sociedade em centro e periferia, em regiões superdesenvolvidas e regiões carentes de desenvolvimento”. LUHMANN, N. Ökologische Kommunikation: Kann die moderne Gesellschaft sich auf ökologische Gefährdungen einstellen? Westdeutscher Verlag, Opladen, 1986, p. 168, tradução nossa. 24LUHMANN, N. Die Politik der Gesellschaft. Suhrkamp, Frankfurt, 2000. 20“É por fundamentos políticos que se persiste na segmentação regional do sistema político da sociedade mundial em Estados, apesar de permanente perigo de guerra; e são fundamentos econômicos que forçam a diferenciação da sociedade em centro e periferia, em regiões superdesenvolvidas e regiões carentes de desenvolvimento”. LUHMANN, N. Ökologische Kommunikation: Kann die moderne Gesellschaft sich auf ökologische Gefährdungen einstellen? Westdeutscher Verlag, Opladen, 1986, p. 168, tradução nossa. 21NEVES, Marcelo. “Os Estados no centro e os Estados na periferia. Alguns problemas com a concepção de Estados da sociedade mundial em Niklas Luhmann”, Revista de Informação Legislativa, ano 52, n. 206, 2015, p. 112. 22“No Estado democrático de direito, portanto, não se trata apenas da autonomia do direito. Faz-se necessária, adicionalmente, a autopoiese da política. Partindo-se da definição da política como a esfera das decisões coletivamente vinculantes [...] ou da generalização da influência (autoridade, reputação, liderança)”. NEVES, Marcelo. “Os Estados no centro e os Estados na periferia. Alguns problemas com a concepção de Estados da sociedade mundial em Niklas Luhmann”, Revista de Informação Legislativa, ano 52, n. 206, 2015, p. 112. 23LUHMANN, N. Beobachtungen der Moderne. VS Verlag für Sozialwissenschaften, Wiesbaden, 1992. g , , , , p 22“No Estado democrático de direito, portanto, não se trata apenas da autonomia do direito. Faz-se necessária, adicionalmente, a autopoiese da política. Partindo-se da definição da política como a esfera das decisões coletivamente vinculantes [...] ou da generalização da influência (autoridade, reputação, liderança)”. NEVES, Marcelo. “Os Estados no centro e os Estados na periferia. Alguns problemas com a concepção de Estados da sociedade mundial em Niklas Luhmann”, Revista de Informação Legislativa, ano 52, n. 206, 2015, p. 112. 23LUHMANN, N. Beobachtungen der Moderne. VS Verlag für Sozialwissenschaften, Wiesbaden, , ç g , , , , p LUHMANN, N. Beobachtungen der Moderne. VS Verlag für Sozialwissenschaften, Wiesbaden, 992. superdesenvolvidas e regiões carentes de desenvolvimento”. LUHMANN, N. Ökologische Kommunikation: Kann die moderne Gesellschaft sich auf ökologische Gefährdungen einstellen? Westdeutscher Verlag, Opladen, 1986, p. 168, tradução nossa. 21NEVES, Marcelo. “Os Estados no centro e os Estados na periferia. Alguns problemas com a concepção de Estados da sociedade mundial em Niklas Luhmann”, Revista de Informação Legislativa, ano 52, n. 206, 2015, p. 112. 22“No Estado democrático de direito, portanto, não se trata apenas da autonomia do direito. Faz-se necessária, adicionalmente, a autopoiese da política. Partindo-se da definição da política como a esfera das decisões coletivamente vinculantes [ ] ou da generalização da influência g, p , , p , ç 21NEVES, Marcelo. “Os Estados no centro e os Estados na periferia. Alguns problemas com concepção de Estados da sociedade mundial em Niklas Luhmann”, Revista de Informaçã Legislativa, ano 52, n. 206, 2015, p. 112. Cadernos de Dereito Actual Nº 21. Núm. Ordinario, (2023) j p p g p Ora, um observador deve dizer como observa: isso sigfnifica dizer qual é a distinção que se usa para observar algo; como algo é observado em oposição a outra coisa. O observador tem que fazer isso porque os outros também o observam e porque ele próprio é o resultado da observação. O observador vê o mundo (metaforicamente) com seus próprios olhos; considera o objeto de sua observação como um objeto, portanto, trata a si como sujeito da observação. Mas é justamente aqui que o objeto se rebela, porque o objeto também se considera como sujeito da observação. Nesse sentido, passa a tratar o outro como um objeto. Certamente não foi muito longe dessas considerações que Heinz von Foerster dizia o observador é aquele que constrói um mundo19. Com efeito: se o observador é aquele que constrói um mundo, o observador não é uma figura ingênua e inofensiva, ele é responsável pelo mundo que constrói e sua responsabilidade não deriva de normas ou valores, deriva do fato de ser parte do mundo que ele mesmo 290 Westdeutscher Verlag, Opladen, 1986, p. 168, tradução nossa. 21NEVES, Marcelo. “Os Estados no centro e os Estados na periferia. Alguns problemas com a concepção de Estados da sociedade mundial em Niklas Luhmann”, Revista de Informação Legislativa, ano 52, n. 206, 2015, p. 112. 22“No Estado democrático de direito, portanto, não se trata apenas da autonomia do direito. Faz-se necessária, adicionalmente, a autopoiese da política. Partindo-se da definição da política como a esfera das decisões coletivamente vinculantes [...] ou da generalização da influência (autoridade, reputação, liderança)”. NEVES, Marcelo. “Os Estados no centro e os Estados na periferia. Alguns problemas com a concepção de Estados da sociedade mundial em Niklas Luhmann”, Revista de Informação Legislativa, ano 52, n. 206, 2015, p. 112. 23LUHMANN, N. Beobachtungen der Moderne. VS Verlag für Sozialwissenschaften, Wiesbaden, 1992 HUMANO Sabemos que a sociedade só pode ameaçar-se por dentro, a partir de si mesma; sabemos também que, de fora, o meio ambiente só pode ameaçá-la por sua extrema complexidade. Isso significa que a alteridade ameaçadora que deveria recair sobre a sociedade, a monstruosidade sob a qual a sociedade deveria sucumbir, é uma construção interna da própria sociedade. E também sabemos, então, que assim como não há ontologia da natureza, também não há ontologia da alteridade, não há algo que seja absolutamente outro, assim como não há algo que seja ontologicamente não-outro. Esta é uma conclusão a que, depois das vergonhosas certezas das primeiras décadas do século XVI, chegou até a Teologia. E depois: se o outro é uma construção semântica, se é um produto que se realiza, que se forma, que se materializa na comunicação social, uma função particularmente relevante do conhecimento é justamente observar como se constrói a alteridade do outro. Sobre a estrutura da comunicação, no processo elementar da comunicação social, o outro não é o destinatário da comunicação — como se costuma dizer —, mas é o seu início: a função do outro consiste no fato de que ele, por meio do entendimento, possibilita a comunicação; o que por sua vez, só assim pode ser atribuída àquele que ativou o próprio ato de comunicar e que, para que isso aconteça, deve ser ele próprio observado pelo outro como aquele que pretende comunicar, portanto, como outro do outro. A comunicação torna as duas posições simétricas, ambas atuando como alteridade: a comunicação, nesses termos, em todos os sentidos e em todas as direções, sempre parte do outro. E, de fato: i) se o outro não apreende o ato de comunicar; ii) se o outro não apreende o sentido de comunicar como o ato aquilo que se pretende comunicar; iii) se o outro com o qual se está comunicando não apreende o outro como um outro que pretende comunicar; consequentemente, não poderá haver comunicação. Na comunicação, então, ambos se observam como o outro. Essa estrutura de comunicação atinge sua plenitude, ou seja, manifesta-se com todo o potencial de expansão de que é capaz, na sociedade moderna; por isso, caracteriza a modernidade da sociedade moderna, portanto, caracteriza a forma de sua complexidade. Com efeito, esta sociedade realiza a sua própria forma de diferenciação interna que lhe permite desenvolver-se como uma sociedade do mundo. Cadernos de Dereito Actual Nº 21. Núm. Ordinario, (2023) Neste sentido, sobre uma questão que nos permite entender a diferença entre o que Luhmann chamou das “Wissen des Nichtwissens”, o saber do não-saber, trinta anos atrás, e o que pensamos que deveríamos chamar das “Wissen des Nichtwissens”, o saber do não-sei, hoje. , j A sociedade do mundo, como Luhmann nos lembrou muitas vezes, opera com base na inclusão universal: o mundo está presente na comunicação. Espaço e tempo não são mais dimensões ameaçadoras, o observador pode controlá-los. A sociedade inclui todas as diferenças. E, no entanto, nosso não saber hoje é não saber da alteridade do outro, ou seja, da diferença. O que não sabemos, o que produz insegurança, o que é utilizado como recurso comunicativo, como tema político, como espaço de medo, não é o outro indivíduo, como dizia Luhmann24, mas sua alteridade. Esta atitude tem uma história muito recente. De fato, nos últimos anos, nossa insegurança surgiu de nosso “Nichtwissen” (não saber) em relação aos desenvolvimentos catastróficos a que as tecnologias estavam expondo o globo. Esses desdobramentos sempre têm o mesmo potencial Raffaele De Giorgi; Aparecida Luzia Alzira Zuin. O direito ao alimento (…) destrutivo; nosso desconhecimento é ainda mais extenso. Todavia, a fonte de nossa insegurança não é mais a mesma, tem um caráter completamente diferente e desencadeia formas extremamente ameaçadoras de sua absorção. Partindo dessa apresentação, eis que no percurso deste trabalho temos: como é possível que os direitos humanos sejam direitos e sejam humanos? Como é possível que a dignidade humana seja intocável? Como é possível pensar que a moral prescreve o bem? Ou, ainda com relação ao tema desenvolvido: como é possível haver fome em um mundo que desperdiça alimentos? Como é possível, em pleno século XXI, abordarmos sobre insegurança alimentar pela qualidade da comida e não pela quantidade de comida? Estas inquietações orientam as reflexões seguintes, sobretudo, no recorte espacial — o lugar aqui é a Amazônia brasileira. 25LIMA, F.R.S. e FINCO, M. “Teoria Sistêmica e Direitos Humanos: o Supremo Tribunal Federal e o direito à saúde”, Revista Pensamento Jurídico, v. 13, n. 2, 2019. 26LIMA, F.R.S. e FINCO, M. Op. Cit., p. 1-25. 27LIMA, F.R.S. e FINCO, M. Op. Cit., p. 24. 28NEITSCH, J. "O filósofo do ‘não saber’", entrevista de Raffaele de Giorgi à Joana Neitsch, Gazeta do Povo, 2013. 29LUHMANN, N. e DE GIORGI, R. Teoria della società. Franco Angeli, Milano, 2013. Cadernos de Dereito Actual Nº 21. Núm. Ordinario, (2023) O argumento que justifica a proposta é de que a sociedade moderna avançou tentando abolir os perigos iminentes de várias catástrofes humanitárias, como o caso em destaque atualmente, da Amazônia brasileira; contudo, não conseguiu encontrar sustentação ou equilíbrio favorável para eliminar os complexos problemas sociais que produz. Desse modo, das estruturas de decisão, fomenta-se o risco. p Ademais, perpassa, ainda, pela proposta de entender como o direito humano à alimentação condensa uma particular simbiose de futuro e sociedade, e permite refletir o futuro desta sociedade complexa. O que caracteriza a nossa sociedade é, entre tantas outras constatações, uma diferença profunda entre o não saber que temos nessa sociedade e as formas diferentes de não saber que havia em outras sociedades. O não saber que temos nessa sociedade é de outra natureza, diferente do não saber em outras sociedades. Por isso, nós sabemos que quanto mais sabemos, tanto maior é o não saber. Esta sociedade descobriu muitas tecnologias favoráveis ao aumento da produção alimentícia e não sabe a reação de determinados alimentos produzidos à saúde humana, por exemplo. Cuidar desse modo do não saber é tentar experimentar possibilidades para construir futuros. Nesse contexto, do mesmo modo, o direito ao alimento “é uma tecnologia social, é uma ordem através da qual se tenta reconstituir a ordem social”28, ou seja, esta ordem social seria ter alimento para todos, para não haver fome. O direito à alimentação, neste caso específico, opera com relação ao futuro a partir da seguinte forma: (1) se acontece algo que está violando o direito humano (neste caso, a fome é violência), então se produz uma consequência, uma sanção; logo, (2) a segurança que temos não é que o direito ao alimento (para todos) realizará alguma justiça (acabar com a fome da humanidade), mas (3) que o direito ao alimento será produzido e transformado com base em direitos (constitucional, econômico, político, à saúde, etc.), que (4) as decisões (elaborar políticas públicas para o combate à fome, por exemplo) serão tomadas com base em direitos, assim como serão anuladas com base em direitos29. Por isso, o direito é afirmado enquanto técnica que serve para produzir expectativas estáveis. De qualquer maneira, o direito tem a ver com o futuro e com o não saber. O que significa: o direito ao alimento tem a ver com o futuro da humanidade, e não com o não saber produzir alimento para a humanidade. HUMANO O mundo é o limite da sociedade e esse limite está presente na comunicação. O limite da possibilidade de construção da alteridade, então, é o mundo. Contudo, o mundo está presente justamente na comunicação que se dá na sociedade do mundo. A comunicação não conhece limites além dos limites da sociedade entendida precisamente como o universo da comunicação social. Nesse ínterim, o tratamento sobre o direito ao alimento como direito humano, sob a perspectiva da Teoria Sistêmica, apresenta algumas contribuições, as quais 292 Cadernos de Dereito Actual Nº 21. Núm. Ordinario, (2023) Cadernos de Dereito Actual Nº 21. Núm. Ordinario, (2023) 293 om base em Lima e Finco25, complementam a discussão e o entendimento sobre a uestão da comunicação. a) uma teoria que entende a sociedade mundial como comunicação pode destacar as funções latentes dos direitos humanos e a natureza deles, além de servir como valores, ou seja, ideais (nível estrutural)26; a) uma teoria que entende a sociedade mundial como comunicação pode destacar as funções latentes dos direitos humanos e a natureza deles, além de servir como valores, ou seja, ideais (nível estrutural)26; lores, ou seja, ideais (nível estrutural)26; b) na modernidade, o direito humano ao alimento — mais que normas básicas ou valores irrenunciáveis — pode (deveria?) ser entendido como referência indispensável para indicar o que é o humano nos seres humanos, tanto para atribuir um sentido concreto à palavra humanidade, além das características biológicas, como para identificar os limites além dos quais a sociedade reage na forma de decisões, indicando situações de violações que não são possíveis de aceitar, e, desta maneira, deduzir o que concretamente se entende por direitos humanos, também questionando se é possível pensar em normas indispensáveis (nível semântico)27. b) na modernidade, o direito humano ao alimento — mais que normas básicas ou valores irrenunciáveis — pode (deveria?) ser entendido como referência indispensável para indicar o que é o humano nos seres humanos, tanto para atribuir um sentido concreto à palavra humanidade, além das características biológicas, como para identificar os limites além dos quais a sociedade reage na forma de decisões, indicando situações de violações que não são possíveis de aceitar, e, desta maneira, deduzir o que concretamente se entende por direitos humanos, também questionando se é possível pensar em normas indispensáveis (nível semântico)27. q p p p ( ) Assim entendemos ser o direito ao alimento, um direito humano irrenunciável, direito imprescindível para a dignidade da pessoa humana, portanto, inaceitável que não seja atendido a todos os seres humanos. O argumento que justifica a proposta é de que a sociedade moderna avançou tentando abolir os perigos iminentes de várias catástrofes humanitárias, como o caso em destaque atualmente, da Amazônia brasileira; contudo, não conseguiu encontrar sustentação ou equilíbrio favorável para eliminar os complexos problemas sociais que produz. Desse modo, das estruturas de decisão, fomenta-se o risco. O direito ao alimento (…) O direito ao alimento (…) possibilidade que é sempre mais expandida no presente: o risco30. Risco não é no campo do negativo, é uma técnica de construir vínculos com o futuro. E qual a característica dessa técnica moderna que se chama de risco? É aquela segundo a qual se sabe querer evitar um dano futuro. Frente a condições como essas, o direito não tem possibilidades, não pode fazer nada. A única possibilidade do direito é permitir uma ação ou proibir. Frente ao risco, o direito pode só proibir31. Por exemplo, se alguém disser que comer hambúrguer todo dia tem-se o risco de ficar doente, então, a única maneira de evitar ficar doente, é a proibição de comer hambúrguer todos os dias. Daí que, todas as ações têm uma abertura diante do futuro, a única possibilidade que poderia ter o direito seria proibir. No caso do direito ao alimento, que se utiliza de muitas técnicas propondo a produção e o consumo de alimentos saudáveis, não para impedir os riscos de as pessoas comerem hambúrguer todos os dias, mas para transformar este risco social em riscos médicos. A produção e o consumo de alimentos saudáveis não vão diminuir os riscos, mas, se acontece um dano (por comer hambúrguer todos os dias), transforma-se este dano em problema de saúde/médico. Assim, a produção e o consumo de alimento saudável incrementam o risco. Na realidade, a alternativa ao risco é somente com outro risco. Pode-se evitar os riscos de ter um acidente cardiovascular, por exemplo, comendo somente verduras. Mas, se comer somente verduras, pode não obter calorias suficientes para manutenção do corpo (a depender de sua necessidade física para determinadas atividades cotidianas). Essas são algumas questões que produzem os efeitos de sentido do risco. g q q p Justamente sobre essa sociedade, sobre a descrição de sua estrutura, de sua diferença, circulam muitos estereótipos e/ou clichês. Um dos mais difundidos é a que a caracteriza como uma sociedade de risco. Luhmann sinalizava com isso o primeiro problema: qual é a distinção que nos permite indicar o risco? Qual é a outra parte cujo risco é a outra parte? Não poderia ser segurança, mesmo que segurança fosse o truísmo implausível que todos aceitavam. Por que os gregos e romanos não tinham sequer um termo que indicassem o que seria o risco? 30DE GIORGI, R. Direito, democracia e risco. Vínculos com o futuro. Fabris, Porto Alegre, 1998. 31DE GIORGI, R. “O risco na sociedade contemporânea”, Seqüência Estudos Jurídicos e Políticos, v. 15, n. 28, 1994, p. 37-49. 32DE GIORGI, R. “O risco na sociedade contemporânea”, Revista de Direito Sanitário, v.9, n.1, 2008, p. 43. O direito ao alimento (…) Por que o risco é encontrado pela primeira vez em documentos notariais do final do século XIII? Por que o termo risco é o mesmo em todas as línguas que expressam esse conceito?32 Em outras palavras: por que o risco é a representação de uma questão típica da sociedade moderna? Não poderia ser apenas uma questão de uso moderno da tecnologia. O risco tem a ver com a estrutura da temporalidade da sociedade moderna, com a forma de sua diferenciação. Tem a ver com o futuro. Tem a ver com não saber. O risco tem a ver com a forma da alteridade. Cadernos de Dereito Actual Nº 21. Núm. Ordinario, (2023) Dessa maneira, a sociedade enfrenta o futuro através do direito. Esta possibilidades de construir o futuro através do direito agora têm diante de si outr Raffaele De Giorgi; Aparecida Luzia Alzira Zuin. , , , , p DE GIORGI, R. “O risco na sociedade contemporânea”, Revista de Direito Sanitário, v.9, n.1, 008, p. 43. 30DE GIORGI, R. Direito, democracia e risco. Vínculos com o futuro. Fabris, Porto Alegre, 199 31DE GIORGI, R. “O risco na sociedade contemporânea”, Seqüência Estudos Jurídicos Políticos, v. 15, n. 28, 1994, p. 37-49. Cadernos de Dereito Actual Nº 21. Núm. Ordinario, (2023) Cadernos de Dereito Actual Nº 21. Núm. Ordinario, (2023) 295 evitar. Mas os riscos, aqueles danos futuros que poderiam ter sido evitados se tivessem sido tomadas decisões diferentes. Este risco condensa33 uma simbiose particular entre o futuro e a sociedade porque permite a construção de estruturas nos processos de transformação dos sistemas, especificando o aparecimento de ordens nas estruturas dos sistemas. Risco é, com efeito, uma construção comunicacional que descreve a possibilidade de arrependimento, no futuro, em relação a uma escolha que produziu um dano que se queria evitar. Dessa forma, o risco está ligado ao significado da comunicação e é relevante por este aspecto, não pelos vestígios que existam na consciência. Isso fica evidente quando se observa que a produção de alimentos no Brasil não está voltada para as reais necessidades do povo brasileiro, mas orientada conforme o ritmo das exportações. Por isso, a soberania alimentar e a reforma agrária popular são pautas que se chocam com os interesses do imperialismo e da burguesia nativa. Ademais, o risco da sociedade moderna se observa quando, embora a produção agrícola brasileira tenha batido recordes, exportando mais de 46 bilhões de reais somente em janeiro de 2022, seja também um país que voltou ao mapa da fome, com mais de 33 milhões de brasileiros famintos e metade da população vivendo com algum grau de insegurança alimentar34. Essa contradição ocorre porque a agroindústria brasileira não produz alimentos para alimentar o povo brasileiro, mas mercadorias para exportação, o alimento como commodities. Então, o que acontece? Ou, como isso acontece? — tomando como base a pergunta de Niklas Luhmann. Acontece que os direitos humanos35, tal como o direito ao alimento para a dignidade da pessoa humana, poderiam coexistir por séculos com o modelo da antiga e conhecida escravidão; porém, ocorre que apesar dos inúmeros depoimentos em defesa dos direitos humanos, eles convivem com as formas atuais de escravidão (escravidão moderna): com a fome na Região Norte da Amazônia, onde se instala o agronegócio, transformando alimentos em commodities; com a morte dos indígenas (guardiões da floresta); com os milhares de desempregados que vivem na informalidade sem direitos trabalhistas; com a população negra sendo a maior população carcerária; dentre outros. Assim, a produção da fome é um projeto político e de interesse econômico que serve a algum propósito; é vista como estrutura de poder. 33DE GIORGI, R. “O direito na sociedade do risco”, Revista Opinião Jurídica, v. 3, n. 5, 2005, p. 388-389. 34BERTOTTI, R. "O 25 de julho e a política de fome do governo Bolsonaro", CUT Brasil, 2022. 35DE GIORGI, R. “Por uma ecologia dos direitos humanos”, Revista Opinião Jurídica, ano 13, n. 20, 2017a. p. 325. 36DE GIORGI, R. “O direito na sociedade do risco”, Revista Opinião Jurídica, v. 3, n. 5, 2005. 37LUHMANN, N. Sistemas sociais: esboço de uma teoria geral. Vozes, Petrópolis, 2016. 3. DA PRODUÇÃO EXCEDENTE DE COMIDA À FOME COMO ESTRUTURA DE PODER A partir desses exemplos e reflexões, retornamos ao direito à alimentação, à produção e ao projeto da fome como estrutura de poder com a problemática que norteia os estudos: por que o Brasil, país que produz tantos alimentos, voltou ao mapa da fome? Por que a Amazônia brasileira, região do agronegócio, tem as cidades com maior grau de insegurança alimentar, do Brasil? Para esta problematização, são necessários os estudos sobre as modalidades na Teoria da Sociedade Complexa — porque as incertezas da vida contemporânea são fonte de reflexão sobre o direito — segundo as quais se tomam decisões de diferentes sistemas sociais e estas decisões se encontram em uma Teoria do Risco. Nesse sentido, o que está em jogo não são os perigos naturais que possam ocorrer, como chuvas, enchentes, terremotos e seca extrema, típicos dos fenômenos naturais da Terra. Afinal, os perigos são aqueles danos futuros que não se podem 294 Cadernos de Dereito Actual Nº 21. Núm. Ordinario, (2023) É uma engrenagem que se movimenta, se retroalimenta, autorreproduz; e isso não tem nada a ver com o perigo igual ao dos antepassados com a fome, tem a ver com o risco. O risco produzido pela sociedade moderna é que amplia o potencial das decisões, duplica a possibilidade de escolha, racionaliza a incerteza (no sentido de que permite ativar mecanismos para sua absorção), bifurca os caminhos da ação possível e duplica suas bifurcações. As alternativas se multiplicam e, em relação ao futuro, essa multiplicação é racional36. p ç O significado é, assim, a categoria que descreve os fenômenos gerados a partir da atualização constante da distinção entre espaço marcado e espaço não marcado, entre atual e virtual. Pode-se, portanto, na perspectiva de Luhmann37, compreendê-lo como uma modalidade, por meio da qual se opera o gerenciamento da atenção de um observador em sua relação com o observado, isto é, como um lugar de articulação entre o que é atualizado (o Brasil no contexto da fome) àquilo que ocupa o centro das atenções a todo momento e que, ao seu redor, permanece como um elo de mera possibilidade (combater a fome). Segundo dados do Inquérito Nacional sobre Insegurança Alimentar no Contexto da Pandemia de Covid-19 no Brasil, no ano de 2022, apenas quatro entre 10 famílias (15,5% da população brasileira) conseguiram acesso pleno à alimentação Raffaele De Giorgi; Aparecida Luzia Alzira Zuin. O direito ao alimento (…) no país. Os dados do relatório, elaborado pela Rede Brasileira de Pesquisa em Soberania e Segurança Alimentar e Nutricional (Rede Penssan) com apoio da Oxfam Brasil e outras organizações, mostram que a situação piorou muito desde a publicação dos primeiros dados, em 202138. Confirma-se, o Brasil é um país de contrastes sociais, formado por um padrão de consumo onde grande parcela da população se encontra em situação de insegurança alimentar ou fome. E, nessa realidade, a mídia legitima e naturaliza o modelo de produção capitalista do espaço, sem se preocupar com o social, tampouco propondo um debate sobre os números apresentados. Logo, a fome passa invisível. Aqui está a questão do tempo. Luhmann procurava afastar-se da ideia de um passado, presente e futuro completamente distintos. A sociedade atual conhece a si própria através dos meios de comunicação de massa. Por isso, a sociedade sempre atua no presente, não há passado nem futuro. O futuro se conjuga no passado e no presente. A pergunta continua. 38MONCAU, G. "Fome se alastra no Brasil: 6 em cada 10 famílias não têm acesso pleno a alimentos", Brasil de Fato, 2022. 39CARDOSO, M.A. "Alimentação no início da vida: janela para o futuro", Jornal da USP, 2018. 40G1RO. "Geografia da fome: Região Norte do Brasil é a mais impactada pela insegurança alimentar", G1RO, 2022. 41NITAHARA, A. "Censo Agropecuário: Brasil tem 5 milhões de estabelecimentos rurais", Agência Brasil, 2019. 42NAÇÕES UNIDAS BRASIL. "Pequenos agricultores familiares produzem mais de um terço dos alimentos no mundo", Nações Unidas Brasil, 2021. 43NITAHARA, A. "Censo Agropecuário: Brasil tem 5 milhões de estabelecimentos rurais", Agência Brasil, 2019. Cadernos de Dereito Actual Nº 21. Núm. Ordinario, (2023) Há razões para esses resultados? Os motivos são conhecidos? Por histórico, o Brasil já reduziu em 82,1% o número de pessoas subalimentadas no período entre os anos de 2002 a 2014, cuja queda foi a maior registrada entre as seis nações mais populosas do mundo, e também superior à média da América Latina, que foi de 43,1%, segundo relatório anual sobre a fome, "Estado da Insegurança Alimentar no Mundo – 2015", publicado pela Organização das Nações Unidas (ONU) para a Alimentação e a Agricultura (FAO), o Fundo Internacional para o Desenvolvimento Agrícola (Ifad) e o Programa Alimentar Mundial (PAM)39. ( ) Entretanto, decisões dos sistemas político, econômico e jurídico fomentaram o aumento da fome nos últimos anos, conforme dados da Rede Penssan40, somente no Norte brasileiro, especificamente na Região Amazônica, 71,6% das famílias sofrem algum tipo de insegurança alimentar, por exemplo. Em todo o país, aproximadamente 33,1 milhões de brasileiros vivem em situação de fome no ano de 2022, 14 milhões a mais que em 2020. Algumas dessas decisões dos sistemas político, econômico e jurídico, a título de exemplo: (i) Desmonte das políticas públicas que fomentam o investimento na produção agrícola familiar, ainda que dados do Instituto Brasileiro de Geografia e Estatística (IBGE) indiquem que 77% (3.897.408) dos estabelecimentos rurais do Brasil no ano de 2017 são classificados como agricultura familiar e correspondem a uma área de 80,89 milhões de hectares, ou seja, 23% da área total do país41; (i) Desmonte das políticas públicas que fomentam o investimento na produção agrícola familiar, ainda que dados do Instituto Brasileiro de Geografia e Estatística (IBGE) indiquem que 77% (3.897.408) dos estabelecimentos rurais do Brasil no ano de 2017 são classificados como agricultura familiar e correspondem a uma área de 80,89 milhões de hectares, ou seja, 23% da área total do país41; (ii) Desvalorização da agricultura familiar em detrimento do agronegócio, isto é, o não reconhecimento da importância da agricultura familiar para a produção de alimento como direito humano, ainda que, segundo a Organização das Nações Unidas, com dados do ano de 2021, a agricultura familiar seja responsável por 80% de toda a produção de alimentos consumidos no planeta42. ç (iii) Criminalização dos movimentos sociais e de luta pelo direito à terra. Cadernos de Dereito Actual Nº 21. Núm. Ordinario, (2023) Cadernos de Dereito Actual Nº 21. Núm. Ordinario, (2023) 297 (iv) Regulamentação da Lei n.º 13.986, de 07 de abril de 202044, conhecida como “Lei do Agro”, resultado da conversão da Medida Provisória n.º 897/2019, que criou mais facilidades para a burguesia rural no acesso ao crédito, isenção de contribuições previdenciárias, entre outros, além da abertura de financiamentos setoriais com capital estrangeiro. Como exposto anteriormente, a fome que avança pelo Brasil atingiu percentualmente mais os moradores da Região Norte até o ano de 2022, justamente na Amazônia: 71,6% sofrem com a insegurança alimentar e a fome extrema faz parte do cotidiano de 25,7% das famílias — índice maior que a média nacional, de aproximadamente 15%45. Quais os vestígios podem existir na consciência quando deparados com esses números? Retorna-se ao risco, haja vista que o risco estabelece a necessidade de um cálculo do tempo segundo condições que nem a racionalidade, nem o cálculo da utilidade, nem a estatística podem fornecer indicações úteis. Nestas condições de não-saber, aquilo que realmente pode-se saber é que cada redução ou minimização do risco aumenta o próprio risco. Disto deriva a necessidade de decidir em condições que se saiba que outra decisão poderia evitar o dano que devia verificar-se. Mas se isto devia acontecer, aconteceria no futuro. É por isso que a sociedade moderna representa o futuro como risco. Se o representa, o constrói. O risco, então, é uma característica estrutural da complexidade da sociedade moderna, de sua temporalização, da simbiose com o futuro, a paradoxalidade do presente, da ecologia do não-saber46. Assim, a perspectiva dos direitos humanos descrita por um ponto de vista sistêmico tem em vista focar na sua função interior e para a sociedade mundial. Então, pode-se pensar no direito ao alimento como direito humano, porque é sabido que durante muito tempo a capacidade de produção de alimentos foi mais do que suficiente para satisfazer toda a população mundial. Hoje, é mais que o dobro do necessário. Com isso, se a real função dos Direitos Humanos consiste na concretização das condições que estabilizam a forma de modernidade da sociedade moderna47, evidentemente, não se pode negar que eles tenham uma função evolutiva: bloqueiam a involução, impedem que o passado inunde o presente com seus escombros — a fome, por exemplo. 44BRASIL. “Lei n.º 13.986, de 07 de abril de 2020”, Diário Oficial da União, 20.8.2020. Brasília, 2020. 45G1RO. "Geografia da fome: Região Norte do Brasil é a mais impactada pela insegurança alimentar", G1RO, 2022. 46DE GIORGI, R. “O direito na sociedade do risco”, Revista Opinião Jurídica, v. 3, n. 5, 2005, p. 383-394. 47DE GIORGI, R. Op. Cit., p. 390. 48DE GIORGI, R. “Periferias da modernidade”, Revista Direito Mackenzie, v. 11, n. 2, 2017b. 49DE GIORGI, R. “Por uma ecologia dos direitos humanos”, Revista Opinião Jurídica, ano 13, n. 20, 2017a, p. 337. Cadernos de Dereito Actual Nº 21. Núm. Ordinario, (2023) Segundo censo agropecuário do IBGE no ano de 2017, a agricultura familiar no Brasil cobre 3.897.408 estabelecimentos rurais; são 77% dos estabelecimentos agropecuários do país, como dito, que empregam mais de 10,1 milhões de pessoas (67% do total censitário), responsáveis por parcela significativa da oferta de alimentos básicos à mesa dos brasileiros e brasileiras43; (iii) Criminalização dos movimentos sociais e de luta pelo direito à terra. Segundo censo agropecuário do IBGE no ano de 2017, a agricultura familiar no Brasil cobre 3.897.408 estabelecimentos rurais; são 77% dos estabelecimentos agropecuários do país, como dito, que empregam mais de 10,1 milhões de pessoas (67% do total censitário), responsáveis por parcela significativa da oferta de alimentos básicos à mesa dos brasileiros e brasileiras43; 296 4. CONSIDERAÇÕES FINAIS A questão da alteridade do outro, tal como é colocada neste trabalho, opõe- se à modernidade da sociedade atual, porque a última violenta a estrutura sedimentada da comunicação e a semântica que a torna possível: ilude-se de que pode deter a forma da modernidade. Eis que, o desconhecimento da alteridade depende do fato de ela ser impenetrável. E é impenetrável tendo em vista que se esconde no olhar do outro, “se esconde no que se esconde”. O olhar é como um feixe invisível que se liga ao ambiente, é a imaterialidade que repousa sobre um objeto e usa a luz para lhe dar um contorno, é como uma impressão silenciosa que envolve o objeto sobre o qual repousa e o torna visível. O olhar é o que nos faz dizer diante do mundo: é a construção de um observador. O outro não sabe que é o construtor que, com seu olhar, se deixa construir por um “eu” ou por um “nós” — e se liga a ele em toda a sua dependência, sem perceber que ele mesmo é outro. Ele se ilude de que não é um guardião e acredita ser um prisioneiro, como em uma história terrível e assustadora de Dürrenmatt50 que há meio século lançou uma luz espectral sobre o paradoxo constitutivo da semântica da sociedade contemporânea, a semântica do guardião e o prisioneiro, de fato. Ainda hoje, o outro é tratado como o objeto: como o que se supõe sem palavra. Aquilo sobre o qual se acredita que à luz da observação caia. Porém, na realidade é precisamente a sua opacidade que permite à luz delinear os contornos do observador. O outro, nesta perspectiva, também observa: ele observa seu ser observado; portanto, deve agir, mesmo que a palavra seja tirada dele; incluir, mesmo que o sujeito da observação o exclua. O outro tem uma fé, mesmo que o sujeito da observação o trate como um infiel; ele tem suas razões, mesmo que o sujeito da observação o trate como desprovido de razão; tem a sua interioridade, ainda que o sujeito da observação a trate como uma exterioridade selvagem51. Daí a sociedade de risco é uma fórmula que serve para despertar alarme, uma fórmula que inclui, de forma autocontraditória, múltiplos conteúdos, muito diferentes entre si, que funcionam sempre quando são usados para produzir, precisamente, alarme no presente e desespero quanto ao futuro. 50DÜRRENMATT, F. Der Auftrag: oder Vom Beobachten des Beobachters der Beobachter. Novelle in vierundzwanzig Sätzen. Diogenes Verlag, Zurique, 1988. 51Como o tema da esplêndida realização cinematográfica de 2017 de Guillermo del Toro, “A Forma da Água” (“The Shape of Water”). Cadernos de Dereito Actual Nº 21. Núm. Ordinario, (2023) , p p No texto “Periferias da modernidade”48, os direitos humanos funcionam na forma de catálogo, tal como uma lista estruturada de direitos que possam nos servir de referência para programas políticos. Os catálogos fornecem esquemas de referência úteis para a estabilização das possibilidades do agir nessa sociedade moderna, nos diferentes âmbitos (subsistemas). Não por acaso, há declarações de direitos (catálogos) de diferentes tipos: direitos econômicos, sociais e culturais, direito à saúde, direito alimentar, entre outros. A inclusão pode ser universal, ou, ao menos, ser crescente, mais ampla, sem discussão acerca dos problemas de reivindicações de direitos, a partir do aumento ou maior acesso aos direitos. Ou seja, os direitos humanos são os requisitos mínimos para a inclusão. A inclusão, porém, tem outro lado: a exclusão. Por isso, a proposta é operar de acordo com uma “ecologia dos Direitos Humanos”, ou seja, “uma perspectiva que observe como no ambiente da sociedade, a sociedade aloca as alteridades que produz”49. Tal perspectiva poderia permitir-nos observar e perguntar: qual é a função dos direitos humanos? Aqui a reflexão é: e quais são as possibilidades de futuro que se podem Raffaele De Giorgi; Aparecida Luzia Alzira Zuin. O direito ao alimento (…) construir a partir dessas observações? Como o caso do direito ao alimento como direito humano. construir a partir dessas observações? Como o caso do direito ao alimento com direito humano. 4. CONSIDERAÇÕES FINAIS A imprecisão semântica53 do risco foi redefinida através da construção política dos riscos, das ameaças, da incumbência de novas barbaridades, de inimigos que poderiam ter sido detidos e controlados, de fato, apenas com o acionamento de políticas de segurança. Tornou-se possível extrair consentimento do medo p O processo de avanço civilizatório jurídico, econômico, político e educacional causou impactos na estrutura de sistemas sociais54, seja em virtude dos novos direitos que vão sendo solicitados para atenderem às demandas complexas, seja pela legitimação de poder através da economia, da política e da educação, ou ainda porque a função de redução da complexidade, inerente à função de cada sistema, incrementa a própria complexidade. É É pela comunicação que se origina essa complexidade, haja vista ser ela que estrutura um sistema social. Afinal, por meio da comunicação faz-se as coisas presentes no mundo funcionarem, terem valores positivos ou negativos, garantias ou não, funcionais ou disfuncionais. Significa que os sentidos das coisas residem nos modos como os sujeitos os determinam para se interagirem. Comunicação é, então, o processo interativo, a ligação, o estar com o outro no sentido de ser igual aqui, agora, ali, alhures; orienta e dá funcionalidade à universalidade; dá sentido e serve como premissa para a elaboração de toda a experiência humana; liberta ou oprime as pessoas. Logo, o sentido das coisas se apresenta como excedente das referências de um dado experimento ulterior nas possibilidades de experimentar; o sentido é a própria imanência das coisas. p p Segundo Luhmann55, a sociedade é um grande sistema, no interior do qual as outras relações sociais se operam e se reconstroem a partir de seus próprios elementos (autopoiesis). Nessa perspectiva, pode-se citar o direito alimentar, por exemplo, enquanto subsistema do Direito, possui mecanismos de auto-organização produzidos pela sociedade para se comunicar sobre o direito e o acesso ao alimento saudável, dentro de outro sistema, caso do Sistema Nacional de Segurança Alimentar e Nutricional (Lei n.º 11.346, de 15 de setembro de 200656). ( Dentre as indeterminações merecedoras de análises, encontra-se a (in)certeza no saber orientado à questão da segurança alimentar concernente ao poder econômico ou status social. 4. CONSIDERAÇÕES FINAIS E como a comunicação do risco amplia a percepção do risco e atualiza sua ameaça, o simples fato de falar do presente como o tempo da sociedade de risco espalha insegurança, incerteza e medo. A fórmula, tanto mais difundida quanto mais sustentada por apelos e proclamações, tem a função de produzir como primeira consequência a representação da necessidade de assegurar-se, de construir segurança: uma representação que simultaneamente evoca o outro, segundo o qual no passado sentíamos-nos seguros, estávamos mesmo mais seguros e, por isso, agora é preciso, de fato, repor a segurança, readquirir aquela segurança que se acredita perdida devido ao risco iminente, à ameaça. Dessa forma, uma confusão semântica particular é ativada, que se multiplica em inúmeras outras confusões semânticas. A consequência é que a indeterminação semântica do risco repousa na indeterminação semântica da segurança. Alastra-se uma semântica de alarme que, nas últimas décadas, tem oferecido grandes espaços de conquista para protestar, para movimentos que tremeram, de tempos em tempos, contra qualquer uma das ameaças reais que esta sociedade tem produzido contra si mesma: o risco ecológico, o risco de grandes obras de destruição que afetam o clima, a produção imensa de resíduos sólidos, a construção de ferrovias, hidrelétricas e gasodutos, a extração de petróleo, a produção de emissões de dióxido de carbono, a 298 Cadernos de Dereito Actual Nº 21. Núm. Ordinario, (2023) Cadernos de Dereito Actual Nº 21. Núm. Ordinario, (2023) 299 energia nuclear, entre outros. Os protestos sobre esses riscos eram setoriais, delimitados ou tão universalizados que poderiam esmorecer, evaporar, consumir-se. Mas um fato é sempre reiterado, caracterizado em cada época: os protestos podiam ser facilmente absorvidos e reprimidos, às vezes até violentamente52. energia nuclear, entre outros. Os protestos sobre esses riscos eram setoriais, delimitados ou tão universalizados que poderiam esmorecer, evaporar, consumir-se. Mas um fato é sempre reiterado, caracterizado em cada época: os protestos podiam ser facilmente absorvidos e reprimidos, às vezes até violentamente52. Agora, por outro lado, os espaços estreitos ocupados por esses protestos — que de qualquer forma resultou no surgimento de formas de civilização de comportamento relacionadas ao meio ambiente — ficaram praticamente vazios ou tendem a se esvaziar. A racionalidade antes envolta em ameaças reais foi substituída por uma racionalidade política que espalhou alarme para fins de produção de consenso: extorquiu consentimento para políticas que garantiam segurança contra riscos construídos com base em velhas ideologias contra-iluministas. 52DE GIORGI, R. Direito, democracia e risco. Vínculos com o futuro. Fabris, Porto Alegre, 1998. 53DE GIORGI, R. Op. Cit., p. 91. 54LUHMANN, N. Sistemas sociais: esboço de uma teoria geral. Vozes, Petrópolis, 2016. 55LUHMANN, N. Op.cit. 56BRASIL. “Lei n.º 11.346, de 15 de setembro de 2006”, Diário Oficial da União, 18.9.2006, Brasília, 2006. 4. CONSIDERAÇÕES FINAIS Nessa modalidade de saber se predispõe a ideia de quem tem poder econômico está mais propenso à segurança, à comunicação e informação e educação alimentar; que a produção alimentícia serve ao consumo responsável e, consequentemente, as tecnologias para impedir catástrofes de natureza alimentar transformam inseguranças econômicas de natureza social (desigualdade, pobreza, discriminação, doenças e outras) em garantias à dignidade da pessoa humana. Então, se há insegurança alimentar em virtude do desperdício de alimentos e isso causa danos irreversíveis à saúde de seres humanos, educa-se para a produção, fornecimento, criam-se índices de preços que não podem ser ultrapassados para Raffaele De Giorgi; Aparecida Luzia Alzira Zuin. O direito ao alimento (…) O direito ao alimento (…) alimentos básicos, ensina-se o consumo responsável. Contudo, enquanto não forem superadas as condições de desperdício, de desinformação, a insegurança se torna algo normal e aceitável, sem a devida consciência da ameaça que afeta alguns humanos pelo mundo. alimentos básicos, ensina-se o consumo responsável. Contudo, enquanto não forem superadas as condições de desperdício, de desinformação, a insegurança se torna algo normal e aceitável, sem a devida consciência da ameaça que afeta alguns humanos pelo mundo. Estruturas sociais desta natureza quando se auto-operam a partir de si mesmas para as próprias operações, tornam-se instáveis e imprevisíveis. As instabilidades/imprevisibilidades podem indicar, novamente, que os conceitos de segurança e educação alimentar, dentre outros que fazem parte desse arcabouço sistemático e/ou estrutural, estão destituídos de qualquer sentido a favor de operar a estabilidade. Daí, pode-se aferir acerca das probabilidades negativas (falta de confiança) constantes nas certezas dispostas nas medidas ditas objetivas, sofisticadas tecnologicamente à superação da ameaça; se os conceitos estão desprovidos de valores para a reversão da catástrofe (como morte por inação, por exemplo) também não se encontra confiança, segurança e estabilidade para o tratamento da complexidade que a própria complexidade produz. A instabilidade e a imprevisibilidade, nesse caso, conferem a complexidade, pessoas morrem de fome quando a produção e o consumo de alimentos estão sendo utilizados de modo irresponsável. A irresponsabilidade/responsabilidade, por sua vez, está correlata à decisão da sociedade com relação ao seu futuro. Se no passado não é mais plausível a busca pelas respostas aos problemas do presente, significa que o passado não conferia estabilidade. Por isso, o problema imanente aqui se refere ao futuro. Nesse modelo de sociedade, “a única possibilidade que temos para construir vínculos com o futuro é o risco”. 4. CONSIDERAÇÕES FINAIS Afinal, o risco na sociedade contemporânea tem a função de racionalizar o medo — o medo segundo o qual as pessoas possam morrer de fome, embora haja desperdício de alimentos; uma sociedade cujo intento é defender mais comunicação e educação, e, no entanto, há mais desinformação; mais saber, maior o não saber. Porquanto, risco significa imputar um eventual dano futuro a uma decisão, na certeza de que outra decisão poderia evitar que o dano da fome se produzisse como uma estrutura de poder, por exemplo. O risco depende sempre da decisão de quem age. Oras, a ação do agente produtor da insegurança alimentar — a fome, a morte de humanos por inação — por mais complexo que seja, parte das escolhas da própria sociedade (por isso mesmo, complexa), a mesma que produz os sistemas das leis de segurança alimentar, do conteúdo da educação para alimentação e consumo responsável dos alimentos, orientações para a produção e o fornecimento de alimentos adequados, entre outros. 5. REFERÊNCIAS BERTOTTI, R. "O 25 de julho e a política de fome do governo Bolsonaro", CUT Brasil [online], 2022 [consult. 26 mai. 2023]. Disponível em https://www.cut.org.br/artigos/o-25-de-julho-e-a-politica-de-fome-do- governo-bolsonaro-53c8 g BORGES, J.L. Ficções. Companhia das Letras, São Paulo, 2007. BORGES, J.L. Ficções. Companhia das Letras, São Paulo, 2007. BORGES, J.L. Ficções. Companhia das Letras, São Paulo, 2007. BRASIL. “Lei n.º 11.346, de 15 de setembro de 2006”, Diário Oficial da União [online], 18.9.2006, Brasília, 2006 [consult. 26 mai. 2023]. Disponível em https://www.planalto.gov.br/ccivil_03/_ato2004-2006/2006/lei/l11346.htm BRASIL. “Lei n.º 11.346, de 15 de setembro de 2006”, Diário Oficial da União [online], 18.9.2006, Brasília, 2006 [consult. 26 mai. 2023]. Disponível em https://www.planalto.gov.br/ccivil_03/_ato2004-2006/2006/lei/l11346.htm BRASIL. “Lei n.º 13.986, de 07 de abril de 2020”, Diário Oficial da União [online], 20.8.2020. Brasília, 2020 [consult. 26 mai. 2023]. Disponível em https://www.planalto.gov.br/ccivil_03/_ato2019-2022/2020/lei/l13986.htm BRASIL. “Lei n.º 13.986, de 07 de abril de 2020”, Diário Oficial da União [online], 20.8.2020. Brasília, 2020 [consult. 26 mai. 2023]. Disponível em https://www.planalto.gov.br/ccivil_03/_ato2019-2022/2020/lei/l13986.htm BRASIL. “Lei n.º 13.986, de 07 de abril de 2020”, Diário Oficial da União [online], 20.8.2020. Brasília, 2020 [consult. 26 mai. 2023]. Disponível em https://www.planalto.gov.br/ccivil_03/_ato2019-2022/2020/lei/l13986.htm BRÖKER, F. M; Teil der Welt. Fraktale einer Ethik – oder: Heinz von Foersters T it d W lt C l A H id lb 2014 p p g BRÖKER, F. M; Teil der Welt. Fraktale einer Ethik – oder: Heinz von Foersters Tanz mit der Welt, Carl-Auer, Heidelberg 2014. , , g CARDOSO, M.A. "Alimentação no início da vida: janela para o futuro", Jornal da USP [online], 2018 [consult. 26 mai. 2023]. Disponível em https://jornal.usp.br/artigos/alimentacao-no-inicio-da-vida-janela-para-o- futuro/ , , g CARDOSO, M.A. "Alimentação no início da vida: janela para o futuro", Jornal da USP [online], 2018 [consult. 26 mai. 2023]. Disponível em https://jornal.usp.br/artigos/alimentacao-no-inicio-da-vida-janela-para-o- futuro/ 300 Cadernos de Dereito Actual Nº 21. Núm. Ordinario, (2023) Cadernos de Dereito Actual Nº 21. Núm. Ordinario, (2023) 301 DE GIORGI, R. “O risco na sociedade contemporânea”, Seqüência Estudos Jurídicos e Políticos [online], v. 15, n. 28, pp. 45-54, 1994 [consult. 26 mai. 2023]. Disponível em doi https://doi.org/10.5007/%x DE GIORGI, R. “O risco na sociedade contemporânea”, Seqüência Estudos Jurídicos e Políticos [online], v. 15, n. 28, pp. 45-54, 1994 [consult. 26 mai. 2023]. Disponível em doi https://doi.org/10.5007/%x ___________. Direito, democracia e risco. Vínculos com o futuro. Fabris, Porto Alegre, 1998. ___________.. “O direito na sociedade do risco”, Revista Opinião Jurídica [online], v. 3, n. 5, pp. 383-394, 2005 [consult. 26 mai. 2023]. Disponível em doi https://doi.org/10.12662/2447-6641oj.v3i5.p383-394.2005 ___________. “O risco na sociedade contemporânea”, Revista de Direito Sanitário [online], v.9, n.1, pp. 37-49, 2008 [consult. 26 mai. 2023]. Disponível em doi https://doi.org/10.11606/issn.2316-9044.v9i1p37-49 ___________.. “Por uma ecologia dos direitos humanos”, Revista Opinião Jurídica [online], ano 13, n. 20, pp. 324-340, 2017a [consult. 26 mai. 2023]. Disponível em doi https://doi.org/10.12662/2447-6641oj.v15i20.p324- 340.2017 ___________. “Periferias da modernidade”, Revista Direito Mackenzie [online], v. 11, n. 2, pp. 39-47, 2017b [consult. 26 mai. 2023]. Disponível em doi https://doi.org/10.5935/2317-2622/direitomackenzie.v11n2p39-47 p p g ___________. Direito, democracia e risco. Vínculos com o futuro. Fabris, Porto Alegre, 1998. ___________.. “O direito na sociedade do risco”, Revista Opinião Jurídica [online], v. 3, n. 5, pp. 383-394, 2005 [consult. 26 mai. 2023]. Disponível em doi https://doi.org/10.12662/2447-6641oj.v3i5.p383-394.2005 _______. “O risco na sociedade contemporânea”, Revista de Direito Sanitário [online], v.9, n.1, pp. 37-49, 2008 [consult. 26 mai. 2023]. Disponível em doi https://doi.org/10.11606/issn.2316-9044.v9i1p37-49 p // g/ / p ___________.. “Por uma ecologia dos direitos humanos”, Revista Opinião Jurídica [online], ano 13, n. 20, pp. 324-340, 2017a [consult. 26 mai. 2023]. Disponível em doi https://doi.org/10.12662/2447-6641oj.v15i20.p324- 340.2017 ___________. “Periferias da modernidade”, Revista Direito Mackenzie [online], v. 11, n. 2, pp. 39-47, 2017b [consult. 26 mai. 2023]. Disponível em doi https://doi.org/10.5935/2317-2622/direitomackenzie.v11n2p39-47 Ü p g p ______.. “Por uma ecologia dos direitos humanos”, Revista Opinião Jurídica [online], ano 13, n. 20, pp. 324-340, 2017a [consult. 26 mai. 2023]. Disponível em doi https://doi.org/10.12662/2447-6641oj.v15i20.p324- 340.2017 “Pe ife ias da mode nidade” Re ista Di eito Macken ie [online] 340.2017 ___________. “Periferias da modernidade”, Revista Direito Mackenzie [online], v. 11, n. 2, pp. 39-47, 2017b [consult. 26 mai. 2023]. Disponível em doi https://doi.org/10.5935/2317-2622/direitomackenzie.v11n2p39-47 Ü DÜRRENMATT, F. Der Auftrag: oder Vom Beobachten des Beobachters der Beobachter. Novelle in vierundzwanzig Sätzen. Diogenes Verlag, Zurique, 1988. FOERSTER, H. von. e BRÖKER, F. M. Cadernos de Dereito Actual Nº 21. Núm. Ordinario, (2023) Teil der Welt: Fraktale einer Ethik - oder Hein von Foersters Tanz mit der Welt, Carl-Auer, Heidelberg, 2014. G1RO. "Geografia da fome: Região Norte do Brasil é a mais impactada pela insegurança alimentar”, G1RO [online], 2022 [consult. 26 mai. 2023]. Disponível em: https://g1.globo.com/ro/rondonia/noticia/2022/06/08/geografia-da-fome- regiao-norte-do-brasil-e-a-mais-impactada-pela-inseguranca- alimentar.ghtml LIMA, F.R.S. e FINCO, M. “Teoria Sistêmica e Direitos Humanos: o Supremo Tribunal Federal e o direito à saúde”, Revista Pensamento Jurídico [online], v. 13, n. 2, 2019 [consult. 26 mai. 2023]. Disponível em https://fadisp.com.br/revista/ojs/index.php/pensamentojuridico/article/view /186 LUHMANN, Niklas. Sociologia do Direito I. Edições Tempo Brasileiro, Rio de Janeiro, 1983. ___________. Ökologische Kommunikation: Kann die moderne Gesellschaft sich auf ökologische Gefährdungen einstellen?. Westdeutscher Verlag, Opladen, 1986. ___________. Beobachtungen der Moderne. VS Verlag für Sozialwissenschaften, Wiesbaden, 1992. ___________. Die Soziologie und der Mensch. Westdeutscher Verlag, Opladen, 1995. ___________. Die Politik der Gesellschaft. Suhrkamp, Frankfurt, 2000. ___________. Introdução à teoria dos sistemas. Vozes, Petrópolis, 2011. ___________. Sistemas sociais: esboço de uma teoria geral. Vozes, Petrópolis, 2016. ___________. e DE GIORGI, R. Teoria della società. Franco Angeli, Milano, 2013. MONCAU, G. "Fome se alastra no Brasil: 6 em cada 10 famílias não têm acesso pleno a alimentos", Brasil de Fato [online], 2022 [consult. 26 mai. 2023]. Disponível em: https://www.brasildefato.com.br/2022/06/08/fome-se-alastra-no- brasil-6-em-cada-10-familias-nao-tem-acesso-pleno-a-comida Õ NAÇÕES UNIDAS BRASIL. "Pequenos agricultores familiares produzem mais de um terço dos alimentos no mundo", Nações Unidas Brasil [online], 2021 [consult. 26 mai. 2023]. Disponível em https://brasil.un.org/pt-br/125880-pequenos- agricultores-familiares-produzem-mais-de-um-ter%C3%A7o-dos-alimentos- no-mundo NAÇÕES UNIDAS BRASIL. "Pequenos agricultores familiares produzem mais de um terço dos alimentos no mundo", Nações Unidas Brasil [online], 2021 [consult. 26 mai. 2023]. Disponível em https://brasil.un.org/pt-br/125880-pequenos- agricultores-familiares-produzem-mais-de-um-ter%C3%A7o-dos-alimentos- no-mundo Raffaele De Giorgi; Aparecida Luzia Alzira Zuin. O direito ao alimento (…) O direito ao alimento (…) NEITSCH, J. "O filósofo do ‘não saber’", entrevista de Raffaele de Giorgi à Joana Neitsch, Gazeta do Povo [online], 2013 [consult. 26 mai. 2023]. Disponível em https://www.gazetadopovo.com.br/vida-publica/justica- direito/entrevistas/o-filosofo-do-nao-saber-7th9nbpgys33b19wynr7l7iz7/ NEVES, Marcelo. “Os Estados no centro e os Estados na periferia. Alguns problemas com a concepção de Estados da sociedade mundial em Niklas Luhmann”, Revista de Informação Legislativa [online], ano 52, n. 206, 2015 [consult. 26 mai. 2023]. Disponível em https://www12 senado leg br/ril/edicoes/52/206/ril v52 n206 p111 pdf ] p https://www12.senado.leg.br/ril/edicoes/52/206/ril_v52_n206_p111.pdf p g p p NITAHARA, A. "Censo Agropecuário: Brasil tem 5 milhões de estabelecimentos rurais", Agência Brasil [online], 2019 [consult. 26 mai. 2023]. Disponível em https://agenciabrasil.ebc.com.br/geral/noticia/2019-10/censo-agropecuario- brasil-tem-5-milhoes-de-estabelecimentos-rurais SARAMAGO, J. Ensaio sobre a cegueira. Companhia das Letras, São Paulo, 1995. ___________. Ensaio sobre a lucidez. Companhia das Letras, São Paulo, 2004. ___________. As intermitências da morte. Companhia das Letras, São Paulo, 2005. SARAMAGO, J. Ensaio sobre a cegueira. Companhia das Letras, São Paulo, 1995. . Ensaio sobre a lucidez. Companhia das Letras, São Paulo, 2004. Ensaio sobre a lucidez. Companhia das Letras, São Paulo, 2004. ___________. Ensaio sobre a lucidez. Companhia das Letras, São Paulo, 2004. ___________. As intermitências da morte. Companhia das Letras, São Paulo, 2005. As intermitências da morte. Companhia das Letras, São Paulo, 2005. 302
https://openalex.org/W4248303768	https://peerj.com/articles/11810v0.3/submission	English	null	Peer Review #1 of "Genome-wide identification of candidate aquaporins involved in water accumulation of pomegranate outer seed coat (v0.2)"	null	2,021	cc-by	22,460	Manuscript to be reviewed Genome-wide identiﬁcation of candidate aquaporins involved in water accumulation of pomegranate outer seed coat Jianjian Liu Equal first author, 1, 2 , Gaihua Qin E Jianrong Zhao Corresp. 1 Jianjian Liu Equal first author, Jianrong Zhao Corresp. 1 1 College of Resource and Environment, Anhui Science and Technology University, Fengyang, China 2 Institute of Horticultural Research(Key Laboratory of Genetic Improvement and Ecophysiology of Horticultural Crop, Anhui Province), Anhui Academy of Agricultural Sciences, Hefei, China 2 Institute of Horticultural Research(Key Laboratory of Genetic Improvement and Ecophysiology of Horticultural Crop, Anhui Province), Anhui Academy of Agricultural Sciences, Hefei, China Key Laboratory of Fruit Quality and Developmental Biology, Anhui Academy of Agricultural Sciences, Hefei, China 3 Key Laboratory of Fruit Quality and Developmental Biology, Anhui Academy of Agricultural Sciences, Hefei, China 4 State Key Laboratory for Managing Biotic and Chemical Threats to the Quality and Safety of Agro-products, Zhejiang Academy of Agricultural Sciences, Hangzhou, China 4 State Key Laboratory for Managing Biotic and Chemical Threats to the Quality and Safety of Agro-products, Zhej Hangzhou, China 4 State Key Laboratory for Managing Biotic and Chemical Threats to the Quality and Safety of Agro-products, Zhejiang Academy of Agricultural Sciences, Hangzhou, China Corresponding Author: Jianrong Zhao Email address: zhaojr@ahstu.edu.cn Corresponding Author: Jianrong Zhao Corresponding Author: Jianrong Zhao Email address: zhaojr@ahstu.edu.cn Aquaporins (AQPs) are a class of highly conserved integral membrane proteins that facilitate the uptake and transport of water and other small molecules across cell membranes. However, little is known about AQP genes in pomegranate (Punica granatum L.) and their potential role in water accumulation of the outer seed coat. We identiﬁed 38 PgrAQP genes in the pomegranate genome and divided them into ﬁve subfamilies based on a comparative analysis. Purifying selection played a role in the evolution of PgrAQP genes and a whole-genome duplication event in Myrtales may have contributed to the expansion of PgrTIP, PgrSIP, and PgrXIP genes. Transcriptome data analysis revealed that the PgrAQP genes exhibited diﬀerent tissue-speciﬁc expression patterns. Among them, the transcript abundance of PgrPIPs were signiﬁcantly higher than that of other subfamilies. The mRNA transcription levels of PgrPIP1.3, PgrPIP2.8, and PgrSIP1.2 showed a signiﬁcant linear relationship with water accumulation in seed coats, indicating that PgrPIP1.3/PgrPIP2.8 located in the plasma membrane and PgrSIP1.2 proteins located on the tonoplast may be involved in water accumulation and contribute to the cell expansion of the outer seed coat, which then develops into juicy edible ﬂesh. Overall, our results provided not only information on the characteristics and evolution of PgrAQPs, but also insights on the genetic improvement of outer seed coats. Jianjian Liu Equal first author, Jianrong Zhao Corresp. 1 PgrPIP1.3/PgrPIP2.8 located in the plasma membrane and PgrSIP1.2 proteins located on the tonoplast may be involved in water accumulation and contribute to the cell expansion of the outer seed coat, which then develops into juicy edible ﬂesh. Overall, our results provided not only information on the characteristics and evolution of PgrAQPs, but also insights on the genetic improvement of outer seed coats. PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) Manuscript to be reviewed Manuscript to be reviewed 1 Title: Genome-wide identification of candidate aquaporins involved in water 2 accumulation of pomegranate outer seed coat 3 Jianjian Liu1,2 †, Gaihua Qin2,3, †, Chunyan Liu2,3, Xiuli Liu4, Jie Zhou4, Jiyu Li2,3，Bingxin Lu1, Jianrong 4 Zhao1, * 5 1. College of Resource and Environment, Anhui Science and Technology University, Fengyang, China. 6 2. Institute of Horticultural Research, Anhui Academy of Agricultural Sciences (Key Laboratory of 7 Genetic Improvement and Ecophysiology of Horticultural Crop, Anhui Province), Hefei, China. 8 3. Key Laboratory of Fruit Quality and Developmental Biology, Anhui Academy of Agricultural Sciences, 9 Hefei, China. 1 Title: Genome-wide identification of candidate aquaporins involved in water 2 accumulation of pomegranate outer seed coat 3 Jianjian Liu1,2 †, Gaihua Qin2,3, †, Chunyan Liu2,3, Xiuli Liu4, Jie Zhou4, Jiyu Li2,3，Bingxin Lu1, Jianrong 4 Zhao1, * 5 1. College of Resource and Environment, Anhui Science and Technology University, Fengyang, China. 6 2. Institute of Horticultural Research, Anhui Academy of Agricultural Sciences (Key Laboratory of 7 Genetic Improvement and Ecophysiology of Horticultural Crop, Anhui Province), Hefei, China. 8 3. Key Laboratory of Fruit Quality and Developmental Biology, Anhui Academy of Agricultural Sciences, 9 Hefei, China. 10 4. State Key Laboratory for Managing Biotic and Chemical Threats to the Quality and Safety of Agro- 11 products, Zhejiang Academy of Agricultural Sciences, Hangzhou, China. 12 † These authors have contributed equally to this work and share first authorship. 13 * Corresponding author: 14 Jianrong Zhao, College of Resource and Environment, Anhui Science and Technology University, 3 Jianjian Liu1,2 †, Gaihua Qin2,3, †, Chunyan Liu2,3, Xiuli Liu4, Jie Zhou4, Jiyu Li2,3，Bingxin Lu1, Jianrong 4 Zhao1, * 5 1. College of Resource and Environment, Anhui Science and Technology University, Fengyang, China. 6 2. Institute of Horticultural Research, Anhui Academy of Agricultural Sciences (Key Laboratory of 7 Genetic Improvement and Ecophysiology of Horticultural Crop, Anhui Province), Hefei, China. 5 1. College of Resource and Environment, Anhui Science and Technology University, Fengyang, China. 6 2. Institute of Horticultural Research, Anhui Academy of Agricultural Sciences (Key Laboratory of 7 Genetic Improvement and Ecophysiology of Horticultural Crop, Anhui Province), Hefei, China. 8 3. Key Laboratory of Fruit Quality and Developmental Biology, Anhui Academy of Agricultural Sciences, 9 Hefei, China. 8 3. Key Laboratory of Fruit Quality and Developmental Biology, Anhui Academy of Agricultural Sciences, 9 Hefei, China. 9 Hefei, China. 10 4. Manuscript to be reviewed State Key Laboratory for Managing Biotic and Chemical Threats to the Quality and Safety of Agro- 11 products, Zhejiang Academy of Agricultural Sciences, Hangzhou, China. 12 † These authors have contributed equally to this work and share first authorship. 13 * Corresponding author: 14 Jianrong Zhao, College of Resource and Environment, Anhui Science and Technology University, 15 Fengyang, China. 16 E-mail: zhaojr@ahstu.edu.cn 17 18 Abstract 19 Aquaporins (AQPs) are a class of highly conserved integral membrane proteins that 20 facilitate the uptake and transport of water and other small molecules across cell 21 membranes. However, little is known about AQP genes in pomegranate (Punica 22 granatum L.) and their potential role in water accumulation of the outer seed coat. We PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) 10 4. State Key Laboratory for Managing Biotic and Chemical Threats to the Quality and Safety of Agro- 11 products, Zhejiang Academy of Agricultural Sciences, Hangzhou, China. 10 4. State Key Laboratory for Managing Biotic and Chemical Threats to the Quality and Safety of Agro- 11 products, Zhejiang Academy of Agricultural Sciences, Hangzhou, China. These authors have contributed equally to this work and share first authorship. Manuscript to be reviewed 23 identified 38 PgrAQP genes in the pomegranate genome and divided them into five 24 subfamilies based on a comparative analysis. Purifying selection played a role in the 25 evolution of PgrAQP genes and a whole-genome duplication event in Myrtales may 26 have contributed to the expansion of PgrTIP, PgrSIP, and PgrXIP genes. Transcriptome 27 data analysis revealed that the PgrAQP genes exhibited different tissue-specific 28 expression patterns. Among them, the transcript abundance of PgrPIPs were 29 significantly higher than that of other subfamilies. The mRNA transcription levels of 30 PgrPIP1.3, PgrPIP2.8, and PgrSIP1.2 showed a significant linear relationship with water 31 accumulation in seed coats, indicating that PgrPIP1.3/PgrPIP2.8 located in the plasma 32 membrane and PgrSIP1.2 proteins located on the tonoplast may be involved in water 33 accumulation and contribute to the cell expansion of the outer seed coat, which then 34 develops into juicy edible flesh. Overall, our results provided not only information on 35 the characteristics and evolution of PgrAQPs, but also insights on the genetic 36 improvement of outer seed coats. 23 identified 38 PgrAQP genes in the pomegranate genome and divided them into five 24 subfamilies based on a comparative analysis. Purifying selection played a role in the 25 evolution of PgrAQP genes and a whole-genome duplication event in Myrtales may 26 have contributed to the expansion of PgrTIP, PgrSIP, and PgrXIP genes. Transcriptome 27 data analysis revealed that the PgrAQP genes exhibited different tissue-specific 28 expression patterns. Among them, the transcript abundance of PgrPIPs were 29 significantly higher than that of other subfamilies. The mRNA transcription levels of 30 PgrPIP1.3, PgrPIP2.8, and PgrSIP1.2 showed a significant linear relationship with water 31 accumulation in seed coats, indicating that PgrPIP1.3/PgrPIP2.8 located in the plasma 32 membrane and PgrSIP1.2 proteins located on the tonoplast may be involved in water 33 accumulation and contribute to the cell expansion of the outer seed coat, which then 34 develops into juicy edible flesh. Overall, our results provided not only information on 35 the characteristics and evolution of PgrAQPs, but also insights on the genetic 36 improvement of outer seed coats. 38 Introduction 39 Pomegranate (Punica granatum L.) is an important economic fruit tree species due 40 to its functional and nutraceutical properties, and it is widely consumed as a fruit, juice, 41 wine, and medicine (Johanningsmeier & Harris 2011; Patel et al. 2008). Pomegranate is 42 native to Iran, the northern India side of the Himalayan Mountains, and is widely 43 planted in Mediterranean-like climates around the world, including Tunisia, Turkey, 43 planted in Mediterranean-like climates around the world, including Tunisia, Turkey, PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) Manuscript to be reviewed 44 Spain, Egypt, Iran, Morocco, the USA, China, India, Argentina, Israel, and South Africa 45 (Qin et al. 2017). The size of the juicy outer seed coat determines the edible quality of 46 the fruit, so it has become an important fruit characteristic. Notably, the morphological 47 characteristics of the pomegranate seed showed a compressed inner seed coat and an 48 expanded outer seed coat, making it an appealing model for studying development of 49 seed coats (Luo et al. 2020; Niu et al. 2018; Qin et al. 2020). 44 Spain, Egypt, Iran, Morocco, the USA, China, India, Argentina, Israel, and South Africa 45 (Qin et al. 2017). The size of the juicy outer seed coat determines the edible quality of 46 the fruit, so it has become an important fruit characteristic. Notably, the morphological 47 characteristics of the pomegranate seed showed a compressed inner seed coat and an 48 expanded outer seed coat, making it an appealing model for studying development of 49 seed coats (Luo et al. 2020; Niu et al. 2018; Qin et al. 2020). 44 Spain, Egypt, Iran, Morocco, the USA, China, India, Argentina, Israel, and South Africa 45 (Qin et al. 2017). The size of the juicy outer seed coat determines the edible quality of 46 the fruit, so it has become an important fruit characteristic. Notably, the morphological 47 characteristics of the pomegranate seed showed a compressed inner seed coat and an 48 expanded outer seed coat, making it an appealing model for studying development of 49 seed coats (Luo et al. 2020; Niu et al. 2018; Qin et al. 2020). 50 The development of the seed coat is accompanied by the import of organic and 51 inorganic nutrients, including sugars, organic acids, cellulose, and water in developing 52 seeds (Qin et al. 2020; Uçar & Karagöz 2009; Zarei et al. 2016; ZHOU et al. 2007b). The 53 symplastic pathway is one of the main pathways by which water moves, which is 54 mediated by integral membrane proteins called aquaporins (AQPs), a major intrinsic 55 protein family (Adams & Wendel 2005; Lian et al. 2004; Suga et al. 2002). AQPs have 56 been shown to transport small molecules such as carbon dioxide, glycerol, ammonia, 57 urea, hydrogen peroxide, and selenite (Ivanov et al. 2007; Yasui et al. 1999; Zwiazek et 58 al. 2017a). Manuscript to be reviewed 50 The development of the seed coat is accompanied by the import of organic and 51 inorganic nutrients, including sugars, organic acids, cellulose, and water in developing 52 seeds (Qin et al. 2020; Uçar & Karagöz 2009; Zarei et al. 2016; ZHOU et al. 2007b). The 53 symplastic pathway is one of the main pathways by which water moves, which is 54 mediated by integral membrane proteins called aquaporins (AQPs), a major intrinsic 55 protein family (Adams & Wendel 2005; Lian et al. 2004; Suga et al. 2002). AQPs have 56 been shown to transport small molecules such as carbon dioxide, glycerol, ammonia, 57 urea, hydrogen peroxide, and selenite (Ivanov et al. 2007; Yasui et al. 1999; Zwiazek et 58 al. 2017a). 59 The general AQP structure is highly conserved in plants, and it is predicted to 60 consist of six transmembrane (TM) α-helices (H1 to H6) and two half-helices 61 (Deshmukh et al. 2015; Lee et al. 2005; Tornroth-Horsefield et al. 2006). These 62 transmembrane α-helices are linked by five short loops (Loops A to E), Loops B and E 63 contain the signature sequence Asparagine-Proline-Alanine (NPA) motif, which has the 64 primary function of forming water-selective channels (Chaumont et al. 2001; Gupta & Manuscript to be reviewed 65 Sankararamakrishnan 2009; Wallace & Roberts 2004). Another important secondary 66 structure known as an aromatic/arginine (ar/R) selectivity filter is formed by four 67 residues contributed by transmembrane helices H2/H5 and the loop LE (Azad et al. 68 2016; Deshmukh et al. 2015; Hove & Bhave 2011; Tornroth-Horsefield et al. 2006). These 69 two structures predominantly determine the specificity of solute transport and 70 transport rate (Deshmukh et al. 2015; Lee et al. 2005; Tornroth-Horsefield et al. 2006). 71 According to sequence similarity and protein subcellular localization, AQPs in higher 72 plant can be classified into five distinct subfamilies: plasma membrane intrinsic proteins 73 (PIPs), tonoplast intrinsic proteins (TIPs), nodulin 26-like intrinsic proteins (NIPs) 74 (Pommerrenig et al. 2015), small basic intrinsic proteins (SIPs), and unrecognized (or X) 75 intrinsic proteins (XIPs)(Chaumont et al. 2001; Danielson & Johanson 2010; Kaldenhoff 76 & Fischer 2006; Khabudaev et al. 2014; Quigley et al. 2002). 65 Sankararamakrishnan 2009; Wallace & Roberts 2004). Another important secondary 66 structure known as an aromatic/arginine (ar/R) selectivity filter is formed by four 67 residues contributed by transmembrane helices H2/H5 and the loop LE (Azad et al. 68 2016; Deshmukh et al. 2015; Hove & Bhave 2011; Tornroth-Horsefield et al. 2006). These 69 two structures predominantly determine the specificity of solute transport and 70 transport rate (Deshmukh et al. 2015; Lee et al. 2005; Tornroth-Horsefield et al. 2006). 71 According to sequence similarity and protein subcellular localization, AQPs in higher 72 plant can be classified into five distinct subfamilies: plasma membrane intrinsic proteins 73 (PIPs), tonoplast intrinsic proteins (TIPs), nodulin 26-like intrinsic proteins (NIPs) 74 (Pommerrenig et al. 2015), small basic intrinsic proteins (SIPs), and unrecognized (or X) 75 intrinsic proteins (XIPs)(Chaumont et al. 2001; Danielson & Johanson 2010; Kaldenhoff 76 & Fischer 2006; Khabudaev et al. 2014; Quigley et al. 2002). 77 The AQP gene family has been widely studied in numerous plant species, such as 78 Arabidopsis thaliana (Quigley et al. 2002), Oryza sativa (Sakurai et al. 2005), Glycine max 79 (Zhang et al. 2013), cotton (Li et al. 2019), Populus trichocarpa (Gupta & 80 Sankararamakrishnan 2009), and grape (Fouquet et al. 2008), by blasting whole genome 81 sequences. Previous studies have demonstrated that AQPs in higher plants play 82 important roles in various physiological and development processes, such as cell and 83 tissue expansion, fiber development, flower pollination, and seed development (Azad et 84 al. Manuscript to be reviewed 86 vulgaris), PvPIPs played a role in the import of water and nutrients by phloem-mediated 87 transport and water recycling in the xylem in developing seeds (Terashima & Ono 2002). 88 In rice, OsPIP1;1 and OsPIP1;3 functions as water channels. Over-expression of 89 OsPIP1;1 could increase rice yield and seed germination. Similarly, the overexpression 90 of OsPIP1;3 promoted the speed of seed germination under drought conditions (Liu et 91 al. 2013; Liu et al. 2007). In Pisum sativum L., PsNIP1 showed high expression level in 92 developing pea seed coats, and the overexpression of PsNIP1 increased the uptake of 93 water and glycerol (Schuurmans et al. 2003; Zhou et al. 2007a). Interestingly, the highly 94 correlations between the expression of AtTIP3/AtTIP1 and seed germination stages (a 95 rapid imbibition of desiccated tissues and embryo growth, respectively) was observed, 96 which also provided insight into the influence of AQPs on the seed developmental 97 process (Gattolin et al. 2011). However, little is known about the role of AQPs on seed 98 coat development. 86 vulgaris), PvPIPs played a role in the import of water and nutrients by phloem-mediated 87 transport and water recycling in the xylem in developing seeds (Terashima & Ono 2002). 88 In rice, OsPIP1;1 and OsPIP1;3 functions as water channels. Over-expression of 89 OsPIP1;1 could increase rice yield and seed germination. Similarly, the overexpression 90 of OsPIP1;3 promoted the speed of seed germination under drought conditions (Liu et 91 al. 2013; Liu et al. 2007). In Pisum sativum L., PsNIP1 showed high expression level in 92 developing pea seed coats, and the overexpression of PsNIP1 increased the uptake of 93 water and glycerol (Schuurmans et al. 2003; Zhou et al. 2007a). Interestingly, the highly 94 correlations between the expression of AtTIP3/AtTIP1 and seed germination stages (a 95 rapid imbibition of desiccated tissues and embryo growth, respectively) was observed, 96 which also provided insight into the influence of AQPs on the seed developmental 97 process (Gattolin et al. 2011). However, little is known about the role of AQPs on seed 98 coat development. 99 In this study, we identified 38 PgrAQP genes in the pomegranate genome, then 100 conducted gene structure, phylogenetics, and evolutionary divergence analyses. The 101 expression patterns of PgrAQP genes in different tissues and seed developmental stages 102 were analyzed,and the potential function of PgrPIP genes in response to drought stress 103 was also identified. Manuscript to be reviewed 2004; Eisenbarth & Weig 2005; Gattolin et al. 2011; Soto et al. 2008; Soto et al. 2010; 85 Vander Willigen et al. 2006; Wudick et al. 2014). For instance, in French bean (Phaseolus 77 The AQP gene family has been widely studied in numerous plant species, such as 78 Arabidopsis thaliana (Quigley et al. 2002), Oryza sativa (Sakurai et al. 2005), Glycine max 79 (Zhang et al. 2013), cotton (Li et al. 2019), Populus trichocarpa (Gupta & 80 Sankararamakrishnan 2009), and grape (Fouquet et al. 2008), by blasting whole genome 81 sequences. Previous studies have demonstrated that AQPs in higher plants play 82 important roles in various physiological and development processes, such as cell and 83 tissue expansion, fiber development, flower pollination, and seed development (Azad et 84 al. 2004; Eisenbarth & Weig 2005; Gattolin et al. 2011; Soto et al. 2008; Soto et al. 2010; 85 Vander Willigen et al. 2006; Wudick et al. 2014). For instance, in French bean (Phaseolus PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) Manuscript to be reviewed 107 PgrAQP. In addition, this study also provided useful information for further 108 investigation of the molecular mechanism by which PgrAQP functions in seed coat 109 development. 107 PgrAQP. In addition, this study also provided useful information for further 108 investigation of the molecular mechanism by which PgrAQP functions in seed coat 109 development. Manuscript to be reviewed Furthermore, the candidate genes contributing to the water 104 accumulation in the seed coat were selected based on a correlation analysis of gene 105 expression and water accumulation in seed coats. Our findings provided fundamental 106 information about the gene structure, phylogenetics, and evolutionary divergence of PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) Manuscript to be reviewed 127 PSORT (http://www.genscript.com/psort/wolf_ psort.html) Cell-PLoc 2.0 128 (http://www.csbio.sjtu.edu.cn/bioinf/Cell-PLoc-2/) 129 2.2 Sequence alignment of PgrAQP genes, phylogenetic analysis, and classification 130 The predicted plant AQP protein sequences were aligned using the ClustalW tool 131 in MEGA7.0 (Kumar et al. 2016). Arabidopsis thaliana, grape (Vitis vinifera), Populus 132 trichocarpa and eucalyptus (Eucalyptus grandis) AQP protein sequences were collected 133 from NCBI (https://www.ncbi.nlm.nih.gov/). A phylogenic tree was constructed with 134 MEGA7.0 using the neighbor-joining method and bootstrap parameter with 1,000 135 replicates. For this study, Pomegranate AQPs were named according to the sequence 136 homology and phylogenetic relationships with Populus trichocarpa . According to the 137 name of the best hit gene in Populus trichocarpa, the subfamily classification (PIP, NIP, 138 SIP, TIP, and XIP) and corresponding names of AQPs are shown in Fig. 1. TBtools 139 software was used to determine the localization of PgrAQP genes on pomegranate 140 chromosomes. 127 PSORT (http://www.genscript.com/psort/wolf_ psort.html) Cell-PLoc 2.0 128 (http://www.csbio.sjtu.edu.cn/bioinf/Cell-PLoc-2/) 111 2.1 Identification of the PgrAQP genes in the pomegranate genome 112 The whole genome sequences of pomegranate were downloaded from NCBI 113 Genome (https://www.ncbi.nlm.nih.gov/genome/?term=Punica+granatum+L). Predicted 114 protein sequences were obtained using TBtools with the following sets: Sequence 115 Toolkit: Batch translate CDS to protein (Chen et al. 2020). The amino acid sequences of 116 the AQPs in Arabidopsis and Populus trichocarpa were employed as queries to blast 117 searches against the whole-genome sequences in pomegranate using TBtools with a 118 query over 50% and e-value less than 0.01 (Quigley et al. 2002). The candidate 119 sequences were submitted to NCBI for EST blast searches. Finally, the AQP protein 120 sequences of pomegranate were examined to verify the presence of the characteristic 121 MIP and transmembrane helical domains using the SMART program 112 The whole genome sequences of pomegranate were downloaded from NCBI 113 Genome (https://www.ncbi.nlm.nih.gov/genome/?term=Punica+granatum+L). Predicted 114 protein sequences were obtained using TBtools with the following sets: Sequence 115 Toolkit: Batch translate CDS to protein (Chen et al. 2020). The amino acid sequences of 116 the AQPs in Arabidopsis and Populus trichocarpa were employed as queries to blast 117 searches against the whole-genome sequences in pomegranate using TBtools with a 118 query over 50% and e-value less than 0.01 (Quigley et al. 2002). The candidate 119 sequences were submitted to NCBI for EST blast searches. Finally, the AQP protein 120 sequences of pomegranate were examined to verify the presence of the characteristic 121 MIP and transmembrane helical domains using the SMART program 122 (http://smart.embl-heidelberg.de/) and TMHMM 123 (http://www.cbs.dtu.dk/services/TMHMM/) (Finn et al. 2014; Letunic et al. 2020). The 124 information of PgrAQP gene family (protein length, molecular weight, and isoelectric 125 point) were determined using Expasy (https://web.expasy.org/compute_pi/) (Artimo et 126 al. 2012). The subcellular localization of the PgrAQP proteins was predicted by WoLF PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) Manuscript to be reviewed 148 motifs, ar/R filter, and Froger’s residue positions were manually identified based on 149 multiple sequence alignments of pomegranate AQPs with heterologous AQPs of 150 Arabidopsis (Kumar et al. 2016). TBtools software was used to construct a chromosome 151 localization map of AQP family genes. 148 motifs, ar/R filter, and Froger’s residue positions were manually identified based on 149 multiple sequence alignments of pomegranate AQPs with heterologous AQPs of 150 Arabidopsis (Kumar et al. 2016). TBtools software was used to construct a chromosome 151 localization map of AQP family genes. 141 2.3 Gene structure and conserved motif analysis of PgrAQPs 142 The conserved motifs in the proteins were identified using Multiple Expectation 143 Maximization for Motif Elucidation (MEME v4.12.0, http://meme-suite.org/tools/meme) 144 with the following parameters: maximum number of motifs, 10; width of optimum 145 motif, ≥15 and ≤50 (Bailey et al. 2009). The gene exon–intron structures of AQPs were 146 confirmed using the GSDS database by aligning the open reading frame (ORF) with 147 their corresponding genomic sequences (Hu et al. 2015a). Sequences representing NPA 142 The conserved motifs in the proteins were identified using Multiple Expectation 143 Maximization for Motif Elucidation (MEME v4.12.0, http://meme-suite.org/tools/meme) 144 with the following parameters: maximum number of motifs, 10; width of optimum 145 motif, ≥15 and ≤50 (Bailey et al. 2009). The gene exon–intron structures of AQPs were 146 confirmed using the GSDS database by aligning the open reading frame (ORF) with 147 their corresponding genomic sequences (Hu et al. 2015a). Sequences representing NPA 52 2.4 Gene duplication and synteny analysis of the PgrAQP gene famil 153 Gene duplication events of PgrAQP genes were analyzed using TBtools with the 154 following parameters: Blast Compare Two Seqs, Outfml: Table, NumofThreads: 2, E- 155 value:1e-5, NumofHits: 5, NumofAligns: 5; File Merger For MCScanX, Merge Mode: 156 GtfGff2 SimGxf (Chen et al. 2020). Duplication types were divided into whole-genome 157 (WGD), segmental, and tandem duplications (Qiao et al. 2015). Tandem duplicated 158 genes were defined as two homologous genes within a range of 100-kb and sequence 159 alignment coverage over 75% (Gu et al. 2002; Wang et al. 2010; Yang et al. 2008). The 160 nonsynonymous substitution ratios (Ka), synonymous substitution ratios (Ks), and 161 Ka/Ks ratios of the PgrAQP family genes were calculated using the TBtools simple Ka/Ks 162 calculator program, with the following parameters: Simple Ka/Ks Calculator (NG). The 163 Ks value was used to calculate the divergence time of duplication events (T=Ks/2R Mya, 164 Millions of years), where R is the rate of synonymous substitutions, R =1.5 × 10-8 165 substitutions per synonymous site per year for dicotyledonous plants and T refers to 166 divergence time (Edlund et al. 2004). The microsyntenic relationship of AQP genes 167 between pomegranate, grape, Arabidopsis, and eucalyptus was determined using 168 TBtools Dual Systeny Plot for MCscanX program (Chen et al. 2020). 153 Gene duplication events of PgrAQP genes were analyzed using TBtools with the 154 following parameters: Blast Compare Two Seqs, Outfml: Table, NumofThreads: 2, E- 155 value:1e-5, NumofHits: 5, NumofAligns: 5; File Merger For MCScanX, Merge Mode: 156 GtfGff2 SimGxf (Chen et al. 2020). Duplication types were divided into whole-genome 157 (WGD), segmental, and tandem duplications (Qiao et al. 2015). Tandem duplicated 158 genes were defined as two homologous genes within a range of 100-kb and sequence 159 alignment coverage over 75% (Gu et al. 2002; Wang et al. 2010; Yang et al. 2008). The 160 nonsynonymous substitution ratios (Ka), synonymous substitution ratios (Ks), and 161 Ka/Ks ratios of the PgrAQP family genes were calculated using the TBtools simple Ka/Ks 162 calculator program, with the following parameters: Simple Ka/Ks Calculator (NG). The 163 Ks value was used to calculate the divergence time of duplication events (T=Ks/2R Mya, 164 Millions of years), where R is the rate of synonymous substitutions, R =1.5 × 10-8 165 substitutions per synonymous site per year for dicotyledonous plants and T refers to 166 divergence time (Edlund et al. 2004). Manuscript to be reviewed Manuscript to be reviewed 169 2.5 Plant materials and analysis of transcript profiles of PgrAQP genes 170 Two major pomegranate cultivars, ‘Dabenzi’ (a hard-seeded cultivar) and ‘Tunisia’ 171 (a soft-seeded cultivar), were planted in Anhui Province (Hefei, 31°51′9.05″N, 172 117°06′34.33″E), China and grown under the same fertilization and irrigation conditions. 173 Seeds from five fruits of the pomegranate cultivar ‘Dabenzi’ and ‘Tunisia’ were 174 collected at 50, 95, and 140 days after pollination (DAP). For outer seed coats, the seeds 175 from fruits collected 95 DAP and 140 DAP was used because it was difficult to visually 176 distinguish the inner and outer seed coats of seeds were collected at 50 DAP. Three 177 biological replicates were taken for RNA sequencing. For each treatment, the fresh 178 weights of the total inner and outer seed coats were recorded and immediately frozen in 179 liquid nitrogen. Part of the inner and outer seed coats were stored at -80°C until they 180 were used for transcriptome sequencing. RNA extraction was performed as described 181 previously (Qin et al. 2020). Briefly, the total RNA was isolated using a Plant RNeasy 182 Mini Elute Cleanup kit (Qiagen, Beijing, China) and the sequencing libraries were 183 prepared using the NEBNext UltraTM RNA Library Prep Kit for Illumina (New England 184 Biolabs, USA) following the manufacturer’s protocols. Transcriptome sequencing was 185 conducted using an Illumina HiSeq 2000 platform. 52 2.4 Gene duplication and synteny analysis of the PgrAQP gene famil The microsyntenic relationship of AQP genes 167 between pomegranate, grape, Arabidopsis, and eucalyptus was determined using 168 TBtools Dual Systeny Plot for MCscanX program (Chen et al. 2020). 169 2.5 Plant materials and analysis of transcript profiles of PgrAQP genes 186 The remaining samples were used for the measurement of water accumulation. The 187 water accumulation (%) in the different seed coats was determined by the fresh weight 188 (FW) and drought weight (DW) using the formula (FW-DW)/FW×100. The freeze- 189 drying analysis was performed using GOLD-SIM FD83 (SIM International group co. ltd, PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) Manuscript to be reviewed 190 USA). The experiments were repeated three times. 191 The abundances of pomegranate AQP transcripts, in the root, flower, leaf, and three 192 developmental stages of the peel and seed coat (inner and outer seed coat), were 193 collected from the microarray data previously published by Qin et al., (Qin et al. 2017). 194 Transcriptional abundances of AQP genes were estimated using the fragments per 195 kilobase of exon per million mapped reads (FRKM) method, and a heat map was 196 generated based on the log2 FRKM transformation value using TBtools (Hu et al. 2018), 197 with the following protocol: Starting from the “Graphics”, click “Heatmap lllustrator”, 198 select the Heatmap button, then set input files in each filed, click “Start” and graph will 199 be generated. 200 2.6 Verification of PgrAQP genes function responding to water deficit in 201 pomegranate root 202 For the analysis of PgrAQP genes expression patterns in response to water deficit, 203 the PEG treatment experiment was performed. For the hydroponic culture experiment, 204 the full Hoagland nutrient solution contained 20% (v/v) polyethylene glycol 6000 205 (PEG6000). The solution pH was adjusted to 5.5. The roots samples were collected after 206 0 h, 1 h, 6 h, 12 h and 24 h of treatment, and were quickly frozen in liquid nitrogen and 207 stored at -80 for subsequent RNA isolation. 208 For performing qRT-PCR analysis, approximately 2 ug of NDA-eliminated total 209 RNA from pomegranate roots were used to synthesize cDNA using a reverse 210 transcription kit (TaKaRa). qRT-PCR was performed on the Applied Biosystems (ABI) p 190 USA). The experiments were repeated three times. 190 USA). The experiments were repeated three times. 191 The abundances of pomegranate AQP transcripts, in the root, flower, leaf, and three 192 developmental stages of the peel and seed coat (inner and outer seed coat), were 193 collected from the microarray data previously published by Qin et al., (Qin et al. 2017). 194 Transcriptional abundances of AQP genes were estimated using the fragments per 195 kilobase of exon per million mapped reads (FRKM) method, and a heat map was 196 generated based on the log2 FRKM transformation value using TBtools (Hu et al. 2018), 197 with the following protocol: Starting from the “Graphics”, click “Heatmap lllustrator”, 198 select the Heatmap button, then set input files in each filed, click “Start” and graph will 199 be generated. Manuscript to be reviewed 200 2.6 Verification of PgrAQP genes function responding to water deficit in 201 pomegranate root 200 2.6 Verification of PgrAQP genes function responding to water deficit in 201 pomegranate root 202 For the analysis of PgrAQP genes expression patterns in response to water deficit, 203 the PEG treatment experiment was performed. For the hydroponic culture experiment, 204 the full Hoagland nutrient solution contained 20% (v/v) polyethylene glycol 6000 205 (PEG6000). The solution pH was adjusted to 5.5. The roots samples were collected after 206 0 h, 1 h, 6 h, 12 h and 24 h of treatment, and were quickly frozen in liquid nitrogen and 207 stored at -80 for subsequent RNA isolation. 202 For the analysis of PgrAQP genes expression patterns in response to water deficit, 203 the PEG treatment experiment was performed. For the hydroponic culture experiment, 204 the full Hoagland nutrient solution contained 20% (v/v) polyethylene glycol 6000 205 (PEG6000). The solution pH was adjusted to 5.5. The roots samples were collected after 206 0 h, 1 h, 6 h, 12 h and 24 h of treatment, and were quickly frozen in liquid nitrogen and 207 stored at -80 for subsequent RNA isolation. 208 For performing qRT-PCR analysis, approximately 2 ug of NDA-eliminated total 209 RNA from pomegranate roots were used to synthesize cDNA using a reverse 210 transcription kit (TaKaRa). qRT-PCR was performed on the Applied Biosystems (ABI) Manuscript to be reviewed 211 StepOne Plus PCR system (Applied Biosystems) using the LightCylcer 96 SYB GREEN I 212 Master (Roche, Indianapolis, IN, USA) in a 20 ul reaction solution. The PgrAQP genes 213 that showed higher expressed in the roots from the RNA-Seq data were selected for 214 qRT-PCR analysis. The relative transcript abundance of each gene was normalized to 215 the pomegranate actin (OWM91407) with the cycle threshold (Ct) 2(-ΔΔCt) method. Three 216 biological replicates and technical replicates were used for each gene. All the specific 217 primers used for each target gene are listed in Table. S1. 218 2.7 Statistical analysis 219 The data were analyzed by ANOVA (SPSS 16.0; SPSS Inc., Chicago, IL, USA), 220 followed by Turkey’s test (P<0.05) to determine differences of inner and outer seed coats 221 The data represent the mean ± SE of three independent biological replicates. 222 3. Results 223 3.1 Genome-wide identification of PgrAQP genes in pomegranate reveal each 224 Aquaporin subfamily presents particular physicochemical characteristics 225 A total of 38 PgrAQP members were identified in the pomegranate genome via a 226 genome-wide search using the AQP protein sequences in Arabidopsis and Populus 227 trichocarpa as queries (Quigley et al. 2002). A subsequent conserved domain analysis 228 also confirmed all of the predicted AQPs (Fig. S1). The characteristics of PgrAQP family 229 genes are shown in Table. S2, including the gene ID, protein length, relative molecular 230 weight, transmembrane domains (TMDs), subcellular localization, and calculated 211 StepOne Plus PCR system (Applied Biosystems) using the LightCylcer 96 SYB GREEN I 212 Master (Roche, Indianapolis, IN, USA) in a 20 ul reaction solution. The PgrAQP genes 213 that showed higher expressed in the roots from the RNA-Seq data were selected for 214 qRT-PCR analysis. The relative transcript abundance of each gene was normalized to 215 the pomegranate actin (OWM91407) with the cycle threshold (Ct) 2(-ΔΔCt) method. Three 216 biological replicates and technical replicates were used for each gene. All the specific 217 primers used for each target gene are listed in Table. S1. 211 StepOne Plus PCR system (Applied Biosystems) using the LightCylcer 96 SYB GREEN I 212 Master (Roche, Indianapolis, IN, USA) in a 20 ul reaction solution. The PgrAQP genes 213 that showed higher expressed in the roots from the RNA-Seq data were selected for 214 qRT-PCR analysis. Manuscript to be reviewed 232 length from 245 to 359 amino acid residues, with a relative molecular weight of 22.86 to 233 35.18 kDa, and a calculated pI ranging from 5.06 to 10.16. The average PI value of 234 PgrTIPs was less than other AQPs due to the loss of basic residues in the C-terminal 235 domain. The grand average of hydrophobicity index (GRAVY) was used to evaluated 236 protein hydrophobicity and hydrophilicity. The results showed the GRAVY of PgrAQP 237 proteins were all positive, ranging from 0.203 to 0.967, which indicated that all of 238 PgrAQP proteins were hydrophobic. Furthermore, the lowest average of GRAVY value 239 (0.44) was found in the PgrPIP subfamily, suggesting that this subfamily has better 240 interaction with water molecules. 241 The predicted transmembrane domains (TMDs) showed that most PgrAQP genes 242 (28 of 38, 73.7%) contained six TMDs, 2.7%, 15.8%, and 7.8% of PgrAQP genes contained 243 four, five, and seven TMDs, respectively (Fig. S2). Based on subcellular localization 244 predicted by WoLF PSORT, most PgrPIP and PgrNIP proteins were predicted to 245 localize in plasma membranes, while only PgrPIP2.2 was found in the chloroplast. All 246 PgrTIP proteins were predicted to localized to vacuoles. For PgrSIP proteins, PgrSIP1.1 247 and PgrSIP1.2 were found in the chloroplast and vacuole, respectively. PgrXIP proteins 248 were predicted to localized in plasma membranes. 241 The predicted transmembrane domains (TMDs) showed that most PgrAQP genes 242 (28 of 38, 73.7%) contained six TMDs, 2.7%, 15.8%, and 7.8% of PgrAQP genes contained 243 four, five, and seven TMDs, respectively (Fig. S2). Based on subcellular localization 244 predicted by WoLF PSORT, most PgrPIP and PgrNIP proteins were predicted to 245 localize in plasma membranes, while only PgrPIP2.2 was found in the chloroplast. All 246 PgrTIP proteins were predicted to localized to vacuoles. For PgrSIP proteins, PgrSIP1.1 247 and PgrSIP1.2 were found in the chloroplast and vacuole, respectively. PgrXIP proteins 248 were predicted to localized in plasma membranes. Manuscript to be reviewed The relative transcript abundance of each gene was normalized to 215 the pomegranate actin (OWM91407) with the cycle threshold (Ct) 2(-ΔΔCt) method. Three 216 biological replicates and technical replicates were used for each gene. All the specific 217 primers used for each target gene are listed in Table. S1. 219 The data were analyzed by ANOVA (SPSS 16.0; SPSS Inc., Chicago, IL, USA), 220 followed by Turkey’s test (P<0.05) to determine differences of inner and outer seed coats. 221 The data represent the mean ± SE of three independent biological replicates. 222 3. Results 223 3.1 Genome-wide identification of PgrAQP genes in pomegranate reveal each 224 Aquaporin subfamily presents particular physicochemical characteristics 225 A total of 38 PgrAQP members were identified in the pomegranate genome via a 226 genome-wide search using the AQP protein sequences in Arabidopsis and Populus 227 trichocarpa as queries (Quigley et al. 2002). A subsequent conserved domain analysis 228 also confirmed all of the predicted AQPs (Fig. S1). The characteristics of PgrAQP family 229 genes are shown in Table. S2, including the gene ID, protein length, relative molecular 230 weight, transmembrane domains (TMDs), subcellular localization, and calculated 231 isoelectric point (pI). The 38 predicted PgrAQP genes encoded proteins that varied in PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) Manuscript to be reviewed 253 identified. An unrooted phylogenetic tree was constructed based on the alignments of 254 their amino acid sequences in MEGA 7 using the neighbor-joining method (Kumar et al. 255 2016). By comparing amino acid sequences of PgrAQPs with APQs from three other 256 plant species, 38 PgrAQPs were divided into five different subfamilies, that is, 14 257 PgrTIPs, 13 PgrPIPs, eight PgrNIPs, two PgrSIPs, and one PgrXIP (Fig. 1). The PgrPIPs 258 divided into two major subgroups, PgrPIP1s and PgrPIP2s, which comprised five and 259 eight members, respectively. Furthermore, PgrNIPs formed six subgroups (PgrNIP1, 260 PgrNIP1, PgrNIP4, PgrNIP5, PgrNIP7, and PgrNIP8) in pomegranate. As the largest sub- 261 family, the TIPs members were classified into five subgroups, consisting of seven 262 PgTIP1s, two PgTIP2s, two PgTIP3s, one PgTIP4, and two PgTIP5s. SIPs and XIPs 263 formed one group, containing two and one members, respectively. In the phylogenetic 264 tree, PgrAQP genes were more closely related to AQPs in eucalyptus than that in 265 Arabidopsis and grape, which is in accordance with the evolutionary relationships 266 among these species (Qin et al. 2017). 267 3.3 Gene structure and conserved motif analysis of PgrAQP genes confirm the 253 identified. An unrooted phylogenetic tree was constructed based on the alignments of 254 their amino acid sequences in MEGA 7 using the neighbor-joining method (Kumar et al. 255 2016). By comparing amino acid sequences of PgrAQPs with APQs from three other 256 plant species, 38 PgrAQPs were divided into five different subfamilies, that is, 14 257 PgrTIPs, 13 PgrPIPs, eight PgrNIPs, two PgrSIPs, and one PgrXIP (Fig. 1). The PgrPIPs 258 divided into two major subgroups, PgrPIP1s and PgrPIP2s, which comprised five and 259 eight members, respectively. Furthermore, PgrNIPs formed six subgroups (PgrNIP1, 260 PgrNIP1, PgrNIP4, PgrNIP5, PgrNIP7, and PgrNIP8) in pomegranate. As the largest sub- 261 family, the TIPs members were classified into five subgroups, consisting of seven 262 PgTIP1s, two PgTIP2s, two PgTIP3s, one PgTIP4, and two PgTIP5s. SIPs and XIPs 263 formed one group, containing two and one members, respectively. In the phylogenetic 264 tree, PgrAQP genes were more closely related to AQPs in eucalyptus than that in 265 Arabidopsis and grape, which is in accordance with the evolutionary relationships 266 among these species (Qin et al. 2017). 249 3.2 Phylogenetic characterization of the pomegranate PgrAQP gene family 250 To investigate the evolutionary relationship of pomegranate AQP family genes, a 251 total of 193 AQP protein sequences from the four species studied (38 in pomegranate, 35 252 in Arabidopsis, 33 in grape, 55 in Populus trichocarpa and 40 in eucalyptus) were 250 To investigate the evolutionary relationship of pomegranate AQP family genes, a 251 total of 193 AQP protein sequences from the four species studied (38 in pomegranate, 35 252 in Arabidopsis, 33 in grape, 55 in Populus trichocarpa and 40 in eucalyptus) were 267 3.3 Gene structure and conserved motif analysis of PgrAQP genes confirm the 268 phylogenetic classification 267 3.3 Gene structure and conserved motif analysis of PgrAQP genes confirm the 268 phylogenetic classification 269 Gene structure and conserved motifs can provide information for exploring the 270 evolutionary relationships among a gene family. The number of introns of PgrAQP 271 genes ranged from zero to four, while the length of exons was highly similar for each 272 subfamily (Fig. S3). Among them, most of the PgrTIPs genes had two introns, except for 273 PgrTIP1.6 and PgrTIP1.8, which had only one intron. For PgrPIP genes, most of 269 Gene structure and conserved motifs can provide information for exploring the 270 evolutionary relationships among a gene family. The number of introns of PgrAQP 271 genes ranged from zero to four, while the length of exons was highly similar for each 272 subfamily (Fig. S3). Among them, most of the PgrTIPs genes had two introns, except for 273 PgrTIP1.6 and PgrTIP1.8, which had only one intron. For PgrPIP genes, most of Manuscript to be reviewed 274 members had three introns, whereas PgPIP2.8 contained two introns. The numbers of 275 introns in PgrNIPs ranged from one to four introns. Four out of 10 members had four 276 introns (PgrNIP2.1, PgrNIP4.1, PgrNIP4.2, and PgrNIP3.1), three members had three 277 introns (PgrNIP1.2, PgrNIP5.1, and PgrNIP7.1), and two members had two introns 278 (PgrNIP1.1 and PgrNIP1.3). PgrXIP2.1 had three introns. The SIP family formed a small 279 subfamily, among which, two members had two introns and one member had no 280 introns. 281 By analyzing the intron–exon structure of PgrAQP genes, we found that the 282 number of introns of each AQPs subfamily (PIP, NIP, TIP, SIP, and XIP) are highly 283 conserved and similar when compared with plant species, such as banana, watermelon, 284 chickpea, and sweet orange, suggesting similar intron loss or gain events were 285 experienced in AQP subfamily over the course of evolution (Deokar & Tar'an 2016; Hu 286 et al. 2015b; Martins Cde et al. 2015; Zhou et al. 2019). Multiple sequence alignments 287 showed that PgrNIPs and PgrPIPs were the most diverse (38.5%) and conserved (72.7%) 288 subfamily at the amino acid level, respectively (Table. S3). This finding is consistent 289 with the AQPs from Nicotiana tabacum, suggesting that the function and regulatory 290 mechanism of PgrPIP subfamily genes showed more conserved than PgrNIP subfamily 291 genes in pomegranate (De Rosa et al. 2020; Deokar & Tar'an 2016). Furthermore, the 292 relative conservation of protein sequences and ar/R selectivity filter among the PgrPIP 293 and PgrTIP subfamily genes suggested that these proteins may share a conserved 274 members had three introns, whereas PgPIP2.8 contained two introns. The numbers of 275 introns in PgrNIPs ranged from one to four introns. Four out of 10 members had four 276 introns (PgrNIP2.1, PgrNIP4.1, PgrNIP4.2, and PgrNIP3.1), three members had three 277 introns (PgrNIP1.2, PgrNIP5.1, and PgrNIP7.1), and two members had two introns 278 (PgrNIP1.1 and PgrNIP1.3). PgrXIP2.1 had three introns. The SIP family formed a small 279 subfamily, among which, two members had two introns and one member had no 280 introns. 274 members had three introns, whereas PgPIP2.8 contained two introns. The numbers of 275 introns in PgrNIPs ranged from one to four introns. Manuscript to be reviewed 295 et al. 2019). 296 To detect the structural diversity and provide further support of the grouping of 297 PgrAQPs, a total of 10 conserved motifs were identified (Fig. S3). Generally, motif 298 compositions were conserved within each subfamily. For PgrPIP, eight motifs were 299 found in all family members. Motifs 1, 2, 3, 6, 8, and 10 were common for TIP and NIP 300 subfamily members, except for PgrNIP7.1. Interestingly, motifs 4 and 5 were only 301 identified in the PIP subfamily and similar motifs were found in bread wheat, 302 indicating the PIP subfamily may have unique functions (Madrid-Espinoza et al. 2018). 296 To detect the structural diversity and provide further support of the grouping of 297 PgrAQPs, a total of 10 conserved motifs were identified (Fig. S3). Generally, motif 298 compositions were conserved within each subfamily. For PgrPIP, eight motifs were 299 found in all family members. Motifs 1, 2, 3, 6, 8, and 10 were common for TIP and NIP 300 subfamily members, except for PgrNIP7.1. Interestingly, motifs 4 and 5 were only 301 identified in the PIP subfamily and similar motifs were found in bread wheat, 302 indicating the PIP subfamily may have unique functions (Madrid-Espinoza et al. 2018). Manuscript to be reviewed Four out of 10 members had four 276 introns (PgrNIP2.1, PgrNIP4.1, PgrNIP4.2, and PgrNIP3.1), three members had three 277 introns (PgrNIP1.2, PgrNIP5.1, and PgrNIP7.1), and two members had two introns 278 (PgrNIP1.1 and PgrNIP1.3). PgrXIP2.1 had three introns. The SIP family formed a small 279 subfamily, among which, two members had two introns and one member had no 280 introns. 281 By analyzing the intron–exon structure of PgrAQP genes, we found that the 282 number of introns of each AQPs subfamily (PIP, NIP, TIP, SIP, and XIP) are highly 283 conserved and similar when compared with plant species, such as banana, watermelon, 284 chickpea, and sweet orange, suggesting similar intron loss or gain events were 285 experienced in AQP subfamily over the course of evolution (Deokar & Tar'an 2016; Hu 286 et al. 2015b; Martins Cde et al. 2015; Zhou et al. 2019). Multiple sequence alignments 287 showed that PgrNIPs and PgrPIPs were the most diverse (38.5%) and conserved (72.7%) 288 subfamily at the amino acid level, respectively (Table. S3). This finding is consistent 289 with the AQPs from Nicotiana tabacum, suggesting that the function and regulatory 290 mechanism of PgrPIP subfamily genes showed more conserved than PgrNIP subfamily 291 genes in pomegranate (De Rosa et al. 2020; Deokar & Tar'an 2016). Furthermore, the 292 relative conservation of protein sequences and ar/R selectivity filter among the PgrPIP 293 and PgrTIP subfamily genes suggested that these proteins may share a conserved 294 function in transporting water and other small neutral solutes (Table. S3 and S4) (Zhu 303 3.4 Comparison of substrate-specific residues in PgrAQP proteins 303 3.4 Comparison of substrate-specific residues in PgrAQP proteins 304 The NPA motifs, ar/R selectively filter, and Froger’s positions were identified by 305 multiple sequence alignment between the PgrAQPs and AtAQPs using MEGA 7.0. 306 These highly conserved motifs and positions were critical for the substrate selectively of 307 AQPs (Tornroth-Horsefield et al. 2006). Conserved domain searches using CDD tool 308 from NCBI confirmed all the predicted AQP genes in Pomegranate encoded MIP 309 domains (Fig.S7). As shown in Table. S4, all of PgrTIPs, PIPs, and XIP proteins 310 harbored two conserved NPA domains in both loop B (LB) and loop E (LE). For the 311 PgrNIPs subfamily, except for PgrNIP5.1, the rest of numbers showed the third residue 312 of the first NPA motifs was serine rather than alanine. In addition, PgrSIP1.2 showed 313 the substitution of alanine by threonine in their first NPA motif. 305 multiple sequence alignment between the PgrAQPs and AtAQPs using MEGA 7.0. 306 These highly conserved motifs and positions were critical for the substrate selectively of 307 AQPs (Tornroth-Horsefield et al. 2006). Conserved domain searches using CDD tool 308 from NCBI confirmed all the predicted AQP genes in Pomegranate encoded MIP 309 domains (Fig.S7). As shown in Table. S4, all of PgrTIPs, PIPs, and XIP proteins 310 harbored two conserved NPA domains in both loop B (LB) and loop E (LE). For the 311 PgrNIPs subfamily, except for PgrNIP5.1, the rest of numbers showed the third residue 312 of the first NPA motifs was serine rather than alanine. In addition, PgrSIP1.2 showed 313 the substitution of alanine by threonine in their first NPA motif. 314 The residues of the ar/R selectivity filter and Froger’s position displayed conserved 315 regions within each subfamily, but regions were more variable across different 314 The residues of the ar/R selectivity filter and Froger’s position displayed conserved 315 regions within each subfamily, but regions were more variable across different Manuscript to be reviewed 316 subfamilies (Kayum et al. 2017). For example, all of the members of the PgrPIPs showed 317 conservative residues at the ar/R selectively filter with phenylalanine-histidine- 318 threonine-arginine, which is typical of aquaporin protein structure. The residues of 319 Froger’s position were conserved in PgrPIPs, including S at P2, A at P3, Y at P4, and W 320 at P5, while the P1 position was variable with the Q/M residues. Different residues were 321 observed at the ar/R selectively filter and Froger’s position in PgrNIPs, which had 322 W/G/A/V-V/S/I-A/G-R and F/L/Y-S/T-A-Y/F-I/V/LM residue compositions, respectively. 323 For PgrTIPs members, the residues of P3, P4, and P5 positions were highly conserved, 324 while the P1 and P2 positions were variable residues. The ar/R selectivity filters were 325 variable with H/S/N-I/V-A/P-V/R/C residues. The residues of Froger’s position were 326 conserved in PgrSIP1.1 and PgrSIP1.2, but the ar/R filter showed distinct difference. 316 subfamilies (Kayum et al. 2017). For example, all of the members of the PgrPIPs showed 317 conservative residues at the ar/R selectively filter with phenylalanine-histidine- 318 threonine-arginine, which is typical of aquaporin protein structure. The residues of 319 Froger’s position were conserved in PgrPIPs, including S at P2, A at P3, Y at P4, and W 320 at P5, while the P1 position was variable with the Q/M residues. Different residues were 321 observed at the ar/R selectively filter and Froger’s position in PgrNIPs, which had 322 W/G/A/V-V/S/I-A/G-R and F/L/Y-S/T-A-Y/F-I/V/LM residue compositions, respectively. 323 For PgrTIPs members, the residues of P3, P4, and P5 positions were highly conserved, 324 while the P1 and P2 positions were variable residues. The ar/R selectivity filters were 325 variable with H/S/N-I/V-A/P-V/R/C residues. The residues of Froger’s position were 326 conserved in PgrSIP1.1 and PgrSIP1.2, but the ar/R filter showed distinct difference. 327 The pore diameter and hydrophobicity of AQP proteins determines their substrate 328 specificity (Almasalmeh et al. 2014; Hove & Bhave 2011). The highly conserved amino 329 acid features of AQPs included six transmembrane domains, the NPA domain, and the 330 ar/R selectivity filter(Froger et al. 1998). For example, all of the PgrPIP subfamily 333 Arabidopsis, Brassica rapa, soybean, and chickpea, indicating that the substrate specificity 334 of this subfamily may be more specific than others subfamilies (Deokar & Tar'an 2016; 335 Kayum et al. 2017; Quigley et al. 2002; Zhang et al. 2013). Manuscript to be reviewed Among the different TIP 336 subgroups, the highly conserved NPA motif, ar/R H2, H5, and Forger’s P3 to P5 were PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) Manuscript to be reviewed Furthermore, the highly conserved ar/R filter (H-I-A-V) and Froger’s 338 positions (T-S-A-Y-W) of the PgrTIP1 subfamily were reported to function as urea and 339 H2O2 transporter (Hove & Bhave 2011). In addition, the conserved ar/R filter G-S-G-R 340 residues were found in PgrNIP2.1, and this characteristic was identified as the indicator 341 of Si transporters, indicating that PgrNIP2.1 may be involved in the transport of Si 342 (Deshmukh et al. 2013; Deshmukh et al. 2015; Zhou et al. 2019). In pomegranate, valine 343 was present at position H5 in the ar/R selectivity filters of PgrXIP2.1, suggesting that the 344 hydrophobicity of PgrXIP2.1 is greater than other PgrAQPs subfamily members 345 (Danielson & Johanson 2008; Gupta & Sankararamakrishnan 2009). 346 3.5 Segmental duplication events have contributed to the expansion of the PgrAQP 347 family genes 348 To investigate the localization of PgrAQP genes and duplication events in 349 pomegranate, we anchored the PgrAQPs on chromosomes and conducted a duplication Manuscript to be reviewed 337 observed. Furthermore, the highly conserved ar/R filter (H-I-A-V) and Froger’s 338 positions (T-S-A-Y-W) of the PgrTIP1 subfamily were reported to function as urea and 339 H2O2 transporter (Hove & Bhave 2011). In addition, the conserved ar/R filter G-S-G-R 340 residues were found in PgrNIP2.1, and this characteristic was identified as the indicator 341 of Si transporters, indicating that PgrNIP2.1 may be involved in the transport of Si 342 (Deshmukh et al. 2013; Deshmukh et al. 2015; Zhou et al. 2019). In pomegranate, valine 343 was present at position H5 in the ar/R selectivity filters of PgrXIP2.1, suggesting that the 344 hydrophobicity of PgrXIP2.1 is greater than other PgrAQPs subfamily members 345 (Danielson & Johanson 2008; Gupta & Sankararamakrishnan 2009). 346 3.5 Segmental duplication events have contributed to the expansion of the PgrAQP 347 family genes 348 To investigate the localization of PgrAQP genes and duplication events in 349 pomegranate, we anchored the PgrAQPs on chromosomes and conducted a duplication 350 analysis. The physical position of the PgrAQP genes were found to be unevenly 351 distributed across all pomegranate chromosomes (Fig. 2). The PIP subfamily genes were 352 randomly anchored on chromosomes, except for Chromosome 7, and TIP subfamily 353 genes were found in all chromosomes except Chromosome 4. Aside from Chromosomes 354 2 and 8, NIP group genes were located in each chromosome. Genes in the SIP subfamily 355 were present only on Chromosome 4. 356 We further analyzed the gene duplication modes of PgrAQP genes in pomegranate. 357 As shown in Fig. 2, 47% of the PgrAQP genes had been duplicated by 337 observed. Furthermore, the highly conserved ar/R filter (H-I-A-V) and Froger’s 338 positions (T-S-A-Y-W) of the PgrTIP1 subfamily were reported to function as urea and 339 H2O2 transporter (Hove & Bhave 2011). In addition, the conserved ar/R filter G-S-G-R 340 residues were found in PgrNIP2.1, and this characteristic was identified as the indicator 341 of Si transporters, indicating that PgrNIP2.1 may be involved in the transport of Si 342 (Deshmukh et al. 2013; Deshmukh et al. 2015; Zhou et al. 2019). In pomegranate, valine 343 was present at position H5 in the ar/R selectivity filters of PgrXIP2.1, suggesting that the 344 hydrophobicity of PgrXIP2.1 is greater than other PgrAQPs subfamily members 345 (Danielson & Johanson 2008; Gupta & Sankararamakrishnan 2009). 337 observed. 346 3.5 Segmental duplication events have contributed to the expansion of the PgrAQP 347 family genes 348 To investigate the localization of PgrAQP genes and duplication events in 349 pomegranate, we anchored the PgrAQPs on chromosomes and conducted a duplication 350 analysis. The physical position of the PgrAQP genes were found to be unevenly 351 distributed across all pomegranate chromosomes (Fig. 2). The PIP subfamily genes were 352 randomly anchored on chromosomes, except for Chromosome 7, and TIP subfamily 353 genes were found in all chromosomes except Chromosome 4. Aside from Chromosomes 354 2 and 8, NIP group genes were located in each chromosome. Genes in the SIP subfamily 355 were present only on Chromosome 4. 356 We further analyzed the gene duplication modes of PgrAQP genes in pomegranate. 357 As shown in Fig. 2, 47% of the PgrAQP genes had been duplicated by 356 We further analyzed the gene duplication modes of PgrAQP genes in pomegranate. 357 As shown in Fig. 2, 47% of the PgrAQP genes had been duplicated by Manuscript to be reviewed 358 tandem/segmental duplication events. We found one tandem duplication event in 359 Chromosome 3 (PgrTIP1.2/PgrTIP1.7). There were eight pairs of segmental duplications 360 detected among six chromosomes. As shown in Table 1, the proportion of segmental 361 PgrAQP gene duplications was 88%, indicating that segmental duplication events have 362 played a key role in the expansion of the PgrAQP gene family. To access the selection 363 pressure and the date at which such duplication events occurred, estimation of the Ka 364 and Ks substitution rates of these duplication PgrAQP gene pairs were calculated. A 365 Ka/Ks ratio >1 and <1 indicate positive Darwinian selection or purifying selection, 366 whereas a value of 1 indicates neutral selection. The Ka/Ks ratios of PgrAQP duplication 367 gene pairs showed a Ka/Ks ratio of <1, indicating that these PgrAQP genes have 368 experienced purifying selection during the course of evolution. According to the 369 mathematical formula (T=Ks/2λ) used to calculate the evolutionary date, we assessed 370 the divergence time of these duplication events and found that the gene duplication 371 events occurred approximately 1.74–6.97 million years ago. 358 tandem/segmental duplication events. We found one tandem duplication event in 359 Chromosome 3 (PgrTIP1.2/PgrTIP1.7). There were eight pairs of segmental duplications 360 detected among six chromosomes. As shown in Table 1, the proportion of segmental 361 PgrAQP gene duplications was 88%, indicating that segmental duplication events have 362 played a key role in the expansion of the PgrAQP gene family. To access the selection 363 pressure and the date at which such duplication events occurred, estimation of the Ka 364 and Ks substitution rates of these duplication PgrAQP gene pairs were calculated. A 365 Ka/Ks ratio >1 and <1 indicate positive Darwinian selection or purifying selection, 366 whereas a value of 1 indicates neutral selection. The Ka/Ks ratios of PgrAQP duplication 367 gene pairs showed a Ka/Ks ratio of <1, indicating that these PgrAQP genes have 368 experienced purifying selection during the course of evolution. According to the 369 mathematical formula (T=Ks/2λ) used to calculate the evolutionary date, we assessed 370 the divergence time of these duplication events and found that the gene duplication 371 events occurred approximately 1.74–6.97 million years ago. 372 Further, we constructed the comparative synteny maps of three plants species 373 (pomegranate vs. Arabidopsis, pomegranate vs. grape, and pomegranate vs. Eucalyptus) 374 to explore the evolutionary process of PgrAQP genes (Fig. PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) Manuscript to be reviewed 379 indicates that the expansion of PgrTIPs and VvTIPs/EucTIPs genes may have occurred 380 after that of Arabidopsis, while the PIP genes have been evolutionarily conserved. 381 3.6 Analysis of PgrAQP gene expression profiles of different pomegranate tissues 382 and water deficit 383 Identifying tissue-specific genes is a basic strategy to select candidate genes involved in 384 biological processes. To explore the possible functions of PgrAQP genes in various 385 developmental stages of different organs of pomegranate, a heat map of PgrAQP 386 expression profiles was conducted. The heat map showed various expression patterns 387 of the 38 PgrAQP genes analyzed (Fig. 4A, Table. S6). Most of TIPs, such as PgrTIP1.1, 388 PgrTIP1.4, PgrTIP1.6, PgrTIP1.8, and PgrTIP2.1, showed higher expression in the roots, 389 leaves, and flowers, whereas the transcripts of PgrTIP2.3, PgrTIP3.1, and PgrTIP3.2 390 showed extremely low expression levels. For the NIPs subfamily, PgrNIP4.1 and 391 PgrNIP5.1 were highly expressed in the leaves and roots, respectively. Interestingly, the 392 transcripts of PgrNIP1.3 could only be detected in flowers, indicating that PgrNIP1 393 might be involved in the development of pomegranate flowers, while other members 394 were transcribed at extremely low levels. The accumulation of transcripts of two 395 PgrSIPs was detected in all tissues analyzed, whereas PgrSIP1.2 had higher relative 396 expression levels than that of PgrSIP1.1. In the PIPs subfamily, PgrPIP3, PgrPIP2.3, and 379 indicates that the expansion of PgrTIPs and VvTIPs/EucTIPs genes may have occurred 380 after that of Arabidopsis, while the PIP genes have been evolutionarily conserved. 382 and water deficit 383 Identifying tissue-specific genes is a basic strategy to select candidate genes involved in 384 biological processes. To explore the possible functions of PgrAQP genes in various 385 developmental stages of different organs of pomegranate, a heat map of PgrAQP 386 expression profiles was conducted. The heat map showed various expression patterns 387 of the 38 PgrAQP genes analyzed (Fig. 4A, Table. S6). Most of TIPs, such as PgrTIP1.1, 388 PgrTIP1.4, PgrTIP1.6, PgrTIP1.8, and PgrTIP2.1, showed higher expression in the roots, 389 leaves, and flowers, whereas the transcripts of PgrTIP2.3, PgrTIP3.1, and PgrTIP3.2 390 showed extremely low expression levels. For the NIPs subfamily, PgrNIP4.1 and 391 PgrNIP5.1 were highly expressed in the leaves and roots, respectively. Manuscript to be reviewed 3, Table. S5) and found that 375 24, 28, and 41 orthologous AQP gene pairs were identified, respectively. Remarkably, 376 the numbers of orthologous TIP gene pairs between pomegranate and grape/Eucalyptus 377 were significantly higher than that in Arabidopsis. Nevertheless, the PIP genes were 378 highly conserved within the species. The syntenic relationship detected in TIP genes 372 Further, we constructed the comparative synteny maps of three plants species 373 (pomegranate vs. Arabidopsis, pomegranate vs. grape, and pomegranate vs. Eucalyptus) 374 to explore the evolutionary process of PgrAQP genes (Fig. 3, Table. S5) and found that 375 24, 28, and 41 orthologous AQP gene pairs were identified, respectively. Remarkably, 376 the numbers of orthologous TIP gene pairs between pomegranate and grape/Eucalyptus 377 were significantly higher than that in Arabidopsis. Nevertheless, the PIP genes were 378 highly conserved within the species. The syntenic relationship detected in TIP genes 372 Further, we constructed the comparative synteny maps of three plants species 373 (pomegranate vs. Arabidopsis, pomegranate vs. grape, and pomegranate vs. Eucalyptus) 374 to explore the evolutionary process of PgrAQP genes (Fig. 3, Table. S5) and found that 375 24, 28, and 41 orthologous AQP gene pairs were identified, respectively. Remarkably, 376 the numbers of orthologous TIP gene pairs between pomegranate and grape/Eucalyptus 377 were significantly higher than that in Arabidopsis. Nevertheless, the PIP genes were 378 highly conserved within the species. The syntenic relationship detected in TIP genes Manuscript to be reviewed 400 and seed coats at all three experimental stages, indicating that these genes may play 401 important roles in seed development. 400 and seed coats at all three experimental stages, indicating that these genes may play 401 important roles in seed development. 402 To reveal the potential function of PgrAQPs in response to water deficit in 403 pomegranate root, the transcript levels of seven selected PgrAQPs from PIP subfamily 404 for seedling under 20% PEG supply condition were tested using qRT-PCR. According to 405 the qRT-PCR results, except the PgrPIP1.5, of which the transcript was enhanced in the 406 roots after the 24 h of 20% PEG supply conditions, the expression of the other five 407 paralogues, PgrPIP1.1, PgrPIP1.2, PgrPIP1.3, PgrPIP2.1 and PgrPIP1.4, were 408 significantly repressed under the 20% PEG supply condition (Figure 5). It should be 409 noted that the expression levels of the PgrPIP1.2 was strongly and rapidly decreased 410 under the 20% PEG treatment. In contrast, the expression pattern of PgrPIP2.4 showed 411 barely changed until 12 h after 20% PEG treatment, the down-regulation was observed 412 at 24 h. 413 3.7 Identification of candidate PgrAQPs involved in water accumulation in the outer 414 seed coat of pomegranate 415 To understand water transport and accumulation in pomegranate seed coats, 416 especially of the juicy outer seed coat, we detected water accumulation in the inner and 417 outer seed coats at different developmental stages, as well as the relative transcript 418 levels of PgrAQP genes in corresponding samples. We found that the water content was 419 significantly increased in outer seed coats during seed development in ‘Dabenzi’ and 420 ‘Tunisia,’ and the water that accumulated in the outer seed coats was higher than that in PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) 402 To reveal the potential function of PgrAQPs in response to water deficit in 403 pomegranate root, the transcript levels of seven selected PgrAQPs from PIP subfamily 404 for seedling under 20% PEG supply condition were tested using qRT-PCR. According to 405 the qRT-PCR results, except the PgrPIP1.5, of which the transcript was enhanced in the 406 roots after the 24 h of 20% PEG supply conditions, the expression of the other five 407 paralogues, PgrPIP1.1, PgrPIP1.2, PgrPIP1.3, PgrPIP2.1 and PgrPIP1.4, were 408 significantly repressed under the 20% PEG supply condition (Figure 5). Manuscript to be reviewed It should be 409 noted that the expression levels of the PgrPIP1.2 was strongly and rapidly decreased 410 under the 20% PEG treatment. In contrast, the expression pattern of PgrPIP2.4 showed 411 barely changed until 12 h after 20% PEG treatment, the down-regulation was observed 412 at 24 h. 402 To reveal the potential function of PgrAQPs in response to water deficit in 403 pomegranate root, the transcript levels of seven selected PgrAQPs from PIP subfamily 404 for seedling under 20% PEG supply condition were tested using qRT-PCR. According to 405 the qRT-PCR results, except the PgrPIP1.5, of which the transcript was enhanced in the 406 roots after the 24 h of 20% PEG supply conditions, the expression of the other five 407 paralogues, PgrPIP1.1, PgrPIP1.2, PgrPIP1.3, PgrPIP2.1 and PgrPIP1.4, were 408 significantly repressed under the 20% PEG supply condition (Figure 5). It should be 409 noted that the expression levels of the PgrPIP1.2 was strongly and rapidly decreased 410 under the 20% PEG treatment. In contrast, the expression pattern of PgrPIP2.4 showed 411 barely changed until 12 h after 20% PEG treatment, the down-regulation was observed 412 at 24 h. Manuscript to be reviewed Interestingly, the 392 transcripts of PgrNIP1.3 could only be detected in flowers, indicating that PgrNIP1 393 might be involved in the development of pomegranate flowers, while other members 394 were transcribed at extremely low levels. The accumulation of transcripts of two 395 PgrSIPs was detected in all tissues analyzed, whereas PgrSIP1.2 had higher relative 396 expression levels than that of PgrSIP1.1. In the PIPs subfamily, PgrPIP3, PgrPIP2.3, and 397 PgrPIP2.4 had a low level of transcripts in all tissues and stages analyzed, whereas the 398 remaining members had higher expression levels in all tissues analyzed. Notably, 399 PgrPIP1.3, PgrPIP1.5, PgrPIP2.1, and PgrPIP2.8 had higher expression levels in the peel PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) 399 PgrPIP1.3, PgrPIP1.5, PgrPIP2.1, and PgrPIP2.8 had higher expression levels in the peel 399 PgrPIP1.3, PgrPIP1.5, PgrPIP2.1, and PgrPIP2.8 had higher expression levels in the peel PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) Manuscript to be reviewed 421 the inner seed coats (Fig. 6). Genes, including PgrPIP1.3, PgrPIP2.8, PgrPIP1.5, 422 PgrPIP2.6, PgrPIP2.1, PgrPIP2.2, PgrPIP2.5, and PgSIP1.2, had relatively high levels of 423 transcript accumulation in inner and outer seed coats. Among the abovementioned 424 genes, PgrPIP1.3, PgrPIP2.8, and PgSIP1.2, had high levels of accumulation of 425 transcripts only in the outer seed coat at the later developmental stages (Fig. 5), which 426 indicated that these genes may be involved in the accumulation of water in outer seed 427 coats at the later developmental stages. 428 In Arabidopsis, a plasma membrane aquaporin, AtPIP1;2 is involved in aquaporin- 429 mediated leaf water transport, lateral root formation, and water uptake in root. To 430 further understand the function of PgrPIP1.3, PgrPIP2.8, and PgrSIP1.2 in the water 431 accumulation in the seed coat, the linear relationship between the water content in the 432 seed coat and differences in transcript levels for PrgPIP1.3 and PgrSIP1.2 in seed coat 433 development was determined. Significant positive correlations in both PgrPIP1.3 and 434 PgrSIP1.2 and water content in outer seed coats were observed (Fig. 7). Therefore, it was 435 concluded that PgrPIP1.3 and PgrSIP1.2 might be involved in the accumulation of water 436 in pomegranate seed coats. 428 In Arabidopsis, a plasma membrane aquaporin, AtPIP1;2 is involved in aquaporin- 429 mediated leaf water transport, lateral root formation, and water uptake in root. To 430 further understand the function of PgrPIP1.3, PgrPIP2.8, and PgrSIP1.2 in the water 431 accumulation in the seed coat, the linear relationship between the water content in the 432 seed coat and differences in transcript levels for PrgPIP1.3 and PgrSIP1.2 in seed coat 433 development was determined. Significant positive correlations in both PgrPIP1.3 and 434 PgrSIP1.2 and water content in outer seed coats were observed (Fig. 7). Therefore, it was 435 concluded that PgrPIP1.3 and PgrSIP1.2 might be involved in the accumulation of water 436 in pomegranate seed coats. 437 4. Discussion 438 4.1 PgrAQP identification and structure analysis 439 AQP proteins play an important role in various physiological and developmental 440 processes of different organs and tissues, and function as a transporter for water and/or 441 small neutral solutes in plants. Numerous studies have been conducted on AQPs in PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) 413 3.7 Identification of candidate PgrAQPs involved in water accumulation in the outer 414 seed coat of pomegranate 415 To understand water transport and accumulation in pomegranate seed coats, 416 especially of the juicy outer seed coat, we detected water accumulation in the inner and 417 outer seed coats at different developmental stages, as well as the relative transcript 418 levels of PgrAQP genes in corresponding samples. We found that the water content was 419 significantly increased in outer seed coats during seed development in ‘Dabenzi’ and 420 ‘Tunisia,’ and the water that accumulated in the outer seed coats was higher than that in 415 To understand water transport and accumulation in pomegranate seed coats, 416 especially of the juicy outer seed coat, we detected water accumulation in the inner and 417 outer seed coats at different developmental stages, as well as the relative transcript 418 levels of PgrAQP genes in corresponding samples. We found that the water content was 419 significantly increased in outer seed coats during seed development in ‘Dabenzi’ and 420 ‘Tunisia,’ and the water that accumulated in the outer seed coats was higher than that in 438 4.1 PgrAQP identification and structure analysis 439 AQP proteins play an important role in various physiological and developmental 440 processes of different organs and tissues, and function as a transporter for water and/or 441 small neutral solutes in plants. Numerous studies have been conducted on AQPs in Manuscript to be reviewed 442 plants, such as in Arabidopsis, grape, soybean, rice, and Populus trichocarpa (Fouquet et al. 443 2008; Gupta & Sankararamakrishnan 2009; Quigley et al. 2002; Sakurai et al. 2005; 444 Zhang et al. 2013). However, the genome-wide identification of the AQP gene family in 445 pomegranate was absent due to the limitations of an available genome sequence (Luo et 446 al. 2020; Qin et al. 2017). In this study, 38 PgrAQP genes were identified and 447 characterized in pomegranate. The characteristics of PgrAQP were comparable with 448 other plants species. For instance, the numbers of exons and structures of intron/exons 449 in the PgrAQP genes were highly conserved in different species, such as Arabidopsis, 450 olive, chickpea, Arachis hypogea, and banana (Deokar & Tar'an 2016; Faize et al. 2020; Hu 451 et al. 2015b; Quigley et al. 2002; Shivaraj et al. 2019). Similar to the organizations of 452 intron/exons, PgrAQP proteins of each subfamily possess the same conserved motifs 453 (Fig. 2). These results suggest that the gene structures of PgrAQPs are closely related to 454 homologous genes. 442 plants, such as in Arabidopsis, grape, soybean, rice, and Populus trichocarpa (Fouquet et al. 443 2008; Gupta & Sankararamakrishnan 2009; Quigley et al. 2002; Sakurai et al. 2005; 444 Zhang et al. 2013). However, the genome-wide identification of the AQP gene family in 445 pomegranate was absent due to the limitations of an available genome sequence (Luo et 446 al. 2020; Qin et al. 2017). In this study, 38 PgrAQP genes were identified and 447 characterized in pomegranate. The characteristics of PgrAQP were comparable with 448 other plants species. For instance, the numbers of exons and structures of intron/exons 449 in the PgrAQP genes were highly conserved in different species, such as Arabidopsis, 450 olive, chickpea, Arachis hypogea, and banana (Deokar & Tar'an 2016; Faize et al. 2020; Hu 451 et al. 2015b; Quigley et al. 2002; Shivaraj et al. 2019). Similar to the organizations of 452 intron/exons, PgrAQP proteins of each subfamily possess the same conserved motifs 453 (Fig. 2). These results suggest that the gene structures of PgrAQPs are closely related to 454 homologous genes. 455 Furthermore the TMDs showed that some of the PgrAQPs were lack one 455 Furthermore, the TMDs showed that some of the PgrAQPs were lack one 456 (PgrSIP1.1) or two (PgrNIP5.1, PgrPIP2.8) TMDs (Fig. S3). PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) Manuscript to be reviewed Manuscript to be reviewed y 464 4.2 The evolutionary relationships among the PgrAQP gene family 465 Gene duplication is considered as a major driving force for the evolution of gene 466 families, and several duplication events have been identified over the course of 467 evolution of some plant species. At approximately 117 million years ago (Mya), all core 468 eudicots experienced a genome triplication event (the γ event), including Arabidopsis, 469 Eucalyptus, grape, and pomegranate (Jiao et al. 2012; Jiao et al. 2014). Then, Arabidopsis 470 experienced two recent WGDs (α and β), whereas pomegranate and Eucalyptus 471 underwent a Myrtales lineage-specific WGD event (109.9Mya, M), but grape did not 472 undergo any additional WGDs (Myburg et al. 2014; Qin et al. 2017). In Arabidopsis, a 473 total of 35 AQP genes were identified and further evolution analysis revealed that 474 AtAQPs result from different types of gene duplication, such as γ WGD (1), β WGD (2), 475 α WGD (8), tandem (2), and transposed (4) (Bowers et al. 2003; Zwiazek et al. 2017b). 476 For poplar, tandem duplication (4) and the recent WGD (20) were the major driving 477 forces for 55 AQP genes (Zou & Yang 2019). Manuscript to be reviewed Variation in the number of 457 TMDs to different plant species has been reported (Ayadi et al. 2011; Zhu et al. 2019). 458 For instance, a truncated form of wheat TdPIP2;1 aquaporin, showed no water transport 459 activity. Interestingly, the truncated tdpip2;1 could reach the plasma membrane by 460 interact with the functional TdPIP2;1, and then may affect the functional form and 461 reduce the water transport activity of aquaporin (Ayadi et al. 2011). Therefore, the 462 absence of TMDs may affect the PgrAQPs subcellular localization and water transport 455 Furthermore, the TMDs showed that some of the PgrAQPs were lack one 456 (PgrSIP1.1) or two (PgrNIP5.1, PgrPIP2.8) TMDs (Fig. S3). Variation in the number of 457 TMDs to different plant species has been reported (Ayadi et al. 2011; Zhu et al. 2019). 458 For instance, a truncated form of wheat TdPIP2;1 aquaporin, showed no water transport 459 activity. Interestingly, the truncated tdpip2;1 could reach the plasma membrane by 460 interact with the functional TdPIP2;1, and then may affect the functional form and 461 reduce the water transport activity of aquaporin (Ayadi et al. 2011). Therefore, the 462 absence of TMDs may affect the PgrAQPs subcellular localization and water transport 464 4.2 The evolutionary relationships among the PgrAQP gene family 478 In this study, the 38-member PgrAQP family is comparable to Arabidopsis (33) and 479 Eucalyptus (40), but had a greater number of genes than grape (28). Furthermore, all the 480 PgrNIP, PgrSIP, and PgrXIP subfamilies were found to have a close relationship with 481 the corresponding genes in Eucalyptus AQPs, which is consistent with the evolutionary 482 relationships among the species, suggesting that these AQP subfamilies might have 483 functional conservation in Myrtales (Qin et al. 2017). It is reasonable to deduce that 478 In this study, the 38-member PgrAQP family is comparable to Arabidopsis (33) and 479 Eucalyptus (40), but had a greater number of genes than grape (28). Furthermore, all the 480 PgrNIP, PgrSIP, and PgrXIP subfamilies were found to have a close relationship with 481 the corresponding genes in Eucalyptus AQPs, which is consistent with the evolutionary 482 relationships among the species, suggesting that these AQP subfamilies might have 483 functional conservation in Myrtales (Qin et al. 2017). It is reasonable to deduce that Manuscript to be reviewed 484 WGD events may function as a trigger of PgrAQP family genes expansion. In additional, 485 we identified eight gene duplication events in the PgrAQPs subfamily, including seven 486 segmental duplications and one tandem duplication event (Fig. 2, Table 1). 484 WGD events may function as a trigger of PgrAQP family genes expansion. In additional, 485 we identified eight gene duplication events in the PgrAQPs subfamily, including seven 486 segmental duplications and one tandem duplication event (Fig. 2, Table 1). 487 Interestingly, the Ka/Ks ratio of the eight duplications was < 1, indicating that the 488 evolution of the PgrAQP genes is mediated by large-scale purifying selection, similar to 489 the AQP family in B. rapa and wheat (Kayum et al. 2017; Madrid-Espinoza et al. 2018). 490 The synonymous substitution rate was also used to estimate the evolutionary timescale, 491 and the divergence time of duplicated PgrAQP genes occurred 1.77 to 6.97 million years 492 ago, which is in accordance with the divergence time of BrAQPs. This indicates that 493 duplicated divergence of the PgrAQP genes occurred after the triplication events and 494 Myrtales WGD duplication events (Kayum et al. 2017; Qin et al. 2017). 490 The synonymous substitution rate was also used to estimate the evolutionary timescale, 491 and the divergence time of duplicated PgrAQP genes occurred 1.77 to 6.97 million years 492 ago, which is in accordance with the divergence time of BrAQPs. This indicates that 493 duplicated divergence of the PgrAQP genes occurred after the triplication events and 494 Myrtales WGD duplication events (Kayum et al. 2017; Qin et al. 2017). 495 According to a genome-wide analysis of different organisms, it was hypothesized 496 that the frequency of gene duplication events was important to the evolution of a 497 species (Flagel & Wendel 2009; Lynch & Conery 2000). In cotton and sesame, tandem 498 duplicated genes showed functional differentiation, although they shared structural 499 conservation (Li et al. 2019; WU Xiang-yang 2016). In our study, PgrTIP1.2 and 500 PgrTIP1.7 were identified as tandem duplicated genes, and they were predominantly 501 expressed in roots and outer seed coats in pomegranate, respectively. Our results 502 provided more information for understanding the evolution of plant AQPs. Manuscript to be reviewed 503 4.3 Potential functions of PgrAQP genes 495 According to a genome-wide analysis of different organisms, it was hypothesized 496 that the frequency of gene duplication events was important to the evolution of a 497 species (Flagel & Wendel 2009; Lynch & Conery 2000). In cotton and sesame, tandem 498 duplicated genes showed functional differentiation, although they shared structural 499 conservation (Li et al. 2019; WU Xiang-yang 2016). In our study, PgrTIP1.2 and 500 PgrTIP1.7 were identified as tandem duplicated genes, and they were predominantly 501 expressed in roots and outer seed coats in pomegranate, respectively. Our results 502 provided more information for understanding the evolution of plant AQPs. 503 4.3 Potential functions of PgrAQP genes 503 4.3 Potential functions of PgrAQP genes 504 As the largest AQP subfamily in most plant species, PIPs play a crucial role in 504 As the largest AQP subfamily in most plant species, PIPs play a crucial role in PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) Manuscript to be reviewed 505 water absorption of roots and leaves. Furthermore, PIPs can affect photosynthesis by 506 enhancing the diffusion of CO2 in mesophyll tissue of rice and N. tabacum (Flexas et al. 507 2006; Xu et al. 2019). In this study, PgrPIP subfamily genes had higher expression levels 508 in all analyzed samples compared with other analyzed AQPs. Interestingly, among the 509 RhPIPs, RhPIP1s and RhPIP2s were involved in the expansion of rose petals via an 510 ethylene-dependent pathway (Chen et al. 2013; Ma et al. 2008). The expression analysis 511 showed that PgrPIP1.3 and PgrPIP2.1 had similar expression patterns in pomegranate 512 flowers, suggesting a similar role of PIPs during pomegranate flower development (Fig. 513 4). In the higher plants, TIPs are widely used as markers for vacuolar compartments 514 and function as transporter for small solutes in various tissue (Bienert et al. 2007; Holm 515 et al. 2005; Liu et al. 2003; Porcel et al. 2018). In Arabidopsis, at least six TIP subfamily 516 numbers, including TIP1;1, TIP1;2, TIP2;1, TIP2;2, TIP2;3, and TIP4;1, showed specific 517 expression patterns in roots (Gattolin et al. 2011). In addition, the rice OsTIP2;1 was only 518 detected in roots (Nguyen et al. 2013). In pomegranate, PgrTIP1.2 and PgrTIP2.3 show 519 higher expression patterns in the roots than other organs, indicated that these gene may 520 specifically participate in the absorption and transport of small solutes, such as NH4+, 521 H2O2, and urea, in the roots of pomegranate (Bienert et al. 2007; Holm et al. 2005; Liu et 522 al. 2003). In our study, the XIP and majority of NIPs showed lower expression levels 523 than PIPs and TIPs. Interestingly, we found that PgrNIP5.1 was highly expressed in 524 roots. Such specific root expression of NIPs was reported for AtNIP5;1 and HvNIP2;1 525 (Schnurbusch et al. 2010; Takano et al. 2006), which were involved in boron (B) Manuscript to be reviewed 547 growth. In Arabidopsis, the specific expression pattern and regulatory mechanism 548 showed that AtTIP1;1 plays a critical role in cell expansion (Beebo et al. 2009; Ludevid et 549 al. 1992). Overexpression of ginseng TIP in Arabidopsis resulted in a significant increase 550 of leaf cell sizes compared with the wild type plants (Lin et al. 2007). In rice, OsPIP1;1 is 551 highly expressed in leaves and roots, and overexpression of OsPIP1;1 exhibited a higher 552 germination rate than the control plants (Liu et al. 2013). 547 growth. In Arabidopsis, the specific expression pattern and regulatory mechanism 548 showed that AtTIP1;1 plays a critical role in cell expansion (Beebo et al. 2009; Ludevid et 549 al. 1992). Overexpression of ginseng TIP in Arabidopsis resulted in a significant increase 550 of leaf cell sizes compared with the wild type plants (Lin et al. 2007). In rice, OsPIP1;1 is 551 highly expressed in leaves and roots, and overexpression of OsPIP1;1 exhibited a higher 552 germination rate than the control plants (Liu et al. 2013). 553 In pomegranate, the expanded outer seed coats had higher water accumulation 554 than the rigid inner seed coats in both the hard-seeded cultivar ‘Dabenzi’ and the soft- 555 seeded cultivar ‘Tunisia’ (Fig. 6). Accordingly, significant positive correlations were 556 found between the expression level of PgrSIP1.2, PgrPIP1.3, and PgrPIP2.8 and water 557 content in the seed coats (Fig. 7). PgrSIP1.2 was predicted to be localized in the vacuole. 558 Considering that promotion of cell expansion and maintenance of turgor requires the 559 transfer of substantial amounts of water to cells, it is reasonable to proposed that 560 PgrPIP1.3, PgrPIP2.8, and PgrSIP1.2 might be involved in mediating the water 561 accumulation in the inner and outer seed coats of pomegranate. Manuscript to be reviewed 526 homeostasis, indicating that PgNIP5;1 may participate in B absorption and translocation 527 in pomegranate roots. 526 homeostasis, indicating that PgNIP5;1 may participate in B absorption and translocation 527 in pomegranate roots. 526 homeostasis, indicating that PgNIP5;1 may participate in B absorption and translocation 527 in pomegranate roots. 528 Drought causes tissue dehydration due to an imbalance between plant water 529 uptake and transpiration. Evidence shows that AQPs play an important role in drought 530 tolerance in plants. Ectopically expression of MpPIP2;1 in Arabidopsis has been shown to 531 enhance drought and salinity tolerance. The decrease or increase of the AQPs transcript 532 levels could prevent water losses or helps plants to direct water flow to specific organ 533 under drought stress. In this study, we found most members of the PgrPIP genes were 534 suppressed by drought stress (Fig 5), including PgrPIP1.1, PgrPIP1.2, PgrPIP1.3, 535 PgrPIP2.1 and PgrPIP2.4, suggested their involvement in reduced water losses in 536 Pomegranate plants. Furthermore, the upregulation of the transcript levels of PgrPIP1.5 537 was observed under 20% PEG supply condition, indicated that PgrPIP1.5 might play a 538 critical for tolerance to drought in pomegranate. 541 Plant cell expansion is primarily driven by turgor and requires steady water intake, 542 the rate of tissue growth is primarily restricted due to decreasing of turgor (Peret et al. 543 2012; Picaud et al. 2003; Reuscher et al. 2013). For example, in higher plants, the 544 development of a seed coat was primarily initiated by fertilization and driven by cell 545 expansion and growth (Figueiredo et al. 2016). Accordingly, most of the AQPs were 546 reported to be strongly expressed in tissues that can be hydraulically limited during Manuscript to be reviewed 568 relationships and duplication events analysis, and a whole-genome duplication event in 569 Myrtales may contribute to the expansion of PgrTIP, PgrSIP, and PgrXIP genes. 570 Furthermore, the high expression of PgrPIP1.3, PgrPIP2.8, and PgrSIP1.2 in seed coats 571 and the positive correlation between transcript levels of PgrAQP genes and the water 572 content revealed these three genes may be the potential candidate genes involving in 573 outer seed coat development. Hence, further studies on functions of this three AQP 574 genes are needed for genetic improvement of outer seed coats in pomegranate. 575 576 References 577 Adams KL, and Wendel JF. 2005. Polyploidy and genome evolution in plants. Curr Opin Plant Biol 8:135-141. 578 10.1016/j.pbi.2005.01.001 579 Almasalmeh A, Krenc D, Wu B, and Beitz E. 2014. Structural determinants of the hydrogen peroxide permeability of 580 aquaporins. FEBS J 281:647-656. 10.1111/febs.12653 581 Artimo P, Jonnalagedda M, Arnold K, Baratin D, Csardi G, de Castro E, Duvaud S, Flegel V, Fortier A, Gasteiger E, 582 Grosdidier A, Hernandez C, Ioannidis V, Kuznetsov D, Liechti R, Moretti S, Mostaguir K, Redaschi N, Rossier 583 G, Xenarios I, and Stockinger H. 2012. ExPASy: SIB bioinformatics resource portal. Nucleic Acids Res 584 40:W597-603. 10.1093/nar/gks400 585 Ayadi M, Cavez D, Miled N, Chaumont F, Masmoudi K. Identification and characterization of two plasma membran 586 e aquaporins in durum wheat (Triticum turgidum L. subsp. durum) and their role in abiotic stress toleranc 587 e. Plant Physiol Biochem. 2011 Sep;49(9):1029- 588 39. doi: 10.1016/j.plaphy.2011.06.002. Epub 2011 Jun 17. PMID: 21723739. 589 Azad AK, Ahmed J, Alum MA, Hasan MM, Ishikawa T, Sawa Y, and Katsuhara M. 2016. Genome-Wide 590 Characterization of Major Intrinsic Proteins in Four Grass Plants and Their Non-Aqua Transport Selectivity 591 Profiles with Comparative Perspective. PLoS One 11:e0157735. 10.1371/journal.pone.0157735 592 Azad AK, Sawa Y, Ishikawa T, and Shibata H. 2004. Phosphorylation of plasma membrane aquaporin regulates 593 temperature-dependent opening of tulip petals. Plant Cell Physiol 45:608-617. 10.1093/pcp/pch069 594 Bailey TL, Boden M, Buske FA, Frith M, Grant CE, Clementi L, Ren J, Li WW, and Noble WS. 2009. MEME SUITE: 595 tools for motif discovery and searching. Nucleic Acids Res 37:W202-208. 10.1093/nar/gkp335 596 Beebo A, Thomas D, Der C, Sanchez L, Leborgne-Castel N, Marty F, Schoefs B, and Bouhidel K. 2009. Life with and 597 without AtTIP1;1, an Arabidopsis aquaporin preferentially localized in the apposing tonoplasts of adjacent 598 vacuoles. Plant Mol Biol 70:193-209. Manuscript to be reviewed 10.1007/s11103-009-9465-2 599 Bienert GP, Moller AL, Kristiansen KA, Schulz A, Moller IM, Schjoerring JK, and Jahn TP. 2007. Specific aquaporins 568 relationships and duplication events analysis, and a whole-genome duplication event in 569 Myrtales may contribute to the expansion of PgrTIP, PgrSIP, and PgrXIP genes. 88 39. doi: 10.1016/j.plaphy.2011.06.002. Epub 2011 Jun 17. PMID: 21723739. 589 Azad AK, Ahmed J, Alum MA, Hasan MM, Ishikawa T, Sawa Y, and Katsuhara M. 2016. Genome-Wide 590 Characterization of Major Intrinsic Proteins in Four Grass Plants and Their Non-Aqua Transport Selectivity 591 Profiles with Comparative Perspective. PLoS One 11:e0157735. 10.1371/journal.pone.0157735 589 Azad AK, Ahmed J, Alum MA, Hasan MM, Ishikawa T, Sawa Y, and Katsuhara M. 2016. Genome-Wide 590 Characterization of Major Intrinsic Proteins in Four Grass Plants and Their Non-Aqua Transport Selectivity 591 Profiles with Comparative Perspective. PLoS One 11:e0157735. 10.1371/journal.pone.0157735 591 Profiles with Comparative Perspective. PLoS One 11:e0157735. 10.1371/journal.pone.0157735 592 Azad AK, Sawa Y, Ishikawa T, and Shibata H. 2004. Phosphorylation of plasma membrane aquaporin regulates 593 temperature-dependent opening of tulip petals. Plant Cell Physiol 45:608-617. 10.1093/pcp/pch069 594 Bailey TL Boden M Buske FA Frith M Grant CE Clementi L Ren J Li WW and Noble WS 2009 MEME SUITE: 592 Azad AK, Sawa Y, Ishikawa T, and Shibata H. 2004. Phosphorylation of plasma membrane aquaporin regulates 593 temperature-dependent opening of tulip petals. Plant Cell Physiol 45:608-617. 10.1093/pcp/pch069 592 Azad AK, Sawa Y, Ishikawa T, and Shibata H. 2004. Phosphorylation of plasma membrane aquaporin regulates 593 temperature-dependent opening of tulip petals. Plant Cell Physiol 45:608-617. 10.1093/pcp/pch069 594 Bailey TL, Boden M, Buske FA, Frith M, Grant CE, Clementi L, Ren J, Li WW, and Noble WS. 2009. MEME SUITE: 595 tools for motif discovery and searching. Nucleic Acids Res 37:W202-208. 10.1093/nar/gkp335 594 Bailey TL, Boden M, Buske FA, Frith M, Grant CE, Clementi L, Ren J, Li WW, and Noble WS. 2009. MEME SUITE: 595 tools for motif discovery and searching. Nucleic Acids Res 37:W202-208. 10.1093/nar/gkp335 596 Beebo A, Thomas D, Der C, Sanchez L, Leborgne-Castel N, Marty F, Schoefs B, and Bouhidel K. 2009. Life with and 597 without AtTIP1;1, an Arabidopsis aquaporin preferentially localized in the apposing tonoplasts of adjacent 598 vacuoles. Plant Mol Biol 70:193-209. 10.1007/s11103-009-9465-2 596 Beebo A, Thomas D, Der C, Sanchez L, Leborgne-Castel N, Marty F, Schoefs B, and Bouhidel K. 2009. 562 Conclusions 563 In this study, a total of 38 AQP genes were identified and their characteristics, including 564 protein physicochemical properties, gene structure, phyletic evolutionary, and 565 expression patterns were studied. These PgrAQP genes are distributed across 9 566 pomegranate chromosomes and divided into five subfamilies. Purifying selection were 567 undergone during the evolution of PgrAQP family genes basing on the syntenic 563 In this study, a total of 38 AQP genes were identified and their characteristics, including 564 protein physicochemical properties, gene structure, phyletic evolutionary, and 565 expression patterns were studied. These PgrAQP genes are distributed across 9 566 pomegranate chromosomes and divided into five subfamilies. Purifying selection were 567 undergone during the evolution of PgrAQP family genes basing on the syntenic Manuscript to be reviewed 600 facilitate the diffusion of hydrogen peroxide across membranes. J Biol Chem 282:1183-1192. 601 10.1074/jbc.M603761200 602 Bowers JE, Chapman BA, Rong J, and Paterson AH. 2003. Unravelling angiosperm genome evolution by 603 phylogenetic analysis of chromosomal duplication events. Nature 422:433-438. 10.1038/nature01521 604 Chaumont F, Barrieu F, Wojcik E, Chrispeels MJ, and Jung R. 2001. Aquaporins constitute a large and highly 605 divergent protein family in maize. Plant Physiol 125:1206-1215. 10.1104/pp.125.3.1206 606 Chen C, Chen H, Zhang Y, Thomas HR, Frank MH, He Y, and Xia R. 2020. TBtools: An Integrative Toolkit Developed 607 for Interactive Analyses of Big Biological Data. Mol Plant 13:1194-1202. 10.1016/j.molp.2020.06.009 608 Chen W, Yin X, Wang L, Tian J, Yang R, Liu D, Yu Z, Ma N, and Gao J. 2013. Involvement of rose aquaporin RhPIP1;1 609 in ethylene-regulated petal expansion through interaction with RhPIP2;1. Plant Mol Biol 83:219-233. 610 10.1007/s11103-013-0084-6 611 Danielson JA, and Johanson U. 2008. Unexpected complexity of the aquaporin gene family in the moss 612 Physcomitrella patens. BMC Plant Biol 8:45. 10.1186/1471-2229-8-45 613 Danielson JA, and Johanson U. 2010. Phylogeny of major intrinsic proteins. Adv Exp Med Biol 679:19-31. 614 10.1007/978-1-4419-6315-4_2 615 De Rosa A, Watson-Lazowski A, Evans JR, and Groszmann M. 2020. Genome-wide identification and 616 characterisation of Aquaporins in Nicotiana tabacum and their relationships with other Solanaceae 617 species. BMC Plant Biol 20:266. 10.1186/s12870-020-02412-5 618 Deokar AA, and Tar'an B. 2016. Genome-Wide Analysis of the Aquaporin Gene Family in Chickpea (Cicer arietinum 619 L.). Front Plant Sci 7:1802. 10.3389/fpls.2016.01802 620 Deshmukh RK, Vivancos J, Guérin V, Sonah H, Labbé C, Belzile F, and Bélanger RR. 2013. Identification and 621 functional characterization of silicon transporters in soybean using comparative genomics of major 622 intrinsic proteins in Arabidopsis and rice. Plant Mol Biol 83:303-315. 10.1007/s11103-013-0087-3 623 Deshmukh RK, Vivancos J, Ramakrishnan G, Guérin V, Carpentier G, Sonah H, Labbé C, Isenring P, Belzile FJ, and 624 Bélanger RR. 2015. A precise spacing between the NPA domains of aquaporins is essential for silicon 625 permeability in plants. Plant J 83:489-500. 10.1111/tpj.12904 626 Edlund AF, Swanson R, and Preuss D. 2004. Pollen and stigma structure and function: the role of diversity in 627 pollination. Plant Cell 16 Suppl:S84-97. 10.1105/tpc.015800 628 Eisenbarth DA, and Weig AR. 2005. Dynamics of aquaporins and water relations during hypocotyl elongation in 629 Ricinus communis L. seedlings. J Exp Bot 56:1831-1842. Manuscript to be reviewed Pollen and stigma structure and function: the role of diversity in 627 pollination. Plant Cell 16 Suppl:S84-97. 10.1105/tpc.015800 626 Edlund AF, Swanson R, and Preuss D. 2004. Pollen and stigma structure and function: the role of diversity in 627 pollination. Plant Cell 16 Suppl:S84-97. 10.1105/tpc.015800 628 Eisenbarth DA, and Weig AR. 2005. Dynamics of aquaporins and water relations during hypocotyl elongation in 629 Ricinus communis L. seedlings. J Exp Bot 56:1831-1842. 10.1093/jxb/eri173 628 Eisenbarth DA, and Weig AR. 2005. Dynamics of aquaporins and water relations during hypocotyl elongation in 629 Ricinus communis L. seedlings. J Exp Bot 56:1831-1842. 10.1093/jxb/eri173 630 Faize M, Fumanal B, Luque F, Ramirez-Tejero JA, Zou Z, Qiao X, Faize L, Gousset-Dupont A, Roeckel-Drevet P, Label 631 P, and Venisse JS. 2020. Genome Wild Analysis and Molecular Understanding of the Aquaporin Diversity in 632 Olive Trees (Olea Europaea L.). Int J Mol Sci 21. 10.3390/ijms21114183 630 Faize M, Fumanal B, Luque F, Ramirez-Tejero JA, Zou Z, Qiao X, Faize L, Gousset-Dupont A, Roeckel-Drevet P, Label 631 P, and Venisse JS. 2020. Genome Wild Analysis and Molecular Understanding of the Aquaporin Diversity in 632 Olive Trees (Olea Europaea L.). Int J Mol Sci 21. 10.3390/ijms21114183 633 Figueiredo DD, Batista RA, Roszak PJ, Hennig L, and Kohler C. 2016. Auxin production in the endosperm drives seed 634 coat development in Arabidopsis. Elife 5. 10.7554/eLife.20542 633 Figueiredo DD, Batista RA, Roszak PJ, Hennig L, and Kohler C. 2016. Auxin production in the endosperm drives seed 634 coat development in Arabidopsis. Elife 5. 10.7554/eLife.20542 635 Finn RD, Bateman A, Clements J, Coggill P, Eberhardt RY, Eddy SR, Heger A, Hetherington K, Holm L, Mistry J, 636 Sonnhammer EL, Tate J, and Punta M. 2014. Pfam: the protein families database. Nucleic Acids Res 637 42:D222-230. 10.1093/nar/gkt1223 635 Finn RD, Bateman A, Clements J, Coggill P, Eberhardt RY, Eddy SR, Heger A, Hetherington K, Holm L, Mistry J, 636 Sonnhammer EL, Tate J, and Punta M. 2014. Pfam: the protein families database. Nucleic Acids Res 637 42:D222-230. 10.1093/nar/gkt1223 638 Flagel LE, and Wendel JF. 2009. Gene duplication and evolutionary novelty in plants. New Phytol 183:557-564. 639 10.1111/j.1469-8137.2009.02923.x 638 Flagel LE, and Wendel JF. 2009. Gene duplication and evolutionary novelty in plants. New Phytol 183:557-564. 639 10.1111/j.1469-8137.2009.02923.x 640 Flexas J, Ribas-Carbo M, Hanson DT, Bota J, Otto B, Cifre J, McDowell N, Medrano H, and Kaldenhoff R. 2006. Manuscript to be reviewed Life with and 597 without AtTIP1;1, an Arabidopsis aquaporin preferentially localized in the apposing tonoplasts of adjacent 598 vacuoles. Plant Mol Biol 70:193-209. 10.1007/s11103-009-9465-2 599 Bienert GP, Moller AL, Kristiansen KA, Schulz A, Moller IM, Schjoerring JK, and Jahn TP. 2007. Specific aquaporins 599 Bienert GP, Moller AL, Kristiansen KA, Schulz A, Moller IM, Schjoerring JK, and Jahn TP. 2007. Specific aquaporins PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) Manuscript to be reviewed 10.1093/jxb/eri173 630 Faize M, Fumanal B, Luque F, Ramirez-Tejero JA, Zou Z, Qiao X, Faize L, Gousset-Dupont A, Roeckel-Drevet P, Label 631 P, and Venisse JS. 2020. Genome Wild Analysis and Molecular Understanding of the Aquaporin Diversity in 632 Olive Trees (Olea Europaea L.). Int J Mol Sci 21. 10.3390/ijms21114183 633 Figueiredo DD, Batista RA, Roszak PJ, Hennig L, and Kohler C. 2016. Auxin production in the endosperm drives seed 634 coat development in Arabidopsis. Elife 5. 10.7554/eLife.20542 635 Finn RD, Bateman A, Clements J, Coggill P, Eberhardt RY, Eddy SR, Heger A, Hetherington K, Holm L, Mistry J, 636 Sonnhammer EL, Tate J, and Punta M. 2014. Pfam: the protein families database. Nucleic Acids Res 637 42:D222 230 10 1093/nar/gkt1223 601 10.1074/jbc.M603761200 602 Bowers JE, Chapman BA, Rong J, and Paterson AH. 2003. Unravelling angiosperm genome evolution by 603 phylogenetic analysis of chromosomal duplication events. Nature 422:433-438. 10.1038/nature01521 604 Chaumont F, Barrieu F, Wojcik E, Chrispeels MJ, and Jung R. 2001. Aquaporins constitute a large and highly 605 divergent protein family in maize. Plant Physiol 125:1206-1215. 10.1104/pp.125.3.1206 606 Chen C, Chen H, Zhang Y, Thomas HR, Frank MH, He Y, and Xia R. 2020. TBtools: An Integrative Toolkit Developed 607 for Interactive Analyses of Big Biological Data. Mol Plant 13:1194-1202. 10.1016/j.molp.2020.06.009 608 Chen W, Yin X, Wang L, Tian J, Yang R, Liu D, Yu Z, Ma N, and Gao J. 2013. Involvement of rose aquaporin RhPIP1;1 609 in ethylene-regulated petal expansion through interaction with RhPIP2;1. Plant Mol Biol 83:219-233. 610 10.1007/s11103-013-0084-6 611 Danielson JA, and Johanson U. 2008. Unexpected complexity of the aquaporin gene family in the moss 612 Physcomitrella patens. BMC Plant Biol 8:45. 10.1186/1471-2229-8-45 613 Danielson JA, and Johanson U. 2010. Phylogeny of major intrinsic proteins. Adv Exp Med Biol 679:19-31. 614 10.1007/978-1-4419-6315-4_2 615 De Rosa A, Watson-Lazowski A, Evans JR, and Groszmann M. 2020. Genome-wide identification and 616 characterisation of Aquaporins in Nicotiana tabacum and their relationships with other Solanaceae 617 species. BMC Plant Biol 20:266. 10.1186/s12870-020-02412-5 618 Deokar AA, and Tar'an B. 2016. Genome-Wide Analysis of the Aquaporin Gene Family in Chickpea (Cicer arietinum 619 L.). Front Plant Sci 7:1802. 10.3389/fpls.2016.01802 620 Deshmukh RK, Vivancos J, Guérin V, Sonah H, Labbé C, Belzile F, and Bélanger RR. 2013. Identification and 621 functional characterization of silicon transporters in soybean using comparative genomics of major 626 Edlund AF, Swanson R, and Preuss D. 2004. Manuscript to be reviewed 641 Tobacco aquaporin NtAQP1 is involved in mesophyll conductance to CO2 in vivo. Plant J 48:427-439. 642 10.1111/j.1365-313X.2006.02879.x 643 Fouquet R, Léon C, Ollat N, and Barrieu F. 2008. Identification of grapevine aquaporins and expression analysis in 644 developing berries. Plant Cell Reports 27:1541-1550. 10.1007/s00299-008-0566-1 645 Froger A, Tallur B, Thomas D, and Delamarche C. 1998. Prediction of functional residues in water channels and 646 related proteins. Protein Sci 7:1458-1468. 10.1002/pro.5560070623 647 Gattolin S, Sorieul M, and Frigerio L. 2011. Mapping of tonoplast intrinsic proteins in maturing and germinating 648 Arabidopsis seeds reveals dual localization of embryonic TIPs to the tonoplast and plasma membrane. Mol 649 Plant 4:180-189. 10.1093/mp/ssq051 650 Gu Z, Cavalcanti A, Chen FC, Bouman P, and Li WH. 2002. Extent of gene duplication in the genomes of Drosophila, 651 nematode, and yeast. Mol Biol Evol 19:256-262. 10.1093/oxfordjournals.molbev.a004079 652 Gupta AB, and Sankararamakrishnan R. 2009. Genome-wide analysis of major intrinsic proteins in the tree plant 653 Populus trichocarpa: characterization of XIP subfamily of aquaporins from evolutionary perspective. BMC 654 Plant Biol 9:134. 10.1186/1471-2229-9-134 655 Holm LM, Jahn TP, Moller AL, Schjoerring JK, Ferri D, Klaerke DA, and Zeuthen T. 2005. NH3 and NH4+ permeability 656 in aquaporin-expressing Xenopus oocytes. Pflugers Arch 450:415-428. 10.1007/s00424-005-1399-1 657 Hove RM, and Bhave M. 2011. Plant aquaporins with non-aqua functions: deciphering the signature sequences. 658 Plant Mol Biol 75:413-430. 10.1007/s11103-011-9737-5 641 Tobacco aquaporin NtAQP1 is involved in mesophyll conductance to CO2 in vivo. Plant J 48:427-439. 642 10.1111/j.1365-313X.2006.02879.x 643 Fouquet R, Léon C, Ollat N, and Barrieu F. 2008. Identification of grapevine aquaporins and expression analysis in 644 developing berries. Plant Cell Reports 27:1541-1550. 10.1007/s00299-008-0566-1 645 Froger A, Tallur B, Thomas D, and Delamarche C. 1998. Prediction of functional residues in water channels and 646 related proteins. Protein Sci 7:1458-1468. 10.1002/pro.5560070623 647 Gattolin S, Sorieul M, and Frigerio L. 2011. Mapping of tonoplast intrinsic proteins in maturing and germinating 648 Arabidopsis seeds reveals dual localization of embryonic TIPs to the tonoplast and plasma membrane. Mol 649 Plant 4:180-189. 10.1093/mp/ssq051 650 Gu Z, Cavalcanti A, Chen FC, Bouman P, and Li WH. 2002. Extent of gene duplication in the genomes of Drosophila, 651 nematode, and yeast. Mol Biol Evol 19:256-262. 10.1093/oxfordjournals.molbev.a004079 641 Tobacco aquaporin NtAQP1 is involved in mesophyll conductance to CO2 in vivo. Plant J 48:427-439. 642 10.1111/j.1365-313X.2006.02879.x 643 Fouquet R, Léon C, Ollat N, and Barrieu F. 2008. Manuscript to be reviewed Genome-wide analysis of major intrinsic proteins in the tree plant 653 Populus trichocarpa: characterization of XIP subfamily of aquaporins from evolutionary perspective. BMC 654 Plant Biol 9:134. 10.1186/1471-2229-9-134 652 Gupta AB, and Sankararamakrishnan R. 2009. Genome-wide analysis of major intrinsic proteins in the tree plant 653 Populus trichocarpa: characterization of XIP subfamily of aquaporins from evolutionary perspective. BMC 654 Plant Biol 9:134. 10.1186/1471-2229-9-134 655 Holm LM, Jahn TP, Moller AL, Schjoerring JK, Ferri D, Klaerke DA, and Zeuthen T. 2005. NH3 and NH4+ permeability 656 in aquaporin-expressing Xenopus oocytes. Pflugers Arch 450:415-428. 10.1007/s00424-005-1399-1 655 Holm LM, Jahn TP, Moller AL, Schjoerring JK, Ferri D, Klaerke DA, and Zeuthen T. 2005. NH3 and NH4+ permeability 656 in aquaporin-expressing Xenopus oocytes. Pflugers Arch 450:415-428. 10.1007/s00424-005-1399-1 657 Hove RM, and Bhave M. 2011. Plant aquaporins with non-aqua functions: deciphering the signature sequences. 658 Plant Mol Biol 75:413-430. 10.1007/s11103-011-9737-5 657 Hove RM, and Bhave M. 2011. Plant aquaporins with non-aqua functions: deciphering the signature sequences. 658 Plant Mol Biol 75:413-430. 10.1007/s11103-011-9737-5 659 Hu B, Jin J, Guo AY, Zhang H, Luo J, and Gao G. 2015a. GSDS 2.0: an upgraded gene feature visualization server. 660 Bioinformatics 31:1296-1297. 10.1093/bioinformatics/btu817 661 Hu W, Hou X, Huang C, Yan Y, Tie W, Ding Z, Wei Y, Liu J, Miao H, Lu Z, Li M, Xu B, and Jin Z. 2015b. Genome-Wide 662 Identification and Expression Analyses of Aquaporin Gene Family during Development and Abiotic Stress 663 in Banana. Int J Mol Sci 16:19728-19751. 10.3390/ijms160819728 661 Hu W, Hou X, Huang C, Yan Y, Tie W, Ding Z, Wei Y, Liu J, Miao H, Lu Z, Li M, Xu B, and Jin Z. 2015b. Genome-Wide 662 Identification and Expression Analyses of Aquaporin Gene Family during Development and Abiotic Stress 663 in Banana. Int J Mol Sci 16:19728-19751. 10.3390/ijms160819728 664 Hu W, Hua X, Zhang Q, Wang J, Shen Q, Zhang X, Wang K, Yu Q, Lin YR, Ming R, and Zhang J. 2018. New insights 665 into the evolution and functional divergence of the SWEET family in Saccharum based on comparative 666 genomics. BMC Plant Biol 18:270. 10.1186/s12870-018-1495-y 667 Ivanov, II, Loktyushkin AV, Gus'kova RA, Vasil'ev NS, Fedorov GE, and Rubin AB. 2007. Oxygen channels of 668 erythrocyte membrane. Dokl Biochem Biophys 414:137-140. 10.1134/s160767290703012x 667 Ivanov, II, Loktyushkin AV, Gus'kova RA, Vasil'ev NS, Fedorov GE, and Rubin AB. 2007. Oxygen channels of 668 erythrocyte membrane. Manuscript to be reviewed Identification of grapevine aquaporins and expression analysis in 644 developing berries. Plant Cell Reports 27:1541-1550. 10.1007/s00299-008-0566-1 645 Froger A, Tallur B, Thomas D, and Delamarche C. 1998. Prediction of functional residues in water channels and 646 related proteins. Protein Sci 7:1458-1468. 10.1002/pro.5560070623 647 Gattolin S, Sorieul M, and Frigerio L. 2011. Mapping of tonoplast intrinsic proteins in maturing and germinating 648 Arabidopsis seeds reveals dual localization of embryonic TIPs to the tonoplast and plasma membrane. Mol 649 Plant 4:180-189. 10.1093/mp/ssq051 650 Gu Z, Cavalcanti A, Chen FC, Bouman P, and Li WH. 2002. Extent of gene duplication in the genomes of Drosophila, 651 nematode, and yeast. Mol Biol Evol 19:256-262. 10.1093/oxfordjournals.molbev.a004079 652 Gupta AB, and Sankararamakrishnan R. 2009. Genome-wide analysis of major intrinsic proteins in the tree plant 653 Populus trichocarpa: characterization of XIP subfamily of aquaporins from evolutionary perspective. BMC 654 Plant Biol 9:134. 10.1186/1471-2229-9-134 655 Holm LM, Jahn TP, Moller AL, Schjoerring JK, Ferri D, Klaerke DA, and Zeuthen T. 2005. NH3 and NH4+ permeability 656 in aquaporin-expressing Xenopus oocytes. Pflugers Arch 450:415-428. 10.1007/s00424-005-1399-1 657 Hove RM, and Bhave M. 2011. Plant aquaporins with non-aqua functions: deciphering the signature sequences. 658 Plant Mol Biol 75:413-430. 10.1007/s11103-011-9737-5 659 Hu B, Jin J, Guo AY, Zhang H, Luo J, and Gao G. 2015a. GSDS 2.0: an upgraded gene feature visualization server. 660 Bioinformatics 31:1296-1297. 10.1093/bioinformatics/btu817 661 Hu W, Hou X, Huang C, Yan Y, Tie W, Ding Z, Wei Y, Liu J, Miao H, Lu Z, Li M, Xu B, and Jin Z. 2015b. Genome-Wide 662 Identification and Expression Analyses of Aquaporin Gene Family during Development and Abiotic Stress 663 in Banana. Int J Mol Sci 16:19728-19751. 10.3390/ijms160819728 664 Hu W, Hua X, Zhang Q, Wang J, Shen Q, Zhang X, Wang K, Yu Q, Lin YR, Ming R, and Zhang J. 2018. New insights 665 into the evolution and functional divergence of the SWEET family in Saccharum based on comparative 666 genomics. BMC Plant Biol 18:270. 10.1186/s12870-018-1495-y 667 Ivanov, II, Loktyushkin AV, Gus'kova RA, Vasil'ev NS, Fedorov GE, and Rubin AB. 2007. Oxygen channels of 650 Gu Z, Cavalcanti A, Chen FC, Bouman P, and Li WH. 2002. Extent of gene duplication in the genomes of Drosophila, 651 nematode, and yeast. Mol Biol Evol 19:256-262. 10.1093/oxfordjournals.molbev.a004079 652 Gupta AB, and Sankararamakrishnan R. 2009. Manuscript to be reviewed 640 Flexas J, Ribas-Carbo M, Hanson DT, Bota J, Otto B, Cifre J, McDowell N, Medrano H, and Kaldenhoff R. 2006. PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) Manuscript to be reviewed Dokl Biochem Biophys 414:137-140. 10.1134/s160767290703012x 669 Jiao Y, Leebens-Mack J, Ayyampalayam S, Bowers JE, McKain MR, McNeal J, Rolf M, Ruzicka DR, Wafula E, Wickett 670 NJ, Wu X, Zhang Y, Wang J, Zhang Y, Carpenter EJ, Deyholos MK, Kutchan TM, Chanderbali AS, Soltis PS, 671 Stevenson DW, McCombie R, Pires JC, Wong GK, Soltis DE, and Depamphilis CW. 2012. A genome 672 triplication associated with early diversification of the core eudicots. Genome Biol 13:R3. 10.1186/gb- 673 2012-13-1-r3 674 Jiao Y, Li J, Tang H, and Paterson AH. 2014. Integrated syntenic and phylogenomic analyses reveal an ancient 675 genome duplication in monocots. Plant Cell 26:2792-2802. 10.1105/tpc.114.127597 676 Johanningsmeier SD, and Harris GK. 2011. Pomegranate as a Functional Food and Nutraceutical Source. Annual 677 Review of Food Science and Technology 2:181-201. 10.1146/annurev-food-030810-153709 678 Kaldenhoff R, and Fischer M. 2006. Functional aquaporin diversity in plants. Biochim Biophys Acta 1758:1134-1141. 676 Johanningsmeier SD, and Harris GK. 2011. Pomegranate as a Functional Food and Nutraceutical Source. Annual 677 Review of Food Science and Technology 2:181-201. 10.1146/annurev-food-030810-153709 678 Kaldenhoff R, and Fischer M. 2006. Functional aquaporin diversity in plants. Biochim Biophys Acta 1758:1134-1141. 679 10.1016/j.bbamem.2006.03.012 676 Johanningsmeier SD, and Harris GK. 2011. Pomegranate as a Functional Food and Nutraceutical Source. Annual 677 Review of Food Science and Technology 2:181-201. 10.1146/annurev-food-030810-153709 678 Kaldenhoff R, and Fischer M. 2006. Functional aquaporin diversity in plants. Biochim Biophys Acta 1758:1134-1141. 679 10.1016/j.bbamem.2006.03.012 678 Kaldenhoff R, and Fischer M. 2006. Functional aquaporin diversity in plants. Biochim Biophys Acta 1758:1134-1141. 679 10.1016/j.bbamem.2006.03.012 680 Kayum MA, Park JI, Nath UK, Biswas MK, Kim HT, and Nou IS. 2017. Genome-wide expression profiling of 681 aquaporin genes confer responses to abiotic and biotic stresses in Brassica rapa. BMC Plant Biol 17:23. 680 Kayum MA, Park JI, Nath UK, Biswas MK, Kim HT, and Nou IS. 2017. Genome-wide expression profiling of 681 aquaporin genes confer responses to abiotic and biotic stresses in Brassica rapa. BMC Plant Biol 17:23. PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) Manuscript to be reviewed 682 10.1186/s12870-017-0979-5 683 Khabudaev KV, Petrova DP, Grachev MA, and Likhoshway YV. 2014. A new subfamily LIP of the major intrinsic 684 proteins. BMC Genomics 15:173. 10.1186/1471-2164-15-173 685 Kumar S, Stecher G, and Tamura K. 2016. MEGA7: Molecular Evolutionary Genetics Analysis Version 7.0 for Bigger 686 Datasets. Mol Biol Evol 33:1870-1874. 10.1093/molbev/msw054 687 Lee JK, Kozono D, Remis J, Kitagawa Y, Agre P, and Stroud RM. 2005. Structural basis for conductance by the 688 archaeal aquaporin AqpM at 1.68 A. Proc Natl Acad Sci U S A 102:18932-18937. 689 10.1073/pnas.0509469102 690 Letunic I, Khedkar S, and Bork P. 2020. SMART: recent updates, new developments and status in 2020. Nucleic 691 Acids Res. 10.1093/nar/gkaa937 692 Li W, Zhang D, Zhu G, Mi X, and Guo W. 2019. Combining genome-wide and transcriptome-wide analyses reveal 693 the evolutionary conservation and functional diversity of aquaporins in cotton. BMC Genomics 20:538. 694 10.1186/s12864-019-5928-2 695 Lian HL, Yu X, Ye Q, Ding X, Kitagawa Y, Kwak SS, Su WA, and Tang ZC. 2004. The role of aquaporin RWC3 in drough 696 avoidance in rice. Plant Cell Physiol 45:481-489. 10.1093/pcp/pch058 697 Lin W, Peng Y, Li G, Arora R, Tang Z, Su W, and Cai W. 2007. Isolation and functional characterization of PgTIP1, a 698 hormone-autotrophic cells-specific tonoplast aquaporin in ginseng. J Exp Bot 58:947-956. 699 10.1093/jxb/erl255 700 Liu C, Fukumoto T, Matsumoto T, Gena P, Frascaria D, Kaneko T, Katsuhara M, Zhong S, Sun X, Zhu Y, Iwasaki I, Din 701 X, Calamita G, and Kitagawa Y. 2013. Aquaporin OsPIP1;1 promotes rice salt resistance and seed 702 germination. Plant Physiol Biochem 63:151-158. 10.1016/j.plaphy.2012.11.018 703 Liu HY, Yu X, Cui DY, Sun MH, Sun WN, Tang ZC, Kwak SS, and Su WA. 2007. The role of water channel proteins and 704 nitric oxide signaling in rice seed germination. Cell Res 17:638-649. 10.1038/cr.2007.34 705 Liu LH, Ludewig U, Gassert B, Frommer WB, and von Wiren N. 2003. Urea transport by nitrogen-regulated 706 tonoplast intrinsic proteins in Arabidopsis. Plant Physiol 133:1220-1228. 10.1104/pp.103.027409 707 Ludevid D, Hofte H, Himelblau E, and Chrispeels MJ. 1992. The Expression Pattern of the Tonoplast Intrinsic Protein 708 gamma-TIP in Arabidopsis thaliana Is Correlated with Cell Enlargement. Plant Physiol 100:1633-1639. Manuscript to be reviewed 709 10.1104/pp.100.4.1633 710 Luo X, Li H, Wu Z, Yao W, Zhao P, Cao D, Yu H, Li K, Poudel K, Zhao D, Zhang F, Xia X, Chen L, Wang Q, Jing D, and 711 Cao S. 2020. The pomegranate (Punica granatum L.) draft genome dissects genetic divergence between 712 soft- and hard-seeded cultivars. Plant Biotechnol J 18:955-968. 10.1111/pbi.13260 713 Lynch M, and Conery JS. 2000. The evolutionary fate and consequences of duplicate genes. Science 290:1151-1155 714 10.1126/science.290.5494.1151 690 Letunic I, Khedkar S, and Bork P. 2020. SMART: recent updates, new developments and status in 2020. Nucleic 691 Acids Res. 10.1093/nar/gkaa937 692 Li W, Zhang D, Zhu G, Mi X, and Guo W. 2019. Combining genome-wide and transcriptome-wide analyses reveal 693 the evolutionary conservation and functional diversity of aquaporins in cotton. BMC Genomics 20:538. 694 10.1186/s12864-019-5928-2 695 Lian HL, Yu X, Ye Q, Ding X, Kitagawa Y, Kwak SS, Su WA, and Tang ZC. 2004. The role of aquaporin RWC3 in drought 696 avoidance in rice. Plant Cell Physiol 45:481-489. 10.1093/pcp/pch058 697 Lin W, Peng Y, Li G, Arora R, Tang Z, Su W, and Cai W. 2007. Isolation and functional characterization of PgTIP1, a 698 hormone-autotrophic cells-specific tonoplast aquaporin in ginseng. J Exp Bot 58:947-956. 699 10.1093/jxb/erl255 700 Liu C, Fukumoto T, Matsumoto T, Gena P, Frascaria D, Kaneko T, Katsuhara M, Zhong S, Sun X, Zhu Y, Iwasaki I, Ding 701 X, Calamita G, and Kitagawa Y. 2013. Aquaporin OsPIP1;1 promotes rice salt resistance and seed 702 germination. Plant Physiol Biochem 63:151-158. 10.1016/j.plaphy.2012.11.018 703 Liu HY, Yu X, Cui DY, Sun MH, Sun WN, Tang ZC, Kwak SS, and Su WA. 2007. The role of water channel proteins and 704 nitric oxide signaling in rice seed germination. Cell Res 17:638-649. 10.1038/cr.2007.34 705 Liu LH, Ludewig U, Gassert B, Frommer WB, and von Wiren N. 2003. Urea transport by nitrogen-regulated 706 tonoplast intrinsic proteins in Arabidopsis. Plant Physiol 133:1220-1228. 10.1104/pp.103.027409 707 Ludevid D, Hofte H, Himelblau E, and Chrispeels MJ. 1992. The Expression Pattern of the Tonoplast Intrinsic Protein 708 gamma-TIP in Arabidopsis thaliana Is Correlated with Cell Enlargement. Plant Physiol 100:1633-1639. 706 tonoplast intrinsic proteins in Arabidopsis. Plant Physiol 133:1220-1228. 10.1104/pp.103.027409 707 Ludevid D, Hofte H, Himelblau E, and Chrispeels MJ. 1992. The Expression Pattern of the Tonoplast Intrinsic Protein 708 gamma-TIP in Arabidopsis thaliana Is Correlated with Cell Enlargement. Plant Physiol 100:1633-1639. Manuscript to be reviewed 723 Myburg AA, Grattapaglia D, Tuskan GA, Hellsten U, Hayes RD, Grimwood J, Jenkins J, Lindquist E, Tice H, Bauer D, 724 Goodstein DM, Dubchak I, Poliakov A, Mizrachi E, Kullan AR, Hussey SG, Pinard D, van der Merwe K, Singh 725 P, van Jaarsveld I, Silva-Junior OB, Togawa RC, Pappas MR, Faria DA, Sansaloni CP, Petroli CD, Yang X, 726 Ranjan P, Tschaplinski TJ, Ye CY, Li T, Sterck L, Vanneste K, Murat F, Soler M, Clemente HS, Saidi N, Cassan- 727 Wang H, Dunand C, Hefer CA, Bornberg-Bauer E, Kersting AR, Vining K, Amarasinghe V, Ranik M, Naithani 728 S, Elser J, Boyd AE, Liston A, Spatafora JW, Dharmwardhana P, Raja R, Sullivan C, Romanel E, Alves-Ferreira 729 M, Kulheim C, Foley W, Carocha V, Paiva J, Kudrna D, Brommonschenkel SH, Pasquali G, Byrne M, Rigault 730 P, Tibbits J, Spokevicius A, Jones RC, Steane DA, Vaillancourt RE, Potts BM, Joubert F, Barry K, Pappas GJ, 731 Strauss SH, Jaiswal P, Grima-Pettenati J, Salse J, Van de Peer Y, Rokhsar DS, and Schmutz J. 2014. The 732 genome of Eucalyptus grandis. Nature 510:356-362. 10.1038/nature13308 733 Nguyen MX, Moon S, and Jung KH. 2013. Genome-wide expression analysis of rice aquaporin genes and 734 development of a functional gene network mediated by aquaporin expression in roots. Planta 238:669- 735 681. 10.1007/s00425-013-1918-9 736 Niu J, Cao D, Li H, Xue H, Chen L, Liu B, and Cao S. 2018. Quantitative proteomics of pomegranate varieties with 737 contrasting seed hardness during seed development stages. Tree Genetics & Genomes 14:14. 738 10.1007/s11295-018-1229-1 739 Patel C, Dadhaniya P, Hingorani L, and Soni MG. 2008. Safety assessment of pomegranate fruit extract: acute and 740 subchronic toxicity studies. Food Chem Toxicol 46:2728-2735. 10.1016/j.fct.2008.04.035 741 Peret B, Li G, Zhao J, Band LR, Voss U, Postaire O, Luu DT, Da Ines O, Casimiro I, Lucas M, Wells DM, Lazzerini L, 742 Nacry P, King JR, Jensen OE, Schaffner AR, Maurel C, and Bennett MJ. 2012. Auxin regulates aquaporin 743 function to facilitate lateral root emergence. Nat Cell Biol 14:991-998. 10.1038/ncb2573 744 Picaud S, Becq F, Dedaldechamp F, Ageorges A, and Delrot S. 2003. Cloning and expression of two plasma 745 membrane aquaporins expressed during the ripening of grape berry. Funct Plant Biol 30:621-630. 746 10.1071/FP02116 747 Pommerrenig B, Diehn TA, and Bienert GP. 2015. Metalloido-porins: Essentiality of Nodulin 26-like intrinsic 748 proteins in metalloid transport. Manuscript to be reviewed 709 10.1104/pp.100.4.1633 710 Luo X, Li H, Wu Z, Yao W, Zhao P, Cao D, Yu H, Li K, Poudel K, Zhao D, Zhang F, Xia X, Chen L, Wang Q, Jing D, and 711 Cao S. 2020. The pomegranate (Punica granatum L.) draft genome dissects genetic divergence between 712 soft- and hard-seeded cultivars. Plant Biotechnol J 18:955-968. 10.1111/pbi.13260 713 Lynch M, and Conery JS. 2000. The evolutionary fate and consequences of duplicate genes. Science 290:1151-1155. 714 10.1126/science.290.5494.1151 713 Lynch M, and Conery JS. 2000. The evolutionary fate and consequences of duplicate genes. Science 290:1151-1155. 714 10.1126/science.290.5494.1151 715 Ma N, Xue J, Li Y, Liu X, Dai F, Jia W, Luo Y, and Gao J. 2008. Rh-PIP2;1, a rose aquaporin gene, is involved in 716 ethylene-regulated petal expansion. Plant Physiol 148:894-907. 10.1104/pp.108.120154 715 Ma N, Xue J, Li Y, Liu X, Dai F, Jia W, Luo Y, and Gao J. 2008. Rh-PIP2;1, a rose aquaporin gene, is involved in 716 ethylene-regulated petal expansion. Plant Physiol 148:894-907. 10.1104/pp.108.120154 717 Madrid-Espinoza J, Brunel-Saldias N, Guerra FP, Gutierrez A, and Del Pozo A. 2018. Genome-Wide Identification 718 and Transcriptional Regulation of Aquaporin Genes in Bread Wheat (Triticum aestivum L.) under Water 719 Stress. Genes (Basel) 9. 10.3390/genes9100497 717 Madrid-Espinoza J, Brunel-Saldias N, Guerra FP, Gutierrez A, and Del Pozo A. 2018. Genome-Wide Identification 718 and Transcriptional Regulation of Aquaporin Genes in Bread Wheat (Triticum aestivum L.) under Water 719 Stress. Genes (Basel) 9. 10.3390/genes9100497 720 Martins Cde P, Pedrosa AM, Du D, Goncalves LP, Yu Q, Gmitter FG, Jr., and Costa MG. 2015. Genome-Wide 721 Characterization and Expression Analysis of Major Intrinsic Proteins during Abiotic and Biotic Stresses in 722 Sweet Orange (Citrus sinensis L. Osb.). PLoS One 10:e0138786. 10.1371/journal.pone.0138786 720 Martins Cde P, Pedrosa AM, Du D, Goncalves LP, Yu Q, Gmitter FG, Jr., and Costa MG. 2015. Genome-Wide 721 Characterization and Expression Analysis of Major Intrinsic Proteins during Abiotic and Biotic Stresses in 722 Sweet Orange (Citrus sinensis L. Osb.). PLoS One 10:e0138786. 10.1371/journal.pone.0138786 PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) Manuscript to be reviewed Manuscript to be reviewed Plant Sci 238:212-227. 10.1016/j.plantsci.2015.06.002 749 Porcel R, Bustamante A, Ros R, Serrano R, and Mulet Salort JM. 2018. BvCOLD1: A novel aquaporin from sugar beet 750 (Beta vulgaris L.) involved in boron homeostasis and abiotic stress. Plant Cell Environ 41:2844-2857. 751 10.1111/pce.13416 752 Qiao X, Li M, Li L, Yin H, Wu J, and Zhang S. 2015. Genome-wide identification and comparative analysis of the heat 753 shock transcription factor family in Chinese white pear (Pyrus bretschneideri) and five other Rosaceae 754 species. BMC Plant Biol 15:12. 10.1186/s12870-014-0401-5 755 Qin G, Liu C, Li J, Qi Y, Gao Z, Zhang X, Yi X, Pan H, Ming R, and Xu Y. 2020. Diversity of metabolite accumulation 756 patterns in inner and outer seed coats of pomegranate: exploring their relationship with genetic 757 mechanisms of seed coat development. Hortic Res 7:10. 10.1038/s41438-019-0233-4 758 Qin G, Xu C, Ming R, Tang H, Guyot R, Kramer EM, Hu Y, Yi X, Qi Y, Xu X, Gao Z, Pan H, Jian J, Tian Y, Yue Z, and Xu Y. 759 2017. The pomegranate (Punica granatum L.) genome and the genomics of punicalagin biosynthesis. Plant 760 J 91:1108-1128. 10.1111/tpj.13625 761 Quigley F, Rosenberg JM, Shachar-Hill Y, and Bohnert HJ. 2002. From genome to function: the Arabidopsis 723 Myburg AA, Grattapaglia D, Tuskan GA, Hellsten U, Hayes RD, Grimwood J, Jenkins J, Lindquist E, Tice H, Bauer D, 724 Goodstein DM, Dubchak I, Poliakov A, Mizrachi E, Kullan AR, Hussey SG, Pinard D, van der Merwe K, Singh 725 P, van Jaarsveld I, Silva-Junior OB, Togawa RC, Pappas MR, Faria DA, Sansaloni CP, Petroli CD, Yang X, 726 Ranjan P, Tschaplinski TJ, Ye CY, Li T, Sterck L, Vanneste K, Murat F, Soler M, Clemente HS, Saidi N, Cassan- 727 Wang H, Dunand C, Hefer CA, Bornberg-Bauer E, Kersting AR, Vining K, Amarasinghe V, Ranik M, Naithani 728 S, Elser J, Boyd AE, Liston A, Spatafora JW, Dharmwardhana P, Raja R, Sullivan C, Romanel E, Alves-Ferreira 729 M, Kulheim C, Foley W, Carocha V, Paiva J, Kudrna D, Brommonschenkel SH, Pasquali G, Byrne M, Rigault 730 P, Tibbits J, Spokevicius A, Jones RC, Steane DA, Vaillancourt RE, Potts BM, Joubert F, Barry K, Pappas GJ, 731 Strauss SH, Jaiswal P, Grima-Pettenati J, Salse J, Van de Peer Y, Rokhsar DS, and Schmutz J. 2014. The 732 genome of Eucalyptus grandis. Nature 510:356-362. Manuscript to be reviewed 764 expression analysis of aquaporins in tomato. PLoS One 8:e79052. 10.1371/journal.pone.0079052 765 Sakurai J, Ishikawa F, Yamaguchi T, Uemura M, and Maeshima M. 2005. Identification of 33 rice aquaporin genes 766 and analysis of their expression and function. Plant Cell Physiol 46:1568-1577. 10.1093/pcp/pci172 767 Schnurbusch T, Hayes J, Hrmova M, Baumann U, Ramesh SA, Tyerman SD, Langridge P, and Sutton T. 2010. Boron 768 Toxicity Tolerance in Barley through Reduced Expression of the Multifunctional Aquaporin HvNIP2;1. 769 153:1706-1715. 10.1104/pp.110.158832 %J Plant Physiology 770 Schuurmans JA, van Dongen JT, Rutjens BP, Boonman A, Pieterse CM, Borstlap AC. Members of the aquaporin famil 771 y in the developing pea seed coat include representatives of the PIP, TIP, and NIP subfamilies. Plant Mol Bi 772 ol. 2003 Nov;53(5):633-45. doi: 10.1023/B:PLAN.0000019070.60954.77. PMID: 15010602. 773 Shivaraj SM, Deshmukh R, Sonah H, and Belanger RR. 2019. Identification and characterization of aquaporin genes 774 in Arachis duranensis and Arachis ipaensis genomes, the diploid progenitors of peanut. BMC Genomics 775 20:222. 10.1186/s12864-019-5606-4 776 Soto G, Alleva K, Mazzella MA, Amodeo G, and Muschietti JP. 2008. AtTIP1;3 and AtTIP5;1, the only highly 777 expressed Arabidopsis pollen-specific aquaporins, transport water and urea. FEBS Lett 582:4077-4082. 778 10.1016/j.febslet.2008.11.002 779 Soto G, Fox R, Ayub N, Alleva K, Guaimas F, Erijman EJ, Mazzella A, Amodeo G, and Muschietti J. 2010. TIP5;1 is an 780 aquaporin specifically targeted to pollen mitochondria and is probably involved in nitrogen remobilization 781 in Arabidopsis thaliana. Plant J 64:1038-1047. 10.1111/j.1365-313X.2010.04395.x 782 Suga N, Takada H, Nomura A, Ohga S, Ishii E, Ihara K, Ohshima K, and Hara T. 2002. Perforin defects of primary 783 haemophagocytic lymphohistiocytosis in Japan. Br J Haematol 116:346-349. 10.1046/j.1365- 784 2141.2002.03266.x 785 Takano J, Wada M, Ludewig U, Schaaf G, von Wiren N, and Fujiwara T. 2006. The Arabidopsis major intrinsic 786 protein NIP5;1 is essential for efficient boron uptake and plant development under boron limitation. Plant 787 Cell 18:1498-1509. 10.1105/tpc.106.041640 788 Terashima I, and Ono K. 2002. Effects of HgCl(2) on CO(2) dependence of leaf photosynthesis: evidence indicating 789 involvement of aquaporins in CO(2) diffusion across the plasma membrane. Plant Cell Physiol 43:70-78. 790 10.1093/pcp/pcf001 791 Tornroth-Horsefield S, Wang Y, Hedfalk K, Johanson U, Karlsson M, Tajkhorshid E, Neutze R, and Kjellbom P. 2006. 792 Structural mechanism of plant aquaporin gating. Nature 439:688-694. 10.1038/nature04316 793 Uçar S, and Karagöz S. 2009. Manuscript to be reviewed 10.1038/nature13308 733 N MX M S d J KH 2013 G id i l i f i i d 733 Nguyen MX, Moon S, and Jung KH. 2013. Genome-wide expression analysis of rice aquaporin genes and 734 development of a functional gene network mediated by aquaporin expression in roots. Planta 238:669- 735 681. 10.1007/s00425-013-1918-9 761 Quigley F, Rosenberg JM, Shachar-Hill Y, and Bohnert HJ. 2002. From genome to function: the Arabidopsis 762 aquaporins. Genome Biol 3:RESEARCH0001. 10.1186/gb-2001-3-1-research0001 763 Reuscher S, Akiyama M, Mori C, Aoki K, Shibata D, and Shiratake K. 2013. Genome-wide identification and 763 Reuscher S, Akiyama M, Mori C, Aoki K, Shibata D, and Shiratake K. 2013. Genome-wide identification and PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) Manuscript to be reviewed Manuscript to be reviewed The slow pyrolysis of pomegranate seeds: The effect of temperature on the product 794 yields and bio-oil properties. Journal of Analytical and Applied Pyrolysis 84:151-156. 795 https://doi.org/10.1016/j.jaap.2009.01.005 796 Vander Willigen C, Postaire O, Tournaire-Roux C, Boursiac Y, and Maurel C. 2006. Expression and inhibition of 797 aquaporins in germinating Arabidopsis seeds. Plant Cell Physiol 47:1241-1250. 10.1093/pcp/pcj094 798 Wallace IS, and Roberts DM. 2004. Homology modeling of representative subfamilies of Arabidopsis major intrinsic 799 proteins. Classification based on the aromatic/arginine selectivity filter. Plant Physiol 135:1059-1068. 800 10.1104/pp.103.033415 801 Wang L, Guo K, Li Y, Tu Y, Hu H, Wang B, Cui X, and Peng L. 2010. Expression profiling and integrative analysis of 802 the CESA/CSL superfamily in rice. BMC Plant Biol 10:282. 10.1186/1471-2229-10-282 803 WU Xiang-yang CC-z, Lü Gao-qiang, WANG Xin-yu. 2016. Identification and Characterization of the AQP Gene 804 Family in Sesame. 49:1844-1858. 10.3864/j.issn.0578-1752.2016.10.002 798 Wallace IS, and Roberts DM. 2004. Homology modeling of representative subfamilies of Arabidopsis major intrinsic 799 proteins. Classification based on the aromatic/arginine selectivity filter. Plant Physiol 135:1059-1068. 800 10.1104/pp.103.033415 803 WU Xiang-yang CC-z, Lü Gao-qiang, WANG Xin-yu. 2016. Identification and Characterization of the AQP Gene 804 Family in Sesame. 49:1844-1858. 10.3864/j.issn.0578-1752.2016.10.002 PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) Manuscript to be reviewed 805 Wudick MM, Luu DT, Tournaire-Roux C, Sakamoto W, and Maurel C. 2014. Vegetative and sperm cell-specific 806 aquaporins of Arabidopsis highlight the vacuolar equipment of pollen and contribute to plant 807 reproduction. Plant Physiol 164:1697-1706. 10.1104/pp.113.228700 808 Xu F, Wang K, Yuan W, Xu W, Liu S, Kronzucker HJ, Chen G, Miao R, Zhang M, Ding M, Xiao L, Kai L, Zhang J, and Zhu 809 Y. 2019. Overexpression of rice aquaporin OsPIP1;2 improves yield by enhancing mesophyll CO2 810 conductance and phloem sucrose transport. Journal of Experimental Botany 70:671-681. 811 10.1093/jxb/ery386 805 Wudick MM, Luu DT, Tournaire-Roux C, Sakamoto W, and Maurel C. 2014. Vegetative and sperm cell-sp 806 aquaporins of Arabidopsis highlight the vacuolar equipment of pollen and contribute to plant 807 reproduction. Plant Physiol 164:1697-1706. 10.1104/pp.113.228700 808 Xu F, Wang K, Yuan W, Xu W, Liu S, Kronzucker HJ, Chen G, Miao R, Zhang M, Ding M, Xiao L, Kai L, Zhang 809 Y. 2019. Overexpression of rice aquaporin OsPIP1;2 improves yield by enhancing mesophyll CO 810 conductance and phloem sucrose transport. Journal of Experimental Botany 70:671-681. 811 10.1093/jxb/ery386 812 Yang S, Zhang X, Yue JX, Tian D, and Chen JQ. 2008. Recent duplications dominate NBS-encoding gene ex 813 two woody species. Mol Genet Genomics 280:187-198. 10.1007/s00438-008-0355-0 814 Yasui M, Hazama A, Kwon TH, Nielsen S, Guggino WB, and Agre P. 1999. Rapid gating and anion permea 815 intracellular aquaporin. Nature 402:184-187. 10.1038/46045 816 Zarei A, Zamani Z, Fatahi R, Mousavi A, Salami SA, Avila C, and Cánovas FM. 2016. Differential expression 817 wall related genes in the seeds of soft- and hard-seeded pomegranate genotypes. Scientia Hort 818 205:7-16. https://doi.org/10.1016/j.scienta.2016.03.043 819 Zhang DY, Ali Z, Wang CB, Xu L, Yi JX, Xu ZL, Liu XQ, He XL, Huang YH, Khan IA, Trethowan RM, and Ma HX 820 Genome-wide sequence characterization and expression analysis of major intrinsic proteins in s 821 (Glycine max L.). PLoS One 8:e56312. 10.1371/journal.pone.0056312 822 Zhou Y, Setz N, Niemietz C, Qu H, Offler CE, Tyerman SD, and Patrick JW. 2007a. Aquaporins and unloadi 823 phloem-imported water in coats of developing bean seeds. Plant Cell Environ 30:1566-1577. 824 10.1111/j.1365-3040.2007.01732.x 825 ZHOU Y, SETZ N, NIEMIETZ C, QU H, OFFLER CE, TYERMAN SD, and PATRICK JW. 2007b. Aquaporins and 826 of phloem-imported water in coats of developing bean seeds. 30:1566-1577. 827 https://doi.org/10.1111/j.1365-3040.2007.01732.x 828 Zhou Y, Tao J, Ahammed GJ, Li J, and Yang Y. Manuscript to be reviewed 2019. Genome-wide identification and expression analysis 829 aquaporin gene family related to abiotic stress in watermelon. Genome 62:643-656. 10.1139/ge 830 0061 831 Zou Z, and Yang J. 2019. Genome-wide comparison reveals divergence of cassava and rubber aquaporin 832 genes after the recent whole-genome duplication. BMC Genomics 20:380. 10.1186/s12864-019 833 Zwiazek J, Xu H, Tan X, Navarro-Ródenas A, and Morte A. 2017a. Significance of oxygen transport throug 834 aquaporins. Scientific Reports 7:40411. 10.1038/srep40411 835 Zwiazek JJ, Xu H, Tan X, Navarro-Rodenas A, and Morte A. 2017b. Significance of oxygen transport throu 836 aquaporins. Sci Rep 7:40411. 10.1038/srep40411 837 812 Yang S, Zhang X, Yue JX, Tian D, and Chen JQ. 2008. Recent duplications dominate NBS-encoding gene expansion in 813 two woody species. Mol Genet Genomics 280:187-198. 10.1007/s00438-008-0355-0 812 Yang S, Zhang X, Yue JX, Tian D, and Chen JQ. 2008. Recent duplications dominate NBS-encoding gene expansion in 813 two woody species. Mol Genet Genomics 280:187-198. 10.1007/s00438-008-0355-0 814 Yasui M, Hazama A, Kwon TH, Nielsen S, Guggino WB, and Agre P. 1999. Rapid gating and anion permeability of an 815 intracellular aquaporin. Nature 402:184-187. 10.1038/46045 814 Yasui M, Hazama A, Kwon TH, Nielsen S, Guggino WB, and Agre P. 1999. Rapid gating and anion permeability of an 815 intracellular aquaporin. Nature 402:184-187. 10.1038/46045 816 Zarei A, Zamani Z, Fatahi R, Mousavi A, Salami SA, Avila C, and Cánovas FM. 2016. Differential expression of cell 817 wall related genes in the seeds of soft- and hard-seeded pomegranate genotypes. Scientia Horticulturae 818 205:7-16. https://doi.org/10.1016/j.scienta.2016.03.043 816 Zarei A, Zamani Z, Fatahi R, Mousavi A, Salami SA, Avila C, and Cánovas FM. 2016. Differential expression of cell 817 wall related genes in the seeds of soft- and hard-seeded pomegranate genotypes. Scientia Horticulturae 818 205:7-16. https://doi.org/10.1016/j.scienta.2016.03.043 819 Zhang DY, Ali Z, Wang CB, Xu L, Yi JX, Xu ZL, Liu XQ, He XL, Huang YH, Khan IA, Trethowan RM, and Ma HX. 2013. 820 Genome-wide sequence characterization and expression analysis of major intrinsic proteins in soybean 821 (Glycine max L.). PLoS One 8:e56312. 10.1371/journal.pone.0056312 819 Zhang DY, Ali Z, Wang CB, Xu L, Yi JX, Xu ZL, Liu XQ, He XL, Huang YH, Khan IA, Trethowan RM, and Ma HX. 2013. 820 Genome-wide sequence characterization and expression analysis of major intrinsic proteins in soybean 821 (Glycine max L.). PLoS One 8:e56312. Manuscript to be reviewed Manuscript to be reviewed 10.1371/journal.pone.0056312 g , , g , , , , Q, , g , , , 820 Genome-wide sequence characterization and expression analysis of major intrinsic proteins in soybean 821 (Glycine max L.). PLoS One 8:e56312. 10.1371/journal.pone.0056312 822 Zhou Y, Setz N, Niemietz C, Qu H, Offler CE, Tyerman SD, and Patrick JW. 2007a. Aquaporins and unloading of 823 phloem-imported water in coats of developing bean seeds. Plant Cell Environ 30:1566-1577. 824 10.1111/j.1365-3040.2007.01732.x 825 ZHOU Y, SETZ N, NIEMIETZ C, QU H, OFFLER CE, TYERMAN SD, and PATRICK JW. 2007b. Aquaporins and unloading 826 of phloem-imported water in coats of developing bean seeds. 30:1566-1577. 827 https://doi.org/10.1111/j.1365-3040.2007.01732.x 828 Zhou Y, Tao J, Ahammed GJ, Li J, and Yang Y. 2019. Genome-wide identification and expression analysis of 829 aquaporin gene family related to abiotic stress in watermelon. Genome 62:643-656. 10.1139/gen-2019- 830 0061 831 Zou Z, and Yang J. 2019. Genome-wide comparison reveals divergence of cassava and rubber aquaporin family 832 genes after the recent whole-genome duplication. BMC Genomics 20:380. 10.1186/s12864-019-5780-4 833 Zwiazek J, Xu H, Tan X, Navarro-Ródenas A, and Morte A. 2017a. Significance of oxygen transport through 834 aquaporins. Scientific Reports 7:40411. 10.1038/srep40411 835 Zwiazek JJ, Xu H, Tan X, Navarro-Rodenas A, and Morte A. 2017b. Significance of oxygen transport through 836 aquaporins. Sci Rep 7:40411. 10.1038/srep40411 837 PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) Manuscript to be reviewed Figure 1 Phylogenetic analysis of AQP proteins from pomegranate, grape, Arabidopsis, Populus trichocarpa and Eucalyptus. The tree was generated by the neighbor-joining (NJ) method in MEGA 7.0 with 1,000 bootstrap replicates. Diﬀerent background colors indicate the diﬀerent subfamilies of the AQP proteins. PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) Figure 2 Analysis of chromosomal locations and syntenic relationships of PgrAQP genes. Manuscript to be reviewed Analysis of chromosomal locations and syntenic relationships of PgrAQP genes. The AQP genes in pomegranate were mapped to diﬀerent chromosomes using TBtools, and AQP genes in red and blue represent genes with segmental and tandem duplications, respectively. PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) Figure 3 Synteny analysis of PgrAQP genes between pomegranate and three plant species. (A) Pomegranate and A. thaliana, (B) pomegranate and Vitis vinifera, (C) pomegranate and Eucalyptus grandis. The gray lines indicated collinearity between pomegranate and other species. The red lines highlight the syntenic AQP gene pairs. The chromosome name is indicated at the top of every chromosome. Manuscript to be reviewed Figure 4 Expression analysis of the PgrAQP genes in pomegranate. PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) Manuscript to be reviewed Manuscript to be reviewed Expression analysis of the PgrAQP genes in pomegranate. (A) Expression proﬁle of PgrAQP genes in the cultivated pomegranate cultivar ‘Dabenzi,’ including roots, ﬂowers, leaves, and three stages of the peel, inner, and outer seed coats (50, 95, and 140 days after pollination). (B) Expression proﬁling of PgrAQP genes at diﬀerent developmental stages of the seed coats in pomegranate. The abbreviations are as follows: D: P. granatum ‘Dabenzi,’ T: P. granatum ‘Tunisia,’ O: Outer seed coat, I: Inner seed coat. The number represents the number of days after pollination (DAP). The heat map was generated using TBtools. Expression data were calculated with log2 normalization based on FPKM values. PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) Figure 5 Expression proﬁles of six Pomegranate PIPs roots in response to water deﬁcit Expression proﬁles of six Pomegranate PIPs roots in response to water deﬁcit qRT-PCR was performed to determined the relative transcript level for the six PgrPIP ge Relative expression level was normalized relative to untreated control group (0 h PEG treatment). Error bar represent SE of three independent biological replicated. Asterisk indicate signiﬁcant diﬀerences, P < 0.05. qRT-PCR was performed to determined the relative transcript level for the six PgrPIP genes. Relative expression level was normalized relative to untreated control group (0 h PEG treatment). Error bar represent SE of three independent biological replicated. Asterisk indicate signiﬁcant diﬀerences, P < 0.05. Relative expression level was normalized relative to untreated control group (0 h PEG treatment). Error bar represent SE of three independent biological replicated. Asterisk indicate signiﬁcant diﬀerences, *P < 0.05. PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) Manuscript to be reviewed Manuscript to be reviewed Figure 6 Water accumulation in seed coats of pomegranate. PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) Manuscript to be reviewed Figure 7 Linear regression between gene expression of PgrAQPs and water accumulation. Water accumulation in seed coats of pomegranate. The abbreviations in the sample designations represent the cultivar names: D forP. granatum‘Dabenzi’ and T forP. granatum ‘Tunisia.’ The second letter represents tissue: O and I stands for the outer seed coats and the inner seed coat, respectively. The numbers represent the three stages of seed coat development at diﬀerent days after ﬂowering (DAF). Diﬀerent letters indicate a signiﬁcant diﬀerence (P<0.05). PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) Linear regression between gene expression of PgrAQPs and water accumulation. Linear regression between gene expression of PgrAQPs and water accumulation. Correlation analysis between gene expression of PgrPIP1.3, PgrPIP2.8 and PgrSIP1.2 andwater accumulation in seed coats from P. granatum ‘Dabenzi’ (A, C, E) and P. granatum‘Tunisia’ (B, D, F). Correlation analysis between gene expression of PgrPIP1.3, PgrPIP2.8 and PgrSIP1.2 andwater accumulation in seed coats from P. granatum ‘Dabenzi’ (A, C, E) and P. granatum‘Tunisia’ (B, D, F). PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) Manuscript to be reviewed PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) Manuscript to be reviewed Manuscript to be reviewed Table 1(on next page) Table 1(on next page) Table 1(on next page) The Ka and Ks values of duplicated PgrAQP gene pairs. PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021) Manuscript to be reviewed 1 Table 1 The Ka and Ks values of duplicated PgrAQP gene pairs. 2 Note: Ka, 3 non- 4 synonym 5 ous 6 substituti 7 on rate; 8 Ks, 9 synonym 10 ous 11 substituti 12 on rate; 13 Mya, Duplicated gene pairs Duplicate type Ka Ks Ka/Ks Time (Mya) Purify selection PgrTIP1.2 vs PgrTIP1.5 Segmental 0.074 1.54 0.048 5.12 Yes PgrTIP3.1 vs PgrTIP3.2 Segmental 0.254 1.19 0.213 3.97 Yes PgrPIP1.5 vs PgrPIP1.1 Segmental 0.078 0.86 0.090 2.88 Yes PgrPIP2.5 vs PgrPIP2.2 Segmental 0.123 1.26 0.098 4.21 Yes PgrPIP2.7 vs PgrPIP2.3 Segmental 0.102 1.38 0.074 4.59 Yes PgrPIP2.6 vs PgrPIP2.7 Segmental 0.073 1.30 0.057 4.32 Yes PgrTIP1.2 vs PgrTIP1.1 Segmental 0.096 2.09 0.046 6.97 Yes PgrTIP1.2 vs PgrTIP1.7 Tandem 0.101 0.52 0.194 1.74 Yes Table 1 The Ka and Ks values of duplicated PgrAQP gene pairs. Table 1 The Ka and Ks values of duplicated PgrAQP gene pairs. PeerJ reviewing PDF \| (2020:12:56609:2:0:NEW 8 Jun 2021)
https://openalex.org/W4235824823	https://academiccommons.columbia.edu/doi/10.7916/D8891HWV/download	English	null	Spontaneous Intracranial Hypotension Associated with Kinetic Tremor and Ataxia	Tremor and other hyperkinetic movements	2,016	cc-by	1,361	Funding: None. Financial Disclosures: None. Conflict of Interest: The author reports no conflict of interest. Conflict of Interest: The author reports no conflict of interest. Ethics Statement: All patients that appear on the video have provided written informed consent; authorization for the video-taping and for publication of the videotape was provided. : All patients that appear on the video have provided written informed consent; authorization for the video-taping and for publication of the Ethics Statement: All patients that appear on the video have provided written informed consent; authorization for the video-taping and pachymeningeal enhancement of the infra- and supratentorial compartments without involvement of the leptomeninges, and a sagittal T1 slice showed mild brain sagging (Video segment 2). Following the MRI findings, a lumbar puncture identified low CSF pressure of 60 mmH2O with normal CSF analysis, which included tuberculosis/fungal cultures and cytology. Upon further questioning, he admitted to having occasional orthostatic headaches. On the basis of the clinical presentation and work-up, the diagnosis of spontaneous intracranial hypotension (SIH) was confirmed. Unfortunately, the computed tomography (CT) myelogram failed to show the site of the CSF leak. Later on, the patient decided not to pursue any further work-up, but instead proceeded with conservative management (i.e., increased caffeine intake) with partial response. SIH was first described by Schaltenbrand in 1938. SIH is a syndrome caused by low CSF pressure due to leakage of CSF that mostly occurs at the cervicothoracic or thoracolumbar junction of the spine. Dural weakness could be secondary to an underlying connective tissue disorder, meningeal diverticula, or because of mechanical disruption of the dura, A 68-year-old male presented with a 2-year history of slightly progressive tremor of bilateral hands associated with an unsteady gait. The tremor was elicited by purposeful movements of the hands. Drinking alcohol did not improve the tremor. He reported mild staggering gait with intermittent hesitation upon gait initiation. He denied bulbar symptoms or deafness. There was no family history of mental retardation. On examination, he displayed a mild, fast, and distal kinetic tremor of bilateral hands without a postural or a resting component (Video segment 1). The Archimedes spiral test was normal. There were no signs of Parkinsonism with the exception of mild gait start hesitation. Cerebellar function was abnormal for mild upper limb dysmetria and trunk titubation. The ambulatory examination showed mild broad base gait associated with impaired tandem gait (Video segment 1). Spontaneous Intracranial Hypotension Associated with Kinetic Tremor and Ataxia Richard Salazar1* 1 Department of Neuroscience, Jackson Hospital & Clinic, Montgomery, AL, USA Video Abstracts Tremor and Other Hyperkinetic Movements http://www.tremorjournal.org The Center for Digital Research and Scholarship Columbia University Libraries/Information Services Abstract Background: Spontaneous intracranial hypotension (SIH) is a clinically variable syndrome caused by low cerebrospinal fluid (CSF) pressure due to a non-traumatic CSF leak. Phenomenology Shown: This case describes a 68-year-old gentleman who presents with chronic and slightly progressive kinetic tremor of bilateral hands associated with gait ataxia and gait start hesitation. Educational Value: This case underscores the importance of having a high index of suspicion for the diagnosis of SIH when encountering a patient presenting with late-onset progressive kinetic tremor and gait ataxia syndrome. Keywords: Spontaneous Intracranial Hypotension, Ataxia, Tremor Keywords: Spontaneous Intracranial Hypotension, Ataxia, Tremor Citation: Salazar R Spontaneous intracranial hypotension associated with kinetic tremor and ataxia Tremor Other Hyperkinet Mov 2016; 6 doi: 10 7916 Citation: Salazar R. Spontaneous intracranial hypotension associated with kinetic tremor and ataxia. Tremor Other Hyperkinet Mov. 2016; 6. doi: 10.7916/ D8HQ3ZN5 * To whom correspondence should be addressed. E-mail: ricsalmont@hotmail.com * To whom correspondence should be addressed. E-mail: ricsalmont@ Editor: Elan D. Louis, Yale University, USA Received: December 29, 2015 Accepted: January 22, 2016 Published: March Copyright: ’ 2016 Salazar. This is an open-access article distributed under the terms of the Creative Commons Attribution–Noncommercial–No Derivatives License, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited; that no commercial use is made of the work; and that the work is not altered or transformed. Funding: None. Funding: None. The remainder of the neurological examination, including his mental examination, was normal. A thorough work-up ensued, which included brain magnetic resonance imaging (MRI) studies followed by cerebrospinal fluid (CSF) and serum analysis. Contrast-enhanced brain MRI revealed diffuse, linear, and thick 1 Salazar R Intracranial Hypotension Associated with Kinetic Tremor Video 1. Movement disorders associated with spontaneous intracranial hypotension. This video displays mild kinetic tremor of bilateral hands without postural or resting components. The ambulatory pattern is characterized by gait start hesitation in addition to a mild ataxic component and an abnormal tandem gait (segment 1). A non-contrast brain magnetic resonance imaging (MRI) T1 sagittal slice shows minimal downward displacement of brain structures. Contrast- enhanced brain MRI shows diffuse and linear meningeal enhancement that involves the pachymeninges of both the infra- and supratentorial compartments without evidence of involvement of the leptomeninges (i.e., no abnormal enhancement around the brainstem, within the depth of the cerebral sulci or the Sylvian fissures) (segment 2). Video 1. Movement disorders associated with spontaneous intracranial hypotension. This video displays mild kinetic tremor of bilateral hands without postural or resting components. The ambulatory pattern is characterized by gait start hesitation in addition to a mild ataxic component and an abnormal tandem gait (segment 1). A non-contrast brain magnetic resonance imaging (MRI) T1 sagittal slice shows minimal downward displacement of brain structures. Contrast- enhanced brain MRI shows diffuse and linear meningeal enhancement that involves the pachymeninges of both the infra- and supratentorial compartments without evidence of involvement of the leptomeninges (i.e., no abnormal enhancement around the brainstem, within the depth of the cerebral sulci or the Sylvian fissures) (segment 2). as in the case of craniotomy, spinal surgery, craniospinal trauma, or placement of a ventriculoperitoneal shunt. Medical causes of SIH include dehydration, diabetic coma, uremia, and severe systemic illness. SIH is recognized as a clinically variable syndrome where movement disorders could represent the main presentation.1,2 In addition to orthostatic headaches, a wide variety of neurological manifestations have been described, such as ataxia, vertigo, blurred vision, cranial nerve palsies, bulbar dysfunction, chorea, parkinsonism, and tremor.2,3 Unfortunately, the pathophysiology behind these neurological symptoms is not well understood. Traditionally, the work-up consists of contrast-enhanced brain MRI and lumbar puncture followed by CT/magnetic resonance myelogram or radioisotope cisternography to localize the site of the CSF leak. The Center for Digital Research and Scholarship Columbia University Libraries/Information Services Tremor and Other Hyperkinetic Movements http://www.tremorjournal.org Funding: None. However, the sensitivity of CT myelogram has not been well studied, and many CSF leaks remain occult on all types of spinal imaging studies, as in this case.2 Brain MRI studies in patients with SIH usually show diffuse thickening of the pachymeninges with gadolinium enhancement, engorge- ment of venous sinuses, subdural fluid collections, enlargement of the pituitary gland, and brain sagging. The resolution of these abnormalities on magnetic resonance images correlates with improvement in clinical symptoms. The available therapeutic options include conservative manage- ment, epidural blood patch, and epidural injection of fibrin glue; surgical repair may be necessary in certain cases. Outcomes are usually favorable but sometimes not long-lasting.2 This case highlights the broad clinical spectrum of this still under-recognized condition that at times presents with movement disorders as the main manifestation. 1. Wiesemann E, Berding G, Goetz F, Windhagen A. Spontaneous intracranial hypotension: Correlation of imaging findings with clinical features. Eur Neurol 2006;56(4):204–210. doi: 10.1159/000096487. 2. Mokri B. Movement disorders associated with spontaneous CSF leaks: A case series. Cephalalgia 2014;34(14):1134–1141. doi: 10.1177/0333102414531154. 3. Turgut N, Unlu¨ E, Hamamciog˘lu MK, Gu¨ldiken B, Albayram S. Postural tremor as a manifestation of spontaneous intracranial hypotension. J Clin Neurosci 2010;17(2):255–257. doi: 10.1016/j.jocn.2009.05.024. References 1. Wiesemann E, Berding G, Goetz F, Windhagen A. Spontaneous intracranial hypotension: Correlation of imaging findings with clinical features. Eur Neurol 2006;56(4):204–210. doi: 10.1159/000096487. 2. Mokri B. Movement disorders associated with spontaneous CSF leaks: A case series. Cephalalgia 2014;34(14):1134–1141. doi: 10.1177/0333102414531154. 2. Mokri B. Movement disorders associated with spontaneous CSF leaks: A case series. Cephalalgia 2014;34(14):1134–1141. doi: 10.1177/0333102414531154. 3. Turgut N, Unlu¨ E, Hamamciog˘lu MK, Gu¨ldiken B, Albayram S. Postural tremor as a manifestation of spontaneous intracranial hypotension. J Clin Neurosci 2010;17(2):255–257. doi: 10.1016/j.jocn.2009.05.024. 2
https://openalex.org/W4229052324	https://www.degruyter.com/document/doi/10.1515/nanoph-2022-0190/pdf	English	null	Multitask bidirectional digital coding metasurface for independent controls of multiband and full-space electromagnetic waves	Nanophotonics	2,022	cc-by	7,521	Liang Wei Wu, Hui Feng Ma, Yue Gou, Rui Yuan Wu, Zheng Xing Wang, Qiang Xiao and Tie Jun Cui Multitask bidirectional digital coding metasurface for independent controls of multiband and full- space electromagnetic waves https://doi.org/10.1515/nanoph-2022-0190 Received April 1, 2022; accepted April 27, 2022; published online May 6, 2022 electromagnetic (EM) waves by properly engineering a arranging periodic or aperiodic subwavelength artific meta-atoms [1–3], which can not only inherit the super electromagnetic (EM) waves by properly engineering and arranging periodic or aperiodic subwavelength artificial meta-atoms [1–3], which can not only inherit the superior capabilities of metamaterials in manipulating the EM properties, such as phase [4–7], magnitude [7–9], polari- zation [10, 11], but also overcome the challenges encoun- tered in metamaterials, such as high losses, bulk, difﬁcult manufacturing, and so on. electromagnetic (EM) waves by properly engineering and arranging periodic or aperiodic subwavelength artificial meta-atoms [1–3], which can not only inherit the superior capabilities of metamaterials in manipulating the EM properties, such as phase [4–7], magnitude [7–9], polari- zation [10, 11], but also overcome the challenges encoun- tered in metamaterials, such as high losses, bulk, difﬁcult manufacturing, and so on. Abstract: Full-space metasurfaces demonstrate powerful abilities in manipulating electromagnetic (EM) waves, but most of them are usually single band. Here, a multiband bidirectional digital coding metasurface is proposed for multiple tasks, which can achieve completely different functions in up to six frequency bands when the EM waves are incident from the front and back of the metasurface. As a proof-of-concept, we design and fabricate a dual-band full-space metasurface with integrated functions of reflection, transmission, holographic imaging, and vortex- beam forming, and a six-band full-space metasurface with completely independent holographic imaging functions at different frequency bands. Simulated and measured re- sults are in good agreements with the theoretical pre- dictions, verifying the good performance of the proposed multitask bidirectional digital coding metasurface. Generally, the metasurfaces are mainly classified to two types: reflection [4, 11–13] and transmission [6, 14], which have demonstrated powerful abilities to control the EM waves. However, most of them can only manipulate the EM waves in half space, including polarization-dependent metasurfaces [15–18], which restrict the utilization of space resources. In order to expand the utilization of space re- sources and improve the information capacity, the concept of full-space metasurface has been proposed in recent years [19–22], but they usually work in a single frequency or are direction dependent, which severely limit the design freedom. A spin-encoded wavelength-direction multi- tasking Janus metasurface was proposed in Ref. [23], but it can only achieve one single wavelength control of trans- mission waves whether the EM waves are incident from the front or the back [23]. In 2014, the concept of digital coding metasurface was proposed [24–26], which provides a link between the physical world and information science. Nanophotonics 2022; 11(12): 2977–2987 Liang Wei Wu, Hui Feng Ma, Yue Gou, Rui Yuan Wu, Zheng Xing Wang, Qiang Xiao and Tie Jun Cui Multitask bidirectional digital coding metasurface for independent controls of multiband and full- space electromagnetic waves https://doi.org/10.1515/nanoph-2022-0190 Received April 1, 2022; accepted April 27, 2022; published online May 6, 2022 electromagnetic (EM) waves by properly engineering a arranging periodic or aperiodic subwavelength artific meta-atoms [1–3], which can not only inherit the super Many impressive works have been proposed using digital coding metasurfaces, such as EM diffusion [27], microwave ho- lography [28, 29], harmonic modulations [30–32], and wireless communications [33, 34]. The digital coding method simpliﬁes the design and provides a new inspira- tion for the full-space metasurface [29, 35–39]. However, the previously proposed full-space digital coding meta- surfaces still have similar limitations of single band and unidirectional control. Keywords: bidirectional; digital coding metasurface; full space; multi-band; multi-task. This work is licensed under the Creative Commons Attribution 4.0 Open Access. © 2022 Liang Wei Wu et al., published by De Gruyter. International License. Wu et al., published by De Gruyter. This work is licensed under the Creative Commons Attribution 4.0 Open Access. © 2022 Liang Wei Wu et al., published by De Gruyter. This work is licensed under the Creative Commons Attribution 4.0 International License. 1 Introduction EM waves are incident from the front and the back of the metasurface, respectively. By adjusting the geometric pa- rameters of unit cells, the working frequency band can be customized and extended to six frequency bands at most, and the metasurface can perform completely different functions in each band. As a proof of concept, a dual-band full-space metasurface with integrated functions of reflectarray, transmitarray, holographic imaging and vor- tex beam generation, as well as a six-band full-space metasurface with independent holographic imaging func- tion of Arabic numerals “1” to “6” at each frequency band are designed and fabricated. The measured results show a good agreement with the theoretical predictions and full- wave simulations, verifying the excellent performance of the proposed multitask bidirectional digital coding meta- surface. This work may provide an efficient way to expand the utilization of space resources and improve information capacity of metadevices. EM waves are incident from the front and the back of the metasurface, respectively. By adjusting the geometric pa- rameters of unit cells, the working frequency band can be customized and extended to six frequency bands at most, and the metasurface can perform completely different functions in each band. As a proof of concept, a dual-band full-space metasurface with integrated functions of reflectarray, transmitarray, holographic imaging and vor- tex beam generation, as well as a six-band full-space metasurface with independent holographic imaging func- tion of Arabic numerals “1” to “6” at each frequency band are designed and fabricated. The measured results show a good agreement with the theoretical predictions and full- wave simulations, verifying the excellent performance of the proposed multitask bidirectional digital coding meta- surface. This work may provide an efficient way to expand the utilization of space resources and improve information capacity of metadevices. The unit cell of metasurface is illustrated in Figure 2a, which is a four layer structure. The layer 1 and 4 are similar and composed of three metal strips, while the layer 2 and layer 3 are the same with three slits etched on a metal surface. The dielectric substrates are Rogers 4350B with relative permittivity of 3.66 and loss tangent of 0.0037, whose thickness is 0.508 mm. The air spacing between the adjacent dielectric substrates is g = 5 mm, and the period of unit cell is p = 15 mm. 1 Introduction The other geometric parameters of a, b, w1, w2, s1, s2, R1, R2, F1, F2, T1, and T2 are variables, which can be designed according to the working frequencies and 1 Introduction Metasurface, as a two-dimensional (2D) planar version of metamaterial, provides an easy way to control the *Corresponding authors: Hui Feng Ma and Tie Jun Cui, State Key Laboratory of Millimeter Waves, School of Information Science and Engineering, Southeast University, Nanjing, 210096, China; and Institute of Electromagnetic Space, Southeast University, Nanjing, 210096, China, E-mail: hfma@seu.edu.cn (H. F. Ma), tjcui@seu.edu.cn (T.J. Cui). https://orcid.org/0000-0002-5862- 1497 (T.J. Cui) Liang Wei Wu, Yue Gou, Rui Yuan Wu, Zheng Xing Wang and Qiang Xiao, State Key Laboratory of Millimeter Waves, School of Information Science and Engineering, Southeast University, Nanjing, 210096, China; and Institute of Electromagnetic Space, Southeast University, Nanjing, 210096, China Liang Wei Wu, Yue Gou, Rui Yuan Wu, Zheng Xing Wang and Qiang Xiao, State Key Laboratory of Millimeter Waves, School of Information Science and Engineering, Southeast University, Nanjing, 210096, China; and Institute of Electromagnetic Space, Southeast University, Nanjing, 210096, China Liang Wei Wu, Yue Gou, Rui Yuan Wu, Zheng Xing Wang and Qiang Xiao, State Key Laboratory of Millimeter Waves, School of Information Science and Engineering, Southeast University, Nanjing, 210096, China; and Institute of Electromagnetic Space, Southeast University, Nanjing, 210096, China Here, we propose a multitask bidirectional digital coding metasurface that can achieve independent controls of both reflection and transmission waves in different fre- quency bands, and perform different functions when the 2978 L.W. Wu et al.: Multitask bidirectional digital coding metasurface 2978 x-polarized EM waves are incident from the back of the metasurface (along z direction), which is not shown in Figure 1. When y-polarized EM waves are incident from the front of the metasurface, the EM waves will be totally re- ﬂected at the layer 2 and independently controlled at fre- quencies of f3 and f4, respectively. Similarly, when y-polarized EM waves are incident from the back of the metasurface, the independent control of dual-band reﬂection waves at other two different frequencies of f5 and f6 can be achieved. In addition, by adjusting the cor- responding geometric parameters of unit cells, these three groups of working frequencies ((f1, f2), (f3, f4), (f5, f6)) can be designed to be the same or totally different. As a proof of concept, a dual-band full-space metasurface with multifunctional integration of reﬂectarray, transmitarray, holographic imaging and vortex beam generation, and a six-band full-space metasurface with independent holo- graphic imaging function at six different frequencies are designed and fabricated, respectively. 2 Principe of operation Figure 1 shows the schematic of the proposed multi-task bidirectional full-space digital coding metasurface, which is composed of four layers. When the x-polarized EM waves are incident from the front of the metasurface (along −z direction), the EM waves can be effectively transmitted and independently controlled at frequencies of f1 and f2, respectively. Because of the structural reciprocity, the same conclusion can also be obtained when the Figure 1: Schematic diagram of the proposed multitask bidirectional digital coding metasurface for independent control of multiband full-space EM waves. Figure 1: Schematic diagram of the proposed multitask bidirectional digital coding metasurface for independent control of multiband full-space EM waves. L.W. Wu et al.: Multitask bidirectional digital coding metasurface 29 2979 Figure 2: The configuration of unit cell and its EM responses. (a) The configuration of unit cell. (b, c) The amplitude and phase characteristics of the unit cell under (b) x- and (c) y-polarized incidences. (d–g) The simulated amplitude distribution of electric ﬁelds on the ﬁrst layer with R1 = 14 mm, R2 = 5 mm under x-polarized incidence at (d) 8.4 GHz and (e) 15.9 GHz, and under y-polarized incidence at (f) 8.4 GHz and (g) 15.9 GHz. (h–k) The simulated amplitude distribution of electric ﬁeld on the second layer with T1 = 14.8 mm, T2 = 6.4 mm under x-polarized incidence at (h) 8.4 GHz and (i) 15.9 GHz, and under y-polarized incidence at (j) 8.4 GHz and (k) 15.9 GHz. . . u et al.: ult tas b d ect o al d g tal cod g etasu ace 979 Figure 2: The configuration of unit cell and its EM responses. Figure 2: The configuration of unit cell and its EM responses. (a) The configuration of unit cell. (b, c) The amplitude and phase characteristics of the unit cell under (b) x- and (c) y-polarized incidences. (d–g) The simulated amplitude distribution of electric ﬁelds on the ﬁrst layer with R1 = 14 mm, R2 = 5 mm under x-polarized incidence at (d) 8.4 GHz and (e) 15.9 GHz, and under y-polarized incidence at (f) 8.4 GHz and (g) 15.9 GHz. (h–k) The simulated amplitude distribution of electric ﬁeld on the second layer with T1 = 14.8 mm, T2 = 6.4 mm under x-polarized incidence at (h) 8.4 GHz and (i) 15.9 GHz, and under y-polarized incidence at (j) 8.4 GHz and (k) 15.9 GHz. 2 Principe of operation g g p (a) The configuration of unit cell. (b, c) The amplitude and phase characteristics of the unit cell under (b) x- and (c) y-polarized incidences. (d–g) The simulated amplitude distribution of electric ﬁelds on the ﬁrst layer with R1 = 14 mm, R2 = 5 mm under x-polarized incidence at (d) 8.4 GHz and (e) 15.9 GHz, and under y-polarized incidence at (f) 8.4 GHz and (g) 15.9 GHz. (h–k) The simulated amplitude distribution of electric ﬁeld on the second layer with T1 = 14.8 mm, T2 = 6.4 mm under x-polarized incidence at (h) 8.4 GHz and (i) 15.9 GHz, and under y-polarized incidence at (j) 8.4 GHz and (k) 15.9 GHz. functionalities of metasurface. It is worth mentioning that the parameters of R (R1, R2) and F (F1, F2) represent the lengths of metal strips on the ﬁrst and fourth layers, respectively. unit cell has the high transmittance at both 8.4 GHz (−3.2 dB) and 15.9 GHz (−2.8 dB), while the 180° trans- mission phase differences at 8.4 and 15.9 GHz can be achieved by changing T1 and T2, respectively. Hence, ac- cording to the knowledge of coding metamaterials [24], the unit cell can be used to realize dual-band 1-bit transmission phase code to independently manipulate the transmitted x-polarized waves at 8.4 and 15.9 GHz, respectively. How- ever, when y-polarized waves are incident along −z direc- tion, the waves will be totally reﬂected by unit cell, and the 180° reﬂection phase differences at 8.4 and 15.9 GHz also can be achieved by changing R1 and R2, respectively, as shown in Figure 2c. Hence, dual-band 1-bit reﬂection phase For the first case, to design a dual-band full-space metasurface, we first fix partial parameters as follows: a = 1.75 mm, b = 1.5 mm, w1 = 1.5 mm, w2 = 1 mm, s1 = 2 mm, and s2 = 1 mm, and then adjust the R1, R2, F1, F2, T1, and T2 to meet the required transmission and reﬂection phases. Here, the ﬁrst and fourth layers are set to be the same, i. e. R1 = F1 and R2 = F2. Figure 2b shows the transmission amplitude and phase distributions of unit cell under x-polarized incidence with the changes of T1 and T2. The 2980 L.W. 2 Principe of operation Considering that the layer 2 is composed of metal sheet decorated with slits, no EM response means all the incoming waves are totally re- ﬂected, so the phase responses of y-polarized reﬂection waves at 8.4 and 15.9 GHz are mainly dependent on the long and short metal strips on layer 1, respectively, which are consistent with the results shown in Figure 2c. The same conclusion can be made when y-polarized waves incoming from back side, and because the y-polarized waves incoming from front and back sides are isolated by layers 2 and 3, the forward and backward reﬂection waves can be independently manipulated. For the y-polarized waves, when they are incident from the front of metasurface, the reﬂected directive radiation and holographic imaging are constructed at 8.4 and 15.9 GHz, respectively. For directive radiation at 8.4 GHz, a y-polarized rectangular waveguide as a feeding source of reﬂectarray is placed 300 mm away from the center of metasurface, and the reﬂected beam is deﬂected to the direction of −17° in the xoz plane whose calculated 1-bit reﬂection phase distribution is given in Figure 3c. The simulated far-ﬁeld radiation pattern and gain are shown in Figure 3i. The result veriﬁes that the EM waves are reﬂected and deﬂected to −17°, with gain of 22.5 dB and side lobe level of −14.5 dB, whose 3-dB gain bandwidth can reach 2.9 GHz (34.5%). For holographic imaging at 15.9 GHz, the required 1-bit reﬂection phase distribution of a holographic image of letter “Y” under the normally incident plane waves is shown in Figure 3d, whose focal plane is z = 188 mm. The simulated result is also in good agreement with the calculations, as shown in Figure 3j. When the y-polarized EM waves are incident from the back side of metasurface, the reﬂected orbital-angular-momentum (OAM) beams with topological charges of l = +1 and l = −1 are realized at 8.4 and 15.9 GHz, respectively. The incident waves are generated by a rectangular wave guide placed 300 mm away from the center of metasurface, and the re- ﬂected beam is deﬂected to −17° in the xoz plane. 2 Principe of operation Wu et al.: Multitask bidirectional digital coding metasurface 2980 from the center of metasurface, and the corresponding calculated 1-bit transmission phase distribution is given in Figure 3a. The simulated far-ﬁeld radiation pattern and gain are demonstrated in Figure 3g. The result shows that the maximum gain of 19.8 dB and the side lobe level of −14.3 dB can be achieved at 8.4 GHz, whose 3-dB gain bandwidth is 2 GHz (23.8%). For holographic imaging at 15.9 GHz, an x-polarized plane waves are normally incident to the metasurface along −z direction, and the focal plane of transmission holographic image is designed at z = −188mm, and the 1-bit holographic phase distribution of a letter “X” is given in Figure 3b, which is calculated by the weighted Gerchberg–Saxton algorithm [40]. The simulated holographic image has a good agreement with the theoretical calculation, as shown in Figure 3h. code can be realized to independently manipulate the re- ﬂected y-polarized waves at 8.4 and 15.9 GHz, respectively. It should be pointed out that the efﬁciency of transmission mode is only about 50% due to the limitation of trans- mission amplitude, but the efﬁciency of reﬂection mode can be as high as 97% because the reﬂection amplitude can reach near to 100%. To illustrate the working mechanism of unit cell, the simulated amplitude responses of electric fields on the layer 1 and 2 are given in Figure 2d–k for x- and y-polarized waves incident from the front of unit cell, respectively. For x-polarized waves, metal strips on layer 1 almost have no EM responses at both 8.4 and 15.9 GHz, as shown in Figure 2d and e, while the strong resonances are excited at 8.4 and 15.9 GHz by the long and short slits on layer 2, respectively, as shown in Figure 2h and i. Hence, because layers 2 and 3 are the same, the amplitude and phase re- sponses of x-polarized transmission waves at 8.4 and 15.9 GHz are mainly related to the long and short slits on these two layers, respectively, which are consistent with the results shown in Figure 2b. For y-polarized waves, the strong resonances are excited by the long and short metal strips on layer 1 at 8.4 and 15.9 GHz, respectively, as shown in Figure 2f and g, while no any EM response is excited by layer 2, as shown in Figure 2j and k. 2 Principe of operation Figure 3e and f show the 1-bit reﬂection phase distributions at 8.4 and 15.9 GHz, and the simulated far-ﬁeld radiation patterns and near-ﬁeld phase distributions of reﬂection beams are illustrated in Figure 3k and l, respectively, which verify that +1st-order and −1st-order OAM beams are generated at 8.4 and 15.9 GHz, respectively. Hence, six different functions are achieved by metasurface at two frequencies of 8.4 and 15.9 GHz. Based on the above mentioned unit cell, a dual-band full-space metasurface with multi-functional integration of reflectarray, transmitarray, holographic imaging, and vortex beam generation is designed and simulated. The metasurface is composed of 25 × 25 unit cells and designed to work at 8.4 and 15.9 GHz, respectively. Figure 3 shows the distributions of coding sequences for different func- tions and the simulated results. For the x-polarized waves, the transmitted directive radiation and holographic imaging are constructed at 8.4 and 15.9 GHz, respectively. For directive radiation at 8.4 GHz, an x-polarized rectangular waveguide antenna as a feeding source of transmitarray is placed 300 mm away L.W. Wu et al.: Multitask bidirectional digital coding metasurface 2981 Figure 3: The phase distributions and simulated results of the dual-band full-space digital coding metasurface. (a–f) The 1-bit phase distribution of (a) the transmitarray at 8.4 GHz, (b) the transmitted imaging at 15.9 GHz, (c) the reﬂectarray at (d) the reﬂected imaging at 15.9 GHz, (e) the reﬂected beam with l = +1 OAM at 8.4 GHz and (f) the reﬂected beam with l = −1 OAM at 1 (g) 2D far-ﬁeld radiation pattern and gain bandwidth of the transmitarray under the x-polarized incidence. (h) The holographic images “X” on the plane of z = −188mm at 15.9 GHz. (i) 2D far-ﬁeld radiation pattern and gain bandwidth of the reﬂectarray under the y-po incidence. (j) The holographic images of letter “Y” on the plane of z = 188 mm at 15.9 GHz. (k–l) 2D far-ﬁeld radiation patterns and di t ib ti f (k) l 1 OAM b d (l) l 1 OAM b th l f 600 d th l i d i id l d Figure 3: The phase distributions and simulated results of the dual-band full-space digital coding metasurface. 2 Principe of operation According to the electric-field responses demonstrated in Figure 2d–k, three couples of working frequencies (f1, f2), (f3, f4) and (f5, f6) can be almost independently controlled by geometric parameters (R1, R2), (F1, F2), and (T1, T2), respectively. Hence, the metasurface can actually be designed to work at up to six completely different fre- quencies by adjusting these parameters. Based on this point, a metasurface that can work at six different fre- quencies has been further designed and demonstrated. Figure 4a–C) shows the amplitude and phase responses of unit cell to x-polarized transmission waves, y-polarized reﬂection waves along −z direction (front side) and y-polarized reﬂection waves along +z direction (back side), respectively, in which the partial geometric parameters of unit cell are ﬁxed as a = 1.75 mm, b = 2.2 mm, w1 = 1.5 mm, w2 = 1 mm, s1 = 1 mm, and s2 = 0.6 mm. The results show that dual-band 1-bit transmission metasurface can be constructed at 7.7 and 15 GHz for x-polarized wav changing the variables T1 and T2, as shown in Figure and dual-band 1-bit reﬂection metasurface can be structed at 8.4 and 15.9 GHz for y-polarized waves inci from the front side by changing the variables R1 and R shown in Figure 4b, and another dual-band 1-bit reﬂec metasurface can be constructed at 9 and 16.5 GHz y-polarized waves incident from the back side by chan the variables F1 and F2, as shown in Figure 4c. A six-band full-space metasurface composed of 25 unit cells has been designed to verify the prop concept, which can achieve independent holographic ages of Arabic numerals “1” to “6” at six different quencies. Figure 5a–f show the required 1-bit p distributions of metasurface for the holographic imag Arabic numeral “1”, “2”, “3”, “4”, “5”, and “6” at 7.7, 8 15, 15.9 and 16.5 GHz, respectively. The transmis Figure 4: The unit cell characteristics o six-band full-space digital coding metasurface. (a–c) The transmission amplitude and phase characteristics for (a) x-polarize incidences along −z direction, (b) y-polarized incidences along −z direct and (c) y-polarized incidences along z direction. Figure six-ba metas (a–c) phase incide y-pola and (c direct Figure 4: The unit cell characteristics of the six-band full-space digital coding metasurface. Figure 4: The unit cell characteristics of the six-band full-space digital coding metasurface. 2 Principe of operation (a–f) The 1-bit phase distribution of (a) the transmitarray at 8.4 GHz, (b) the transmitted imaging at 15.9 GHz, (c) the reﬂectarray at 8.4 GHz, (d) the reﬂected imaging at 15.9 GHz, (e) the reﬂected beam with l = +1 OAM at 8.4 GHz and (f) the reﬂected beam with l = −1 OAM at 15.9 GHz. (g) 2D far-ﬁeld radiation pattern and gain bandwidth of the transmitarray under the x-polarized incidence. (h) The holographic images of letter “X” on the plane of z = −188mm at 15.9 GHz. (i) 2D far-ﬁeld radiation pattern and gain bandwidth of the reﬂectarray under the y-polarized incidence. (j) The holographic images of letter “Y” on the plane of z = 188 mm at 15.9 GHz. (k–l) 2D far-ﬁeld radiation patterns and phase distributions of (k) l = +1 OAM beam and (l) l = −1 OAM beam on the plane of z = −600mm under the y-polarized incidence along z direction. Figure 3: The phase distributions and simulated results of the dual-band full-space digital coding metasurface. (a–f) The 1-bit phase distribution of (a) the transmitarray at 8.4 GHz, (b) the transmitted imaging at 15.9 GHz, (c) the reﬂectarray at 8.4 GHz, (d) the reﬂected imaging at 15.9 GHz, (e) the reﬂected beam with l = +1 OAM at 8.4 GHz and (f) the reﬂected beam with l = −1 OAM at 15.9 GHz. (g) 2D far-ﬁeld radiation pattern and gain bandwidth of the transmitarray under the x-polarized incidence. (h) The holographic images of letter “X” on the plane of z = −188mm at 15.9 GHz. (i) 2D far-ﬁeld radiation pattern and gain bandwidth of the reﬂectarray under the y-polarized incidence. (j) The holographic images of letter “Y” on the plane of z = 188 mm at 15.9 GHz. (k–l) 2D far-ﬁeld radiation patterns and phase distributions of (k) l = +1 OAM beam and (l) l = −1 OAM beam on the plane of z = −600mm under the y-polarized incidence along z direction. 2982 L.W. Wu et al.: Multitask bidirectional digital coding metasurface Figure 4: The unit cell characteristics of the six-band full-space digital coding metasurface. (a–c) The transmission amplitude and phase characteristics for (a) x-polarized incidences along −z direction, (b) y-polarized incidences along −z direction, and (c) y-polarized incidences along z direction. 2 Principe of operation (a–c) The transmission amplitude and phase characteristics for (a) x-polarized incidences along −z direction, (b) y-polarized incidences along −z direction, and (c) y-polarized incidences along z direction. According to the electric-field responses demonstrated in Figure 2d–k, three couples of working frequencies (f1, f2), (f3, f4) and (f5, f6) can be almost independently controlled by geometric parameters (R1, R2), (F1, F2), and (T1, T2), respectively. Hence, the metasurface can actually be designed to work at up to six completely different fre- quencies by adjusting these parameters. Based on this point, a metasurface that can work at six different fre- quencies has been further designed and demonstrated. Figure 4a–C) shows the amplitude and phase responses of unit cell to x-polarized transmission waves, y-polarized reﬂection waves along −z direction (front side) and y-polarized reﬂection waves along +z direction (back side), respectively, in which the partial geometric parameters of unit cell are ﬁxed as a = 1.75 mm, b = 2.2 mm, w1 = 1.5 mm, w2 = 1 mm, s1 = 1 mm, and s2 = 0.6 mm. The results show that dual-band 1-bit transmission metasurface can be constructed at 7.7 and 15 GHz for x-polarized wave by changing the variables T1 and T2, as shown in Figure 4a, and dual-band 1-bit reﬂection metasurface can be con- structed at 8.4 and 15.9 GHz for y-polarized waves incident from the front side by changing the variables R1 and R2, as shown in Figure 4b, and another dual-band 1-bit reﬂection metasurface can be constructed at 9 and 16.5 GHz for y-polarized waves incident from the back side by changing the variables F1 and F2, as shown in Figure 4c. A six-band full-space metasurface composed of 25 × 25 unit cells has been designed to verify the proposed concept, which can achieve independent holographic im- ages of Arabic numerals “1” to “6” at six different fre- quencies. Figure 5a–f show the required 1-bit phase distributions of metasurface for the holographic images of Arabic numeral “1”, “2”, “3”, “4”, “5”, and “6” at 7.7, 8.4, 9, 15, 15.9 and 16.5 GHz, respectively. The transmission L.W. Wu et al.: Multitask bidirectional digital coding metasurface 2983 Figure 5: The phase distributions and holographic imaging results of the six-band full-space digital coding metasurface. L.W. Wu et al.: Multitask bidirectional digital coding metasurface Figure 5: The phase distributions and holographic imaging results of the six-band full-space digital coding metasurface. 3 Experimental verification holographic images of Arabic numerals “1” and “4” at 7.7 and 15 GHz for x-polarized waves are illustrated in Figure 5g and j, respectively, the reﬂection holographic images of Arabic numerals “2” and “5” at 8.4 and 15.9 GHz for y-polarized waves in forward space are illustrated in Figure 5h and k, respectively, and the reﬂection holo- graphic images of Arabic numerals “3” and “6” at 8.4 and 15.9 GHz for y-polarized waves in backward space are illustrated in Figure 5i and l, respectively. All the simulated results show a good agreement with the calculations, verifying the ability of metasurface to independently work at six different frequencies. It is worth mentioning that all holographic images are designed in different focal planes, as marked in Figure 5g–l. holographic images of Arabic numerals “1” and “4” at 7.7 and 15 GHz for x-polarized waves are illustrated in Figure 5g and j, respectively, the reﬂection holographic images of Arabic numerals “2” and “5” at 8.4 and 15.9 GHz for y-polarized waves in forward space are illustrated in Figure 5h and k, respectively, and the reﬂection holo- graphic images of Arabic numerals “3” and “6” at 8.4 and 15.9 GHz for y-polarized waves in backward space are illustrated in Figure 5i and l, respectively. All the simulated results show a good agreement with the calculations, verifying the ability of metasurface to independently work at six different frequencies. It is worth mentioning that all holographic images are designed in different focal planes, as marked in Figure 5g–l. The above-mentioned two metasurfaces are fabricated and measured, respectively. Figure 6a and b illustrate the experimental setups of the far-ﬁeld and near-ﬁeld mea- surements, respectively. In the far-ﬁeld experiment, the metasurface sample is placed on a rotational platform together with a rectangular waveguide feed. The distance between the rectangular waveguide feed and metasurface is 300 mm for both transmitarray and reﬂectarray. A receiving horn antenna (Receiver) is placed 15 m away from the metasurface to receive the far-ﬁeld signal, as shown in Figure 6a. In the near-ﬁeld experiment, a feeding horn is placed 3 m away from the sample to generate normally incident plane waves and a rectangular wave probe is located at focal planes of holographic images to measure near-ﬁeld electric-ﬁeld distributions, as shown in Figure 6b. The absorbing materials are wrapped around the sample to reduce the inﬂuence of edge diffraction introduced by incident waves. 2 Principe of operation (a–f) The 1-bit phase distributions for (a) transmission imaging of numeral “1” at 7.7 GHz, (b) reﬂection imaging of numeral “2” at 8.4 GHz, (c) reﬂection imagingof numeral “3” at9GHz,(d) transmissionimagingofnumeral“4” at 15GHz, (e) reﬂectionimagingofnumeral“5” at 15.9GHz, (f) reﬂection imaging of numeral “6” at 16.5 GHz. (g–l) The calculated and simulated results of holographic images of (g) numeral “1” at 7.7 GHz, (h) numeral “2” at 8.4 GHz, (i) numeral “3” at 9 GHz, (j) numeral “4” at 15 GHz, (k) numeral “5” at 15.9 GHz, and (l) numeral “6” at 16.5 GHz. Figure 5: The phase distributions and holographic imaging results of the six-band full-space digital coding metasurface. (a–f) The 1-bit phase distributions for (a) transmission imaging of numeral “1” at 7.7 GHz, (b) reﬂection imaging of numeral “2” at 8.4 GHz, (c) reﬂection imagingof numeral “3” at9GHz,(d) transmissionimagingofnumeral“4” at 15GHz, (e) reﬂectionimagingofnumeral“5” at 15.9GHz, (f) reﬂection imaging of numeral “6” at 16.5 GHz. (g–l) The calculated and simulated results of holographic images of (g) numeral “1” at 7.7 GHz, (h) numeral “2” at 8.4 GHz, (i) numeral “3” at 9 GHz, (j) numeral “4” at 15 GHz, (k) numeral “5” at 15.9 GHz, and (l) numeral “6” at 16.5 GHz. 2984 L.W. Wu et al.: Multitask bidirectional digital coding metasurface 2984 3 Experimental verification Figure 6: The experiment setups of the far-field and near-field experiments. (a) The far-field experiment setup. (b) The near-field experiment setup. Figure 7 demonstrates the experimental results of dual-band full-space digital coding metasurface. For x-polarized incidence along −z direction, the measured far- ﬁeld radiation pattern of metasurface as a directive trans- mitarray at 8.4 GHz is illustrated in the left ﬁgure of Figure 7a, which shows that the radiation gain is 19.4 dB and side lobe is −12.9 dB. The 3-dB gain bandwidth can reach 2 GHz, as shown in the right ﬁgure of Figure 7a. The measured near-ﬁeld distribution of transmission waves at 15.9 GHz is illustrated in Figure 7b. An expected holo- graphic image of letter “X” is obtained at the focal plane of z = −188mm. For the y-polarized incidence along the −z direction, the measured far-ﬁeld radiation pattern of met- asurface as a directive reﬂectarray at 8.4 GHz is shown in the left ﬁgure of Figure 7c, showing that the radiation beam is deﬂected to θ = −17° with radiation gain of 22.1 dB and side lobe of −13.4 dB. The 3-dB gain bandwidth can reach 2.8 GHz as shown in the right ﬁgure of Figure 7c. The measured near-ﬁeld distribution of reﬂection waves at 15.9 GHz is illustrated in Figure 7d. An expected holo- graphic image of letter “Y” is obtained at the focal plane of z = 188 mm. For y-polarized incidence along +z direction, the OAM beams with topological charges of l = +1 and l = −1 are generated at 8.4 and 15.9 GHz, whose measured amplitude and phase distributions on the plane of z = −600 mm are shown in Figure 7e and f, respectively. All the far-ﬁeld and near-ﬁeld measured results are in good agreement with the calculated and simulated results shown in Figure 3. Figure 6: The experiment setups of the far-field and near-field experiments. (a) The far-field experiment setup. (b) The near-field experiment setup. Figure 6: The experiment setups of the far-field and near-field experiments. (a) The far-field experiment setup. (b) The near-field experiment setup. L.W. Wu et al.: Multitask bidirectional digital coding metasurface 2 Figure 7: The experimental results of the dual-band full space digital coding metasurface. (a) 2D far-field and gain bandwidth of the 1 bit transmitarray under the x-polarized incidence. 3 Experimental verification (b) The holographic image of letter “X” on the plane of z = −188mm at 15.9 GHz. (c) 2D far-ﬁeld and gain bandwidth of the 1 bit reﬂectarray under the y-polarized incidence. (d) The holographic image of letter “Y” on the plane of z = 188 mm at 15.9 GHz. (e, f) The amplitude and phase distributions of (e) l = +1 OAM beam at 8.4 GHz, and (f) l = −1 OAM beam at 15.9 GHz. Figure 7: The experimental results of the dual-band full space digital coding metasurface. (a) 2D far-field and gain bandwidth of the 1 bit transmitarray under the x-polarized incidence. (b) The holographic image of letter “X” on the plane of z = −188mm at 15.9 GHz. (c) 2D far-ﬁeld and gain bandwidth of the 1 bit reﬂectarray under the y-polarized incidence. (d) The holographic image of letter “Y” on the plane of z = 188 mm at 15.9 GHz. (e, f) The amplitude and phase distributions of (e) l = +1 OAM beam at 8.4 GHz, and (f) l = −1 OAM beam at 15.9 GHz. Figure 8: The experimental results of the six-band full space digital coding metasurface. (a) Numeral “1” at 7.7 GHz. (b) Numeral “2” at 8.4 GHz. (c) Numeral “3” at 9 GHz. (d) Numeral “4” at 15 GHz. (e) Numeral “5” at 15.9 GHz. (f) Numeral “6” at 16.5 GHz. Figure 8: The experimental results of the six-band full space digital coding metasurface. (a) Numeral “1” at 7.7 GHz. (b) Numeral “2” at 8.4 GHz. (c) Numeral “3” at 9 GHz. (d) Numeral “4” at 15 GHz. (e) Numeral “5” at 15.9 GHz. (f) Numeral “6” at 16.5 GHz. 2986 L.W. Wu et al.: Multitask bidirectional digital coding metasurface 2986 Figure 8 illustrates the experimental results of the six- band full-space metasurface. For the x-polarized inci- dence, the measured transmission Arabic numerals “1” and “4” on focal planes of z = −156mm and z = −200mm at 7.7 and 15 GHz are illustrated in Figure 8a and d, respec- tively. For y-polarized incidence along −z direction, the measured transmission Arabic numerals “2” and “5” on focal planes of z = 142 mm and z = 198 mm at 8.4 and 15.9 GHz are illustrated in Figure 8b and e, respectively. References [1] N. Yu, P. Genevet, M. A. Kats, et al., “Light propagation with phase discontinuities: generalized laws of reﬂection and refraction,” Science, vol. 334, no. 6054, pp. 333–337, 2011. [2] S. B. Glybovski, S. A. Tretyakov, P. A. Belov, et al., “Metasurfaces: from microwaves to visible,” Phys. Rep., vol. 634, pp. 1–72, 2016. [3] C. L. Holloway, E. F. Kuester, J. A. Gordon, et al., “An overview of the theory and applications of metasurfaces: the two- dimensional equivalents of metamaterials,” IEEE Antenn. Propag. Mag., vol. 54, no. 2, pp. 10–35, 2012. Propag. Mag., vol. 54, no. 2, pp. 10–35, 2012. [4] G. Zheng, H. Mühlenbernd, M. Kenney, et al., “Metasurface holograms reaching 80% efﬁciency,” Nat. Nanotechnol., vol. 10, no. 4, pp. 308–312, 2015. [5] K. Chen, G. W. Ding, G. W. Hu, et al., “Directional Janus metasurface,” Adv. Mater., vol. 32, no. 2, p. 1906352, 2020. 3 Experimental verification For y-polarized incidence along +z direction, the measured reﬂection Arabic numerals “3” and “6” on focal planes of z = −150 mm and z = −191 mm at 9 and 16.5 GHz are illus- trated in Figure 8c and f, respectively. All the measured results are in good agreement with the calculated and simulated results shown in Figure 5, except for some slight differences, which may be caused by the accuracy of sample processing, assembly accuracy of multilayer plates and accuracy of quasi plane wave generated by feeding horn in experiment. Peaks Project in Jiangsu Province (XCL-077), the Project for Jiangsu Specially-Appointed Professor, the China National Postdoctoral Program for Innovative Talents (BX20200080). Conﬂict of interest statement: The authors declare no conﬂicts of interest regarding this article. 4 Conclusions [6] L. W. Wu, H. F. Ma, Y. Gou, et al., “High-transmission ultrathin Huygens’ metasurface with 360 phase control by using double- layer transmitarray elements,” Phys. Rev. Appl., vol. 12, no. 2, p. 024012, 2019. We have presented a multitask bidirectional digital coding metasurface that can achieve the independent multi-band controls of both reflection and transmission waves whether the EM waves are incident from the front or back of the metasurface. Further, the metasurface can independently work in up to six frequency bands in full space at the same time with completely different functions. As proofs of concept, a dual-band full-space metasurface and a six- band full-space metasurface are both simulated and fabricated with good performances. In addition, the coding of metasurface can also be easily extended to 2 bit by increasing the layers of unit structure. The proposed mul- titask bidirectional digital coding metasurface has power- ful ability in manipulation of both reﬂection and transmission EM waves, which is incapable for previously reported metasurfaces and may provide a simple way to extend the functionality and information capacity of high- efﬁciency metadevices. [7] L. X. Liu, X. Q. Zhang, M. Kenney, et al., “Broadband metasurfaces with simultaneous control of phase and amplitude,” Adv. Mater., vol. 26, no. 29, pp. 5031–5036, 2014. [8] Y. Yao, R. Shankar, M. A. Kats, et al., “Electrically tunable metasurface perfect absorbers for ultrathin mid-infrared optical modulators,” Nano Lett., vol. 14, no. 11, pp. 6526–6532, 2014. [9] K. Fan, J. Y. Suen, X. Liu, et al., “All-dielectric metasurface absorbers for uncooled terahertz imaging,” Optica, vol. 4, no. 6, pp. 601–604, 2017. [10] N. K. Grady, J. E. Heyes, D. R. Chowdhury, et al., “Terahertz metamaterials for linear polarization conversion and anomalous refraction,” Science, vol. 340, no. 6138, pp. 1304–1307, 2013. [11] S. C. Jiang, X. Xiong, Y. S. Hu, et al., “Controlling the polarization state of light with a dispersion-free metastructure,” Phys. Rev. X, vol. 4, no. 2, p. 021026, 2014. [12] S. L. Sun, Q. He, S. Xiao, et al., “Gradient-index meta-surfaces as a bridge linking propagating waves and surface waves,” Nat. Mater., vol. 11, no. 5, pp. 426–431, 2012. [13] S. Sun, K.-Y. Yang, C.-M. Wang, et al., “High-efﬁciency broadband anomalous reﬂection by gradient meta-surfaces,” Nano Lett., vol. 12, no. 12, pp. 6223–6229, 2012. Author contribution: All the authors have accepted responsibility for the entire content of this submitted manuscript and approved submission. 4 Conclusions Yi, S. W. Qu, K. B. Ng, et al., “Terahertz wavefront control on both sides of the cascaded metasurfaces,” IEEE Trans. Antenn. Propag., vol. 66, no. 1, pp. 209–216, 2017. [34] X. Wan, Q. Zhang, T. Y. Chen, et al., “Multichannel direct transmissions of near-ﬁeld information,” Light Sci. Appl., vol. 8, p. 60, 2019. [21] X. H. Zhang, M. B. Pu, Y. H. Guo, et al., “Colorful metahologram with independently controlled images in transmission and reﬂection spaces,” Adv. Funct. Mater., vol. 29, no. 22, p. 1809145, 2019. [35] L. Zhang, R. Y. Wu, G. D. Bai, et al., “Transmission-reﬂection- integrated multifunctional coding metasurface for full-space controls of electromagnetic waves,” Adv. Funct. Mater., vol. 28, no. 33, p. 1802205, 2018. [22] Y. Chen, J. Gao, and X. Yang, “Direction-controlled bifunctional metasurface polarizers,” Laser Photon. Rev., vol. 12, no. 12, p. 1800198, 2018. [36] L. Bao, X. J. Fu, R. Y. Wu, et al., “Full-space manipulations of electromagnetic wavefronts at two frequencies by encoding both amplitude and phase of metasurface,” Adv. Mater. Technol., vol. 6, no. 4, p. 2001032, 2021. [23] H. X. Xu, C. Wang, G. Hu, et al., “Spin-encoded wavelength- direction multitasking Janus metasurfaces,” Adv. Opt. Mater., vol. 9, no. 11, p. 2100190, 2021. [24] T. J. Cui, M. Q. Qi, X. Wan, et al., “Coding metamaterials, digital metamaterials and programmable metamaterials,” Light Sci. Appl., vol. 3, no. 10, p. e218, 2014. [37] H. L. Wang, H. F. Ma, M. Chen, et al., “A reconﬁgurable multifunctional metasurface for full-space control of electromagnetic waves,” Adv. Funct. Mater., vol. 31, no. 25, p. 2100275, 2021. [25] T. J. Cui, “Microwave metamaterials,” Natl. Sci. Rev., vol. 5, no. 2, pp. 134–136, 2018. [38] L. Bao, Q. Ma, R. Y. Wu, et al., “Programmable reﬂection- transmission shared-aperture metasurface for real-time control of electromagnetic waves in full space,” Adv. Sci., vol. 8, no. 15, p. 2100149, 2021. [26] T. J. Cui, S. Liu, and L. L. Li, “Information entropy of coding metasurface,” Light Sci. Appl., vol. 5, no. 11, p. e16172, 2016. [27] L. H. Gao, Q. Cheng, J. Yang, et al., “Broadband diffusion of terahertz waves by multi-bit coding metasurfaces,” Light Sci. Appl., vol. 4, no. 9, p. e324, 2015. [39] R. Y. Wu, L. Zhang, L. Bao, et al., “Digital metasurface with phase code and reﬂection-transmission amplitude code for ﬂexible full- space electromagnetic manipulations,” Adv. Opt. Mater., vol. 7, no. 8, p. 4 Conclusions Research funding: This work was supported by the National Key Research and Development Program of China (2017YFA0700200, 2017YFA0700201, and 2017YFA0700202), the National Natural Science Foundation of China (62071117 and 61831006), and the 111 Project (111-2-05), the Six Talent Author contribution: All the authors have accepted responsibility for the entire content of this submitted manuscript and approved submission. [14] C. Pfeiffer and A. Grbic, “Metamaterial Huygens’ surfaces: tailoring wave fronts with reﬂectionless sheets,” Phys. Rev. Lett., vol. 110, no. 19, p. 197401, 2013. [15] A. Pors, O. Albrektsen, I. Radko, et al., “Gap plasmon-based metasurfaces for total control of reﬂected light,” Sci. Rep., vol. 3, p. 2155, 2013. Research funding: This work was supported by the National Key Research and Development Program of China (2017YFA0700200, 2017YFA0700201, and 2017YFA0700202), the National Natural Science Foundation of China (62071117 and 61831006), and the 111 Project (111-2-05), the Six Talent [16] W. T. Chen, K. Y. Yang, C. M. Wang, et al., “High-efﬁciency broadband meta-hologram with polarization-controlled dual images,” Nano Lett., vol. 14, no. 1, pp. 225–230, 2014. 2987 L.W. Wu et al.: Multitask bidirectional digital coding metasurface by a reconﬁgurable anisotropic digital coding metasurface,” Adv. Opt. Mater., vol. 8, no. 22, p. 2001065, 2020. [17] J. P. B. Mueller, N. A. Rubin, R. C. Devlin, et al., “Metasurface polarization optics: independent phase control of arbitrary orthogonal states of polarization,” Phys. Rev. Lett., vol. 118, no. 11, p. 113901, 2017. [30] L. Zhang, X. Q. Chen, S. Liu, et al., “Space-time-coding digital metasurfaces,” Nat. Commun., vol. 9, p. 4334, 2018. [31] J. Y. Dai, J. Zhao, Q. Cheng, et al., “Independent control of harmonic amplitudes and phases via a time-domain digital coding metasurface,” Light Sci. Appl., vol. 7, p. 90, 2018. [18] S. Li, Z. Wang, S. Dong, et al., “Helicity-delinked manipulations on surface waves and propagating waves by metasurfaces,” Nanophotonics, vol. 9, no. 10, pp. 3473–3481, 2020. [32] L. Zhang, X. Q. Chen, R. W. Shao, et al., “Breaking reciprocity with space-time-coding digital metasurfaces,” Adv. Mater., vol. 31, no. 41, p. 1904069, 2019. [19] T. Cai, G. M. Wang, S. W. Tang, et al., “High-efﬁciency and full- space manipulation of electromagnetic wave fronts with metasurfaces,” Phys. Rev. Appl., vol. 8, no. 3, p. 034033, 2017. [33] T. J. Cui, S. Liu, G. D. Bai, et al., “Direct transmission of digital message via programmable coding metasurface,” Research, vol. 2019, p. 2584509, 2019. [20] H. 4 Conclusions 1801429, 2019.p. [28] L. L. Li, T. J. Cui, W. Ji, et al., “Electromagnetic reprogrammable coding-metasurface holograms,” Nat. Commun., vol. 8, p. 197, 2017. [40] Z. C. Wang, X. M. Ding, K. Zhang, et al., “Huygens metasurface holograms with the modulation of focal energy distribution,” Adv. Opt. Mater., vol. 6, no. 12, p. 1800121, 2018. [29] L. W. Wu, H. F. Ma, R. Y. Wu, et al., “Transmission-reﬂection controls and polarization controls of electromagnetic holograms
https://openalex.org/W2474597586	https://flore.unifi.it/bitstream/2158/1294170/1/ijerph-13-00631.pdf	English	null	Respiratory Health in Waste Collection and Disposal Workers	International journal of environmental research and public health/International journal of environmental research and public health	2,016	cc-by	5,395	Respiratory Health in Waste Collection and Disposal Workers Luigi Vimercati 1,, Antonio Baldassarre 1,,†, Maria Franca Gatti 1,†, Luigi De Maria 1, Antonio Caputi 1, Angelica A. Dirodi 1, Francesco Cuccaro 2 and Raffaello Maria Bellino 3 Luigi Vimercati 1,, Antonio Baldassarre 1,,†, Maria Franca Gatti 1,†, Luigi De Maria 1, Antonio Caputi 1, Angelica A. Dirodi 1, Francesco Cuccaro 2 and Raffaello Maria Bellino 3 1 Interdisciplinary Department of Medicine, Occupational Medicine “B. Ramazzini”, University of Bari Medical School, Bari 70124, Italy; mariafranca.gatti@gmail.com (M.F.G.); luidemale@gmail.com (L.D.M anto.caputi@gmail.com (A.C.); angelica.dirodi@gmail.com (A.A.D.) 1 Interdisciplinary Department of Medicine, Occupational Medicine “B. Ramazzini”, University of Bari Medical School, Bari 70124, Italy; mariafranca.gatti@gmail.com (M.F.G.); luidemale@gmail.com (L.D.M.); anto.caputi@gmail.com (A.C.); angelica.dirodi@gmail.com (A.A.D.) 2 Health Local Unit of Barletta-Andria-Trani, Barletta 76121, Italy; francesco_cuccaro@hotmail.com 2 Health Local Unit of Barletta-Andria-Trani, Barletta 76121, Italy; francesco_cuccaro@ 3 H l h L l U i f B i B i 70122 I l ff ll b lli @ il * Correspondence: luigi.vimercati@uniba.it (L.V.); antonio.baldassarre@uniba.it (A.B.); Tel.: +39-080-5478256 (L.V.); +39-080-5478256 (A.B.) † These authors contributed equally to this work. Academic Editor: Paul B. Tchounwou Received: 6 April 2016; Accepted: 21 June 2016; Published: 24 June 2016 ved: 6 April 2016; Accepted: 21 June 2016; Published: Abstract: Waste management, namely, collection, transport, sorting and processing, and disposal, is an issue of social concern owing to its environmental impact and effects on public health. In fact, waste management activities are carried out according to procedures that can have various negative effects on the environment and, potentially, on human health. The aim of our study was to assess the potential effects on respiratory health of this exposure in workers in the waste management and disposal ﬁeld, as compared with a group of workers with no occupational exposure to outdoor pollutants. The sample consisted of a total of 124 subjects, 63 waste collectors, and 61 ofﬁce clerks. Informed consent was obtained from all subjects before inclusion in the study. The entire study population underwent pulmonary function assessments with spirometry and completed two validated questionnaires for the diagnosis of rhinitis and chronic bronchitis. Statistical analyses were performed using STATA 13. Spirometry showed a statistically signiﬁcant reduction in the mean Tiffenau Index values in the exposed workers, as compared with the controls, after adjusting for the confounding factors of age, BMI, and smoking habit. Similarly, the mean FEV1 values were lower in the exposed workers than in the controls, this difference being again statistically signiﬁcant. Respiratory Health in Waste Collection and Disposal Workers The FVC differences measured in the two groups were not found to be statistically signiﬁcant. We ran a cross-sectional study to investigate the respiratory health of a group of workers in the solid waste collection and disposal ﬁeld as compared with a group of ofﬁce workers. In agreement with most of the data in the literature, our ﬁndings support the existence of a prevalence of respiratory deﬁcits in waste disposal workers. Our data suggest the importance of adopting preventive measures, such as wearing speciﬁc individual protection devices, to protect this particular category of workers from adverse effects on respiratory health. Keywords: occupational exposure; bioaerosol; endotoxins; occupational respiratory disease; waste workers International Journal of Environmental Research and Public Health 1. Introduction Waste management, namely, collection, transport, sorting and processing, and disposal, is an issue of social concern owing to its environmental impact and effects on public health. In fact, waste management activities are carried out according to procedures that can have various negative effects on the environment and, potentially, on human health. They pose risks resulting from the emissions or Int. J. Environ. Res. Public Health 2016, 13, 631; doi:10.3390/ijerph13070631 www.mdpi.com/journal/ijerph www.mdpi.com/journal/ijerph 2 of 8 Int. J. Environ. Res. Public Health 2016, 13, 631 release of hazardous chemical agents and biological agents, from the types of exposure to these agents, and from the susceptibility of the populations exposed to them. release of hazardous chemical agents and biological agents, from the types of exposure to these agents, and from the susceptibility of the populations exposed to them. Epidemiological studies of the short- and long-term effects of exposure to waste on public health have been focused above all on identifying any associations between residence in the vicinity of a landﬁll and adverse effects on health [1–3]. Meanwhile, less attention has been paid to the risk posed in subjects who are directly involved in waste management, especially as regards the effects on the respiratory tract. Workers in the waste disposal ﬁeld, in particular, can be exposed to bioaerosols, gases, and vapours. Waste sorting, collection, and recycling, procedures that have now been adopted by nearly all EU member states, have actually introduced new risk proﬁles for garbage collectors, largely attributable to the inhalation of bioaerosols generated by the decomposition of organic waste [4,5]. This process can lead to the formation of various biological agents including bacteria, fungi, and volatile compounds such as endotoxins, β(1-3)-glucans, and mycotoxins that can provoke inﬂammation of the airways [6–8]. In particular, endotoxins, the components of the cell wall of Gram-negative bacteria, are considered to be the most powerful proinﬂammatory component present in bioaerosols [9]. Exposure to low concentrations of endotoxins seems to induce an inﬂammatory response of the upper airways, through neutrophil activation and the release of cytokines such as IL6 and IL8 and TNF-alpha, the main factors triggering inﬂammatory processes [10,11]. The levels of these proinﬂammatory mediators were found to be increased in patients with a diagnosis of chronic obstructive respiratory diseases [12]. β(1-3)-glucans are an integral part of the wall of fungal spores, but are also present in some bacteria. 1. Introduction Experimental studies in animals have shown that the inﬂammatory response is characterised by an increased number of eosinophils [13] and that repeated exposure to β(1-3)-glucans exacerbates the inﬂammatory response to endotoxins via a synergic action [14]. Exposure to bioaerosols, therefore, seems to induce an inﬂammatory response at the level of the airways mucosa, probably due to a complex interaction among microorganisms or cell wall components of these microorganisms and the host immune system [5]. When assessing the respiratory health risk in solid waste disposal workers, it is also important to take into account the physical effort and muscle work that these activities entail, during lifting and manhandling of the various containers. This induces a ventilation response consisting of an increased ﬂow volume and respiratory frequency [15,16] that will increase the amount of bioaerosol inhaled, including not only airborne organic compounds, but also dust particles and vehicle exhaust fumes and gases. These factors have been reported to be responsible for a higher incidence of respiratory diseases in this worker category [6,7,17–19], namely, inﬂuenza-like disorders (rhinitis, conjunctivitis, cough, headache) [20], as well as more severe obstructive disorders such as chronic obstructive bronchopneumonia [21] and allergic bronchopulmonary aspergillosis [22,23], and in some cases restrictive lung diseases, which are especially related to the exposure to elements or chemicals such as Al, Si, carbon black, TiO2, silicon oxide, talcum powder, asbestos, and other ﬁbres [24]. The aim of our study was to assess the potential effects on respiratory health of this exposure in workers in the waste management and disposal ﬁeld, as compared with a group of workers with no occupational exposure to outdoor pollutants. 2. Materials and Methods The research was carried out on a working population of 375 employees of two waste collection and disposal companies, 300 waste collectors (exposed), and 75 clerks (controls). Both exposed and control subjects were randomly selected to reduce the possibility of bias in self-selection. Upon inclusion in the study, a completed questionnaire as regards medical and working history, as well as various confounding factors, was collected from all workers in the presence of a physician. The response rate was higher for exposed (94%) than for controls (91%). All recruited subjects, both exposed and non-exposed, had a negative clinical history for previous respiratory diseases (inclusion criteria); only for non-exposed subjects did we adopt as exclusion criteria previous or current 3 of 8 us or of 124 Int. J. Environ. Res. Public Health 2016, 13, 631 (inclusion criteria); only for non-exp current occupational exposure to ou occupational exposure to outdoor pollutants. The ﬁnal sample consisted of a total of 124 subjects, 63 exposed and 61 controls (Figure 1). The investigation was carried out over a period from July to November 2013, and the two groups were studied at the same time to eliminate any confounding factor introduced by seasonal variations. The waste disposal activities of the exposed workers were carried out 5 days a week for 6 h/day and consisted of emptying different-sized containers (wheely-bins, dumpsters) manually or mechanically with waste compactors, sometimes with ground assistance from the mechanised refuse collection service. Informed consent was obtained from all subjects before inclusion in the study. All subjects agreed to the processing of their personal data, treated as sensitive data. All subjects were informed that data from the research protocol would be processed anonymously and collectively, applying proper scientiﬁc methods and for scientiﬁc purposes, in accordance with the principles of the Helsinki Declaration. Our study is in accordance with good clinical practice guidelines. Both spirometric and questionnaire were performed in agreement with the health surveillance protocol and health promotion’s campaigns (Italian Legislative Decree N˝ 81 of 2008, in transposition of several European Directives). from July to November 2013, and the two groups were studied at the same time to eliminate any confounding factor introduced by seasonal variations. 2. Materials and Methods The waste disposal activities of the exposed workers were carried out 5 days a week for 6 h/day and consisted of emptying different-sized containers (wheely-bins, dumpsters) manually or mechanically with waste compactors, sometimes with ground assistance from the mechanised refuse collection service. Informed consent was obtained from all subjects before inclusion in the study. All subjects agreed to the processing of their personal data, treated as sensitive data. All subjects were informed that data from the research protocol would be processed anonymously and collectively, applying proper scientific methods and for scientific purposes, in accordance with the principles of the Helsinki Declaration. Our study is in accordance with good clinical practice guidelines. Both spirometric and questionnaire were performed in agreement with the health surveillance protocol and health promotion’s campaigns (Italian Legislative Decree N° 81 of 2008, in transposition of several European Directives). Figure 1. Participation and recruitment ﬂow chart. Figure 1. Participation and recruitment ﬂow chart. Figure 1. Participation and recruitment ﬂow chart. 3 Figure 1. Participation and recruitment flow chart. All the recruited subjects then completed two questionnaires validated for the diagnosis of rhinitis [25] and chronic bronchitis [26]. The first one probed for the presence, frequency, and severity of rhinitis symptoms like rhinorrhoea, sneezing, nasal congestion, and pruritus, while the All the recruited subjects then completed two questionnaires validated for the diagnosis of rhinitis [25] and chronic bronchitis [26]. The ﬁrst one probed for the presence, frequency, and severity of rhinitis symptoms like rhinorrhoea, sneezing, nasal congestion, and pruritus, while the second was aimed at diagnosing chronic bronchitis, verifying whether the subject suffers from coughing, catarrh, dyspnoea, or other symptoms suggesting bronchial asthma, such as wheezing, coughing up mucous, and a feeling of chest oppression. Rhinitis symptoms deﬁnition was based on the International Chronic 4 of 8 Int. J. Environ. Res. Public Health 2016, 13, 631 Rhinitis (ICR) working group recommendation: the occurrence of two or more symptoms (nasal obstruction, rhinorrhea, sneezing, or itchy nose) on most days during the past year. Chronic bronchitis symptoms deﬁnition is based on the occurrence of cough that lasts for at least 3 months, 2 years in a row, associated with catarrh, dyspnea, wheezing, and a feeling of chest oppression. The entire study population then underwent pulmonary function assessments. 2. Materials and Methods Spirometry was performed with the Pony FX spirometer (version 1.7, COSMED srl, Albano Laziale, Rome, Italy) to assess the respiratory function and to identify any obstructive or restrictive deﬁcits. All respiratory function tests were performed by the same operator, with the subject in a sitting position with his nose closed by a clip, following the American Thoracic Society (ATS) guidelines. The pulmonary function test was performed three times for each subject, and we chose the best one. The most important aspects of spirometry are the forced vital capacity (FVC), which is the volume delivered during an expiration made as forcefully and completely as possible starting from full inspiration, and the forced expiratory volume (FEV1) in one second, which is the volume delivered in the ﬁrst second of an FVC manoeuvre. Spirometric airﬂow limitations were deﬁned according to the Tiffenau Index as (FEV1/FVC) < 70% or an FEV1 < 80% of predicted values. A restrictive ventilatory pattern is characterised by a proportional reduction in FEV1 and FVC < 80% with a normal FEV1/FVC [27]. Statistical analyses were performed via analysis of variance (ANOVA) to seek statistically signiﬁcant differences between the exposed group (waste disposal workers) and the non-exposed group (ofﬁce workers) for the continuous variables, after adjusting the models for confounding factors (age, BMI, and smoking habit). Other confounding factors, such as socioeconomic status (SES) and educational level were not considered relevant, because all participants belonged to classes IIIM (skilled manual—exposed) or IIINM (skilled non-manual—controls) and had completed secondary education. Only three participants, among the controls, had a university degree. We did not study the confounding interaction of cardiovascular diseases. Logistic regression models were applied to assess dichotomous spirometric and symptoms variables (rhinitis and bronchitis) related to exposure, again after adjusting for the same confounding factors. Statistical analyses were performed using STATA version 13 Software (STATA Corporation, College Station, TX, USA). Table 1. Characteristics of participants. Characteristics Exposed Controls p Values N 63 61 Sex M 63 61 F 0 0 Age Average 53 51 0.38 SD 7 7 Median 54 52 Range 32–66 36–63 BMI Average 26.40 26.00 0.71 SD 11.06 4.36 Median 26.37 25.25 Range 16.90–34.84 18.36–40.14 Smoking habit Yes 17 (27.00%) 13 (21.30%) 0.19 No 31 (49.20%) 38 (62.30%) Former 15 (23.80%) 10 (16.40%) SD = Standard deviation; Signiﬁcance p < 0.05. Table 2. Spirometric test results. Spirometric Values Exposed Controls p Values N 63 61 FVC Average 4.76 4.56 0.114 SD 0.99 1.09 Median 4.67 4.49 % mean 115.06 114.2 % SD 22.42 18.06 % median 113 112 FEV1 Average 3.53 3.64 0.046 SD 0.73 0.85 Median 3.42 3.50 % mean 105.89 111.03 % SD 19.05 17.53 % median 107 110 TIFFENAU INDEX Average 75.08 79.93 0.001 SD 9.06 3.89 Median 76.09 80.12 Range 32.62–93.43 67.57–87.31 SD = Standard deviation; Signiﬁcance p < 0.05. SD = Standard deviation; Signiﬁcance p < 0.05. Table 3. Association between occupational exposure to waste and spirometric alterations using logistic regression. OR 95% CI p Value 8.4 1.8–38.9 0.006 7.9 * 1.7–37.0 0.008 * Adjusted for age, BMI, and smoking habit. * Adjusted for age, BMI, and smoking habit. Table 4. Association between occupational exposure to waste and symptoms in questionnaire. Symptoms Exposed (N = 63) Controls (N = 61) OR 95% CI p Value N % N % Rhinitis symptoms 11 17.4 19 31.1 0.41 0.17–1.00 0.051 Bronchitis symptoms 12 19.0 11 18.0 1.09 0.43–2.76 0.847 3. Results All 124 study participants were male Caucasians (mean age 53 years for exposed workers and 51 for controls) and worked in the Apulia region in Southern Italy. As regards smoking habits, a confounding factor, 50.8% of exposed subjects were smokers or formerly smokers, while only 37.7% of controls had a history of smoking. Characteristics of both groups are shown in Table 1. The pulmonary function test was performed three times for each subject, and we chose the best one. Spirometry showed, as reported in Table 2, a statistically signiﬁcant reduction in the mean Tiffenau Index values in the exposed workers (75.08%) compared with the controls (79.93%) after adjusting for the confounding factors of age, BMI, and smoking habit. Similarly, the mean FEV1 values were lower in the exposed workers (3.53 L) than in the controls (3.64 L), this difference again being statistically signiﬁcant. The FVC differences measured in the two groups were not found to be statistically signiﬁcant (controls, 4.56; exposed, 4.76). The exposed workers also showed more spirometric alterations, which were statistically signiﬁcant, compared with the non-exposed group (OR = 8.42; 95% CI: 1.8–38.9), again after adjusting for age, BMI, and smoking habit (OR = 7.9; 95% CI: 1.7–37.0), as reported in Table 3. When analysing the questionnaires on respiratory symptoms, rhinitis symptoms were found to be prevalent in the non-exposed group (31.1%) as compared with the exposed subjects (17.4%) (OR = 0.41; 95% CI: 0.17–1.00), whereas lower airway obstruction symptoms, investigated in the questionnaire for diagnosing bronchitis, were slightly more prevalent in the exposed group (19.0%) than in the controls (18.0%), as shown in Table 4. In both cases, the results were not statistically signiﬁcant, even after adjusting for the confounding factors. 5 of 8 Int. J. Environ. Res. Public Health 2016, 13, 631 Table 1. Characteristics of participants. Characteristics Exposed Controls p Values N 63 61 Sex M 63 61 F 0 0 Age Average 53 51 0.38 SD 7 7 Median 54 52 Range 32–66 36–63 BMI Average 26.40 26.00 0.71 SD 11.06 4.36 Median 26.37 25.25 Range 16.90–34.84 18.36–40.14 Smoking habit Yes 17 (27.00%) 13 (21.30%) 0.19 No 31 (49.20%) 38 (62.30%) Former 15 (23.80%) 10 (16.40%) SD = Standard deviation; Signiﬁcance p < 0.05. Table 2. Spirometric test results. 3. Results Spirometric Values Exposed Controls p Values N 63 61 FVC Average 4.76 4.56 0.114 SD 0.99 1.09 Median 4.67 4.49 % mean 115.06 114.2 % SD 22.42 18.06 % median 113 112 FEV1 Average 3.53 3.64 0.046 SD 0.73 0.85 Median 3.42 3.50 % mean 105.89 111.03 % SD 19.05 17.53 % median 107 110 TIFFENAU INDEX Average 75.08 79.93 0.001 SD 9.06 3.89 Median 76.09 80.12 Range 32.62–93.43 67.57–87.31 SD = Standard deviation; Signiﬁcance p < 0.05. Table 3. Association between occupational exposure to waste and spirometric alterations u logistic regression. OR 95% CI p Value 8 4 1 8 38 9 0 006 Table 1. Characteristics of participants. 4. Discussion We ran a cross-sectional study to investigate the respiratory health of a group of workers in the solid waste collection and disposal ﬁeld as compared with a group of ofﬁce workers. The spirometric tests performed showed a signiﬁcant reduction in the Tiffenau Index and FEV 1 values in the exposed Int. J. Environ. Res. Public Health 2016, 13, 631 6 of 8 workers, after adjusting for the confounding factors. Meanwhile, no statistically signiﬁcant differences were found among the FVC values. These results suggest the prevalence of an obstructive pulmonary pattern and seem to be in agreement with the pathogenic hypothesis whereby cytokines, such as IL6, IL8, and TNF-alpha, which are produced as a result of contact with bacterial endotoxins of bioaerosols, may be the main mediators of the inﬂammatory response involved in the development of chronic obstructive lung diseases. This ﬁnding is in agreement with the results obtained by Athanasiou et al., who demonstrated a statistically signiﬁcant reduction in FVC and FEV1 in a group of 104 municipal solid waste disposal workers in the city of Keratsini (Grecia) compared with a control group of 80 ofﬁce workers [28]. Comparable results were obtained in a study conducted in India in 96 solid waste disposal workers at an open landﬁll. A signiﬁcant reduction in FEV1 and the Tiffenau Index values was found in the exposed workers compared with the controls [15]. In our study, the results of the allergological questionnaires did not reveal statistically signiﬁcant differences between the two study populations, even if we found a higher prevalence of rhinitis symptoms in the controls, related to possible indoor air pollution. This evidence is not in agreement with Athanasiou et al. who showed that respiratory symptoms, investigated by means of a validated questionnaire, were also signiﬁcantly more common in the exposed group than the control group. Moreover, in the exposed group studied by Ray et al. [15], there was a statistically signiﬁcant prevalence of respiratory symptoms. We did not ﬁnd differences in the allergological questionnaires because our control group is represented by ofﬁce clerks. For these subjects, we can assume a possible exposure to indoor pollutants, such as those emitted from ofﬁce equipment (volatile organic chemicals (VOCs), ozone, and particulate matter), biological contaminants (bacteria, virus, fungi including moulds), and other chemicals (ofﬁce furniture, cleaning, and consumer products). References 1. Dolk, H.; Vrijheid, M.; Armstrong, B.; Abramsky, L.; Bianchi, F.; Garne, E.; Nelen, V.; Robert, E.; Scott, J.E.; Stone, D.; et al. Risk of congenital anomalies near hazardous-waste landﬁll sites in Europe: The eurohazcon study. Lancet 1998, 352, 423–427. [CrossRef] 1. Dolk, H.; Vrijheid, M.; Armstrong, B.; Abramsky, L.; Bianchi, F.; Garne, E.; Nelen, V.; Robert, E.; Scott, J.E.; Stone, D.; et al. Risk of congenital anomalies near hazardous-waste landﬁll sites in Europe: The eurohazcon study. Lancet 1998, 352, 423–427. [CrossRef] 2. Elliott, P.; Briggs, D.; Morris, S.; de Hoogh, C.; Hurt, C.; Jensen, T.K.; Maitland, I.; Richardson, S.; Wakeﬁeld, J.; Jarup, L. Risk of adverse birth outcomes in populations living near landﬁll sites. Brit. Med. J. 2001, 323, 363–368. [CrossRef] [PubMed] 2. Elliott, P.; Briggs, D.; Morris, S.; de Hoogh, C.; Hurt, C.; Jensen, T.K.; Maitland, I.; Richardson, S.; Wakeﬁeld, J.; Jarup, L. Risk of adverse birth outcomes in populations living near landﬁll sites. Brit. Med. J. 2001, 323, 363–368. [CrossRef] [PubMed] 3. Jarup, L.; Briggs, D.; de Hoogh, C.; Morris, S.; Hurt, C.; Lewin, A.; Maitland, I.; Richardson, S.; Wakeﬁeld, J.; Elliott, P. Cancer risks in populations living near landﬁll sites in Great Britain. Brit. J. Cancer 2002, 86, 1732–1736. [CrossRef] [PubMed] 3. Jarup, L.; Briggs, D.; de Hoogh, C.; Morris, S.; Hurt, C.; Lewin, A.; Maitland, I.; Richardson, S.; Wakeﬁeld, J.; Elliott, P. Cancer risks in populations living near landﬁll sites in Great Britain. Brit. J. Cancer 2002, 86, 1732–1736. [CrossRef] [PubMed] 4. Sikes, P.; Morris, R.H.; Allen, J.A.; Wilsmith, J.D.; Jones, K.P. Workers’ exposure to dust, endotoxin and beta-(1-3) glucan at four large scale composting facilities. Waste Manag. 2011, 31, 423–430. [CrossRef] [PubMed] 4. Sikes, P.; Morris, R.H.; Allen, J.A.; Wilsmith, J.D.; Jones, K.P. Workers’ exposure to dust, endotoxin and beta-(1-3) glucan at four large scale composting facilities. Waste Manag. 2011, 31, 423–430. [CrossRef] [PubMed] 5. Heldal, K.K.; Halstensen, A.S.; Thorn, J.; Eduard, W.; Halstensen, T.S. Airway inﬂammation in waste handlers exposed to bioaerosols assessed by induced sputum. Eur. Resp. J. 2003, 21, 641–645. [CrossRef] 5. Heldal, K.K.; Halstensen, A.S.; Thorn, J.; Eduard, W.; Halstensen, T.S. Airway inﬂammation in waste handlers exposed to bioaerosols assessed by induced sputum. Eur. Resp. J. 2003, 21, 641–645. [CrossRef] 6. Douwes, J.; Wouters, I.; Dubbeld, H.; van Zwieten, L.; Steerenberg, P.; Doekes, G.; Heederik, D. 4. Discussion 7 of 8 Int. J. Environ. Res. Public Health 2016, 13, 631 Our data, which need to be conﬁrmed with larger samples, suggest the importance of adopting preventive measures such as wearing speciﬁc individual protection devices, to protect this particular category of workers from adverse effects on respiratory health. 5. Conclusions Waste management is an issue of social concern owing to its environmental impact and effects on public health. In fact, waste management activities are carried out according to procedures that can have various negative effects on the environment and, potentially, on human health. On this basis, we ran a cross-sectional study to evaluate respiratory health effects in a group of workers in the solid waste collection and disposal ﬁeld, compared with a group of administrative employees. Our study showed a signiﬁcant reduction in the Tiffenau Index and FEV 1 values in the exposed workers as compared to the controls, after adjusting for confounding factors. These data suggest the importance of adopting preventive measures, such as wearing speciﬁc Personal Protection Equipments (PPEs), to protect this category of workers from adverse effects on respiratory health. Author Contributions: Luigi Vimercati conceived and designed the study, prepared and edited the manuscript; Antonio Baldassarre performed statistical analysis, prepared and edited the manuscript; Maria Franca Gatti performed spirometry; Luigi De Maria conducted questionnaire administration; Antonio Caputi entered data; Angelica A. Dirodi entered data; Francesco Cuccaro: performed statistical analysis; Raffaello Maria Bellino conceived and designed the study. All authors read and approved the ﬁnal draft of the manuscript. Conﬂicts of Interest: The authors declare no conﬂict of interest. Conﬂicts of Interest: The authors declare no conﬂict of interest. 4. Discussion In a recent study made in 25 waste handlers at a landﬁll near Oslo, the spirometric FEV1 values were signiﬁcantly lower at the end of the workweek than at the beginning of the week. In addition, cytological examination of sputum showed an increased percentage of neutrophils and IL-8, supporting the hypothesis that an etiopathogenic mechanism of an aspeciﬁc inﬂammatory nature, mediated by neutrophils, underlies the spirometric alterations [5]. In a study by Gea de Meer, only a slight reduction in FEV1 was observed at the end of the workweek in a group of 16 organic waste loaders, only 6 of whom had a positive history for respiratory symptoms, which was determined by questionnaire. The bronchial challenge test subsequently performed with methacholine showed a signiﬁcantly reduced FEV1 in the symptomatic subjects only. The authors concluded that the occupational exposure could be responsible for exacerbating airway inﬂammation only in those subjects with preexisting respiratory symptoms [29]. In agreement with most of the data in the literature, our ﬁndings support the existence of a prevalence of respiratory deﬁcits, as evidenced by spirometry, in waste disposal workers. Although this conclusion is not based on a correlation with the respiratory symptoms investigated in the questionnaires, it seems plausible to assume that the altered ventilatory function parameters obtained in our study may be considered as a preclinical indicator of respiratory diseases in subjects who have not yet shown the onset of clinical symptoms. It should be noted that our study, which was not conducted in a very large population, did not include environmental monitoring data on the concentrations of allergens, endotoxins, dust, and toxic substances in the areas surrounding the exposed group’s work zones. In particular, it would be useful to measure the concentration of endotoxins and β(1-3)-glucans that seem to play a key role, even via a synergic action, in inducing an inﬂammatory response of the upper airways. However, it is reasonable to suppose that there will be no signiﬁcant differences in the concentrations values measured in other European nations with comparable waste management policies compared to those adopted in Italy [30]. It must also be borne in mind that waste collection, like other prevalently outdoor working activities, can expose workers to urban pollutants (i.e., particulate matter, nitrous oxide, carbon dioxide, ozone, and carbon monoxide) [31–33], known to play a causal or concausal role in the onset of respiratory symptoms and spirometric alterations. References Airway function and respiratory t i it ti k J O E i M d 1996 38 522 527 [C R f] [P bM d] 18. Zuskin, E.; Mustajbegovic, J.; Schachter, E.N.; Kern, J.; Pavicic, D.; Budak, A. Airway function and respiratory symptoms in sanitation workers. J. Occup. Environ. Med. 1996, 38, 522–527. [CrossRef] [PubMed] 19. Ivens, U.I.; Ebbehoj, N.; Poulsen, O.M.; Skov, T. Season, equipment, and job function related to gastrointestinal problems in waste collectors. Occup. Environ. Med. 1997, 54, 861–867. [CrossRef] [PubMed] 20. Wouters, I.M.; Hilhorst, S.K.; Kleppe, P.; Doekes, G.; Douwes, J.; Peretz, C.; Heederik, D. Upper airway inﬂammation and respiratory symptoms in domestic waste collectors. Occup. Environ. Med. 2002, 59, 106–112. [CrossRef] [PubMed] 21. Matheson, M.C.; Benke, G.; Raven, J.; Sim, M.R.; Kromhout, H.; Vermeulen, R.; Johns, D.P.; Walters, E.H.; Abramson, M.J. Biological dust exposure in the workplace is a risk factor for chronic obstructive pulmonary disease. Thorax 2005, 60, 645–651. [CrossRef] [PubMed] 22. Kramer, M.N.; Kurup, V.P.; Fink, J.N. Allergic bronchopulmonary aspergillosis from a contaminated dump site. Am. Rev. Resp. Dis. 1989, 140, 1086–1088. [CrossRef] [PubMed] 23. Allmers, H.; Huber, H.; Baur, X. Two year follow-up of a garbage collector with allergic bronchopulmonary aspergillosis (abpa). Am. J. Ind. Med. 2000, 37, 438–442. [CrossRef] 24. Falcon-Rodriguez, C.I.; Osornio-Vargas, A.R.; Sada-Ovalle, I.; Segura-Medina, P. Aeroparticles, composition, and lung diseases. Front. Immunol. 2016, 7, 3. [CrossRef] [PubMed] 5. Medical Research Council’s Committee on Environmental and Occupational Health. Questionnair Respiratory Symptoms; Medical Research Council: London, UK, 1986. 6. Minette, A.; Aresini, G.; Sanna-Randaccio, F.; Seaton, A.; Smodt, U.; Teculescu, U. Promemoria Relativ Questionario Della CECA, 3rd ed.; Commissione delle Comunità Europee: Luxembourg, Lussemburgo, 1 27. ATS. Standardization of Spirometry. Available online: https://www.thoracic.org/statements/resources/ pfet/PFT2.pdf (accessed on 2 February 2016). 28. Athanasiou, M.; Makrynos, G.; Dounias, G. Respiratory health of municipal solid waste workers. Occup. Med. 2010, 60, 618–623. [CrossRef] [PubMed] 29. De Meer, G.; Heederik, D.; Wouters, I.M. Change in airway responsiveness over a workweek in organic waste loaders. Int. Arch. Occup. Envir. Health 2007, 80, 649–652. [CrossRef] [PubMed] 30. Neumann, H.D.; Balfanz, J.; Becker, G.; Lohmeyer, M.; Mathys, W.; Raulf-Heimsoth, M. Bioaerosol exposure during refuse collection: Results of ﬁeld studies in the real-life situation. Sci. Total Environ. 2002, 293, 219–231. [CrossRef] 31. Vimercati, L.; Gatti, M.F.; Baldassarre, A.; Nettis, E.; Favia, N.; Palma, M.; Martina, G.L.; Di Leo, E.; Musti, M. References Upper airway inﬂammation assessed by nasal lavage in compost workers: A relation with bio-aerosol exposure. Am. J. Ind. Med. 2000, 37, 459–468. [CrossRef] 7. Thorn, J.; Beijer, L.; Rylander, R. Airways inﬂammation and glucan exposure among household waste collectors. Am. J. Ind. Med. 1998, 33, 463–470. [CrossRef] . Viegas, S.; Veiga, L.; Figueiredo, P.; Almeida, A.; Carolino, E.; Viegas, C. Assessment of workers’ exposur aﬂatoxin B1 in a Portuguese waste industry. Ann. Occup. Hyg. 2015, 59, 173–178. [CrossRef] [PubMed] 8. Viegas, S.; Veiga, L.; Figueiredo, P.; Almeida, A.; Carolino, E.; Viegas, C. Assessment of workers’ exposure to aﬂatoxin B1 in a Portuguese waste industry. Ann. Occup. Hyg. 2015, 59, 173–178. [CrossRef] [PubMed] 9. Rylander, R.; Jacobs, R. ICOH committee on organic dust—Endotoxin in the environment: A criteria document. Int. J. Occup. Environ. Health 1997, 3, 1–48. 9. Rylander, R.; Jacobs, R. ICOH committee on organic dust—Endotoxin in the environment: A document. Int. J. Occup. Environ. Health 1997, 3, 1–48. 10. Michel, O. Human challenge studies with endotoxins. Int. J. Occup. Environ. Health 1997, 3, 18–25. 11. Ulmer, A.J. Biochemistry and cell biology of endotoxins. Int. J. Occup. Environ. Health 1997, 3, 8–17. 12. Ling, S.H.; van Eeden, S.F. Particulate matter air pollution exposure: Role in the development and exacerbation of chronic obstructive pulmonary disease. Int. J. Chronic Obstr. Pulm. Dis. 2009, 4, 233–243. [CrossRef] Int. J. Environ. Res. Public Health 2016, 13, 631 8 of 8 13. Fogelmark, B.; Thorn, J.; Rylander, R. Inhalation of (1->3)-beta-d-glucan causes airway eosinophilia. Mediat. Inﬂamm. 2001, 10, 13–19. [CrossRef] [PubMed] 14. Fogelmark, B.; Sjostrand, M.; Rylander, R. Pulmonary inﬂammation induced by repeated inhalations of beta(1,3)-D-glucan and endotoxin. Int. J. Exp. Pathol. 1994, 75, 85–90. [PubMed] 15. Ray, M.R.; Roychoudhury, S.; Mukherjee, G.; Roy, S.; Lahiri, T. Respiratory and general health impairments of workers employed in a municipal solid waste disposal at an open landﬁll site in Delhi. Int. J. Hyg. Environ. Health 2005, 208, 255–262. [CrossRef] [PubMed] 16. Yang, C.Y.; Chang, W.T.; Chuang, H.Y.; Tsai, S.S.; Wu, T.N.; Sung, F.C. Adverse health effects among household waste collectors in Taiwan. Environ. Res. 2001, 85, 195–199. [CrossRef] [PubMed] 17. Sigsgaard, T.; Malmros, P.; Nersting, L.; Petersen, C. Respiratory disorders and atopy in Danish refuse workers. Am. J. Respir. Crit. Care Med. 1994, 149, 1407–1412. [CrossRef] [PubMed] 18. Zuskin, E.; Mustajbegovic, J.; Schachter, E.N.; Kern, J.; Pavicic, D.; Budak, A. © 2016 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC-BY) license (http://creativecommons.org/licenses/by/4.0/). References Occupational exposure to urban air pollution and allergic diseases. Int. J. Environ. Res. Public Health 2015, 12, 12977–12987. [CrossRef] [PubMed] 32. Nordenhall, C.; Pourazar, J.; Ledin, M.C.; Levin, J.O.; Sandstrom, T.; Adelroth, E. Diesel exhaust enhances airway responsiveness in asthmatic subjects. Eur. Resp. J. 2001, 17, 909–915. [CrossRef] 33. Behndig, A.F.; Larsson, N.; Brown, J.L.; Stenfors, N.; Helleday, R.; Duggan, S.T.; Dove, R.E.; Wilson, S.J.; Sandstrom, T.; Kelly, F.J.; et al. Proinﬂammatory doses of diesel exhaust in healthy subjects fail to elicit equivalent or augmented airway inﬂammation in subjects with asthma. Thorax 2011, 66, 12–19. [CrossRef] [PubMed] © 2016 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC-BY) license (http://creativecommons.org/licenses/by/4.0/).
https://openalex.org/W1979870607	https://zenodo.org/records/2271585/files/article.pdf	English	null	The photic reactions of tadpoles in relation to the Bunsen‐Roscoe law	The journal of experimental zoology/Journal of experimental zoology	1,921	public-domain	16,017	Translation by Jose F. Nonidez Cornell Medical College, New York Zoological Laboratory, Harvard University. NINE FIGURES CONTENTS I. Introduction ........................................................ 235 11. Historical review.. ................................................. 237 111. Material and methods.. ...... ................... 238 1. Selection of animals for ................... 239 2. Apparatus.. ... ... .............................. 241 3. Method.. .............. ................... 242 IV. Dynamics of sensitivity to fight. .................................... 244 V. Photosensitivity of the skin 8s compared with the eye. .............. 252 1. The skin ....................................................... 252 2. The eye ............... ... ................. 259 VI. Velocity of changes in the receptors with higher light intensities and the duration of secondary processes following sensitization.. ...... 261 VJI. Photosensory recovery. ............................................. 268 IX. Discussion .......................................................... 273 1. The nature of the process.. .................................... 273 2. The photoreceptors.. ........................................... 275 X. Conclusions ......................................................... 277 XI. Bibliography.. ...................................................... 278 V111. The process of fatigue.. ............. .......................... 270 Las reacciones f6ticas de 10s renacuajos en relaci6n con la ley de Bunsen-Roscoe. Los renacuajos de Rana clamitans que han permanecido durante algdn tiempo en la oscuridad cuando se les ilumina stibitamente con una luz de suficiente intensidad responden a1 estimulo avanzando r6pidamente hacia delante. El tiempo que transcurre entre la aplicacidn del estimulo y la respuesta varia con la intensidad de la luz. A causa de esta relaci6n definida el autor supone la existencia de una substancia fotoquincica pre- sente en 10s fotoreceptores, cuya substancia sufre ciertos cambios definidos durante la iluminaci6n. El autor ha llevado A cabo un estudio para determinar hasta que punto la ley de Bunsen- Roscoe es aplicable 6 10s cambios que tienen lugar en 10s foto- receptores. La raz6n de las intensidades extremas empleadas fu6 1:500; comparhndose la sensibilidad del ojo con la de la pie1 y tambi6n con la sensibilidad cuando se iluminan ambos 6rganos. Los cambios experimentados en 10s fotoreceptores durante la ilumi- nacibn proceden de acuerdo con la ley de Bunsen-Roscoe. Con una serie media de iluminaciones, 10s experimentos demuestran que 10s productos de la intensidad por el tiempo coinciden. Con mayores intensidades tiene lugar una desviaci6n en 10s productos de la intensidad por el tiempo, la cual es gradual y parece pro- ceder con una constancia definida. El autor presenta una representach empirica, la cual satisface bien 10s datos actuales obtenidos. El proceso de la fatiga y el recobro de la sensibilidad fotosensorial son tambih objeto de discusi6n. AUTHOR'S ABSTRACT OF THIS PAPER IESUED BY THE BIBLIOQRAPHIC SERVICE, AUQUST 5 I. INTRODUCTION The principle of transference of energy implies that for a given amount of energy used in the incitement of a reaction there must be evoked an equivalent effect. This principle has led to the development of a number of laws in general chemistry, which furnish expressions of the relation of the amount of energy received, to the effect produced, and the progression of this effect with the time of action and active masses. In 1862 Bunsen and Roscoe showed that in order to produce equal shades of darkness on silver-chloride papers exposed to 235 236 VASIL OBRESHKOVE various intensities of illumination, the time required was in- versely proportional to the intensity of the light. The inten- sity of light and the time of action varied in such a manner in the production of the same effect that the product of these two variables was always constant. This can be expressed: I X T = K , where 1 stands for the intensity of light, T the time of action, and K the constant effect produced. Much has been written regarding the responses of animals to light, and it is now believed that light acts as a stimulus to the photoreceptors through a change in a chemical substance with which the nerve endings are in immediate contact. Parker and his students have described a number of fishes and amphib- bians possessing photoreceptors in the skin. This work has been recently extended to animals belonging to other groups. It seems advisable, therefore, to undertake a study with the aim of determining (1) to what extent the Bunsen-Roscoe law is applicable to animals which respond to light; (2) the dynamic nature of the responses, and, (3) the extent to which chemical changes produced in receptors are analogous to those in general chemistry. In sensory reflexes, however, the apparent effect produced -let us say as a change in a specific peripheral receptor-is pre- ceded by several other steps, namely: 1) a diffusion of the sub- stance newly formed in the receptor to the specific nerve endings; 2) transmission of the impulse from the nerve endings to the adjustor and thus to the muscles, and, 3) contraction of the muscles. When in the test of the validity of the transference of energy, these intervening steps occupy a relatively long period, they must be separated from the energy-receiving period, analyzed, and appropriate values given to them. I. INTRODUCTION We have reason to believe, however, that a species whose responses are characterized by rapid changes in the receptors after stimulation and a very short secondary period is an ani- mal better adapted to receive and respond to a specific stimulus than one in which these changes are slow. When the specific PHOTIC REACTIONS OF TADPOLES 237 energy involved is small in amount, a critical quantitative study of the responses may enable us to substitute for the terms 'strongly phototropic ' and 'slightly phototropic, ' heretofore used in describing degrees of sensitivity, a numerical value for the intensity-time product. A series of such studies on various organisms, if successful, should enable us to arrange them in the order of their photosensitivities. This in turn may throw some light on the organization of the animal and on the extent to which this state serves it in its life. Quantitative studies have already been made by Hecht ('18, '19a, b, '20) on the dynamic nature of the responses of Ciona and of Mya. The animals used in the present study were tadpoles of Rana clamitans. These were especially appro- priate in such a study, since they remain undisturbed in the dark for long periods, and in sudden illumination produce clear-cut responses by forward swimming or by movements of the tail. My thanks are due Prof. G. H. Parker, who suggested this problem and supervised the work and for whom, both as teacher and as adviser, I have the most sincere respect. His character- istic way of making suggestions and offering criticisms has served as an inspiration during the time of my association with him. 11. HISTORICAL REVIEW Interest in the photodynamic behavior of animals and plants and the extent of agreement with the Bunsen-Roscoe law has been shown only comparatively recently. Blaauw ('09) was one of the first to undertake an investigation of this subject. In the production of a curvature in the seedlings of Avena sativa when exposed to light of various strengths, he found that, the value of the product of the intensity of light and the time re- quired to cause the same number of the plants to bend toward the illuminated side were nearly constant. Similar observations have been recorded by Froschel ('09) with work on other plants. These animals were found to be strongly phototropic. When lights Loeb and Northrop ('17) used the larvae of barnacles. 238 VASIL OBRESHKOVE of equal intensity were placed one on each side of a rectangular aquarium containing the larvae of barnacles, collected in a blackened pipette, the animals when freed oriented themselves in their swimming in such a manner as to form a trail at an angle of 90" to a line connecting the two lights. When the lights were of unequal intensity, the trail formed was toward the weaker light. The degree of deflection was found to be dependent upon the ratio of the two intensities used, thus showing in a relative way that the effect produced is dependent on the amount of energy received. Loeb and Wasteneys ('17) experimented with Eudendrium, and the results were similar to those obtained by Blaauw. y In some recent publications Hecht ('18, '19a, b, '20) has recorded a series of readings representing the pho tic sensitivity of Ciona intestinalis and Mya when the time of exposure and the intensities of light were varied. The results obtained leave no doubt that the velocity of change in the receptors during illumi- nation is dependent upon the intensity and the time of action. He has pointed out that to produce in an animal, a reflex due to light the time required is inversely proportional to the intensity of the light and therefore that the product of these two is acon- stant quantity for all conditions. 111. MATERIAL AND METHODS Tadpoles of Rana clamitans are very abundant in the ponds in the vicinity of Cambridge and Boston. The tadpoles transform the next season after hatching, thus covering two breeding seasons. Hence during any month of the year there may be found in the ponds tadpoles of various sizes; some that were hatched during the preceding spring and others that are about to transform. Animal dealers in Boston keep the tadpoles of Rana clami- tans for sale throughout the year, and from these sources the greater part of the material for the experimental work was secured. On several occasions, however, the animals were brought directly from the ponds into the laboratory. Such tadpoles usually were found much overfed and inactive and PHOTIC REACTIONS OF TADPOLES 239 were not favorable material for immediate experimental work. These individuals were kept in the laboratory for at least one week before they were subjected to experiment. In that time the mud and other matter which filled the stomach and intestine when the animal was captured were discharged and the animals became accustomed to the laboratory condition, were easily stimulated, and reacted with regularity to light. 1. Selection of animals for experimentation Although the melanophores in the tadpoles of Rana clamitans are not subject to as much contraction and expansion as is ob- served in other tadpoles and in certain fishes, yet during the course of the experimentation, as well as when tadpoles were just brought from the pond or from the general aquarium, they showed considerable variation in the state of these bodies. A knowledge of the condition which induces contraction and ex- pansion of the melanophores and the general state of the animals when in either condition proved of importance early in the work since tadpoles darker in color from expanded pigment cells showed a slower reaction time than those that were lighter in color. It was decided, therefore, that that condition of the melanophores is an important factor to be kept in mind for a study like the one undertaken, where the reaction-time of the individuals to light constituted the only measurement for the examination of the validity of a law. Tadpoles caught in the shallow water of a pond where the background and bottom were chiefly white clay, and tadpoles which had been left in white porcelain dishes in a light room for several days were found to be light in color. On the other hand, tadpoles which were captured from deep portions of the pond, and had been concealed under grass or debris, were dark brown in appearance. Light colored tadpoles, however, which had been placed in 0.2 per cent chloretone solution for anesthetizing, preparatory to the removal of the eyes, expanded the melano- phores in the course of a few minutes. Such animals, when placed in the dark room, remained dark for several days, and in 240 VASIL OBRESHKOVE most cases this condition was retained for much longer periods. Light-colored tadpoles, however, which had not been subjected to anesthetics, when removed to the dark room, remained in the light condition. The work was confined, therefore, to tadpoles as nearly as possible of the same color, and those some- what pale in appearance were chosen. For this reason, when blind animals were needed for some phases of the problem itwas found advisable to remove the eyes without previously placing the individuals in chloretone. It is very possible that the slower reaction-time of animals with expanded melanophores is due to a nervous condition of the individual which may be associated with the melanophores in an expanded state. 1. Selection of animals for experimentation There is another and more plausible explanation of this condition, which will be considered in connection with another phase of the problem. Preliminary experiments revealed the necessity of avoiding another difficulty. Individuals differing in size showed variation in the regularity with which they responded. The smaller individuals were more active and more orderly in their responses. Tadpoles between 4 and 5 em. in length were found to be the most favorable. Tadpoles which were chosen for the experimental work on the basis of size and coloration were kept in the darkroom through- out the period of experimentation. Previous to each experi- ment, however, they were tested to determine whether they were in normal state, free from any infection, such as saprolegnia, and whether all of them were of equal photosensitivity. This was done in the following manner. Five or six tadpoles were placed in a white porcelain dish about 37 em. in diameter and with just sufficient water to cover them. Such individuals were then repeatedly stimulated with a constant intensity of light at intervals of one minute or longer. Only those tadpoles were chosen which, with no exception, showed activity upon the application of the light. Those which showed indifference to the light stimulus were removed and placed in a special dish and from time to time tested during the next few days. These tadpoles usually died soon after being isolated from the chosen stock. Previous to each experiment the selected animals were PHOTIC REACTIONS OF TADPOLES 241 observed in the dark with the aid of a constant dim light, to determine the frequency of the spontaneous movements. When their state of activity was such as to leave no doubt that their movements were due to the applied stimulus of light, the experi- mental work was begun. This usually took place when the time of rest in the dark was several times longer than the time required to produce a reaction after the stimulus had been applied. Then any movement after the stimulus was applied could not be questioned as having been the result of the effect of light. 2. Apparatus The apparatus for the first set of experiments (fig. 1) con- sisted of a small rectangular table (a, b), on which was placed a bicycle tire (c) and then a rectangular board (n). On this rested a round white porcelain dish (v) of the size already mentioned, containing a number of tadpoles of equal photosensitivity as previously determined. A short distance in front of the table and slightly above the level of the vessel was a dim light of 1 candlepower inclosed in a box (a). The front of the box was made of a number of layers of thin translucent paper, which allowed just sufficient light to escape for the observation of the animals. The legs of the table rested upon individual mats (m) each formed, from the floor up, of, 1) a rectangular rubber plate 3.5 cm. in thickness; 2) a tile of the same size and thick- ness; 3) a pad of excelsior, and, 4) a small square board. It was believed that such an arrangement would absorb any vibra- tion which might occur during the experimentation. The light (s) was from above. By using a flash-light bulb of 0.5 candlepower and Mazda tungsten electric bulbs of 10 and 40 watts and by varying the distance of the bulb from the tadpole, a series of intensities was secured ranging from 0.3 candleimeter to 500 candle-meters. The amount of light which these bulbs produced was accurately measured by means of a Lummer- Brodhun photometer. 242 VASIL OBRESHKOVE 3. Method If tadpoles subjected to various intensities of light are respon- sive with any degree of regularity, they ought to furnish measure- ments of the photic reactions for an examination of the Bunsen- 6 ! Fig. 1 Diagram of apparatus used for testing the sensitivity of tadpoles of Rana clamitans to various intensities of light. a,b, small rectangular table; c, bicycle tire; n, rectangular board; u, porcelain dish; q, dim light inclosed in a box; m, mat; s, source of light. Fig. 1 Diagram of apparatus used for testing the sensitivity of tadpoles of Rana clamitans to various intensities of light. a,b, small rectangular table; c, bicycle tire; n, rectangular board; u, porcelain dish; q, dim light inclosed in a box; m, mat; s, source of light. Roscoe law, especially if any relation between the degree of stimulation and the time which elapses before a response is noted can be ascertained. The intensities generally used were 0.3, 0.71, 1.2, 5, 10, 15, 20 and 30 candle-meters. To these, in the latter nmt, of the work. were ~ . d d ~ d 60. 200. m d ,500 rtmdle- 243 PHOTIC REACTIONS OF TADPOLES meters. In the first set of experiments the photosensitivity of tadpoles with eye and skin was tested. The use of such low degrees of stimulation seems to be in marked contrast with the methods used in the study of the reactions of amphibians to light by other workers. A re- view of the extensive literature on the subject reveals that the intensities used in various investigations range approxi- mately from 200 candle-meters to 1000 candle-meters, and in some cases much higher degrees of stimulation were used. The instances where low intensities were used are few, but in all cases the object of the studies was to test the orientation of the animals in relation to a source of light and no attempt was made to determine extensively any quantitative relation which might exist between the degree of stimulation and response. Cole and Dean (’17), working with the tadpoles of Rana clamitans, ob- served photokinesis when three intensities were employed, 500 c.P., 100 c.P., and 48 c.p. (distance from source of stimulus was not given). THE JOURNAL OF EXPERIMENTAL ZOOLOGY, VOL. 34, NO. 2 3. Method The present study revealed that, in addition to this, these tadpoles, when kept in the dark and then suddenly and repeatedly illuminated by relatively low intensities within a certain range, show a period elapsing between the moment of stimulation and response which can be accurately measured by a stop-watch. This can be called the reaction-time and includes a sensitization period, during which light energy is taken up by the receptors and thus initiates a change which acts as stimulus to the nerve endings, and also the time for certain secondary processes, such as diffusion in the receptors to the nerve endings, transmission of the impulse to the adjustor and effector and the time occupied by the muscles in their contraction. This secon- dary period involving the processes just enumerated is dis- cussed by certain authors under the term of ‘latent period.’ Others use this term to designate the total time from the appli- cation of the stimulus to the observable response. In this discussion we shall use reaction-time to designate the time occupied by sensitization, that is, the minimum period during which the light must act in order to produce a change in the receptors of sufficient strength to initiate a reflex, plus the time occupied by any secondary processes. 244 VASIL OBRESHKOVE If animals that were to be repeatedly stimulated were allowed one minute of rest in the dark before stimulation was renewed, it was found possible to secure from fifteen to twenty readings from the same individual without any evidence of exhaustion. Now and then in the course of the experiment inconsistencies occurred. These were of the type of premature response or delayed response in the midst of a series of readings otherwise characterized by marked regularity. These irregular responses were recorded, however, and it will be evident from an inspec- tion of the results that they were not numerous enough to throw doubt on the numerical values obtained. An insight into the cause of these occasional irregularities is difficult to obtain. Premature responses were associated at times with discharges from the intestine. IV. DYNAMICS OF SENSITIVITY TO LIGHT When tadpoles of Rana clamitans are subjected to a stimulus of 5 candle-meters, the light being applied from above, the period which elapsed between the time of the application of the stimulus and that of the response had approximately a fixed value. Re- peated stimulations at one-minute intervals revealed only slight changes in their period. These slight deviations from the mean of the reaction-time for this and other intensities were only such as were to be expected from normal law of error, as was demonstrated by a study of frequency curves. The reac- tion time for a set of ninety-eight readings at 5 candle-meters showed an average of 24.1 seconds. Since not all of the readings were taken during the same day, and since this part of the work was done during the winter months, the variations in temperatures were recorded. Obser- vations based on approximately 2,000 readings with various intensities of light showed no evidence that a temperature varia- tion between 16°C. and 19°C. has an observable effect on the reaction-time. This is not surprising in view of the fact that the velocity of reaction in photic operations is dependent upon the actual amount of light absorbed per unit of time and this amount 245 PHOTIC REACTIONS OF TADPOLES varies very little with the rise and fall of temperature. Kot- withstanding this fact, after some preliminary work, the tempera- ture during the process of experimentation was regulated to approximately 19°C. With a constant intensity of 10 candle-meters, the reactjon- time was considerably reduced. From the general course of the readings it became evident that the stronger light was more effective, for the animals responded to it with a greater degree of regularity. The reaction-time, which was 24.1 seconds at 5 candle-meters, being in each case the average of about 90 read- ings, was now reduced to 12.1 seconds. This definite relation be- tween the strength of stimulus and the effect produced was re- tained with an intensity of 30 candle-meters. The average reaction-time with this intensity for ninety-three readings was 6.1 seconds. The actual values obtained with these three in- tensities are given in table 1 with their average reaction-times, standard deviation, and probable error. Firom an inspection of the table, it is evident that a stronger stimulus requires less time to produce the same effect, namely, a simple swimming movement, than a weaker one. TABLE 1 TABLE 1 Reaction-times in seconds of tadpoles of Rana clamitans to lights of different intensities. Temperature, 16" to 19°C. Source of light above TABLE 1 Reaction-times in seconds of tadpoles of Rana clamitans to lights of different intensities. Temperature, 16" to 19°C. Source of light above 0.3 No ?ffect TABLE 1 Reaction-times in seconds of tadpoles of Rana clamitans to lights of different intensities. Temperature, 16" to 19°C. IV. DYNAMICS OF SENSITIVITY TO LIGHT In addition to the values obtained with 5, 10, and 30 candle- meters other intensities were used. These were 0.3, 0.7, 1.2, 15, and 20 candle-meters and with each intensity about 100 readings were secured. In all the experiments in which the relation of the reaction-time to the intensity was studied, this relation between stimulus and effect produced was retained. The weaker the light, the longer was the period required to pro- duce a reaction. With the higher intensities more uniformity of responses was obtained, showing that the energy is more effective as a stimulus, judging from the gradual decrease in the standard deviation and probable error (table 1). The probable errors with the more effective intensities are small.. In order to determine whether the differences between the average reaction- times for the various intensities are of sufficient magnitude for stat,istical value, the probable errors of these differences were computed on the basis that the probable error of the difference 246 VASIL OBRESHKOVE TABLE 1 Source of light above 0.7 1.2 5 267 195 45 125 29 24 227 362 46 252 31 18 135 74195 90 24 20 81 219107 70 26 24 390 225 61 45 16 20 76 140185 136 20 10 135 333 83 35 20 18 136 216197 95 21 25 lU7 I90 75 110 20 34 110 92 83 85 45 15 160 182107 110 22 24 321 245 134 85 34 16 89 130231 67 30 12 230 95 66 55 21 20 157 120127 98 35 33 91 220126 56 42 25 170 147 60 62 23 25 290 336102 50 8 25 175 80104 68 22 26 242 112100 62 23 13 306 130271 60 20 36 382 50 105 145 25 9 164 217 55 54 15 27 200 155 99 85 38 16 llb 135113 97 31 20 340 195 48 48 25 22 155 115 44 90 13 14 124 145 54 87 45 9 105 120 69 60 29 9 198 357 158 89 26 24 60 82 55 95 19 13 95 60137 110 51 22 135 141 95 44 34 22 70 421 45 55 20 46 60 205 72 45 29 13 142 181 39 100 31 18 171 134 60 105 24 39 197 198110 115 26 14 97 71 47 78 16 30 114 210 55 122 20 63 -~~~ INTGNSITY OF LIQHT IN CANDLE-METERS 10 12 7 9 13 15 16 16 17 15 6 11 12 20 19 16 9 18 11 9 10 16 8 10 9 18 12 10 11 13 6 8 8 8 16 17 15 14 10 9 12 12 16 6 14 17 9 7 17 11 11 17 16 18 10 6 17 14 6 7 10 13 8 13 11 6 16 9 10 6 8 12 12 7 7 17 10 12 25 10 19 15 5 7 6 7 5 7 8 7 6 6 16 6 14 16 7 8 LO 4 14 4 LO 15 6 16 6 3 5 5 15 9 11 6 17 7 7 12 10 4 13 10 12 6 9 5 5 18 5 7 8 9 12 11 4 4 4 4 5 6 5 10 10 15 10 9 12 3 6 8 6 6 11 5 8 5 10 10 11 8 3 10 20 3 8 3 5 6 3 5 5 4 4 6 7 13 8 9 4 10 5 11 9 5 3 16 10 6 6 6 3 14 9 13 6 7 6 ; 6 5 9 6 9 6 9 12 7 11 7 5 15 5 4 10 7 6 14 6 8 5 7 12 6 5 7 10 9 6 6 7 5 5 5 5 11 5 10 17 5 6 15 13 10 8 30 5 4 5 6 5 6 8 10 8 8 6 10 6 5 6 7 5 9 4 5 3 8 4 5 10 5 5 7 5 5 5 5 5 3 4 2 6 5 9 3 8 9 9 5 6 6 4 6 7 10 4 7 4 11 9 2 2 5 6 7 5 9 3 11 7 5 6 2 8 4 8 5 3 6 10 4 4 6 8 9 247 PHOTIC REACTIONS OF TADPOLES PHOTIC REACTIONS OF TADPOLES TABLE I-Continued 15 10 11 9 1 0 10 14 9 1 3 4 15 8.57 3.62 0.251 0.3 20 __- 15 8 8 7 6 6 6 5 1 0 8 5 5 1 2 9 6 6 9 5 1 3 4 12 15 -- 7.62 3.36 0.233 No effect .21 .30 .59 180 !05 Average. . . . . . . . . . . Standard deviation Probable error of mean. . . . . . . . . . . 426105 11: 162108 5! 60 14: INTENBITY OF LIQHT IN CANDLE-METER5 173.06 87.91 6.29 0.7 I 1.2 -- 93.10 45.93 3.32 45 5 20 14 21 55 30 9 14 22 18 35 24 33 14 34 28 9 41 33 24.14 10.38 0.711 10 9 25 10 17 11 10 I1 15 13 11 I1 12 20 9 I1 13 10 9 12.14 4.14 0.284 30 5 8 6 7 11 3 8 11 8 9 4 7 4 6.16 2.30 0.162 TABLE I-Continued between the constants of two uncorrelated series is where El and E2 are the constants of the two series to be com- pared. By this means it was found that the probable error of the difference between the means of the reaction-times at 5 and 10 candle-meters is 0.74. Since the difference between the average reaction times at these two intensities is 12, which is more than 16 times the probable error 0.74, and since a similar condition is true for the other higher and more effective inten- sities of illuminations, it is obvious that the values obtained are of significance in a discussion of the photodynamic nature of the responses. Between 0.71 candle-meter and 0.3 candle-meter ; inter- mediate intensities were used which do not appear in any of the tables, but it was noticed that in the lower range a point was reached where the animal no longer responded to the stimulus. This point in tadpoles of Rana clamitans was found to be 0.3 candle-meter. This with the other facts presented leads to the conclusion that in this animal there is an optimum intensity to 248 VASIL OBRESHKOVE light for the most prompt form of response. As the light in- tensities fall below this optimum, the effectiveness of the stimulus diminishes to cease at a certain point regardless of the length of time of exposure. Within the range of effectiveness, however, the principle of transference of energy was found to hold, for the action of light upon the photoreceptors produced an effect. This effect was equivalent to the amount of energy received, and it, was measured in terms of t.he time necessary at a definite inten- sity to produce a response. From the data thus far given, we have no way of studying the exact changes which occur in the sense organs. It seems obvious, however, that light must be absorbed in order to act as a stimulus and that some energy must be used up in order to initiate a change in the receptors. This in fact is a law governing photo- chemical phenomena. This law was first demonstrated by Grotthus (1819). It was later confirmed by others (Draper, '41). Lasareff ('07) showed that the Grotthus law is a quantitative photochemical law and that there is a definite relation between the amount of light absorbed per unit of time and the velocity of chemical change produced. between the constants of two uncorrelated series is In the light of the observations made thus far and on a basis of our knowledge of the mechanism of other types of receptors better known to us, we are forced to assume that the changes which occur in the sense organs in tadpoles of Rana clamitans during illumination are of a chemical kind. The reason for this assumption will become clearer when additional data are presented. For a discussion, then, of the dynamic process in the photo- receptors we must consider inore critically the relation of the reaction-time to the intensity of light. On the supposition that the relation is dependent on the velocity of a chemical change and assuming the validity of the Bunsen-Roscoe law, the product of these two variables should be a constant quantity for all intensities. This leads us to a discussion of table 2, which contains a summary of the results entered in table 1. In table 2 each figure in the second column represents the average of about ninety readings, and in the third column are given the intensity-reaction-time products. 249 PHOTIC REACTIONS OF TADPOLES PHOTIC REACTIONS OF TADPOLES From the apparent nature of the process during illumination, the reaction-time in the table represents a duration of action and the intensity represents the amount of energy which con- trols the velocity of change in the receptors concerned in the responses to light. In order that a swimming movement be produced, it appears that a definite material change must be produced in the sense organs as a stimulus. That the products of the intensity of light and the time of action agree well with each other for all intensities, can be little questioned from the numerical values entered in the third column. It will be noticed, however, that with the 20- and 30-candle-meter intensities, this TABLE 2 TABLE 2 X reaction-time products I X T LIGHT INTENSITY IN CANDLE-METERS 1 0.3 0.7 1.2 5.0 10.0 15.0 20.0 30.0 No effect 173.0 93.1 24.1 12.1 8.5 7 . 6 6.1 121.1 111.7 120.5 121.0 127.5 152.0 183.0 product is somewhat large. As a matter of fact, this change in the intensity-reaction-time product is not abrupt, but follows a definite course. It is characterized by a gradual increase with the increase in the degree of stimulation. These products with the first five intensities are well within the range of the Bunsen- Roscoe law. The deviation. which becomes so pronounced after an intensity of illumination of 20 candle-meters is reached, occurs in other experiments in similar manner, and will be dis- cussed later in more detail. product is somewhat large. As a matter of fact, this change in the intensity-reaction-time product is not abrupt, but follows a definite course. It is characterized by a gradual increase with the increase in the degree of stimulation. These products with the first five intensities are well within the range of the Bunsen- Roscoe law. The deviation. which becomes so pronounced after an intensity of illumination of 20 candle-meters is reached, occurs in other experiments in similar manner, and will be dis- cussed later in more detail. When we proceed to examine the results obtained and to plot graphically the values of the intensities on the abscissa and the values of the corresponding reaction-times on the ordinates, a 250 VASIL OBRESHKOVE smooth curve is obtained (fig. 2). This curve is hyperbolic (curve A) and indicates that as one of the variables increases, the other decreases-in this case, according to a constancy, a relation which is retained throughout the whole series of experi- ments. By the method of least squares from (x - a) (y - b) = c, computing for three constants and employing the summary computing for three constants and employing the summary L OQ. of reaction-he Snknsi fy - Candle meters i C ill h d i bl d h l i Fig. 2 Curve A illustrates the data in table 2 and represents the relation Curve B gives the logarithmic between the intensity of light and reaction-time. relation between the intensity of light and reaction-time of the same data. Fig. TABLE 2 2 Curve A illustrates the data in table 2 and represents the relation Curve B gives the logarithmic between the intensity of light and reaction-time. relation between the intensity of light and reaction-time of the same data. results in table 2, it was found that the data can be represented by the formula : (x - 0.15)(y - 2.85) = 94.48 or Y = 94*48 + 2.85 x - 0.15 251 PHOTIC REACTIONS O F TADPOLES Henri and Henri ('12) carried out a series of studies on the effect of ultraviolet rays upon cyclops. From their data it appears that in producing a reflex, the intensity-reaction-time products do not give a constant, but that the hyperbolic curve obtained can be expressed by the formula: a y = -2 + k. 2 Hecht ('18, '19, '20), in studying the photosensitivity of Ciona and Mya to different intensities of light emitted from an electric bulb, reports that the products of the sensitization periods and corresponding intensities agree well with each other for all in- tensities. The Bunsen-Roscoe ( '62) law was applied originally to the influ- ence of light on silver chloride. In the responses of tadpoles, there is a striking parallelism between that which has been demon- strated to be true for a purely chemical phenomenon 'in vitro' and the changes occurring 'in vivo,' namely, in the receptors of the tadpoles. p The Bunsen-Roscoe law may be expressed : ( l ) R T x I = K , where RT is the time of action, I , the intensity of the light, and K, the constant effect produced. Equation (1) may be put in the following forms : where RT is the time of action, I , the intensity of the light, and K, the constant effect produced. Equation (1) may be put in the following forms : K (2) RT = - I (3) log RT = log K - log I , (4) log RT = - log I. (4) log RT = - log I. TABLE 2 In the last form of the equation (4) the logarithm of the reaction-time is plotted as function of the logarithm of the corresponding intensity, and if the data in table 2 are plotted in accordance with this form, the points give an approximately straight line (B), thus indicating that in the production of a constant effect the velocity of change in the receptors proceeds at a uniform rate. With the gradual increase in the intensities of illumination there is an acceleration in the rate of the changes in the sense organs. 252 VASIL OBRESHKOVE V. PHOTOSENSITIVITY OF THE SKIN AS COMPARED WITH THAT OF THE EYE 1, The skin 1, The skin 1, The skin In terms of the Bunsen-Roscoe law, which was demonstrated as being valid in the photosensitivity of R. clamitans tadpoles, it is easy to see that the sensitivity of one animal as compared with another, or of two types of photoreceptors in the same ani- mal, is dependent upon t,he rapidity with which changes of sufficient strength are produced in the sense organs. Photo- receptors, therefore, are comparable to photographic plates, whose relative sensitivity is determined by the intensity of light and time of exposure necessary to produce the same degree of change on the various plates. In this the Bunsen-Roscoe law finds its first applicability in making possible the classification of degrees of sensitivity in terms of their corresponding intensity- reaction-time products. When tadpoles with eye and skin intact were subjected to light, it was at first supposed that the responses were those due to stimuli received through the eye. The next problem that naturally suggested itself was to determine the sensitivity of the skin as compared with that of the eye. In reviewing the extensive literature on the responses of amphibians to light, it is found to be the general opinion that responses may be brought about by stimulating either the eye or the skin. Among some of the workers advocating this view are Graber ('84), Dubois ( 'go), Parker ( '03), Reese ( 'OS), Pearse ('lo), Laurens (,14), and more recently Cole and Dean (,17). All of this work has been purely qualitative and in no case has an attempt been made to determine quantitatively the amount of energy necessary to stimulate, 1) the skin alone, 2) the eye alone, and, 3) both eye and skin. The only exception to this is found in the observations of Reese and of Pearse, that when the eyes of Necturus were removed, a greater period elapsed before orientation than normally. The short reaction-time that occurs when the eyes are present is ascribed by these workers to stimulations received through those organs. 253 PHOTIC REACTIONS O F TADPOLES The eyes of tadpoles of Rana clamitans were removed by cutting the skin around them very carefully with a small sharp dissecting knife and then placing curved scissors under the eyeball in such a way that all muscles and nerves were cut with one stroke and the eye removed. V. PHOTOSENSITIVITY OF THE SKIN AS COMPARED WITH THAT OF THE EYE Such tadpoles were allowed a few days in which to recover before being subject to experi- mentation. When the work with blind tadpoles was begun, it was intended to use the same number and range of intensities as were used for testing the photosensitivity of tadpoles with eyes and skin. Very early in the work, however, it became evident that the reaction-time of blind tadpoles was the same as that of tadpoles with eyes and skin illuminated. When blind animals were repeatedly stimulated with light of 1.2-candle-meter intensity, the individual readings corresponded very closely with those obtained from the animals with eyes and skin stimulated. Simi- larly, it was shown that both in tadpoles without eyes and in tadpoles with eyes and skin, the same period elapses between the stimulation and response under the influence of 15 candle- meters. Table 3 gives the actual observations at these two intensities, as well as the probable error, standard deviation, and the intensity-time products. These products agree well with each other, and likewise with those of corresponding inten- sities when eye and skin were illuminated. These surprising results suggested that with light falling vertically upon laterally placed eyes, such as those in tadpoles, the eyes may not have received the full stimulus applied and hence the similarity of results. For this reason the experiments with other intensities were not carried further in the manner described, but the subsequent tests and apparatus were modified so that the stimulus was applied from the side. The individuals to be tested were placed in a rectangular glass aquarium 38 em. long and 10 em. wide (fig. 3, m). By a tightly fitting plate t,his width was subdivided longitudinally into two compartments, one of which was just wide enough to allow freedom of movement of animals forward and backward, but always parallel to the side of the aquarium. At a distancein 254 VASIL OBRESHKOVE front of the aquarium sufficient to bring about the desired inten- sity of illumination was placed the electric bulb (n) properly connected with a switch within easy reach. The arrangement of the remainder of the apparatus was the same as when the light came from above. front of the aquarium sufficient to bring about the desired inten- sity of illumination was placed the electric bulb (n) properly connected with a switch within easy reach. V. PHOTOSENSITIVITY OF THE SKIN AS COMPARED WITH THAT OF THE EYE The arrangement of the remainder of the apparatus was the same as when the light came from above. TABLE 3 Reaction-times in seconds of eyeless tadpoles of Rana clamitans to lights of differ- ent intensities. Temperature, 19°C. Source of light above ent intensities. Temperature, 19 C. Source of light above INTENSITY OF LIGHT IN CANDLE-METERS - 1 8 96 144 97 85 67 93 73 102 45 103 240 119 180 180 47 45 Average. .................... Standard deviation.. ....... 1.2 42 95 99 71 40 85 104 96 75 141 105 55 290 65 81 233 72 135 61 49 87 134 45 91 202 47 86 188 90 115 119 47 97 105 117 196 173 101 92 156 171 71 104 136 50 111 58 151 64 45 116 159 - 168 65 55 98 48 60 154 10 1 44 143 67 71 68 55 116 56 167 258 132 ,105.5 . 52.75 Probable error of mean.. .................. 3.81 Intensity X reaction-time.. .............. .126.6 5 5 8 6 7 18 15 15 5 5 5 5 8 9 9 9 11 5 8 9 - 13 13 5 9 8 8 9 4 7 4 6 5 11 7 5 5 16 12 4 10 15 - 3 7 7 4 8 5 6 4 4 5 5 4 10 5 3 4 5 7 5 5 - 7.9 3.38 0.24 119.1 - 11 11 6 5 6 13 8 5 12 11 7 11 6 10 10 18 8 7 8 10 - __ 9 16 9 7 9 5 7 6 11 14 9 11 5 10 9 Usually two individuals at a time were placed in the aquarium. These were kept far enough apart to avoid any shadow effect. After the application of the stimulus the time of reaction of the individual to move first was recorded. With an apparatus thus arranged the eye as well as the side of the animal was illuminated. 255 PHOTIC REACTIONS O F TADPOLES The sensitivity of animals was tested under the two conditions: 1) skin alone and, 2) eye and skin. Such experiments furnished further data for a test of the validity of the Bunsen-Roscoe law as applying to the photosensitivity of this animal. The sensitivity of animals was tested under the two conditions: 1) skin alone and, 2) eye and skin. V. PHOTOSENSITIVITY OF THE SKIN AS COMPARED WITH THAT OF THE EYE Such experiments furnished further data for a test of the validity of the Bunsen-Roscoe law as applying to the photosensitivity of this animal. In table 4 are given the reaction-times of tadpoles to lights of different intensities when eye and skin are exposed to light (A) and after the eyes had been removed (B), the light in both instances being applied from the side. The reaction-times of Fig. 3 Diagram of apparatus for testing the sensitivity of the eye as compared with the skin. m, rectangular glass aquarium; n, light. Fig. 3 Diagram of apparatus for testing the sensitivity of the eye as compared with the skin. m, rectangular glass aquarium; n, light. animals with eye and skin were so similar to those obtained from corresponding intensities with blind tadpoles either when the source of stimulus was from the side or from above, that it was not found necessary to carry the work beyond five inten- sities, only four of which were effective. This is apparent by comparing the results in table 4 wiih those in tables 1 and 3. That with the gradual increase in the intensity of illumination the stimulus becomes more and more effective for bringing about a prompt form of response is again demonstrated here by the gradual decrease in the standard deviation and probable error 256 VASIL OBRESHKOVE 256 VASIL OBRESHKOVE TABLE 4 Reaction-tim.es in seconds of tadpoles of Rana clamitans to lights of different intensities: A , with eye and skin exposed to light; B, with skin only exposed to light. Light applied from the side A. With eye and skin ex posed. ............... 0.3 No effect Average. ........................... Standard deviation. ................ Probable error of mean. ............ Intensity X reaction-time.. ......... V. PHOTOSENSITIVITY OF THE SKIN AS COMPARED WITH THAT OF THE EYE INTENSITY OF LIGHT IN CANDLE-PETERS 5 39 21 16 43 22 28 23 22 18 35 46 13 11 28 4 4 6 34 24 45 37 13 42 23 10 11 6 10 54 24 16 4 3 6 14 7 13 15 19 12 10 19 13 29 14 35 15 17 22.61 12.70 1.28 113.0 10 11 9 18 23 11 16 8 12 15 6 8 10 14 22 11 13 13 15 15 12 14 6 7 9 12 16 10 15 12 12 12 13 8 11 9 20 9 12 10 13 8 8 8 2 0 16 15 10 12.28 3.94 0.39 122.8 15 8 9 8 8 14 11 8 8 5 10 10 4 10 4 14 8 10 8 9 9 8 13 14 10 9 6 9 12 7 7 8 7 6 8 7 6 5 11 5 5 8 10 11 9 4 7 5 8.34 2.57 0.26 125.1 20 TABLE 4 Reaction-tim.es in seconds of tadpoles of Rana clamitans to lights of different intensities: A , with eye and skin exposed to light; B, with skin only exposed to light. Light applied from the side accompanying the increase of the intensity of illumination. The intensity-time products agree well with each other under the two conditions and with those recorded in table 2. With a light intensity of 20 candle-meters the characteristic deviation already mentioned again becomes evident. 257 PHOTIC REACTIONS O F TADPOLES PHOTIC REACTIONS O F TADPOLES ~ 5 10 26 14 20 15 17 4 24 35 11 9 2 3 3 5 1 3 9 21 16 15 10 16 11 8 14 29 20 10 7 30 31 9 13 21 30 12 11 11 19 13 16 32 16 9 18 2 0 4 0 2 4 5 1 2 2 5 2 8 6 2 1 4 0 1 4 9 20 13 16 23 2 5 4 0 5 8 32 45 18 8 27 18 16 14 21 17 21 11 19 22 21 9 35 10 16 22 25 24 4 12 25 32 20 20 29 15 15 7 23 21 23 7 TABLE 4-Continued ~ 15 20 10 13 12 5 5 10 8 5 5 6 7 5 3 3 16 6 9 8 8 10 6 7 17 11 5 12 6 14 9 11 9 4 6 12 4 7 5 14 5 9 5 8 8 9 4 4 9 9 3 5 10 9 5 3 4 1 1 8 4 11 3 7 8 5 14 7 10 4 12 9 9 6 9 14 14 5 4 5 8 5 10 4 9 9 16 10 8 4 17 6 8 6 0 . 3 ~ Average ............................ Standa.rd deviation.. ............... Probable error of mean. ............ Intensity X reaction-time.. ......... No effect 23.82 13.25 8.51 7.15 8.31 6.03 3.78 2.38 0.80 0.59 0.38 0.24 119.1 132.5 127.6 143.0 INTENSITY OF LIGHT I N CANDLE-METER8 In figure 4 are plotted the data with blind tadpoles as re- corded in table 4 In order to secure a wider range of comparison TABLE 4-Continued In figure 4 are plotted the data with blind tadpoles as re- corded in table 4. In order to secure a wider range of comparison to this was added the results with 1.2 candle-meters in table 3. When the average reaction-times are plotted on the ordinates as functions of the intensity of light, the plotted points give a hyperbola (curve A). The logarithm of the average reaction- times plotted as function of the logarithm of the intensity, gives a straight line (B). From the hyperbolic representation it is clear that the reaction t,ime holds a definite relation to the inten- 258 VASIL OBRESHKOVE sity and that the relation is inverse. PHOTIC REACTIONS O F TADPOLES 259 PHOTIC REACTIONS O F TADPOLES Curve B characterizes the steady rate of increase in the velocity of change in the recep- tors with an increase of intensity of illumination. These graphic representations and the intensity-reaction-time products furnish additional evidence that the changes in the photoreceptors in this animal proceed according to the Bunsen-Roscoe law. LO$ of reaction-time Sntensik- Candle melers Fig. 4 Curve A illustrates the data in table 4. The reaction-time is plotted on the ordinates, as function of intensity, plotted on the abscissa. Curve B rep- resents the logarithmic relation between the reaction-time and intensity of illumination of the same data. Fig. 4 Curve A illustrates the data in table 4. The reaction-time is plotted on the ordinates, as function of intensity, plotted on the abscissa. Curve B rep- resents the logarithmic relation between the reaction-time and intensity of illumination of the same data. If, now, the curves in figure 2, illustrating the relation of the intensity of. illumination to the reaction-time of tadpoles with eye and skin, and the curves in figure 4, giving this relation of records obtained from blind tadpoles, are plotted with regard to the same axis, they will be seen to coincide almost exactly. This is apparent from an inspection of the data and curves, and for this reason no such graphic comparison has been under- taken. PHOTIC REACTIONS O F TADPOLES 2. The eye From the recorded data and graphic representations it appears that when the eyes and skin of tadpoles are illuminated, the responses are those due to stimuli received through the skin. To test this further, rays of known intensity were concentrated upon the eye only. To avoid stimulation of other parts of the body, the following method was employed. A box measuring 6 L S F > y , , ) Fig. 5 Diagram of apparatus for testing the sensitivity of the eye. a,b, abox constructed of heavy black cardboard; k , tightly fitting hollow cone; n, light. Fig. 5 Diagram of apparatus for testing the sensitivity of the eye. a,b, abox constructed of heavy black cardboard; k , tightly fitting hollow cone; n, light. 28 em. in length and 9 em. in height and in width was con- structed of heavy black cardboard (fig. 5, a, b). One end of the box was so modified by a cylindrical projection as toreceive a tightly fitting hollow cone (R) terminating in a small opening, slightly less than the diameter of the eye of the tadpole. A light bulb (n) of 0.3 candlepower was enclosed in the box and the distance from the end of the cone was so adjusted as to give a light intensity of 5 candle-meters at the position of the animal, this being the intensity which proved effective with the experi- 260 VASIL OBRESHKOVE ments on eye and skin and on skin alone. The rest of the appara- tus was the same as shown in figure 3. With the aid of a dim light, the end of the cone was closely applied in the direction of the eye lying next to the side of the aquarium, so that when the light was turned on, the rays reached the eye only. Such an arrangement made it possible to illuminate the eye by an accu- rately measured intensity of light. Several individuals, which had previously responded with a great degree of regularity, were used. Repeated stimulations of the eye at one-minute intervals produced no response. If, however, the cone was slightly rotated or its position moved forward or back in the process of adjustment, occasional responses were observed. Such reactions were attributed to the stimulus of motion rather than to the influence of light, for the same degree of stimulus accompanied with no motion led to no response. 2. The eye The sensitivity of the eye was studied in detail. In the eye of the tadpoles, however, we find an apparatus adapted for receiving light stimulus. It is not to be questioned that during illumination chemical changes are occurring in the retina. In fact, if we consider the electrical changes in the retina due to light of a photochemical order, as demonstrated by Dewar and McKendrick ( '74), Waller ( 'OO), Einthoven and Jolly ('OS), we have a chemical change in the retina in a given time dependent upon the intensity of illumination and measured by the amount of electrical current produced. This has been demonstrated by all the workers on ret,inal changes. The lack of responsiveness of Rana clamitans tadpoles when the stimulus is applied only to the eye and the orderly responses when the skin is stimulated becomes clear when we consider the dif3erence in specialization and organization of these two types of receptors. The skin receptors are connected with the lower reflex centers, while the eye is a more specialized analyzer for image-forming functions. The distinction in function is made clear by Pavlov, who pointed out that in the lower centers the reflexes are un- conditional; that is, knowing the nature of the stimulus, the responses can be predicted. When the stimulus is applied to a higher, more specialized sense organ, the occurrence or a 26 1 PHOTIC REACTIONS OF TADPOLES visible response is dependent on other conditional factors. Again the failure of the animals to produce a reflex when the eyes are illuminated may be due to the inaccuracy of the visual apparatus. An example of this is found in the mollusk Pecten. In Pecten along the free edge of the mantle are situated a number of eyes, each one histologically representing an elaborate visual apparatus. Patten ('87) found in the eyes of Pecten a true cornea, lens, ciliary muscles, retinal structures, and nerve fibers, all occupying the same position as in the eye of the higher vertebrates. Uex- kull ('12) reports that even the image of an enemy, such as a starfish, is not sufficient to initiate a response in Pecten, unless the image formed in the retina is reinforced by a movement of the object. Similar observations on Pecten have been recorded by Wenrich ('16). 2. The eye The experimental evidence, therefore, under the three condi- tions: 1) eye and skin, 2) skin alone, and 3) eye alone, in terms of the amount of energy necessary to act as a stimulus as ex- hibited by the intensity-reaction-time products, shows that this amount of energy is constant. Repeated stimulations of the eye with intensities which were effective when the photoreceptors of the skin were illuminated, led to no responses. It appears therefore, that, under the conditions of the experiments, the tadpoles of Rana clamitans are stimulated through the skin and that the eyes take no part in the recorded responses to light. VI. VELOCITY OF CHANGES IN THE RECEPTORS WITH HIGHER LIGHT INTENSITIES AND THE DURATION OF SECONDARY PROCESSES FOLLOWING SENSITIZATION The steady increase in the rate of changes in the photoreceptors with the gradual increase in the intensity of illumination was demonstrated graphically when the logarithm of the intensity of light was plotted as function of the corresponding reaction-times This relation was found to be linear (figs. 2 and 4). The more exact relation between the velocity of change in the sense organs and the dependence of this on the intensity factor, as well as the exact range within which this relation holds true, can be studied by an employment of the reciprocity law as derived from the Bunsen-Roscoe law formula. 262 VASIL OBRESHKOVE The Bunsen-Roscoe law can be expressed: (1) I x T = K , where 1 stands for the intensity of light, T indicates the reaction- time, and K the constant effect produced. By transposing, this equation can be written : where the constant effect, K , produced has been considered as unity. In this form we have a reciprocal relation and for practi- cal purposes it can be interpreted to mean that the constant amount of work done divided by T, the time in which the work is done, expresses the velocity of change or the rapidity with which this is done. The greater T becomes, the slower becomes the velocity of change, and this whole relation from the nature of the equation is dependent upon the intensity of light. A graphic representation, therefore, of the data secured under the conditions: 1) sensitivity of eye and skin and, 2) skin alone, when the reciprocal of the reaction-time is plotted as function of the intensity (equation 2) produces a curve which illustrates the velocity of chemical change. This is shown in figure 6, where for curve A the data in table 2 were employed; for curve B the average reaction-times of blind tadpoles as shown in table 4 were used in addition to the results obtained with 1.2 candle-meters as recorded in table 3. It is obvious from the two curves, that the velocity of change in the receptors is approximately linear up to 15 candle-meters, above which a gradual deviation occurs. This has already been observed in the tables on intensity-reaction-time products in three sets of readings, under three different conditions (tables 1, 2, 3). VI. VELOCITY OF CHANGES IN THE RECEPTORS WITH HIGHER LIGHT INTENSITIES AND THE DURATION OF SECONDARY PROCESSES FOLLOWING SENSITIZATION In reflex actions due to light, Hecht ('18) observed in Ciona intestinalis a delay in the response after sensitization during which the animal need not be illuminated in order to respond. This secondary period he has termed the 'latent period,' and for Ciona he has assigned to it a value of 1.7 seconds. In subse- quent researches Hecht ('19 a, b) has demonstrated the striking , ) PHOTIC REACTIONS OF TADPOLES 263 phenomenon that the actual sensitization period is extremely brief, while the time occupied by secondary processes may vary between one and four seconds, depending upon the duration of illumination and temperature. Henri et Henri ('12), in study- ing the excitability of Cyclops by ultraviolet light, find that the animal responds in 0.2 to 0.1 second after cessation of illumi- Fig. 6 Curve A represents the relation of the reciprocals of the average reaction-times plotted as function of intensity when eye and skin are illuminated from above. Curve B shows the relation of the reciprocals of the reaction- times plotted as function of intensity when blind tadpoles are illuminated from the side. Fig. 6 Curve A represents the relation of the reciprocals of the average reaction-times plotted as function of intensity when eye and skin are illuminated from above. Curve B shows the relation of the reciprocals of the reaction- times plotted as function of intensity when blind tadpoles are illuminated from the side. nation. Since only the duration of illumination represents the actual time occupied for the necessary photochemical processes in the peripheral organs, the deviation in the reciprocal curves in figure 6 may be regarded as indicative of a secondary process in the sense organs, nervous system, and muscles, which as yet has not been analyzed in the study of the light reactions of tadpoIes. 264 VASIL OBRESHKOVE For, as the intensity factor increases and the reaction-time becomes short, a secondary period, for secondary processes after sensitization, of appreciable value may cause an increase in the intensity -reac tion-time products. From an0 ther standpoint it may be claimed that in the lower range of intensities the Bunsen- Roscoe lam holds good, but with higher illumination the velocity of changes taking place in the receptors is no longer proportional to the intensity factor. In such a case a proportionally longer period is required to produce the amount of stimulating substance necessary in order to activate the nerve endings. TABLE 5 TABLE 5 TABLE 5 Reaction-times in seconds of blinded Rana clamitans tadpoles to light intensities of 60, 200, and 600 candle-meters. Source of light above I INTENSITY 01” LIGHT IX CANDLE-METERS 60 2.3 3.0 2.5 3.0 3.0 2.9 2.3 3 2 2.4 (over time) 3.0 (over time) 3.1 3.0 3.2 3.0 3.1 2.9 (over time) 4.0 3.0 3.2 3.1 3.1 200 1.4 1.4 1.8 1.2 2.0 1.4 1.4 1.4 2.4 1.6 2.0 1.4 3.0 1.6 1.6 2.0 2.0 1.4 1.6 1.8 1.8 1.6 (2 over time) 1.6 500 0.7 0.7 1.0 0.7 0.7 1.6 0.6 0.8 0.8 0.8 0.8 2.4 0.8 0.6 2.0 0.6 0.6 0.6 0.8 1.0 1.8 0.6 0.8 0.8 1.8 0.8 0.6 0.8 0.7 smaller, assumed values were subtracted from the average of the reaction-times and curves for these values were also plotted. They were 0.2 second and 0.5 second. Such a treatment of the data led to no improvement in the reciprocal curves. With an assumed value of 0.5 second for secondary processes there was a slight improvement in the intensity-time products. But here also the gradual characteristic deviation with the higher illuminations was retained. This becomes clearer from an inspection of table 6. The deviation with the higher intensities Reaction-times in seconds of blinded Rana clamitans tadpoles to light intensities of 60, 200, and 600 candle-meters. Source of light above I INTENSITY 01” LIGHT IX CANDLE-METERS 60 2.3 3.0 2.5 3.0 3.0 2.9 2.3 3 2 2.4 (over time) 3.0 (over time) 3.1 3.0 3.2 3.0 3.1 2.9 (over time) 4.0 3.0 3.2 3.1 3.1 200 1.4 1.4 1.8 1.2 2.0 1.4 1.4 1.4 2.4 1.6 2.0 1.4 3.0 1.6 1.6 2.0 2.0 1.4 1.6 1.8 1.8 1.6 (2 over time) 1.6 500 0.7 0.7 1.0 0.7 0.7 1.6 0.6 0.8 0.8 0.8 0.8 2.4 0.8 0.6 2.0 0.6 0.6 0.6 0.8 1.0 1.8 0.6 0.8 0.8 1.8 0.8 0.6 0.8 0.7 Reaction-times in seconds of blinded Rana clamitans tadpoles to light intensities of 60, 200, and 600 candle-meters. Source of light above smaller, assumed values were subtracted from the average of the reaction-times and curves for these values were also plotted. They were 0.2 second and 0.5 second. Such a treatment of the data led to no improvement in the reciprocal curves. With an assumed value of 0.5 second for secondary processes there was a slight improvement in the intensity-time products. VI. VELOCITY OF CHANGES IN THE RECEPTORS WITH HIGHER LIGHT INTENSITIES AND THE DURATION OF SECONDARY PROCESSES FOLLOWING SENSITIZATION y g In searching for the duration of a period occupied by secondary processes after sensitization, during which the animal need not be illuminated in order to respond to the stimulus, the following method was employed. By increasing the intensity of light, t'he reaction-time was reduced to approximately three seconds. It was found possible to reduce further the reaction-time by employing still higher degrees of stimulation. Thus the average value obtained with 200 candle-meters was found to be 1.7 seconds and with 500 candle-meters 0.76 second. With higher intensities than 500 candle-meters, it was found difficult to measure accurately with a stop-watch the reaction-time, because of its extreme shortness. The actual values obtained with 60, 200, and 500 candle-meters are given in table 5 . If the reaction-time is composed of two periods, a sensitiza- tion period and a period for other processes following sensitization, then from the reaction-time with 500 candle-meters it is seen that 0.76 second must represent both these processes. From this it can be concluded that the duration of this secondary period must be very short and that, whatever its value is, it must lie at or below 0.76 second. If the deviation noted with the higher intensities in the intensity-reaction-time products is due to such a period, then a value lying between 0 and 0.76 second ought to be found which when subtracted from the average reaction-times for the various intensities would yield results capable of right-line plotting. That this is not the case is seen from figure 7, in which the curves were plotted by trhe same method as was employed in figure 6, except that from average PHOTIC REACTIONS OF TADPOLES 265 reaction-times was subtracted 0.7 second, an assumed value for the probable duration of secondary processes. The general course of the curves in figure 7 is the same as that shown in figure 6, and it is characterized by the same deviation in the higher intensities. If, now, the values obtained with 60, 200, and 500 candle-meters are also plotted, the region in the curves representing these intensities assumes an irregular form. Other, TABLE 5 But here also the gradual characteristic deviation with the higher illuminations was retained. This becomes clearer from an inspection of table 6. The deviation with the higher intensities 266 VASIL OBRESHKOVE is, therefore, gradual and it seems to proceed with an approxi- mate constancy. The photodynamic nature of Rana clamitans tadpoles, there- fore, differs from that of Ciona (Hecht, '18) and of Mya (Hecht, '19 a, b, '20) in that nearly all of the reaction-time in the tad- Fig. 7 Curve A represents the relation of the reciprocals of reaction-times of tadpoles with eye and skin, minus a probable secondary period of 0.7 second, plotted on the ordinates, as function of intensity, plotted on the abscissa. Curve B, reciprocals of reaction-times of blind tadpoles, minus 0.7 second, plotted on the ordinates as function of intensity of light. Fig. 7 Curve A represents the relation of the reciprocals of reaction-times of tadpoles with eye and skin, minus a probable secondary period of 0.7 second, plotted on the ordinates, as function of intensity, plotted on the abscissa. Curve B, reciprocals of reaction-times of blind tadpoles, minus 0.7 second, plotted on the ordinates as function of intensity of light. poles represents a sensitization period and that the time occupied for secondary processes seems to be extremely short. In view of the close agreement between the intensity-reaction- time products up to 20 candle-meters and the consistency of deviation with the higher illuminations, it appears that the Bunsen-Roscoe law holds true only within a certain limited 267 PHOTIC REACTIONS OF TADPOLES range. With the higher degrees of stimulation, the dependence of the velocity of change upon the intensity is retained, but there is a slight slowing down in this velocity. This condition is further illustrated by the empirical formula for the curve pre- sented early in this discussion (p. 250). The formula was found to be as follows: 94’48 + 2.85 Y = x - 0.15 in which y is the reaction-time in seconds, x, is the light inten- sity in candle-meters, and 0.15, 2.85 and 94.48 are three con- TABLE 6 Summary table, in seconds of average reaction-times, of products of reaction-times and light intensities, and of these values modi$ed by subtractions (0.5 second and O.Y second) for assumed secondary processes Reaction-time .......... I X T... ............... Reaction-time-0.5sec.172.5 I X T... ............... Reaction-time-0.7 sec. I X T... ............... 0 7 1 2 5 10 15 20 30 60 200 500 - - - - ~ _ _ - - - - 173 93.1 24.1 12.1 8.5 7.6 6.1 2.9 1 .7 0.76 121.1 111.7120.5 121 127.5 152 183 174 340 380 92.6 23.6 11.6 8.0 7.1 5.6 2.4 1.2 0.26 120.7 111.1 118 116 120 142 168 144 240 130 172.3 92.4 23.4 11.4 7.8 6.9 5.4 2.2 1.0 0.06 120.7ll0.8117 114 117 138 162 1132 200 30 ( ) y p I LIGHT INTENBITY IN CANDLE-METERS stants. This formula satisfies well the actual data obtained. The constant a of the original formula (p. 250) is equal to 0.15 in the empirical representation and means that with the lower intensities of light the Bunsen-Roscoe law does not hold true for all values of z, but that there is a certain minimum intensity of light beyond which the light produces no apparent effect upon the organism. This was demonstrated experimentally. The constant b ( = 2.85) denotes that with the higher intensities and as the degree of illumination approaches infinity, a point is reached where the reaction-times are the same regardless of the intensity of illumination. This point in the reaction-time of Rana clamitans, however, is not reached abruptly, but gradually and with a definite constancy. Reaction-time .......... I X T... ............... Reaction-time-0.5sec.172.5 I X T... ............... Reaction-time-0.7 sec. I X T... ............... 0 7 1 2 5 10 15 20 30 60 200 500 - - - - ~ _ _ - - - - 173 93.1 24.1 12.1 8.5 7.6 6.1 2.9 1 .7 0.76 121.1 111.7120.5 121 127.5 152 183 174 340 380 92.6 23.6 11.6 8.0 7.1 5.6 2.4 1.2 0.26 120.7 111.1 118 116 120 142 168 144 240 130 172.3 92.4 23.4 11.4 7.8 6.9 5.4 2.2 1.0 0.06 120.7ll0.8117 114 117 138 162 1132 200 30 I LIGHT INTENBITY IN CANDLE-METERS stants. This formula satisfies well the actual data obtained. The constant a of the original formula (p. 268 Hecht ('18, above. It is t intensities em range of illum field of light in my results are After anima to a continuo they show no taking place in a reduction of the changes p to be of insu starting point, recovery. The followi The animal to with an overh animal respon illumination o to the light sti with different 68 VASIL OBR 268 VASIL OBRESHKOVE Hecht ('18, '19 a, b) observed no such deviation as described above. It is to be noticed, however, that the ratio of the range intensities employed by him is about 1:20, while the ratio of range of illuminations employed in the present study covers a field of light intensities 1:500. With a range of intensities 1:20 my results are in accord with those obtained by Hecht. TABLE 6 250) is equal to 0.15 in the empirical representation and means that with the lower intensities of light the Bunsen-Roscoe law does not hold true for all values of z, but that there is a certain minimum intensity of light beyond which the light produces no apparent effect upon the organism. This was demonstrated experimentally. The constant b ( = 2.85) denotes that with the higher intensities and as the degree of illumination approaches infinity, a point is reached where the reaction-times are the same regardless of the intensity of illumination. This point in the reaction-time of Rana clamitans, however, is not reached abruptly, but gradually and with a definite constancy. VII. PHOTOSENSORY RECOVERY After animals which have been kept in the dark are exposed to a continuous illumination of 10 candle-meters for an hour, they show no sensitiveness to light. In terms of the changes taking place in the photoreceptors, this may be said to be due to a reduction of the photochemical substance to a minimum, and the changes produced in the peripheral sense organs thereafter to be of insufficient rate to act effectively as stimuli. As a starting point, this enables us to study the nature of photosensory recovery. The following method was employed in studying recovery. The animal to be tested was placed in a white porcelain dish and with an overhead light its sensitivity was determined. If the animal responded with regularity, it was exposed to a continuous illumination of 10-candle-meter intensity till no longer sensitive to the light stimulus. The time necessary for this process varied with different individuals from forty to sixty minutes, but for uniformity of procedure it was decided to use one hour of con- tinuous illumination. At the end of that period the animal was tested several times, and if no response was noted within two minutes from the time of the application of the stimulus, the animal was left in the dark ten minutes and its sensitivity again tested. After ten more minutes in the dark its reaction-time was again tried. Four such experiments were made with four animals, the summary results of which are given in table 7 . From an inspection of the table it can be seen that in the course of photosensory recovery, the reaction-time at first is long, then gradually decreases, and finally becomes a constant quantity. L4fter this point, repeated stimulations at ten- minute intervals in the dark revealed no very marked increase From an inspection of the table it can be seen that in the course of photosensory recovery, the reaction-time at first is long, then gradually decreases, and finally becomes a constant quantity. L4fter this point, repeated stimulations at ten- minute intervals in the dark revealed no very marked increase PHOTIC REACTIONS OF TADPOLES 269 or decrease in the reaction-time. This period is reached in fifty minutes. The condition is shown graphically in figure 8, where the data in table 7 are used, the reaction-time being plotted as the function of time in the dark. VII. PHOTOSENSORY RECOVERY The curve also p otted as t e u ct o o t e t e da . e cu ve a so TABLE 7 Summary table of average reaction-times in seconds for photosensory recovery in four tadpoles of Rana clamitans TIME IN DARK minutes 10 20 30 40 50 60 REACTION-TIME seconds 52.3 30.0 20.6 15.2 12.8 12.7 Tme indark- M~nutcs Fig. 8 Curve illustrating the process of photosensory recovery. The reac- tion-time (seconds) is plotted on the ordinates as function of time in the dark (minutes) on the abscissa.. TABLE 7 Summary table of average reaction-times in seconds for photosensory recovery in four tadpoles of Rana clamitans Fig. 8 Curve illustrating the process of photosensory recovery. The reac- tion-time (seconds) is plotted on the ordinates as function of time in the dark (minutes) on the abscissa.. 270 VASIL OBRESHKOVE illustrates the velocity of the process of recovery and the nature of the chemical changes in the receptors. In the beginning the process moves with extreme rapidity. The velocity of change in the sense organs then gradually decreases in rate, as seen in the diminution of the slope in the curve, and finally the curve in its lower course is almost parallel to the abscissa, thus indi- cating that the process of photosensory recovery has reached a state of equilibrium. From the nature of the experimental results obtained, it is evident that the reaction-time at any moment in the course of recovery is an expression of the amount of substance present in the system at that moment. The relation is inverse, for the greater the amount the shorter is the period required to produce a definite change. VIII. THE PROCESS OF FATIGUE When tadpoles are subjected to continuous illumination for a considerable time they become no longer sensitive to light. The course of the process of fatigue was studied, the method employed being similar to that used in studying the process of photosensory recovery. The individuals to be tested were placed in a porcelain dish with the source of illumination above them; whereupon their photosensitivity was tested. Only those tadpoles were used which had been kept previously in the dark and which showed normal behavior. Usually two individuals at a time were placed in the vessel. In such cases only the reaction-time of the individual to respond first was recorded. A light intensity of 10 candle-meters was employed. The individ- uals were exposed for ten minutes to continuous illumination, and at the end of that period, after one minute of darkness, the light was again applied and the reaction-time of the first individual to move was recorded. After another ten minutes of continuous illumination, the process was repeated. Only one reading was taken after the application of the stimulus. Now and then in the course of the experiments irregularities occurred in the midst of readings which otherwise showed uniformity of re- sponse. To check the significance of an occasional deviation another record was taken immediately after the diverse reading, PHOTIC REACTIONS O F TADPOLES 27 1 and the average of these two was considered to represent the true value. Almost without exception, the second reading taken agreed well with the previous uniform results. Five experiments were made, a summary of which is given in table 8. In addition to this, the data are represented graphically in figure 9, the reaction-time being plotted as a function of time g g p TABLE 8 Summary table of average reaction-limes in seconds for tadpoles of Rana clamitans in the process of fatigue. Light intensity, 10 candle-meters REACTION-TIME I TIME IN LIQHT minutes 0 10 20 30 40 8 e c o n d 8 12.0 23.4 52.3 71.0 Over time Fig. 9 Curve illustrating the course of the process of fatigue. The reaction- time, in seconds, is plotted as function of time in the light, in minutes. Fig. 9 Curve illustrating the course of the process of fatigue. The reaction- time, in seconds, is plotted as function of time in the light, in minutes. Fig. 9 Curve illustrating the course of the process of fatigue. VIII. THE PROCESS OF FATIGUE The reaction- time, in seconds, is plotted as function of time in the light, in minutes. 272 VASIL OBRESHKOVE in the light. From the table it is seen that the reaction-time is at first normal and then gradually becomes longer with the increase of continuous illumination. After thirty minutes, although the individuals mere still responsive, they showed a reaction-time suggesting that they were nearing exhaustion. With only one exception, complete fatigue occurred after forty minutes of continuous illumination. From the curve in figure 9 it is seen that the process of fatigue is characterized at first by a slow velocity of change, then by a gradual increase, then by a decrease, and finally by a sudden break, where the animals no longer responded. This is in marked contrast with the velocity of change during photosensory recovery where the rate of change in the receptors at first increases rapidly, then gradually decreases, and finally becomes constant--a pro- cess characteristic of monomolecular reaction. The process of fatigue, however, is not a simple process progressing only in one direction, but it is being opposed also by the return of a certain amount of the decomposed material to its original state, at a rate independent of the intensity of illumination. This was shown in the study of photosensory recovery. Because of this fact, therefore, the observed results furnish us with only a limited quantitative criterion for an analysis of the chemical process in the sense organs during illumination, The dependence of the velocity of change upon the active masses finds close analogy in the retinal electrical changes due to illumination. Dewar and JfcKendrick ('74) showed that the effect of continuous action of light upon the retina of the frog is marked at first by an increase in the electrical change, then by a gradual diminution, and finally by constancy. Similar phenom- ena in the electrical changes of the retina have been shown by Kuehne und Steiner ('go), Waller ('OO), and Einthoveii and Jolly ('OS), who demonstrated that, whether the currents are positive or negative, the retina becomes less and less sensitive to the light after it has been subjected to continuous illumination. Hecht ('18) showed that Ciona intestinalis is fatigued by light when the individuals are repeatedly stimulated after one- minute intervals in the dark. VIII. THE PROCESS OF FATIGUE Rana clamitans tadpoles, re- PHOTIC REACTIONS OF TADPOLES 273 peatedly stimulated at one-minute intervals, revealed no sub- stantial change in the reaction-time. The difference in behavior becomes clear when we recall that in Ciona the process of recovery requires from two to three hours, while in the tadpoles it is only fifty minutes. From this it is not difficult to see that the more rapidly an animal is able to recover decomposed material in the sense organs the less rapidly will it become fatigued. 1. The nature of the process At various points in this paper it has been stated that the process active in the sense organs of Rana clamitans tadpoles during illumination is of a photochemical kind. This conclu- sion was based on the regularity with which responses took place under measured amount of radiant energy and also upon the definite relation between reaction-time and the amount of active mass during the photosensory recovery and fatigue. That the process in the photic sense organs of animals is of a chemical character rather than of the nature of diffusion or some other physical process finds strong support in a comparative study of temperature coefficients of photochemical reactions as compared with those of physical phenomena. No attempt was made to determine the temperature coefficient of the tad- poles in their responses to light, for preliminary study revealed that the animals placed in water below 10°C. become very active and show numerous spontaneous movements at short intervals, while tadpoles placed in water above 22°C. become sluggish and unresponsive. It is a well-known fact from general chemistry, however, that the velocity of a chemical reaction increases with a rise of temperature. The temperature coefficient of a true chemical process is between 2 and 3 for a change of temperature of 10°C. In chemical changes due to light the temperature coefficient has much lower value than in chemical process proper, and it is found to lie between 1.1 and 1.2. This is evident from the fact that the rate of photochemical change where light is the stimulating agent is dependent upon the amount of light 274 VASIL OBRESHKOVE absorbed (Grotthus, 1819; Draper, '41), and this varies very little with a change of temperature. absorbed (Grotthus, 1819; Draper, '41), and this varies very little with a change of temperature. g p Henri et Henri ('12) find that the excitability of Cyclops by ultraviolet light is independent of temperature. Hecht ('19), in studying the photodynamic nature of Mya with respect to the amount of energy received and the length of the reaction- time, assigns a value for the temperature coefficient of 1.1 for a change of 10°C. These two studies are the only ones to my knowledge on the temperature coefficient of animals responsive to light. 1. The nature of the process The magnitude of the temperature coefficient is in close accord with those of chemical reactions due to light, thus affording strong evidence that the sensitivity of animals to light is of a photochemical kind. Osterhout ('17) points out that a process which has a temperature coefficient as high as 2 cannot be considered of physical nature. For photosynthe- sis, he assigns a value of 1.7 for the temperature coefficient. This slightly higher value, he holds, is evidence that the process of photosynthesis is of a combined nature, photochemical and chemical. In the light of the observations made in the study of the photodynamic nature of the responses of Rana clamitans tad- poles, namely, that, with a constant intensity of illumination at one-minute intervals in the dark, the reaction-time shows no marked variation from a value which is also constant, we have a suggestion that the changes in the receptors are of an orderly type. Whatever these changes are, they must involve a photo- chemical substance, normally present in the receptors, a part of which undergoes a change to a different state under the in- fluence of light. For simplicity, at this point of the discussion this may be expressed by assuming a change from A to B, where A is chemically different from B and only in this last state capable of acting as a stimulus for the nerve endings with which it is in immediate contact. Such a hypothetical representation is usually employed for the measurement of photochemical changes in chemistry, and it is variously modified by Weigert ('11) and Sheppard ('14) to meet the requirements of the order of photochemical reactions which they have classified. Two PHOTIC REACTIOSS O F TADPOLES 275 such alternative photochemical systems have been employed by Hecht ('18, '19a, b, 1920) in a discussion of the nature of the photosensitivity of Ciona and hfya. T A E JOTRSAL OF EXPERIMEATAL ZO6LOGY. VOL. 31, KO. 2 2. The photoreceptors The isolation of skin photoreceptors for histological examina- tion has met with difficulty by all workers who have undertaken it. No progress has been made other than regional localization of such organs. It is not difficult, however, to imagine a struc- tural condition for photoreception such as described by Polara ('06). In Holothuria poli he describes pigment cells in close association with nerve terminations. These pigment cells, he suggests, are a part of a sensory apparatus for the reception of light. As a result of continuous illumination, the pigment cells are decolorized and associated with this process, there is a loss in the photosensitivity of the animals, In the darkness pigment reappears and with it photosensitivity is regained. Similar observations have been made by Crozier ('14) in Holo- thuria surinamensis and H. captiva. A fluorescent pigment, confined to the superficial layer of the integument, showed ab- sorption in the blue-green band of the spectrum, thus suggesting that the process of decolorization is of a chemical nature. The photoreceptors in the tadpoles must be so situated in the integument as to be partly or entirely concealed when the melanophores are in an expanded condition. Tadpoles dark in appearance either did not respond when light was applied or showed a reaction-time totally different from tadpoles light in appearance. This behavior was studied in detail and with considerable care. Before the facts of this situation were known, much difficulty was experienced in the experimental work. For this reason, after certain preliminary trials, only tadpoles of the same degree of coloration were used. Photoreceptors in animals are analyzers for appreciating changes in light and darkness. When the animals are living in their natural habitats this type of sense organs must play an appreciable r81e. In species which respond only to a decrease of 276 VASIL OBRESHKOVE light intensity, the appreciation of a shadow may serve as a warning of the approach of an enemy. In species capable of responding to an increase of intensity, the appreciation of the change of illumination must also be of value. It is evident, therefore, that animals better equipped for judging light and darkness and for responding to such stimuli, are better adapted for their life struggle. The more sensitive the receptors are, the more value they must have for the individuals. y In reflexes due to light we recognize two important periods. PHOTIC REACTIONS OF TADPOLES PHOTIC REACTIONS OF TADPOLES 277 2. The photoreceptors These are a sensitization period, during which energy is received for the chemical processes necessary in the sense organs, and a secondary period which is occupied by such processes as the diffusion of chemical products to the nerve endings, transmission of the impulse, and a contraction of the muscles of the organ involved in the response. When the time taken up by the secondary processes is considerable, it must be deducted from the reaction-time in determining the actual duration of the preceding chemical process. In Sudging the relative photosensitivity of animals, a separa- tion of the reaction-time into its two phases, with the assignment of a value for each, is important. Animals, however, whose photoreceptors are characterized by the quickness with which they can execute the chemical processes necessary for the stimu- lation of nerve endings, but also by slowness in the secondary processes, gain little from the quickness with which the first process is executed. In such animals the duration of both processes must be considered in judging photosensitivity, The Bunsen-Roscoe law furnishes us with an important quan- titative means in testing the photosensitivity of animals. By extending this type of experimentation to a large number of animals, it will be possible to gain important insight into the nervous mechanism of these forms. The intensity-time prod- ucts of three such animals have already been recorded: Ciona intestinalis (Hecht, 'IS), Mya arenaria (Hecht, '19 a, b, '20), and Rana clamitans. It is hoped that this work will be further extended. XI. BIBLIOGRAPHY BLAAUW, A. H. 1909 Die Perzeption des Lichtes. Rec. trav .bot. N6erlandais, Bd. 5, S. 209-377. BUNSEN, R., AND ROSCOE, H. E. 1862 Photo-chemical researches. Part 5. On the direct measurement of the chemical action of sunlight. Phil. Trans., London, vol. 153, pp. 139-160. COLE, W. H., AND DEAN, C. F. 1917 The photokinetic reactions of frog tad- poles. Jour. Exp. Zool., vol. 23, pp. 361-370. p p pp CROZIER, W. J. 1914 Orientation of Holothuria by light. Amer. Jour. Physiol., V O ~ . 36, pp. 8-10. DEWAR, J., AND MCKENDRICK, J. G. 1874 On the physiological action of light. Trans. Roy. SOC., Edinburg, vol. 27, pp. 141-142. DRAPER, J. W. 1841 On some analogies between the phenomena of the chemical rays and those of radiant heat. Phil. Mag., London, vol. 19, pp. 195-210. DUBOIS, R. 1890 Sur la perception des radiations lumineuses par la peau, chez les Protees aveugles des grottes de la Carniole. Compt. rend. Acad. Sci., Paris, T. 110, pp. 35s-361. 1908 The form and magnitude of the elec- trical response of the eye to stimulations by light at various intensi- ties. EINTHOVEN, W., AND JOLLY, W. A. Quart. Jour. Exper. Physiol., vol. 1, pp. 373416. FROSCHEL, P. 1909 Untersuchung uber die heliotropische Prasentationszeit. Sitzb. k Akad. Wiss., Wien, math-nat. Cl., Bd. 118, Abt. 1, S. 1247- 1294. GRABER, V. 1884 Grundlinien zur Erforschung des Helligkeits und Farben- sinnes der Tiere. Leipzig. GROTTHUS, T. YON 1819 Ueber die chemische Wirksamkeit des Lichtes und der Elektrizitat. Reprinted in Ostwald’s ‘Klassiker,’ No. 152. HECHT, S. 1918 The photic sensitivity of Ciona intestinalis. Jour. gen. Physiol., vol. 1, pp. 147-166. 1919 a Sensory equilibrium and dark adaptation in Mya arenaria. Jour. gen. Physiol., vol. 1, pp. 5-14-558. 1919 b The effect of temperature on the latent period in the photic response of Mya arenaria. 1920 The photochemical nature of the photosensory process. Jour. Gen. Physiol., vol. 2, pp. 22Cb246. Jour. gen. Physiol., vol. 1, pp. 667-685. HENRI, MME., ET HENRI, V. 1912 Excitation des organismes par lesrayons ultra-violet. Compt. rend. Soc. Biol., Paris, T. 72, pp. 992-996. HENRI, V., ET LARGUIER DES BANCELS, J. 1912 Un nouveau type des temps de r6action. Compt. rend. SOC. Biol., Paris, T. 73, pp. 55-56. KUEHNE, W., UND STEINER, J. 1880 Ueber das electromotorische Verhalten der Netzhaut. Unters. physiol. Inst., Heidelberg, Bd. 3, S. 327-377. LASAREFF, P. X. CONCLUSIONS 1. There is an optimum intensity for the responses in Rana clamitans tadpoles to light. p g 2. At and below 0.3 candle-meter intensity, the light ceases to have a physiological effect, regardless of the time of exposure. 2. At and below 0.3 candle-meter intensity, the light ceases to have a physiological effect, regardless of the time of exposure. 3. With effective light intensities below 20 candle-meters, the changes in the receptors during illumination proceed accord- ing to the Bunsen-Roscoe law. g 4. With intensities higher than 20 candle-meters, a deviation occurs in the intensity-time products which seem to proceed with a definite constancy. y 5. Nearly the whole of the reaction-time in Rana clamitans represents a sensitization period. 6. The eyes are not necessary for the responses of tadpoles to light of the kind used in these experiments. 7. Tadpoles subjected to continuous illumination of definite duration become no longer sensitive to light. g g 8. In the process of fatigue, the reaction-time at any moment has a definite relation to previous illumination. 9. Photosensory recovery after complete exhaustion occurs in the dark in about fifty minutes. 278 VASIL OBRESHKOVE XI. BIBLIOGRAPHY 1907 Ueber das Ausbleichen von Farbstoffen im sichtbaren Spektrum. Ann. Physik, Bd. 24, S. 661-671. LAURENS, H. 1914 The reactions of normal and eyeless amphibian larvae to light. Jour. Exp. Zool., vol. 16, pp. 195-210. 279 PHOTIC REACTIONS OF TADPOLES LOEB, J., AND NORTHROP, J. H. 1917 Heliotropic animals as photometers on the basis of the validity of the Bunsen-Roscoe law. Proc. Nat. Acad. Sci., vol. 3, pp. 539-544. pp 1917 A re-examination of the applicability of the Bunsen-Roscoe law to phenomena of animal heliotropism. Jour. LOEB, J., AND WASTENEYS, H. Exp. ZoOl., VOI. 22, pp. 187-192. MARCELL DONCHEFLDEZENSE 1914 L’image et les reflexes conditionnels dans les traveaux de Pavlov. Paris, 1914. p , pp OSTERHOUT, W. J. V. 1917 Does the temperature coefficient of permeability indicate that it is chemical in nature. Bot. Gaa., vol. 68, pp. 317-320. PARKER G H 1903 Th ki d h i i h i PARKER, G. H. 1903 The skin and the eyes as receptive organs in the reactions of frogs to light. Amer. Jour. Phvsiol., vol. 10, pp. 28-36. PATTEN, W. 1887 The eyes of molluscs and arthropods. Jour. Morph., vol. 1, pp. 67-89. PEARSE, A. S. 1910 The reactions of amphibians to light. Proc. Am. Acad. Arts and Sci., vol. 45, pp. 161-208. POLARA, G. 1906 Sulla connessione delle cellule pigmentate del mesenchima cutaneo dell olothurie con i nervi cutanei e sul loro significato fun- zionale. Boll. Accad. Gioenia, Catania, fasc. 88, pp. 12-24. 1906 Observations on the reactions of Cryptobranchus and Necturus to light and heat. Biol. Bull., vol. 11, pp. 93-99. REESE, A. M. g , SHEPPARD, S. E. 1914 Photochemistry. New York. UEXKULL, J. VON 1912 Studien iiber den Tonus. VI. Die Pilgermuschel. Zeitschr. f. Biol., Bd. 58, S. 305-332. WALLER, A. D. 1900 On the retinal currents of the frog’s eye, excited by light and excited electrically. Phil. Trans., London, vol. 193, pp. 123-163. y pp WEICERT, F. 1911 General theory of photochemical reactions. New York. WEICERT, F. 1911 General theory of photochemical reactions. New York. WENRICH, D. H. 1916 Reactions of bivalve mollusks. Jour. Anim. Behavior, V O ~ . 6, pp. 297-315. y p WENRICH, D. H. 1916 Reactions of bivalve mollusks. Jour. Anim. Behavior, V O ~ . 6, pp. 297-315.
https://openalex.org/W3112331495	https://karolinum.cz/data/clanek/8646/Kinan_56_2_0134.pdf	English	null	Practical ways for coaches to reduce their stress and avoid burnout	Acta Universitatis Carolinae. Kinanthropologica	2,020	cc-by	4,844	134 134 AUC Kinanthropologica, Vol. 56, No. 2, pp. 134–142 ABSTRACT Stress and the potential for burnout are a very real and present concern for coaches at all levels of compe- tition. However, while stress is ever-present, many coaches have never received education or training on how to effectively cope with experienced stress. Therefore, the purpose of this article is to present practical ways for coaches to reduce stress and avoid burnout. A brief review of stress, burnout, and the stressors coaches often face are presented. This is followed by methods to help mitigate stress and avoid burnout including best practices for physical activity, meditation, connecting with mentors, connecting with fam- ily, renewal, setting limits, saying no, continuing education, and some unexpected methods to reduce stress. All coaches will experience stress, and its recognition, combined with the skills and tools needed to moderate its effects aid in longer and more successful coaching careers. Practical ways for coaches to reduce their stress and avoid burnout Robert Pearson, Timothy Baghurst* Florida State University, Tallahassee, USA Florida State University, Tallahassee, USA * Corresponding author: tbaghurst@fsu.edu * Corresponding author: tbaghurst@fsu.edu KEYWORDS coaching; coach; stress management; sleep; meditation; exercise Practical ways for coaches to reduce their stress and avoid burnout y Wake up, coach, plan training, teach, complete other assigned duties and tasks, sleep a few hours, and repeat. This, or a similar pattern, may be familiar to coaches; the grind of a daily routine without balance repeatedly exposes them to stressors that, unless moderated in some way, can lead to burnout (Pearson et al., 2020). Research in the past decade has determined burnout is more than an emotional state; it is an employee’s response to job stressors. These stressors, especially through long-term exposure, result in decreased productivity, feelings of role ambiguity and conflict, job overload, and emotional exhaustion (Purvanova & Muros, 2010; Sisley, 1987; Vladut & Kallay, 2010). Therefore, if a coach is experiencing levels of stress that cannot be balanced or moderated, burnout is likely to occur.t Coaches are highly visible members of the community and often face job perfor- mance demands not found in other professions (Schroeder, 2010; Van Mullem, 2015). Winning or losing is a very visible performance indicator of success, and when com- Coaches are highly visible members of the community and often face job perfor- mance demands not found in other professions (Schroeder, 2010; Van Mullem, 2015). Winning or losing is a very visible performance indicator of success, and when com- © 2020 The Authors. This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, © 2020 The Authors. This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Practical ways for coaches to reduce their stress and avoid burnout 135 Practical ways for coaches to reduce their stress and avoid burnout bined with balancing the vast array of job requirements associated with the coach- ing profession, creates a stressful work environment (Rundle-Thiele & Auld, 2009). Among other duties coaches may serve as a substitute parent, disciplinarian, academic tutor, tactician, mentor, and friend (Davis, 2005). Coaches can be challenged to aid transitioning young adults into a new athletic and academic environment, and perhaps even a new culture or language when working with international athletes (Baghurst et al., 2018). Coaches are also charged with creating group cohesion, a sense of purpose, and developing the life skills of their athletes (Saavedra, 2013). Practical ways for coaches to reduce their stress and avoid burnout These are just a few of the many behaviors, skills, and characteristics expected of those in the coaching profession that must be balanced with winning. Employment as a coach is tenuous at best, and the job expectations of the profes- sion may lead to stress and eventually burnout (Kelley & Baghurst, 2009; Tashman et al., 2010). Stress occurs when the demands of the job or a given situation exceed perceived capabilities and resources for an individual to be successful (Hjälm et al., 2007). Extreme stress, or repeated exposure to stressors, can lead to psychological, physical, and sociological concerns that in turn lead toward burnout. For example, psychological consequences include emotional exhaustion, frustration, reduced per- sonal accomplishment or disillusionment, and negative feelings and beliefs (Freuden- berger, 1974; Maslach & Jackson, 1982; Sonnentag & Jelden, 2009; Stickle & Scott, 2016). Physical ramifications such as fatigue, heart attacks, strokes, ulcers, substance abuse, and increased morbidity may also occur (Frey, 2007; Mazerolle et al., 2008). Socially, the coach may experience negative consequences such as interference with daily living, lack of a positive work-life balance, and difficulty with interpersonal rela- tionships (Olusoga et al., 2009; Tekano et al., 2011). Stress and burnout in coaches is well-documented throughout a variety of sports (Olusoga et al., 2019). With repeated stress, or extreme bouts of stress, and without the appropriate resources to cope with the stress, a coach may leave the profession (Anderson, 2010; Bradford & Keshock, 2009; Coy & Masterson, 2007; Kelley & Ba- ghurst, 2009). Therefore, to battle stress, a coach must make a concerted effort to schedule time toward stress reduction and control (Stickle & Scott, 2016). The nature of the coaching profession is fluid and requires coaches to confront issues as they arise. This means taking time from a current task to battle a new crisis, whether an athlete walks into the office with a personal problem, or a contest needs to be re- scheduled because of weather. Therefore, coaches must consider the time necessary to combat stress as sacred and not to be interrupted, or stress may continue to build. p y Given that stress is ever-present within the coaching profession, and coaches across all demographics and experience levels report experiencing stress of different kinds (Pearson et al., 2020), it is important to consider what and how stress coping mecha- nisms could be adopted by coaches to help mitigate these stressors. Practical ways for coaches to reduce their stress and avoid burnout Therefore, based on previous research and application, and presented in no specific order, we offer a variety of guidelines and suggestions that coaches can utilize to establish a personal tool kit of stress reduction techniques. Meditation Coaches must take time away from everyone and everything. Meditation, or the use of relaxation techniques, can be used to reduce stress. Long-term mediation has been found to improve cognitive function and improve the body’s physiological relaxation as measured by heart rate and galvanic skin response (Singh et al., 2012). If new to mediation, find a place that is comfortable and calming. An office could suffice if there are no interruptions. Listening to a guided meditation tape, soothing music, or a guided app can be helpful. Headspace is a popular app that provides a va- riety of relaxation exercises. To meditate, focus on posture, being comfortable, and breathing. Losing “focus” is normal, but efforts should be made to return to the task at hand, which is to take a break from daily living and working. Mediation can help avoid “future fret” (Small & Vorgan, 2019). Meditation helps with decision making and reduces worry about an unknown future (Stahl & Goldstein, 2019). Exercise Exercise can be a valuable moderator of stress in coaching (Frey, 2007). It is interest- ing that very few empirical articles report coaches using exercise or physical activity Robert Pearson, Timothy Baghurst 136 as a buffer of stress, yet exercise has been found to be very valuable in alleviating stress (Lawrence, 2005). Although age and injury may affect how much or how hard a coach exercises (Baghurst & Diehl, 2016), regular physical exercise even in the form of walking or the use of an elliptical provides physiological and psychological benefits (Michaelis, 2013). Many sports are conducted indoors, and coaches likely spend considerable time in indoor offices. Therefore, coaches may be susceptible to a vitamin D deficiency, which has also been linked to stress levels (Lansdowne & Provost, 1998). Therefore, outdoor activity might be even more beneficial at reducing stress (Olafsdottir et al., 2017). Coaches should consider walks away from the training/competition facility, or consider engaging in low-impact outdoor activities such as yoga or tai chi. Connect with familyhl The conflict between work and family directly or indirectly affects a large portion of the global population (Kossek & Lee, 2017). Even those who are single will experi- ence stress associated with family life, as they may be sons or daughters or are part of a group of friends that functions as a family (Kossek & Lee, 2017).l Coaches are no exception to this work/family conflict. Time management is one of the most difficult challenges of a coach, and sometimes time is given to work over oth- er commitments, which leads to stress and burnout (Pearson et al., 2020). Therefore, it is important to make committed efforts to connect with family on a regular basis. Similar to connecting with a mentor, connecting with family can bring immediate improvement to mood states (Karakas et al., 2004). It may be one of the greatest joys in life and helps to build connections among and between generations toward a more shared future (Murakami & Jacobs, 2017). Connecting deeply with a spouse, partner, family member, or friend helps to relieve stress in many ways, including sharing the challenges and burdens that might be being experienced. The feeling of profound connectedness with someone is calming (Eliuk & Chorney, 2017). So, whether face-to-face, on the phone, the computer, or by any other means, coaches should make efforts to connect regularly with family members, especially if they are those who can aid in reducing experienced stress. Connect with a mentor Whether formal or informal, a mentor can help. Although not a daily activity, convers- ing with a mentor can help serve as a motivator and an impartial advisor (Sciarappa, 2010). In their study, Christie Jr. and Baghurst (2017) reported mentors provided assis- tance with personal and professional success, breaking through barriers, and assisted in developing leadership skills. Further, a strong correlation exists between those in- volved in a mentorship experience and a heightened sense of well-being (Kutsyuruba et al., 2019). Kutsyuruba and colleagues also reported that those who did not connect with a mentor revealed a significantly lower sense of well-being. gi y g Time is a precious commodity in the workplace (Davies, 2014). Past research spe- cific to those employed in the field of education found striking a balance between personal and professional duties is a difficult task (Lester, 2015). However, if a mentor strongly supports a mentee’s healthy work-life balance, the less likely they are to expe- rience associated stress (Nielson et al., 2001). It is important to recognize that mentors may also be busy and experiencing stress themselves. Therefore, be cognizant of how often and when conversations take place. Recognize that within a scale that values time, a mentor’s time is more valuable than that of the mentee’s. Therefore, find ways to show appreciation for the mentor giving Practical ways for coaches to reduce their stress and avoid burnout 137 up that time. Sending a handwritten note of thanks or a small gift goes a long way to ensure that the value is appreciated. i One coach, cited by Baghurst (2020), summarized why finding a mentor and net- working was important. Make an honest effort in getting to know other coaches, especially those that are veterans and have been around awhile. Most coaches will be open to you and give advice and ideas. It will also help you to network and be more involved in your profession. Some of the best tips, advice, scouting reports, and practice procedures came from veteran coaches. Do not be scared or embarrassed to ask for some help (p. 544). Prioritize and set limits It can be easy for coaches to take on too many responsibilities (Lyst, 2019). However, it is important coaches set realistic limits of what you can accomplish (Valcour, 2016). Stress is created by expectations, whether personal or those placed on an individual by others, and is a key element in coaching stress (Radke, 2012). Coaches should consider using day planners or task lists to get through the day. Calendar software or apps can also help to establish schedules and routines, including times for breaks (e.g., relaxation/meditation). Prioritizing tasks also assist in achiev- ing work goals efficiently. Baghurst (2019) presented the suggestion of using the Eisenhower Box to make decisions. Decisions are assigned as Urgent/Important (e.g., meeting a project guideline), Important/Not Urgent (e.g., family time), Not Important/ Urgent (e.g., booking a hotel), and Not Important/Not Urgent (e.g., using social media). Developing a plan and prioritizing accordingly will assist in time management, effi- ciency, and ultimately lower levels of stress. 138 Robert Pearson, Timothy Baghurst Renewal Coaches should take time to rest and renew (Valcour, 2016). Vacation time should be used, not excused. For many coaches, this may prove difficult, especially if coaching multiple sports in a school or club setting or recruiting in the off-season. However, delegation of duties to others such as an assistant on occasion creates a time window and also provides the assistant with opportunity for growth. Sleep is also an important component of renewal. Even a single night of sleep dep- rivation may result in poor decision-making and impact vigilance and motor skills (Stojanoski et al., 2019). Lastella and colleagues (2017) measured the sleep quantity and quality of a coach compared to his athletes and reported that the coach obtained less sleep with poorer quality than his athletes, markedly so before significant games. Rest and renewal serve to reduce exposure to stressors and help avoid burnout (Son- nentag & Jelden, 2009). Spending some time on a favorite hobby can also reduce stress (Sabo, 2011). Ex- amples of such hobbies include, but are certainly not limited to reading, listening to music, knitting, painting or other art projects, playing cards, board games or computer games, or watching a movie. Caution should be given to electronic forms of relaxation such as TV shows, movies, or computer games, as they can be addictive. For example, Carlston et al. (2006) reported online gaming as addictive as gambling; therefore, if selecting a form of relaxation that could be addictive, set boundaries and limits. What- ever activity chosen, it should bring pleasure and reduce stress. Say no y If tasks have been prioritized tasks and limits set, saying no to extra tasks or assign- ments is important, especially if not part of the job description. Saying yes can come at a cost of time and additional stress. A coach may feel pressured into saying yes too frequently, especially if the coach is new to the position or is experiencing job insecu- rity. In addition, saying no might create feelings of worry as it pertains to a relationship with a superior, colleague, or even recruit. However, saying no is an important word to accept and use (Ury, 2007). Practicing saying no may be of benefit. “I just can’t commit to that right now”, “I would love to help, but I have other priorities”, or “Could you ask me again in a week’s time? I am right in the middle of another project” are all example responses that could be offered. Regardless of the phrase used, learning to say no should be re- quired of all coaches, and may be one of the most effective ways for leaders to shape their program or organization (Jamison & Neubauer, 2019). Unexpected ways to alleviate stress A literature search of stress reduction methods for coaches revealed some unexpected methods that may or may not be beneficial based on the individual preferences of the coach. For example, research suggests that chewing gum has been found to lower stress, anxiety, and reduce fatigue (Smith et al., 2012). However, long-term chewing or chomping gum may result in jaw discomfort (Farella et al., 2001). Laughter and humor have been found to reduce stress (Yoshikawa et al., 2018), and viewing a vid- eo of a stand-up routine or reading the comics section of the newspaper will aid in relaxation and mood improvement. A final unexpected method to reduce stress is to watch fish swim in an aquarium (Cracknell et al., 2015). Viewing the animals swim, especially if the population of fish is larger, was found to increase a sense of well-being. Ultimately, there are a variety of ways to reduce stress, but each individual should find ways that work for them. CONCLUSION The consequences of stress in the coaching profession can be severe. Stress exacts a psychological and physical toll on the body that includes a variety of personal, phys- ical, and psychological effects (Bryant, 1992; Ferraro & Nuriddin, 2006; Khan, 2011; Kim et al., 2011; Olusoga et al., 2009; Tekano et al., 2011). Stress also adversely affects decision-making skills and quality of life (Ferraro & Nuriddin, 2006). f When, not if, a coach experiences stress, it may negatively affect work demeanor, attitude, and performance. Knowing stress exists in the coaching profession is the first step to confronting the dangers. Presented were a variety of methods to reduce levels of stress, but it should be noted that what works for each coach will differ. However, coaches should understand that all coaches will experience stress, even if the type that each might experience differs (Pearson et al., 2020). Therefore, coaches must develop stress reduction strategies that mitigate stress, or burnout will almost certainly occur. Continuing educationh The more competent an individual feels about the task at hand, the lower the level of stress experienced (Aldrup et al., 2017). Therefore, coaches should seek oppor- tunities to improve their skillset. Attending clinics, enrolling in academic programs, or even reading or watching educational content aids in developing new knowledge and ideas. Some universities offer courses online without fees, or others offer them for a more formal certificate. For example, the FSU COACH Center (n.d.) at Florida State University offers coaching certifications as well as free online educational con- tent via YouTube (www.youtube.com/fsucoach). Of course, coaches must ensure that Practical ways for coaches to reduce their stress and avoid burnout 139 there is sufficient time to achieve long-term educational commitments. Completing an academic credential can be stressful in and of itself (Hirsh et al., 2020); therefore, support systems should be built-in to aid in ensuring that stress levels do not exceed the resources in place to manage them. REFERENCES Aldrup, K., Klusmann, U., & Lüdtke, O. (2017). Does basic need satisfaction mediate the link between stress exposures and well-being? A diary study among beginning teachers. Learn- ing and Instruction, 50, 21–30. Aldrup, K., Klusmann, U., & Lüdtke, O. (2017). Does basic need satisfaction mediate the link between stress exposures and well-being? A diary study among beginning teachers. Learn- ing and Instruction, 50, 21–30. g Anderson, D. L. (2010). Organization development: The process of leading organizational change. Thousand Oaks, CA: Sage Publications. Baghurst, T. (2019). Learning from coaches. In: T. Baghurst (Ed.), Fundamentals of coaching for sports performance. London: Routledge. Baghurst, T. (2019). Lifestyle wellness. In: T. Tapps, B. Dorrell, M. Symonds, & T. Baghurst (Eds.), Health and wellness: A practical workbook. Dubuque, IA: Kendall Hunt. h Baghurst, T., & Diehl, B. (2016). Shaping up the profession: The importance of modeling health and fitness as coaches. International Sport Coaching Journal, 3(3), 349–354. Robert Pearson, Timothy Baghurst 140 Baghurst, T., Fiaud, V., Tapps, T., & LaGasse, A. (2018). Considerations when coaching the international athlete. International Journal of Kinesiology in Higher Education, 2(3), 76–86. Bradford, S. H., & Keshock, C. M. (2009). Female coaches and job stress: A review of the literature. College Student Journal, 43(1), 196–199. g Bryant, W. K. (1992). Human capital, time use, and other family behavior. Journal of Family and Economic Issues, 13(4), 395–405. Carlston, D. L., Baghurst, T., Hugman, S., Johnson, L., & Snyder, A. (presented 2006, March). Online gaming: Addictive quality among young adults. Presentation at 51st Annual Meeting, Southwestern Psychological Association, Memphis, TN. (Regional). Christie Jr., E., & Baghurst, T. (2017). College mentoring: Alumni views on programme efficacy in shaping leadership ability. International Journal of Evidence Based Coaching and Mentoring, 15(1), 169–185. g Coy, B., & Masterson, G. (2007). Hiring a winning coach and coping with stress. Coach and Athletic Director, 76(9), 32–35. Cracknell, D., White, M. P., Pahl, S., Nichols, W. J., Depledge, M. H. (2016). Marine biota and psychological well-being: A preliminary examination of dose-response effects in an aquarium setting. Environment and Behavior, 48(10), 1242–1269.h Davies, N. (2014). The work-life balance. Nursing Standard, 29(2), 63. Davis, P. G. (2005). Performance appraisal for coaches. Paper presented at the ICCE Coaching Conference Hong Kong, China. Eliuk, K., & Chorney, D. (2017). Calming the monkey mind. International Journal of Higher Education, 6(2), 1–7.f Farella, M., Bakke, M., Michelotti, A., & Martina, R. (2001). REFERENCES Effects of prolonged gum chew- ing on pain and fatigue in human jaw muscles. European Journal of Oral Sciences, 109(2), 81–85. Ferraro, K. F., & Nuriddin, T A. (2006). Psychological distress and mortality: Are women more vulnerable? Journal of Health and Social Behavior, 47(3), 227–241.f f Freudenberger, H. J. (1974). Staff burnout. Journal of Social Issues, 30(1), 159–165. Frey, M. (2007). College coaches’ experiences with stress: Problem solvers have problems, too. The Sport Psychologist, 21(1), 38–57. h g FSU COACH. (n.d.). FSU Coach. Retrieved from https://fsu-coach.fsu.edu. Hirsch, J. D., Nemlekar, P., Phuong, P., Hollenbach, K. A., Lee, K. C., Adler, D. S., & Morello, C. M. (2020). Patterns of stress, coping and health-related quality of life in doctor of phar- macy students. American Journal of Pharmaceutical Education, 84(3), 7547. y J f Hjälm, S., Kenttä, G., Hassménan, P., & Gustafsson, H. (2007). Burnout among elite soccer coaches. Journal of Sport Behavior, 30(4), 415–427.i Jamison, L. S., & Neubauer, B. (2019). Leadership lessons from the field: Leading by saying “No.” Military Medicine, 185(1–2), 12–13. https://doi-org.proxy.lib.fsu.edu/10.1093 /milmed/usz273. Karakas, F., Lee, M. D., & MacDermid, S. M. (2004). A qualitative investigation into the meaning of family well being from the perspective of part time professionals. Equal Oppor- tunities International, 23(1–2), 57–77. Kelley, B. C., & Baghurst, T. (2009). Development of the coaching issues survey. The Sports Psychologist, 23(3), 367–387. Khan, S. (2011). Relationship of job burnout and type-A behavior on psychological health among secretaries. International Journal of Business and Management, 6(6), 31–38. Kim, H., Ji, J., & Kao, D. (2011). Burnout and physical health among social workers: A three- year longitudinal study. Social Work, 56(3), 258–268.ll Kossek, E. E., & Lee, K. H. (2017). Work-family conflict and work-life conflict. In: Oxford Research Encyclopedia of Business and Management. Oxford University Press. Practical ways for coaches to reduce their stress and avoid burnout 141 Kutsyuruba, B., Godden, L., & Bosica, J. (2019). The impact of mentoring on the Canadian early career teachers’ well-being. International Journal of Mentoring and Coaching in Edu- cation, 8(4), 285–309. Lansdowne, A., Provost, S. (1998) Vitamin D3 enhances mood in healthy subjects during win- ter. Psychopharmacology, 135(4), 319–323. Lastella, M., Roach, G. D., Halson, S. L., Gore, C. J., Garvican-Lewis, L. A., & Sargent, C. (2017). Sleep at the helm: A case study of how a head coach sleeps compared to his team. REFERENCES International Journal of Sports Science & Coaching, 12(6), 782–789. Lawrence, D. (2005). The complete guide to exercising away stress. A & C Black Publisher Lawrence, D. (2005). The complete guide to exercising away stress. A & C Black Publishers, 160 p. Lester, J. (2015). Cultures of work-life balance in higher education. Journal of Diversity in Higher Education, 8(3), 139–156. Lester, J. (2015). Cultures of work-life balance in higher education. Journal of Diversity in Higher Education, 8(3), 139–156. Lyst, R. (2019). Coaching collegiate sports. In: T. Baghurst (Ed.), Fundamentals of coaching for sports performance. New York: Routledge. h Maslach, C., & Jackson, S. E. (1986). The Maslach Burnout Inventory (2nd ed.). Palo Alto, CA: Consulting Psychologist Press.l Mazerolle, S. M., Bruening, J. E., & Casa, D. (2008). Work-family conflict part 1: Antecedents of work-family conflict in national collegiate athletic association division I-A certified athletic trainers. Journal of Athletic Training, 43(5), 505–512. Michaelis, B. (2013). Your next big thing: Ten small steps to get moving and get happy. Wolf Street Press. Murakami, K., & Jacobs, R. L. (2017). Connecting dots: Family reminiscence. In: Å. Mäkitalo, P. Linell, & R. Säljö (Eds.), Advances in cultural psychology: Constructing human develop- ment. Memory practices and learning: Interactional, institutional and sociocultural perspec- tives (pp. 293–315). Information Age Publishing. Nielson, T. R., Carlson, D. S., & Lankau, M. J. (2001). The supportive mentor as a means of reducing work-family conflict. Journal of Vocational Behavior, 59(3), 364–381. Olafsdottir, G., Cloke, P., Vögele, C. (2017). Place, green exercise and stress: An exploration of lived experience and restorative effects. Health & Place, 46, 358–365. Olusoga, P., Bentzen, M., & Kentta, G. (2019). Coach burnout: A scoping review. Internation- al Sport Coaching Journal, 6(1), 42–62. Olusoga, P., Butt, J., Hays, K., & Maynard, I. (2009). Stress in elite sports coaching: Identify- ing stressors. Journal of Applied Sports Science, 21(4), 442–459. Pearson, R. T., Baghurst, T., & Mwarumba, M. (2020). Stress and burnout experienced by intercollegiate swimming head coaches. International Sport Coaching Journal, 1–7. https:// doi.org/10.1123/iscj.2019-0030. g j Purvanova, R. K., & Muros, J. P. (2010). Gender differences in burnout: A meta-analysis Jour- nal of Vocational Behavior, 77(2), 168–185. f Radke, K. (2012). Be your best without the stress: It’s not about the medal. Motivational Press.h Rundle-Thiele, S., & Auld, C. (2009). Should I stay or should I go? Retention of junior sport coaches. REFERENCES Annals of Leisure Research, 12(1), 1–21.fh f Saavedra, L. K. (2013). Effective team building: The role of coaches. Strategies, 26(4), 3–6 Sabo, B. (2011). Reflecting on the concept of compassion fatigue. The Online Journal of Issues in Nursing, 16(1).h Schroeder, P. J. (2010). Changing team culture: The perspective of ten successful head coach- es. Journal of Sport Behavior, 33(1), 63–88. f Sciarappa, K. (2010). Mentor connections. Principal, 90(1), 36–38. Singh, Y., Sharma, R., & Talwar, A. (2012). Immediate and long-term effects of meditation on acute stress reactivity, cognitive functions, and intelligence. Alternative Therapies in Health and Medicine, 18(6), 46–53. Sisley, B. L. (1987). Preventing burnout in teacher coaches. Journal of Physical Education, Recreation, & Dance, 58(8), 71–75. Robert Pearson, Timothy Baghurst 142 Small, G., & Vorgan G. (2019). The small guide to anxiety. Humanix Books. Smith, A. P., Chaplin, K., & Wadsworth, E. (2012). Chewing gum, occupational stress, work performance and well-being. An intervention study. Appetite, 58(3), 1083–1086. Sonnentag, S., & Jelden, S. (2009). Job stressors and the pursuit of sport activities: A day-level perspective. Journal of Occupational Health Psychology, 14, 165–181. Stahl, B., & Goldstein, E. (2019). A mindfulness-based stress reduction workbook. New Harbin- ger Publications. Stickle, F. E., & Scott, K. (2016). Leadership and occupational stress. Education, 137(1), 27–38. Stojanoski, B., Benoit, A., Van Den Berg, N., Ray, L. B., Owen, A. M., Zandi, A. S., Quddus, A., Comeau, F. J. E., & Fogel, S. M. (2019). Sustained vigilance is negatively affected by mild and acute sleep loss reflected by reduced capacity for decision making, motor prepa- ration and execution. Sleep, 42(1), 1–9.hf Tashman, L. S., Tenenbaum, G., & Eklund, R. (2010). The effect of perceived stress on the relationship between perfectionism and burnout in coaches. Anxiety, Stress, & Coping, 23(2), 195–212. Tekano, K., Sakamoto, S., & Tanno, Y. (2011). Ruminative and reflective forms of self-focus: Their relationships with interpersonal skills and emotional reactivity under interpersonal stress. Personality and individual Differences, 51(4), 515–520. Ury, W. (2007). The power of a positive no: How to say no and still get to yes. Bantam. Tekano, K., Sakamoto, S., & Tanno, Y. (2011). Ruminative and reflective forms of self-focus: Their relationships with interpersonal skills and emotional reactivity under interpersonal stress. Personality and individual Differences, 51(4), 515–520. , , , , , ( )l Their relationships with interpersonal skills and emotional reactivity under interperso stress. Personality and individual Differences, 51(4), 515–520. REFERENCES Ury, W. (2007). The power of a positive no: How to say no and still get to yes. Bantam. Valcour, M. (2016). Beating burnout. Harvard Business Review, 94(11), 98–101. stress. Personality and individual Differences, 51(4), 515–520. Ury, W. (2007). The power of a positive no: How to say no and still get to yes. Bantam. Valcour M (2016) Beating burnout Harvard Business Review 94(11) 98–101 y ff Ury, W. (2007). The power of a positive no: How to say no and still get to yes. Bantam. Valcour, M. (2016). Beating burnout. Harvard Business Review, 94(11), 98–101. Van Mullem, P. (2015). Developing a systematic evaluation for coaches, a commentary on ‘Coach Evaluation’ by Gilham, Hansen and Brady. International Sport Coaching Journal, 2(2), 203–205. Vladut, C., & Kallay, E. (2010). Work stress, personal life, and burnout. Causes, consequenc- es, possible remedies: A theoretical review. Cognition, Brain, Behavior, 14(3), 261–280. Yoshikawa, Y., Ohmaki, E., Kawahata, H., Maekawa, Y., Ogihara, T., Morishita, R., Aoki, M. (2018). Beneficial effect of laughter therapy on physiological and psychological function in elders. Nursing Open, 6(1), 93–99.
https://openalex.org/W4312382980	http://staitbiasjogja.ac.id/jurnal/index.php/saliha/article/download/365/177	Indonesian	null	Manajemen Keuangan Pada Lembaga Pendidikan Bina Anak Sholeh Yogyakarta	Saliha	2,022	cc-by-sa	3,696	Manajemen Keuangan Pada Lembaga Pendidikan Bina Anak Sholeh Yogyakarta Denas Hasman Nugraha STAI Terpadu Yogyakarta denasnugraha@gmail.com Abstract: Islamic educational institutions are currently developing very quickly as a result of public interest and growing public confidence in these institutions. One of the many educational institutions in Yogyakarta with a network of schools at various levels is the Educational Institution for Bina Anak Sholeh Yogyakarta. Batita, Play Group, Integrated Islamic Elementary School, Integrated Islamic Junior High School, and Integrated Islamic High School are some of the current levels. This study combined a descriptive method with a qualitative approach as its research methodology. For the implementation of operations at each level to be successful, financial management must be done in an organized and orderly fashion. Where financial management becomes crucial due to its role in carrying out educational activities. The primary finding of this study may be noticed in the financial management realization, planning, and supervision phases. At the Islamic Education Institution of Bina Anak Sholeh Yogyakarta, financial management planning is executed in the form of a RAPBS that is put together by all leaders at each level based on that levels' needs while still referencing the school's vision and mission. The implementation of financial management at the Islamic Education Institution of Bina Anak Sholeh Yogyakarta has been modified to the technical guidelines listed in the RAPBS for that year and is used for expenditures on school interests, practice materials, practical tools, financing student activities, and developing the quality of students and teachers. Fairness and efficiency have also been applied to the management of school finances. An honest and reliable public accountant was used to oversee the financial administration at the Islamic Educational Institution of Bina Anak Sholeh Yogyakarta. Reports are produced in compliance with the regulations that are in effect. Keywords: Islamic educational institutions, financial management Abstrak: Lembaga Pendidikan Islam saat ini sudah berkembang sangat pesat, hal ini disebabkan oleh animo masyararakat yang semakin mempercayai terhadap lembaga pendidikan tersebut. Lembaga Pendidikan Bina Anak Sholeh Yogyakarta merupakan salah satu dari sekian banyak lembaga pendidikan di Yogyakarta yang memiliki jejaring sekolah berbagai jenjang yang ada di bawahnya. Diantara jenjang yang ada yaitu Batita, Play Group, SD Islam Terpadu, SMP Islam Terpadu, SMA Islam Terpadu. Metode penelitian yang digunakan dalam penelitian ini adalah metode deskriptif dengan pendekatan kualitatif. Pengelolaan keuangan dengan tertib dan teratur menjadi syarat yang harus untuk menjamin terlaksananya kegiatan pada masing-masing jenjang. Saliha: Jurnal Pendidikan dan Agama Islam Volume 5 No. 2 Juli 2022 ISSN: 2614-1388 \| E-ISSN 2622-7479 DOI : 10.0118/saliha.v5i2.365 Saliha: Jurnal Pendidikan dan Agama Islam Volume 5 No. 2 Juli 2022 ISSN: 2614-1388 \| E-ISSN 2622-7479 DOI : 10.0118/saliha.v5i2.365 Manajemen Keuangan Pada Lembaga Pendidikan Bina Anak Sholeh Yogyakarta Dimana manajemen keuangan menjadi sangat penting karena hubungannya dalam pelaksanaan kegiatan sekolah. Yang menjadi titik pokok dalam penelitian ini adalah dilihat dari perencanaan pengelolaan keuangan, realisasi pengelolaan keuangan dan pengawasan pengelolaan keuangan. Perencanaan pengelolaan keuangan pada Lembaga Pendidikan Islam Bina Anak Sholeh Yogyakarta diimplementasikan dalam bentuk RAPBS yang disusun oleh semua pimpinan yang ada di jenjang masing-masing dengan berdasarkan kebutuhan setiap jenjang juga tetap mengacu pada visi dan misi sekolah. Realisasi pengelolaan keuangan pada Lembaga Pendidikan Islam Bina Anak Sholeh Yogyakarta telah disesuaikan dengan petunjuk teknis yang tertera pada RAPBS pada tahun tersebut dan digunakan untuk belanja kepentingan sekolah, bahan praktik, alat praktik, pembiayaan kegiatan siswa, dan pengembangan mutu siswa dan guru serta telah menerapkan prinsip keadilan dan efisien dalam pengelolaan keuangan sekolah. Bentuk pengawasan terkait pengelolaan keuangan pada Lembaga Pendidikan Islam Bina Anak Sholeh Yogyakarta sudah menggunakan akuntan publik yang kredibel dan sudah pasti keakuratanya. Mengenai pelaporan-pelaporan sudah sesuai dengan standar yang berlaku. Kata kunci: Pengelolaan keuangan, Lembaga Pendidikan Islam \|Published: Juli 30, 2022 \|Accepted: June 28, 2022 \|Submitted: June 27, 2022 Manajemen Keuangan Manajemen Keuangan 1 Agus Harjito, Manajemen Keuangan, Edisi Kedua (Yogyakarta: EKONISIA, 2005). 222 ŚALIĤA\|Jurnal Pendidikan dan Aga Latar Belakang Masalah Lembaga pendidikan Islam merupakan tempat atau suatu badan yang mengelola dan memastikan untuk berlangsungnya proses pendidikan Islam, baik itu kegiatan-kegiatan yang bersifat keterampilan, keagamaan, sosial kebudayaan termasuk penelitian dan keahlian yang berbenntuk dalam hal ini adalah pendidika intelektual, spiritual, serta keahlian atau keterampilan. Tentu dengan proses proses pembudayaan serta dapat mengikat individu yang berda dalam naungannya, sehingga lembaga ini mempunyai kekuatan hukum. Sebagai tempat yang melibatkan banyak orang yang berkumpul, bekerjasama secara rasional dan sistematis, terencana, terorganisasi, terpimpin dan terkendali, dalam memanfaatkan sumber daya, sarana-parasarana, data, dan lain sebagainya yang digunakan secara efisien dan efektif untuk mencapai tujuan pendidikan. Lembaga Pendidikan mempunyai tanggung jawab formal kelembagaan sesuai dengan fungsi dan tujuan yang ditetapkan menurut ketentuan yang berlaku, tanggung jawab keilmuan berdasarkan bentuk, isi, tujuan dan tingkat pendidikan juga tanggung jawab fungsional yang profesional dalam pengelolaan dan pelaksanaan pendidikan yang menerima ketetapan berdasarkan ketentuan jabatannya. Salah satu unsur penting yang akan memastikan terlaksananya fungsi pendidikan dengan baik adalah dari sisi keuangan. Manajemen keuangan menjadi sangat penting karena hubungannya dalam pelaksanaan kegiatan sekolah. Ada beragam sumber dana yang dimiliki oleh suatu lembaga pendidikan, baik dari pemerintah maupun pihak lain. Ketika dana masyarakat atau dana pihak ketigalainnya mengalir masuk, harus dipersiapkan sistem pengelolaan keuangan yang professional dan jujur. Pengelolaan keuangan secara umum sebenarnya telah dilakukan dengan baik oleh semua sekolah. Hanya kadar substansi pelaksanaanya yang beragam antara sekolah yang satu dengan yang lainnya. Sebesar apapun dana yang ada disekolah bila tidak dikelola dengan manajemen yang baik maka sekolah tersebut akan mengalami suatu kemunduran. Manajemen Keuangan sebagai aktifitas memperoleh dana, menggunakan dana dan mengelola asset secara effisien membutuhkan beberapa tujuan atau sasaran.1 222 Denas Hasman Nugraha Lembaga Pendidikan Bina Anak Sholeh Yogyakarta merupakan lembaga pendidikan yang memiliki jejaring sekolah berbagai jenjang yang ada di bawahnya. Diantara jenjang yang ada yaitu Batita, Play Group, SD Islam Terpadu, SMP Islam Terpadu, SMA Islam Terpadu. Oleh karena banyaknya jenjang yang berada di bawah lembaga tersebut juga latar belakang pentingnya mengelola keuangan dengan baik maka penulis akan membahas tentang keuangan Lembaga Pendidikan yang initinya adalah untuk mengetahui perencanaan, perencanaan dan pengawasan dalam pengelolaan keuangan. Adapaun pembahasan penelitian ini adalah: Untuk mengetahui perencanaan dalam pengelolaan keuangan di Lembaga Pendidikan Bina Anak Sholeh Yogyakarta? Untuk mengetahui realisasi anggaran dalam pengelolaan keuangan di Lembaga Pendidikan Bina Anak Sholeh Yogyakarta? Untuk mengetahui pengawasan dalam pengelolaan keuangan di Lembaga Pendidikan Bina Anak Sholeh Yogyakarta? 2 Sugiyono, Metode penelitian pendidikan: (pendekatan kuantitatif, kualitatif dan R & D) (Bandung: Alfabeta, 2 Sugiyono, Metode penelitian pendidikan: (pendekatan kuantitatif, kualitatif dan R & D) (Bandung: Alfabeta, 2011). ŚALIĤA\|Vol. 5 No. 2, Juli 2022 2011). 2 Sugiyono, Metode penelitian pendidikan: (pendekatan kuantitatif, kualitatif dan R & D) (Bandung: Alfabeta, Metode Penelitian Adapun rancangan dalam penelitian ini adalah menggunakan metode deskriptif dengan pendekatan kualitatif. Berkaitan dengan penelitian deskriptif kualitatif. Menurut Sugiyono bahwa metode penelitian kualitatif ini sering disebut metode naturalistik karena penelitiannya dilakukan pada kondisi yang alamiah (natural setting).2 Metode penelitian kualitatif adalah metode penelitian yang digunakan untuk meneliti pada kondisi obyek yang alamiah, dimana peneliti adalah sebagai instrumen kunci, teknik pengumpulan data dilakukan secara gabungan, analisis data bersifat induktif, dan hasil penelitian kualitatif lebih menekankan makna dari pada generalisasi. Metode pengumpulan data yang akan digunakan adalah wawancara, observasi dan dokumentasi. Wawancara digunakan sebagai teknik pengumpulan data apabila peneliti ingin melakukan studi pendahuluan untuk menemukan permasalahan yang harus diteliti. Observasi digunakan agar memungkinkan bagi peneliti nantinya untuk mengenal secara baik linkungan, orang, dan tempat dari penelitian ini. Dokumentasi 223 Manajemen Keuangan dapat berupa catatan, transkrip, buku, surat kabar, majalah, prasasti, notulen rapat, leggger, agenda, dan sebagainya. Menurut Arikunto, langkah langkah analisis data adalah setelah data terkumpul dari hasil pengumpulan data perlu segera digarap oleh staf peneliti, khususnya yang bertugas mengolah data.3 Proses mengatur urutan data, mengorganisasikannya kedalam suatu pola, kategori, dan satuan uraian dasar merupakan analisis data. Berdasarkan pendapat tersebut diatas, semua data yang diperoleh nantinya akan dianalis dengan menggunakan teknik analisa deskriptif kualitatif. 3 Suharsimi Arikunto, Prosedur penelitian: suatu pendekatan praktik (Jakarta: Rineka Cipta, 1992). 224 ŚALIĤA\|Jurnal Pendidikan dan Agam Pengertian Pengelolaan Keuangan Manajemen keuangan merupakan susuatu yang tidak dapat dipisahkan dalam pengelolaan keuangan pada lembaga pendidikan. Manajemen pada dasarnya berasal dari kata kerja manage yang berarti mengatur, mengelola, atau menurusi. Selain itu manajemen juga sering digambarkan sebagai ilmu, profesi, dan seperangkat aturan. Manajemen dipandang sebagai disiplin yang secara metodologis bertujuan untuk memahami mengapa dan bagaimana individu bekerja sama untuk mencapai tujuan dan mengembangkan sistem kerja sama yang lebih menguntungkan bagi kemanusiaan. Gaya manajemen yang diterapkan pada suatu lembaga pendidikan tertentu, akan sangat berpengaruh terhadap upaya peningkatan mutu pendidikan. Pengelolaan akan efektif dan efisien jika didukung oleh sumber daya manusia yang terampil untuk menjalankan lembaga pendidikan, kurikulum yang sesuai dengan tahap perkembangan dan karakteristik siswa, kapasitas dan dedikasi tenaga kependidikan yang andal, infrastruktur yang memadai untuk mendukung kegiatan belajar mengajar, anggaran yang cukup untuk membayar staf sesuai dengan fungsinya, dan tingkat keterlibatan masyarakat yang tinggi. Efektivitas dan efisiensi pada pengelolaan lembaga pendidikan atau sekolah tidak akan sesuai dengan yang ŚALIĤA\|Jurnal Pendidikan dan Agama Islam ŚALIĤA\|Jurnal Pendidikan dan Agama Islam 224 Denas Hasman Nugraha diharapkan dan/atau tidak berjalan sebagaimana mestinya jika salah satu faktor di atas tidak dapat berjalan sesuai dengan yang direncanakan. Manajemen pendidikan adalah proses pemusatan sumber daya pendidikan dalam upaya memenuhi tujuan pendidikan yang telah ditetapkan. Dan untuk memenuhi tujuan pendidikan, setiap aspek manajemen di setiap bagian ini harus diatur seefisien dan seefektif mungkin. Tetapi masalah hari ini adalah bahwa para manajer dan pemimpin kurang memperhatikan pengelolaan keuangan sekolah. Salah satu komponen manajemen di lembaga pendidikan dan sekolah adalah manajemen keuangan. Pengelolaan keuangan dilakukan melalui proses perencanaan, pengorganisasian, pengarahan, pengkoordinasian, pengawasan, dan pengendalian. Mencari dan memutuskan sumber pembiayaan untuk penggunaan dana, pelaporan, auditing, dan pertanggungjawaban adalah beberapa kegiatan pengelolaan keuangan.4 Manajemen keuangan adalah suatu tindakan pengelolaan atau penatausahaan keuangan menurut yang meliputi pencatatan, perencanaan, pelaksanaan, pertanggungjawaban, dan pelaporan.5 Karena perencanaan, pembukuan, pembelanjaan, pengawasan, dan pertanggungjawaban keuangan merupakan langkah awal dalam proses ini maka pengelolaan keuangan lembaga pendidikan dan sekolah dapat dilihat sebagai rangkaian tindakan untuk mengendalikan keuangan lembaga tersebut. Sumber keuangan dan pembiayaan lembaga pendidikan secara garis besar dapat dikategorikan menjadi tiga sumber, yaitu: 1) Pemerintah, baik pemerintah pusat maupun daerah, atau keduanya, yang bersifat umum atau khusus yang ditujukan untuk tujuan pendidikan; 2) Orang tua peserta didik; dan 3) Masyarakat, baik yang mengikat maupun yang tidak mengikat. ( j j g ) 6 H. M Daryanto, Evaluasi pendidikan: komponen MKDK, Cet. 6 (Jakarta: PT Rineka Cipta, 2012). p , j g p ekolah’ (Dirjen Dikdasmen, Direktorat Pendidikan Lanjutan Tingkat Pertama, 2002), Jakarta. 4 Harjito. 5 Departemen Pendidikan Nasional, ‘Manajemen Keuangan. Materi Pelatihan Terpadu Untuk Kepala Ś Ĥ ŚALIĤA\|Vol. 5 No. 2, Juli 2022 7 Nanang Fattah, Landasan manajemen pendidikan (Bandung: PT Remaja Rosdakarya, 1999). g , j p ( g j y , ) 8 ‘Undang-Undang (UU) No. 20 Tahun 2003. Tentang Sistem Pendidikan Nasional’, 2003. ŚALIĤA\|Jurnal Pendidikan dan Agama Islam Pengertian Pengelolaan Keuangan Menurut Daryanto pengelolaan biaya pendidikan meliputi perencanaan, pengorganisasian, penggerakan, dan pengawasan sumber daya keuangan.6 Faktor terpenting dalam penganggaran adalah bagaimana menggunakan uang secara efektif, menetapkannya sesuai skala prioritas, dan mendukung setiap langkah proses pendidikan sehingga dihasilkan lulusan yang unggul. Anggaran adalah salah satu alat ŚALIĤA\|Vol. 5 No. 2, Juli 2022 225 Manajemen Keuangan manajemen, yang berarti anggaran berfungsi sebagai rencana atau perkiraan untuk semua tindakan organisasi di masa depan. Baik biaya langsung maupun biaya tidak langsung merupakan komponen dari pembiayaan pendidikan. Biaya langsung meliputi biaya yang dikeluarkan untuk melaksanakan kegiatan belajar mengajar dan pembelajaran siswa; meliputi biaya bahan pelajaran, sarana dan prasarana pembelajaran, biaya transportasi, dan gaji guru yang dibayarkan oleh pemerintah, orang tua, dan siswa. Sedangkan biaya tidak langsung berupa hilangnya pendapatan dan hilangnya kesempatan siswa untuk belajar.7 Ada dua metode untuk menghitung biaya satuan; pendekatan makro dan teknik mikro. Strategi makro didasarkan pada perhitungan jumlah total dana yang diterima untuk pendidikan dari berbagai sumber, kemudian dibagi dengan jumlah siswa. Strategi mikro difokuskan pada alokasi dana untuk setiap elemen pendidikan yang digunakan siswa. Mencari sumber pendanaan dan mengalokasikan untuk tujuan penyelenggaraan proses pendidikan adalah apa yang dimaksud dengan pembiayaan pendidikan. Ketika membahas pendanaan pendidikan, penting untuk mempertimbangkan isu-isu yang terlibat dalam mengalokasikan dan mendistribusikan dana dari berbagai sumber dengan cara yang paling sesuai dengan kebutuhan lembaga untuk menyelenggarakan proses pendidikan. Berbagai konsep harus dipertimbangkan ketika mengelola keuangan sekolah. Pengelolaan dana pendidikan didasarkan pada prinsip-prinsip keadilan, efisiensi, transparansi, dan tanggung jawab publik, sesuai dengan UU No. 20 Tahun 2003 Pasal 48.8 Selain itu, penting untuk menekankan gagasan efektivitas. Masing-masing prinsip ini adalah transparansi, akuntabilitas, efektivitas, dan efisiensi yang akan di bahas berikutnya. Transparansi dalam bidang manajemen berarti keterbukaan dalam menjalankan kegiatan. Dalam lembaga pendidikan, area pengelolaan keuangan yang transparan adalah keterbukaan dalam pengelolaan keuangan lembaga pendidikan, yaitu pengungkapan sumber dan jumlah dana, perincian penggunaan dan 226 ŚALIĤA\|Jurnal Pendidikan dan Agama Islam Denas Hasman Nugraha pertanggungjawaban harus jelas sehingga mudah bagi pemangku kepentingan. pihak untuk mengetahuinya. Transparansi keuangan sangat diperlukan untuk meningkatkan dukungan orang tua, masyarakat dan pemerintah dalam pelaksanaan semua program pendidikan di sekolah. Selain itu, transparansi dapat menciptakan rasa saling percaya antara pemerintah, masyarakat, orang tua dan warga sekolah dengan memberikan informasi dan menjamin kemudahan akses informasi yang akurat dan memadai. Akuntabilitas adalah keadaan seseorang yang dinilai oleh orang lain atas kualitas kinerjanya dalam menyelesaikan tugas untuk mencapai tujuan yang menjadi tanggung jawabnya. E. Mulyasa, Manajemen Berbasis Sekolah: Konsep, Strategi, Dan Implementasi, Cet. 1 (Bandung: Remaj osdakarya, 2002). 9 E. Mulyasa, Manajemen Berbasis Sekolah: Konsep, Strategi, Dan Implementasi, Cet. 1 (Bandung: Remaj d k 2002) 9 E. Mulyasa, Manajemen Berbasis Sekolah: Konsep, Strategi, Dan Implementasi, Cet. 1 (Bandung: Remaja akarya, 2002). ŚALIĤA\|Vol. 5 No. 2, Juli 2022 9 E. Mulyasa, Manajemen Berbasis Sekolah: Konsep, Strategi, Dan Implementasi, Cet. 1 (Bandung: Remaja Pengertian Pengelolaan Keuangan Akuntabilitas dalam pengelolaan keuangan berarti penggunaan biaya sekolah dapat diperhitungkan sesuai dengan rencana yang telah ditentukan. Berdasarkan rencana yang ditetapkan dan aturan yang berlaku, sekolah membelanjakan uang secara bertanggung jawab. Pertanggungjawaban dapat dilakukan kepada orang tua, masyarakat dan pemerintah. Ada tiga pilar utama yang menjadi prasyarat terwujudnya akuntabilitas, yaitu (1) transparansi pengelola sekolah dengan menerima masukan dalam mengelola sekolah dan melibatkan berbagai konstituen, (2) adanya standar kinerja yang terukur di setiap lembaga. Melaksanakan tugas, fungsi, dan wewenangnya, (3) partisipasi dalam menciptakan lingkungan yang saling menguntungkan dalam menciptakan pelayanan publik dengan prosedur yang lebih mudah, biaya yang lebih rendah, dan pelayanan yang lebih cepat. Efektivitas adalah kesesuaian antara orang yang melakukan kegiatan dengan sasaran yang dituju serta efisiensi penempatan dan penggunaan sumber daya oleh suatu organisasi dalam upaya mencapai tujuan operasional.9 Efektivitas berkorelasi dengan penyelesaian semua tugas utama, pencapaian tujuan, ketepatan waktu, dan keterlibatan aktif anggota, sesuai dengan konsep yang disajikan di atas. Kemudian Mulyasa mengungkapkam suatu tindakan dianggap efektif jika tujuannya dapat dicapai dengan cara terbaik dengan waktu atau sumber daya yang dikeluarkan sesedikit mungkin. Efisiensi adalah ukuran seberapa efektif suatu organisasi menggunakan sumber dayanya yang terbatas. Perbandingan input dan output, y Rosdakarya, 2002). 227 Manajemen Keuangan kekuatan dan hasil, belanja dan pemasukan, pengeluaran, dan kesenangan yang dihasilkan merupakan perspektif tentang efisiensi.10 Efisiensi adalah perbandingan yang terbaik antara masukan (input) dan keluaran (output) atau antara daya dan hasil. Daya yang dimaksud meliputi tenaga, pikiran, waktu, biaya. Prinsip – prinsip manajemen keuangan sangat diperlukan ketika orang tua berperan penting dalam pengelolaan keuangan. Dalam pengelolaan dana dibutuhkan Tranparansi, akuntabilitas, efektif dan efisiensi. Lembaga Pendidikan Bina Anak Sholeh Yogyakarta adalah sebuah Lembaga Pendidikan Islam Swasta yang beralamat di Jl. Wirosaban Barat No. 6, Sorosutan, Yogyakarta. Lembaga Pendidikan BIAS ini sudah berjalan lebih dari 20 tahun, dengan dipimpim oleh Ir. Hj. Lilik Indriati yang bertindak sebagai Direktur dari Lembaga Pendidikan Bina Anak Sholeh. Jenjang pendidikan yang ada tersebar mulai dari Batita, Play Group, SD Islam Terpadu, SMP Islam Terpadu, SMA Islam Terpadu. Dengan berlokasi di beberapa tempat yang berbeda. Jejang Batita dan Play Group beralamat di Jl. Tritunggal, Sorosutan, Umbulharjo, Kota Yogyakarta, Daerah Istimewa Yogyakarta, Jl. Palagan Tentara Pelajar Km 7, 9, Sariharjo, Ngaglik, Karang Moko, Sariharjo, Ngaglik, Kabupaten Sleman, Daerah Istimewa Yogyakarta 55581, Jl. Bias no.101, Jalan Kaliurang km 10.9 D.I. Yogyakarta. Kemudian untuk jenjang TK Islam Terpadu (TK BIAS) Jl. Mendung Warih, Yogyakarta, Daerah Istimewa Yogyakarta, Jl. 10 Agus Dharma, Manajemen Supervisi : Petunjuk Praktis Bagi Para Supervisor (Jakarta: Raja Grafindo Persada, 2000). 10 Agus Dharma, Manajemen Supervisi : Petunjuk Praktis Bagi Para Supervisor (Jakarta: Raja Grafin g j p j g p (J j Persada, 2000). 10 Agus Dharma, Manajemen Supervisi : Petunjuk Praktis Bagi Para Supervisor (Jakarta: Raja Persada, 2000). Pengertian Pengelolaan Keuangan Palagan Tentara Pelajar Km 7, 9, Sariharjo, Ngaglik, Karang Moko, Sariharjo, Ngaglik, Kabupaten Sleman, Daerah Istimewa Yogyakarta 55581, Jl. Kaliurang km 9.8, Sleman, Daerah Istimewa Yogyakarta 55581. Jenjang SD Islam Terpadu (SD BIAS) Jl. Mendung Warih No.155, Giwangan, Kec. Umbulharjo, Kota Yogyakarta, Daerah Istimewa Yogyakarta 55163, Jl. Kaliurang Km 10.9 Ngaglik Sleman DIY 55581. Jenjang SMP Islam Terpadu (SMP BIAS) Jl. Mendung Warih No.145B, Giwangan, Kec. Umbulharjo, Kota Yogyakarta, Daerah Istimewa Yogyakarta 55163. Dan yang terakhir adalah jenjang SMA Islam Terpadu (SMA BIAS) Jl. Pangeran Wirosobo No.17, Sorosutan, Kec. Umbulharjo, Kota Yogyakarta, Daerah Istimewa Yogyakarta 55162. ŚALIĤA\|Jurnal Pendidikan dan Agama Islam 228 Denas Hasman Nugraha Dalam proses perencanaan yang di lakukan adalah proses pembuatan RAPBS, dimana dalam pelaksanaan menjelang akhir semester berjalan masing-masing jenjang menyusun kebutuhan sekolah berdasarkan karakteristik dan kebutuhan masing- masing sekolah. Point yang tertuang dalam RAPBS adalah semua kebutuhan Sekolah jenjang selama 1 tahun ajaran. Yang kemudian akan di bagi menjadi 2 semester. Yang menjadi dasar dalam pembuatan RAPBS ini adalah rasio jumlah siswa, luas sekolah dan kegaiatan-kagiatan yang dilaksanakan. Kemudian setelah di susun maka tahapan selanjutnya adalah pengumpulan seluruh pengajuan kepada bendahara untuk kemudian ditinjau dan observasi pengajuan-pengajuan yang ada. Yang nantinya semua pengajuan yang sudah masuk akan di bawa pada rapat pleno para pimpinan jenjang untuk validasi. Proses yang terjadi pada rapat pleno berkaitan dengan semua kebutuhan dari masing-masing jenjang juga menyesuaikan program-program akan di lakasanakan. Pada pelaksanaan pleno ini RAPBS yang sudah di susun akan di bahas bersama tim lembaga dan seluruh pimpinan jenjang. Kemungkinan yang terjadi pada rapat pleno ini adalah bisa berubahnya RAPBS yang sudah di hasilkan dari masing-masing jenjang dengan tetapi ketika yang diajukan sudah sesuai makan posisinya akan aman tanpa ada perubahan sama sekali. Hasil dari rapat pleno tersebut akan muncul tanggal pencairan anggaran masing-masing jenjang, dan setiap jenjang akan berbeda tanggal pencairan untuk memudahkan pencatatan yang di lakukan oleh bendahara lembaga. Untuk prioritas adalah jenjang yang kebutuhannya paling besar biasanya akan mendapatkan pencairan anggaran paling dulu. Kemudian setelah masing-masing jenjang mendapatkan tanggal pencairan maka tahap selanjtnya adalah melaksanakan aturan yang mengatur tentang pencairan tersebut. Bahwa sekolah jenjang dapat mengajukan pencairan sesuai dengan tanggal yang sudah ditetapkan ketika rapat pleno jika sudah melaporkan seluruh aktifitas keuangan yang sudah terlaksana selama bulan berjalan. ŚALIĤA\|Vol. 5 No. 2, Juli 2022 Pengertian Pengelolaan Keuangan Berdasarkan hasil analisis mengenai hubungan antara teori-teori pengelolaan keuangan sekolah dengan fenomena yang ada di Lembaga Pendidikan Bina Anak Sholeh Yogyakarta maka dapat bahwa perencanaan keuangan sekolah dapat ŚALIĤA\|Vol. 5 No. 2, Juli 2022 ŚALIĤA\|Jurnal Pendidikan dan Agama Islam Kesimpulan Perencanaan pengelolaan keuangan pada Lembaga Pendidikan Islam Bina Anak Sholeh Yogyakarta diimplementasikan dalam bentuk RAPBS yang disusun oleh semua pimpinan yang ada di jenjang masing-masing dengan berdasarkan kebutuhan setiap jenjang juga tetap mengacu pada visi dan misi sekolah. Realisasi pengelolaan keuangan pada Lembaga Pendidikan Islam Bina Anak Sholeh Yogyakarta telah disesuaikan dengan petunjuk teknis` yang tertera pada RAPBS pada tahun tersebut dan digunakan untuk belanja kepentingan sekolah, bahan praktik, alat praktik, pembiayaan kegiatan siswa, dan pengembangan mutu siswa dan guru serta telah menerapkan prinsip keadilan dan efisien dalam pengelolaan keuangan sekolah. Bentuk pengawasan terkait pengelolaan keuangan pada Lembaga Pendidikan Islam Bina Anak Sholeh Yogyakarta sudah menggunakan akuntan publik yang kredibel dan sudah pasti keakuratanya. Mengenai pelaporan-pelaporan sudah sesuai dengan standar yang berlaku. ŚALIĤA\|Vol. 5 No. 2, Juli 2022 229 Manajemen Keuangan diimplementasikan dalam bentuk anggaran pendapatan dan belanja sekolah yang disesuaikan kebutuhan masing-masing jenjang dalam bentuk RAPBS, disusun oleh masing-masing jenjang dengan melibatkan berbagai unsur dalam, pelaksanaanya. Hal ini sesuai dengan Peraturan Pemerintah Republik Indonesia Nomor 19 Tahun 2005 tentang Standar Nasional Pendidikan mewajibkan sekolah untuk menyusun RAPBS, yakni rencana kerja tahunan yang memuat rencana anggaran pendapatan dan belanja satuan pendidikan untuk masa kerja satu tahun. Realisasi penerimaan dan pengeluaran dana sekolah mengacu pada perencanaan yang telah dirancang dalam RAPBS dan sesuai dengan ketentuan yang sudah disepakati oleh Pimpinan yang ada di masing-masing jenjang. Alokasi anggaran yang digunakan telah disesuaikan dengan petunjuk teknis yang tertera pada RAPBS pada tahun tersebut dan direalisasikan untuk belanja kepentingan sekolah, bahan praktik, alat praktik, pembiayaan kegiatan siswa, dan pengembangan mutu siswa dan guru. Hal ini ditegaskan dalam Peraturan Pemerintah Repubik Indonesia Nomor 48 Tahun 2008 tentang Pendanaan Pendidikan bahwa penggunaan dana pendidikan oleh satuan pendidikan dilaksanakan melaui mekanisme yang diatur dalam anggaran dasar dan anggaran rumah tangga satuan pendidikan. Bentuk pengawasan dan evaluasi sudah dilakukan dengan baik, karena untuk penerimaan keuangan harus di pastikan bahwa laporan bulan berjalan sudah di laksanakan sesuai dengan ketentuan yang tertera di RAPBS dan sudah terlaporkan dengan akuntabel. Dari sini bisa di lihat ke untuk menjamin efisiensi keuangan Lembaga benar-benar di laksanakan dengan baik dan terstruktur. Untuk mewujudkan kualitas yang bermutu dalam pengelolaan keuangan sekolah, maka harus dilandaskan berdasarkan prinsip-prinsip yang ada dalam pengelolaan keuangan sekolah, salah satunya adalah penerapan prinsip keadilan dan efisien, dimana prinsip keadilan dalam pengelolaan keuangan sekolah berupa merencanakan pendapatan, alokasi keuangan dan pemberian pelayanan yang optimal kepada siswa, guru, karyawan sebagai bentuk pemerataan pendidikan. Selanjutnya prinsip efisiensi dalam pengelolaan keuangan sekolah berupa pemberdayaan sumber daya uang sekolah dalam mencapai optimalisasi akses, mutu, relevansi, dan daya saing pelayanan pendidikan. ŚALIĤA\|Jurnal Pendidikan dan Agama Islam 230 Denas Hasman Nugraha ‘Undang-Undang (UU) No. 20 Tahun 2003. Tentang Sistem Pendidikan Nasional’, 2003. ŚALIĤA\|Jurnal Pendidikan dan Agama Islam Daftar Pustaka Agus Dharma. Manajemen Supervisi : Petunjuk Praktis Bagi Para Supervisor. Jakarta: Raja Grafindo Persada, 2000. Daryanto, H. M. Evaluasi pendidikan: komponen MKDK. Cet. 6. Jakarta: PT Rineka Cipta, 2012. Departemen Pendidikan Nasional. ‘Manajemen Keuangan. Materi Pelatihan Terpadu Untuk Kepala Sekolah’. Dirjen Dikdasmen, Direktorat Pendidikan Lanjutan Tingkat Pertama, 2002. Jakarta. Fattah, Nanang. Landasan manajemen pendidikan. Bandung: PT Remaja Rosdakarya, 1999. Harjito, Agus. Manajemen Keuangan. Edisi Kedua. Yogyakarta: EKONISIA, 2005. Mulyasa, E. Manajemen Berbasis Sekolah: Konsep, Strategi, Dan Implementasi. Cet. 1. Bandung: Remaja Rosdakarya, 2002. Sugiyono. Metode penelitian pendidikan: (pendekatan kuantitatif, kualitatif dan R & D). Bandung: Alfabeta, 2011. Suharsimi Arikunto. Prosedur penelitian: suatu pendekatan praktik. Jakarta: Rineka Cipta, 1992. ŚALIĤA\|Vol. 5 No. 2, Juli 2022 231 Manajemen Keuangan 232 ŚALIĤA\|Jurnal Pendidikan dan Agama Islam
https://openalex.org/W2129304756	https://hal.archives-ouvertes.fr/hal-01183708/file/Fradet_2013_PlosOneA%20New%20Murine%20Model.pdf	English	null	A New Murine Model of Osteoblastic/Osteolytic Lesions from Human Androgen-Resistant Prostate Cancer	PloS one	2,013	cc-by	8,303	Abstract This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: This work was supported by the CNRS (Edith Bonnelye), Inserm, the University of Lyon, “Ligue Régionale contre le Cancer” (Isère) (EB) http:// www.ligue-cancer.net/ and “Association pour la Recherche sur les Tumeurs de la Prostate (ARTP)” (Edith Bonnelye) http://www.artp.org/. Anais Fradet is supported by the Ligue Nationale contre le Cancer, http://www.ligue-cancer.net/ Baptiste Depalle by a grant from the Région Rhône Alpes "Cible" program and Akeila Bellahcene is a Senior Research Associate from the National Fund for Scientific Research, Belgium. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing interests: Edith Bonnelye is in the list of Plos One Academic editors. This does not alter the authors' adherence to all PLOS ONE policies on sharing data and materials. Edith Bonnelye is in the list of Plos One Academic editors. This does not alter the authors' adherence to all PLOS ONE policies on l interests: Edith Bonnelye is in the list of Plos One Academic editors. This does not alter the authors' adherence to all PLOS ON a and materials. * E-mail: edith.bonnelye@inserm.fr * E-mail: edith.bonnelye@inserm.fr ☯ These authors contributed equally to this work. ☯ These authors contributed equally to this work. Abstract Background: Up to 80% of patients dying from prostate carcinoma have developed bone metastases that are incurable. Castration is commonly used to treat prostate cancer. Although the disease initially responds to androgen blockade strategies, it often becomes castration-resistant (CRPC for Castration Resistant Prostate Cancer). Most of the murine models of mixed lesions derived from prostate cancer cells are androgen sensitive. Thus, we established a new model of CRPC (androgen receptor (AR) negative) that causes mixed lesions in bone. Methods: PC3 and its derived new cell clone PC3c cells were directly injected into the tibiae of SCID male mice. Tumor growth was analyzed by radiography and histology. Direct effects of conditioned medium of both cell lines were tested on osteoclasts, osteoblasts and osteocytes. Results: We found that PC3c cells induced mixed lesions 10 weeks after intratibial injection. In vitro, PC3c conditioned medium was able to stimulate tartrate resistant acid phosphatase (TRAP)-positive osteoclasts. Osteoprotegerin (OPG) and endothelin-1 (ET1) were highly expressed by PC3c while dikkopf-1 (DKK1) expression was decreased. Finally, PC3c highly expressed bone associated markers osteopontin (OPN), Runx2, alkaline phosphatase (ALP), bone sialoprotein (BSP) and produced mineralized matrix in vitro in osteogenic conditions. Conclusions: We have established a new CRPC cell line as a useful system for modeling human metastatic prostate cancer which presents the mixed phenotype of bone metastases that is commonly observed in prostate cancer patients with advanced disease. This model will help to understand androgen-independent mechanisms involved in the progression of prostate cancer in bone and provides a preclinical model for testing the effects of new treatments for bone metastases. Citation: Fradet A, Sorel H, Depalle B, Serre CM, Farlay D, et al. (2013) A New Murine Model of Osteoblastic/Osteolytic Lesions from Human Androgen- Resistant Prostate Cancer. PLoS ONE 8(9): e75092. doi:10.1371/journal.pone.0075092 Citation: Fradet A, Sorel H, Depalle B, Serre CM, Farlay D, et al. (2013) A New Murine Model of Osteoblastic/Osteolytic Lesions from Human Androgen- Resistant Prostate Cancer. PLoS ONE 8(9): e75092. doi:10.1371/journal.pone.0075092 Editor: Vladislav V. Glinskii, University of Missouri-Columbia, United States of America Editor: Vladislav V. Glinskii, University of Missouri-Columbia, United States of America Received June 6, 2013; Accepted August 8, 2013; Published September 19, 2013 Received June 6, 2013; Accepted August 8, 2013; Published September 19, 2013 Copyright: © 2013 Fradet et al. Introduction which prostate cancers are induced to metastasize to bone rely on a complex interplay between prostate cancer cells and the bone microenvironment [4]. Growth of prostate cancer cells alters bone remodeling (formation and resorption) by secreting factors that will directly affect osteoblasts (bone forming cells) and osteoclasts (bone resorbing cells). RANKL (Receptor activator of NF-kB ligand) stimulates osteoclasts differentiation and action while osteoprotegerin (OPG) acts as a decoy receptor for RANK (RANKL receptor). Therefore the balance between RANKL and OPG, that can be both produced by Bone is the most frequent site of prostate carcinoma metastases with bone metastases in up to 80% of advanced disease [1]. Surgical and hormonal therapies have shown beneficial effects only for early-stage hormone-responsive disease. Indeed, if the disease in most cases initially responds, it often progresses and become androgen independent. At that stage, patients with advanced disease often display osteoblastic or mixed lesions in bone [2,3]. The mechanisms by PLOS ONE \| www.plosone.org 1 September 2013 \| Volume 8 \| Issue 9 \| e75092 1 New Androgen-Resistant Bone Metastasis Model Manassas, VA, USA). VCAP were cultured in RPMI medium. PC3 and PC3c cells were routinely cultured in F12K nutrient mixture and DMEM medium (Life technologies, Carlsbad, CA, USA) respectively supplemented with 10% (v/v) fetal bovine serum (FBS; Perbio/Thermo scientific; Rockford, IL, USA) and 1% (v/v) penicillin/streptomycin (Life technologies, Carlsbad, CA, USA) at 37°C in a 5% CO2 incubator. PC3 and PC3c were also cultured upon osteogenic conditions for three weeks in the osteoblast medium supplemented with 50 µg/ml ascorbic acid (Sigma-Aldrich, Buchs, Switzerland). Ten mM sodium β- glycerophosphate (Sigma-Aldrich, Buchs, Switzerland) was added during 1 week at the end of the culture. PC3 and PC3c were continuously (day 1 to day 21) exposed to osteogenic conditions. For the visualization of mineralization, wells were fixed and stained with von Kossa and for ALP [16]. prostate cancer cells, is critical in controlling osteoclast activity and osteolysis in bone metastasis [4-6]. On the other side, pro- osteoblastic molecules can also be produced by prostate cancer cells. In fact, the first clinical studies to specifically target osteoblasts in patients with metastatic prostate cancer was based on endothelin-1 (ET1), a mitogenic factor for osteoblasts that can promote the growth of osteoblasts at metastatic sites [7,8]. Ethics statement The mice used in our study were handled according to the rules of Décret N° 87-848 du 19/10/1987, Paris. The experimental protocol have been reviewed and approved by the Rhone-Alpes Regional Committee on the Ethic of Animal Experiments (Lyon, France) (Register Number: 0121). Animal experiments were routinely inspected by the attending veterinarian to ensure continued compliance with the proposed protocols. SCID mice, 6 weeks age, were housed under barrier conditions in laminar flow isolated hoods. Animals bearing tumor xenografts were carefully monitored for established signs of distress and discomfort and were humanely euthanized. Subcutaneous injections of PC3c cells (106 in 100µl PBS) were also performed in 6- to 8-week-old SCID mice. Animals were euthanized after 12 weeks and tumors were fixed and embedded in paraffin. Animal studies For intra-osseous tumor xenograft experiments (Charles River Laboratories, Wilmington, MA, USA), a small hole was drilled with a 26-gauge sterile needle through the right tibia with the knee flexed in anesthetized 6- to 8-week-old SCID mice. Using a new sterile needle fitted to a 50-µl sterile Hamilton syringe (Hamilton Co.; Bonaduz, GR, Switzerland), a single-cell suspension (6x105 in 15-µl PBS) of PC3 or PC3c cells was carefully injected in the bone marrow cavity. From week 2 after tumor cell inoculation, radiographs of anesthetized mice were weekly taken with the use of MIN-R2000 films (Kodak, Rochester, NY, USA) in an MX-20 cabinet X-ray system (Faxitron X-ray Corp, Tucson, AZ, USA). Animals were euthanized after 6 and 10 weeks for mice injected by PC3 and PC3c cells respectively. Microcomputed tomography analyses were carried out using a micro-CT scanner Skyscan 1174 (Skyscan; Kontich, Belgium). The X-ray tube was set to a voltage of 50 kV and a current of 800 µA. A 0.5 mm aluminum filter was used to reduce beam hardening artifacts. Samples were scanned in 70% ethanol with a voxel size of 20 µm. For each sample, 265 section images were reconstructed with NRecon software (version 1.6.1.8, Skyscan). Three- dimensional modeling and analysis of BV (Bone Volume)/TV (Total Volume) ratio (percentage of bone tissue) were obtained with the CTAn (version 1.9, Skyscan) and CTVol (version 2.0, Skyscan) software. The dissected bones were then processed for histological and histomorphometric analysis. The majority of mixed bone metastases derived from prostate cancer mouse models are androgen sensitive and for that matter do not really mimic the clinical situation. We described the characterization of a new cell line (namely PC3c) that induce mixed skeletal lesions in animals that is derived from the human androgen independent AR-negative cell line PC3, known to induce pure osteolytic bone metastases. Introduction In addition, transforming growth factor β (TGFβ), vascular endothelial growth factor (VEGF) are abundantly expressed by the prostate cancer cells and have a direct effect on osteoblast function [9,10]. The wingless (WNT) pathway that is implicated in osteoblastogenesis has been also implicated in the development of osteoblastic metastasis in prostate cancer [11]. Up-regulation of the WNT-family ligand WNT1 in prostate cancer cells and a decrease in the serum of the WNT antagonist dikkopf-1 (DKK1) expression has been reported in patients with advanced metastatic prostate carcinoma and is associated with osteoblastic lesions [12]. Finally prostate cancer cells that induce bone metastasis also express large amount of bone associated factors like osteopontin (OPN), osteocalcin (OCN) or bone sialoprotein (BSP) secreted in the bone matrix and that will contribute to promote their osteomimicry properties [13]. Cell culture PC3 cell line was obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA). The PC3c cells, a subculture cell line of PC3 was isolated in our laboratory in vitro after single cell population culture. Consequently to spontaneous derivation of the cells, we finally obtained a subculture cell line named PC3c which was chosen based on its epithelial phenotype (Figure S1) [14,15]. The hormone dependent human prostate cancer VCAP cells were a generous gift of Pr M Cecchini (Department of Clinical Research, University of Bern, Bern, Switzerland) and was obtained from the American Type Culture Collection (ATCC, Osteoclastogenesis assay washing, the sections were revealed by 3,3’-diaminobenzidine (Dako, Glostrup, Denmark). Counterstaining was performed using Mayer’s hematoxylin (Merck, Whitehouse Station, NJ, USA). Primary bone marrow cells were obtained after tibia and femur bone marrow flushing from 6-week-old OF1 male mice. Cells were then cultured for 7 days, in differentiation medium: α-MEM medium containing 10% fetal calf serum (Life technologies, Carlsbad, CA, USA), 20 ng/mL of M-CSF (R&D Systems, Minneapolis, MN, USA) and 200 ng/mL of soluble recombinant RANK-L in presence or absence of conditioned medium extracted from PC3 and PC3c (25µg of proteins for each conditions) [17]. Medium was, first, changed every two days then from day 4 every days. After 7 days, mature multinucleated osteoclasts (OCs) were obtained and stained for TRAP activity (Sigma-Aldrich, Buchs, Switzerland), following the manufacturer’s instructions. Multinucleated TRAP- positive cells containing three or more nuclei were counted as OCs. Osteoblastogenesis assay Calvaria of 3-day-old OF-1 mice were dissected then cells were enzymatically isolated by sequential digestion with collagenase, as described previously [18,19]. Cells obtained from the last four of the five digestion steps (populations II-V) were plated onto 24-well plates at 2x104 cells / well. After 24 hours incubation, the medium including α-MEM medium containing 10% fetal bovine serum (Life technologies, Carlsbad, CA, USA) was changed and supplemented with 50µg/ml ascorbic acid (Sigma-Aldrich, Buchs, Switzerland) and with or without conditioned medium (25µg of proteins for each conditions) extracted from PC3 and PC3c. Medium was changed every two days for 15 days. 10mM sodium β- glycerophosphate (Sigma-Aldrich, Buchs, Switzerland) was added during 1 week at the end of the culture. At day 15, when bone mineralized nodules were formed, cells were then fixed and stained with von Kossa for quantification. ALP+ and bone mineralized nodules were then counted on a grid [16]. Results are plotted as the mean number of nodules ± SD of three wells for controls and each condition (PC3, PC3c) and were representative of two independent experiments. Osteocyte cell line MLO-Y4 were a generous gift of Pr L Bonewald (School of Dentistry, University of Missouri, Kansas City, MO, USA) and were cultured as described previously [20]. Real time RT-PCR Total RNA was extracted with Trizol reagent (Life Technologies, Carlsbad, CA, USA) from PC3, PC3c, OBs, OCs and MLO-Y4 cells. Samples of total RNA (1 µg) were reverse- transcribed using random hexamer (Promega, Madison, WI, USA) and the first strand synthesis kit of SuperscriptTM II (Life Technologies, Carlsbad, CA, USA). Real-time RT-PCR was performed on a Roche Lightcycler Module (Roche, Penzberg, Germany) with primers specific for human and mouse (see Tables S1 and S2). Real-time RT-PCR was carried out by using SYBR Green (Qiagen, Hilden, Germany) according to the manufacturer’s instructions with an initial step for 10 min at 95°C followed by 40 cycles of 20 sec at 95°C, 10 sec at Tm (see Tables S1 and S2) and 10 sec at 72°C. We verified that a single peak was obtained for each product using the Lightcycler Roche software. Amplimers were all normalized to corresponding L32 values. Data analysis was carried out using the comparative CT method: in real-time each replicate average genes CT was normalized to the average CT of L32 by subtracting the average CT of L32 from each replicate to give the ∆CT. Results are expressed as Log-2 __CT with ∆∆CT equivalent to the ∆CT of the genes in PC3, PC3c or treated OBs, OCs and MLO-Y4 cells subtracting to the ∆CT of the endogenous control (non-treated OBs, OCs and MLO-Y4 cells respectively). Bone histomorphometry and histology Tibia from animals were fixed, decalcified with 15% EDTA/ 0,4% PFA and embedded in paraffin. Five µm sections were stained with Goldner’s Trichrome and proceeded for histomorphometric analyses to calculate the TB (Tumor Burden)/STV (Soft Tissue Volume) ratio (percentage of tumor tissue). The in situ detection of osteoclasts was carried out on metastatic bone tissue sections using the tartrate-resistant acid phosphatase (TRAP) activity kit assay (Sigma-Aldrich, Buchs, Switzerland). September 2013 \| Volume 8 \| Issue 9 \| e75092 PLOS ONE \| www.plosone.org 2 New Androgen-Resistant Bone Metastasis Model New Androgen-Resistant Bone Metastasis Model Electron microscopy PC3c cells were cultured on glass coverslips, then fixed for 1h in 2% glutaraldehyde in 0.1M of sodium cacodylate buffer at pH7.4. After three rinses in 0.2M saccharose in 0.1M of sodium cacodylate buffer, the cells were postfixed in 1% osmium tetroxyde in 0.15M cacodylate buffer, dehydrated in graded ethanol, then embedded in Epon. Ultrathin sections were counterstained with uranyl acetate and lead citrate, the examined under a 1200 EX JEOL electron microscope (Jeol, Tokyo, Japan). Immunocytochemistry Fourier Transform InfraRed Microspectroscopy (FTIRM) Undecalcified sections (2µm-thick) of tibia embedded in MMA were cut longitudinally with a microtome Polycut (Reichert-Jung, Leica, Germany), and stored between 2 glass slides. FTIRM was performed with a PerkinElmer GXII Auto- image Microscope (Norwalk, CT, USA), equipped with a cooled liquid nitrogen wide band Mercury Cadmium Telluride detector (7800-400 cm-1). Infrared measurements were performed on bone matrix (in cortical bone) around the tumor and on the tumor itself. Infrared measurement of cortical bone from sham mice was also collected. IR spectra were collected in transmission mode, at 4 cm-1 of spatial resolution, and 40 µm X 40 µm of spatial resolution. Contribution of air and MMA were subtracted from the original spectrum. Automatic baseline correction was performed on each IR spectrum with Spectrum software (PerkinElmer, Inc). PC3c tumors and metastatic tibia were fixed and embedded in paraffin. Five µm sections were subjected to immunohistochemistry using rabbit polyclonal antibodies anti human/ mouse osteopontin antibody (Bachem, Bubendorf, Switzerland), anti human Endothelin-1 antibody (Abbiotec, San Diego, CA, USA) and anti human OPG antibody (Abbiotec, San Diego, CA, USA). BSP antibody was a generous gift of Dr L Malaval (University of J Monnet, St Etienne, France). Sections were deparaffinized in methylcyclohexan, hydrated then treated with a peroxidase blocking reagent (Dako, Glostrup, Denmark). Sections were incubated with normal calf serum for 1 hour and incubated overnight at 4°C with primary antibodies (dilution: 1/100). Sections were incubated with secondary antibody HRP- conjugated donkey anti rabbit (Amersham/GE Healthcare; Chalfont St Giles, UK) (dilution 1/300) for 1 hour. After September 2013 \| Volume 8 \| Issue 9 \| e75092 PLOS ONE \| www.plosone.org 3 New Androgen-Resistant Bone Metastasis Model Figure 1. Expression of pro-osteoblastic factors by PC3c cells. Detection by real-time PCR of AR mRNA expression in PC3, PC3c and VCAP cancer cells lines (A), AMACR, PAP (B) and DKK1, ET-1, FGF9, Noggin, OPN, OPG, Runx2 and TGFβ mRNA expression (C and D) in PC3 and PC3c cancer cells lines. Genes expression was assessed by real-time PCR on triplicate samples and normalized against that of the ribosomal protein gene L32 p<0.05; p<0,001, *p<0,0001. doi: 10.1371/journal.pone.0075092.g001 Figure 1. Expression of pro-osteoblastic factors by PC3c cells. Statistical analysis based on its epithelial phenotype (Figure S1). As expected and similarly to the parental PC3 cells, AR could not be detected by real-time PCR in PC3c while it was expressed in the hormone dependant cell line VCaP used as a positive control (Figure 1A). On the other hand, the prostate markers P504S (alpha methylacyl-coA racemase (AMACR)) and the prostatic acid phosphatase (PAP) were expressed in both cell lines confirming the prostate origin of the cells (Figure 1B) [21]. Data were expressed as mean +/- SD, and analyzed statistically by one way analysis of variance (ANOVA) followed by post hoc t-tests or student t-test to assess the differences between groups for in vitro and in vivo studies. Statistical significance was taken as p<0.05. Immunocytochemistry Detection by real-time PCR of AR mRNA expression in PC3, PC3c and VCAP cancer cells lines (A), AMACR, PAP (B) and DKK1, ET-1, FGF9, Noggin, OPN, OPG, Runx2 and TGFβ mRNA expression (C and D) in PC3 and PC3c cancer cells lines. Genes expression was assessed by real-time PCR on triplicate samples and normalized against that of the ribosomal protein gene L32 p<0.05; p<0,001, p<0,0001. doi: 10 1371/journal pone 0075092 g001 Figure 1. Expression of pro-osteoblastic factors by PC3c cells. Detection by real-time PCR of AR mRNA expression in PC3, PC3c and VCAP cancer cells lines (A), AMACR, PAP (B) and DKK1, ET-1, FGF9, Noggin, OPN, OPG, Runx2 and TGFβ mRNA expression (C and D) in PC3 and PC3c cancer cells lines. Genes expression was assessed by real-time PCR on triplicate samples and normalized against that of the ribosomal protein gene L32 p<0.05; p<0,001, p<0,0001. doi: 10.1371/journal.pone.0075092.g001 Results Characterization by real-time PCR of PC3c cells indicated that ET1 and OPG, two factors that have been implicated in the pathogenesis of osteosclerotic bone metastases from prostate cancer are overexpressed compared to the parental cell line PC3 (Figure 1C-D) while other factors such as fibroblast growth PC3c cells induce mixed osteoblastic/osteolytic bone lesions In order to test the property of PC3c to induce bone lesions, intra-tibial injections were performed into male SCID mice. Ten weeks after tumor cell inoculation, radiographic analysis revealed that animals bearing PC3c tumors had bone lesions that included osteolytic and osteoblastic components (Figure 2I) while pure osteolytic lesions were observed in animal bearing PC3 tumors after 6 weeks (Figure 2E). The capacity of PC3 and PC3c to induce pure osteolytic and mixed lesions, respectively, was confirmed using 3D micro-CT reconstruction (Figure 2F-G and J-K) (bone volume, BV/TV, Table 1), histology (Figure 2H and L) and histomorphometric analyses of tibiae (skeletal tumor burden, TB/STV; Table 1). As expected no skeletal lesions were observed after PBS injection (Sham animals) (Figure 2A-D, Table 1). By immunohistochemistry, we confirmed, in vivo, that ET-1 and OPG were highly expressed in PC3c tumors (Figure S1, B, D) when compared with PC3 (Figure S2, A, C). (SOST) and Dentin matrix acidic phosphoprotein 1 (DMP1) expression was stimulated after 24h of treatment with PC3 and PC3c conditioned medium respectively while OPG and RANKL expression was not affected (Figure 3D). PC3c cells induce robust osteoblastic reactions upon osteogenic conditions Because PC3c cells induced new bone formation in vivo, we next tested whether they could produce OBs markers. After immunostaining of bone metastatic tissue sections, OPN and BSP were found expressed in PC3c cells in situ (Figure 4B a and b). Moreover after 3 weeks of culture, in vitro, upon osteogenic conditions including ascorbic acid and β- glycerophosphate (Fig 4C b and d), PC3c cells were revealed to be alkaline phosphatase (ALP)-positive (Fig 4Cc) and were able to form a calcified matrix positive for von Kossa staining (Fig 4C d), while PC3 were ALP-negative and did not induce matrix mineralization (Figure 4C a and b). Expression of ALP after ascorbic acid treatment was confirmed in PC3c cells by real-time PCR (Figure 4D). Similarly, OPN was highly expressed in PC3c compared to PC3 cells while OCN was expressed by both cells lines under these experimental conditions (Figure 4D), suggesting high osteomimicry property of PC3c compared to PC3 cells. Finally, Fourier Transform InfraRed Microspectroscopy (FTIRM) study on tumors obtained after subcutaneous injection of PC3c cells revealed the presence of amides I (mainly C=O stretching) and II (mainly N- H bending) and III (mainly C-N stretching and N-H bending) groups of proteins (Figure S4, see I and II red line) that usually correspond to the organic matrix (90% type I collagen) in bone (Figure S4, see Blue and black line). No phosphate or carbonate molecular vibrations were found, indicating the absence of mineral within the PC3c tumor in vivo (Figure S4). On the other side, new bone matrix obtained from mice tibia injected with PC3c cells showed the presence of mineral (Figure S4). Concomitantly to these result, high amount of Type I Collagen was found to be expressed by PC3c when compared with PC3 cells by real-time PCR in vitro (Figure 5A). Additionally, PC3c cells were shown surrounded by typical type I Collagen fibers in situ as judged by electron microscopy (Figure 5B see arrows and higher magnification). All together, these data suggest higher osteomimicry properties for PC3c compared with PC3 cells, thereby explaining at least in part, Expression of pro-osteoblastic factors by PC3c cells From human androgen-resistant prostate cancer cell line PC3, we obtained after single cell population culture in vitro a new subculture cell line named PC3c cells that was chosen PLOS ONE \| www.plosone.org September 2013 \| Volume 8 \| Issue 9 \| e75092 4 New Androgen-Resistant Bone Metastasis Model factor 9 (FGF9) and TGFβ are similarly expressed in both cell lines [8,22,23]. On the other hand, the expression of DKK1 and Noggin, two osteoblast inhibitors (respectively Wnts and bone morphogenetic protein (BMP) inhibitors), is decreased in PC3c versus PC3 (Figure 1 C-D) [24,25]. Moreover, PC3c similarly to PC3 cells expressed factors known to be implicated in prostate cancer osteomimicry such as OPN and Runx2. All together, these results suggest that PC3c cells may potentially induce osteoblastic lesions when compared with PC3 cells that are known to predominantly exhibit osteolytic lesions in bone. Table 1. Histomorphometric analysis of tibia with metastases induced by injection of PC3 and PC3c cells. BV/TV (%) TB/STV (%) Sham (n=10) 22,4 +/- 2,9 0 PC3 (n=6) 3,6 +/- 4,1 72,6 +/- 6,6* PC3c (n=8) 39,5 +/- 3,0$ 36,6 +/- 4,5*$ BV/TV: bone volume/ total volume. TB/STV: tumor burden/soft tissue volume. Sham were performed as control. n is the number of legs with bone metastases. P<0,05; **P<0,001 compared with Sham; $P<0,001 compared with PC3. doi: 10.1371/journal.pone.0075092.t001 Table 1. Histomorphometric analysis of tibia with metastases induced by injection of PC3 and PC3c cells. BV/TV: bone volume/ total volume. TB/STV: tumor burden/soft tissue volume. Sham were performed as control. n is the number of legs with bone metastases. P<0,05; **P<0,001 compared with Sham; $P<0,001 compared with PC3. doi: 10.1371/journal.pone.0075092.t001 Bone remodeling stimulation by PC3c cells Given these data, we next asked whether PC3c could alter the bone resorbing cells, the osteoclasts (OCs) and the bone forming cells, the osteoblasts (OBs). Treatment of primary mouse bone marrow cells with RANKL, macrophage colony stimulating factor (M-CSF) and with the conditioned medium of PC3c stimulated more the formation of tartrate resistant acid phosphatase (TRAP)-positive multinucleated OCs compared with that observed with the conditioned medium of PC3 cells and untreated cells (Ct) (Figure 3A). On the other side, treatment of primary mouse calvaria cells cultured in osteogenic conditions with the conditioned medium of PC3c had less inhibitory effect on OB differentiation than conditioned medium of PC3 cells compared with untreated cells (Ct) (Figure 3B). Indeed, a high number of OBs was visualized using OPN immunostaining in vivo (Figure 4A a-b). Interestingly, PC3 conditioned medium stimulated OPG and RANKL expression by primary OBs while PC3c conditioned medium decreased OPG production leading to a stronger increase of RANKL/OPG ratio by OBs treated with PC3c conditioned medium compared with that of PC3 cells (Figure 3C). Consistent with these in vitro results, TRAP staining of tibial sections of metastatic legs from animals bearing PC3c showed high number of TRAP-positive multinucleated OCs compared with that observed in PC3 and Sham animals (Figure S3). Finally, semi-quantitative PCR performed on the osteocyte cell line, MLO-Y4, allowed us to show that sclerostin September 2013 \| Volume 8 \| Issue 9 \| e75092 PLOS ONE \| www.plosone.org 5 New Androgen-Resistant Bone Metastasis Model Figure 2. Induction of lytic and mixed bone lesions by PC3 and PC3c cells respectively after intratibial injection. (A) PC3 and PC3c cells were inoculated into male SCID mice; 10 weeks post inoculation, radiography revealed pure osteolytic lesions in mice injected with PC3 cells (n=6) (E) and mixed lesions in mice injected with PC3c cells (n=8) (I) compared to mice injected with PBS (n=10) (A) (see (lysis) and white arrows (formation)). (B,C- F,G-J,K) Three-dimensional micro-CT reconstructions of tibiae and (D, H, L) histology after Goldner’s Trichrome staining confirmed the radiography results. T: Tumor; NB: New Bone. doi: 10.1371/journal.pone.0075092.g002 New Androgen-Resistant Bone Metastasis Model Figure 2. Induction of lytic and mixed bone lesions by PC3 and PC3c cells respectively after intratibial injection. Bone remodeling stimulation by PC3c cells (A) PC3 and PC3c cells were inoculated into male SCID mice; 10 weeks post inoculation, radiography revealed pure osteolytic lesions in mice injected with PC3 cells (n=6) (E) and mixed lesions in mice injected with PC3c cells (n=8) (I) compared to mice injected with PBS (n=10) (A) (see (lysis) and white arrows (formation)). (B,C- F,G-J,K) Three-dimensional micro-CT reconstructions of tibiae and (D, H, L) histology after Goldner’s Trichrome staining confirmed the radiography results. T: Tumor; NB: New Bone. doi: 10.1371/journal.pone.0075092.g002 their capacity to induce mixed osteoblastic/osteolytic bone lesions. their capacity to induce mixed osteoblastic/osteolytic bone lesions. Discussion In this study, we have established and characterized a new androgen-independent prostate cancer cell line (PC3c) which rapidly gives mixed bone lesions in male SCID mice, 10 weeks PLOS ONE \| www.plosone.org 6 September 2013 \| Volume 8 \| Issue 9 \| e75092 6 Figure 3. Stimulation of osteoclasts and osteoblasts by PC3c cells in vitro. (A) Primary mouse bone marrow cells were ultured in the presence of RANKL and M-CSF and treated or not (Ct) with conditioned medium obtained from PC3 and PC3c cells. More OCs (white arrow) were formed in cultures treated with PC3c conditioned medium compared to cultures treated with PC3 onditioned medium and Ct (ANOVA, p<0.0001). (B) Primary mouse calvaria cell cultures were treated from day 1-21 with onditioned medium obtained from PC3 and PC3c cell. Mineralized bone nodules were present and visualized by von Kossa taining at day 21 (see mineral in black, white arrows). Mineralized bone nodule formation was decreased when primary cells were reated with conditioned medium from any of the PC3/PC3c cells (compared with non-treated (Ct) cells); the decrease was less when PC3c cell conditioned medium was used (compared with PC3) (ANOVA, p<0.001 versus Ct and versus PC3). (C) PC3 onditioned media stimulated the expression of OPG and RANKL in primary OBs compared with non-treated (Ct) while PC3c onditioned media only inhibits the expression of OPG compared with Ct leading to an higher RANKL/OPG ratio in PC3c conditions. D) Detection by real-time PCR of SOST, DMP1, OPG and RANKL mRNA expression in MLO-Y4 cells treated with PC3 and PC3c onditioned medium. Results are plotted as the mean number of OC ± SD and OB nodules ± SD of three wells for controls and each ondition and are representative of two independent experiments Genes expression was assessed by real time PCR on triplicate New Androgen-Resistant Bone Metastasis Model New Androgen-Resistant Bone Metastasis Model Figure 3. Stimulation of osteoclasts and osteoblasts by PC3c cells in vitro. (A) Primary mouse bone marrow cells were cultured in the presence of RANKL and M-CSF and treated or not (Ct) with conditioned medium obtained from PC3 and PC3c cells. More OCs (white arrow) were formed in cultures treated with PC3c conditioned medium compared to cultures treated with PC3 conditioned medium and Ct (ANOVA, p<0.0001). (B) Primary mouse calvaria cell cultures were treated from day 1-21 with conditioned medium obtained from PC3 and PC3c cell. Discussion Mineralized bone nodules were present and visualized by von Kossa staining at day 21 (see mineral in black, white arrows). Mineralized bone nodule formation was decreased when primary cells were treated with conditioned medium from any of the PC3/PC3c cells (compared with non-treated (Ct) cells); the decrease was less when PC3c cell conditioned medium was used (compared with PC3) (ANOVA, p<0.001 versus Ct and versus PC3). (C) PC3 conditioned media stimulated the expression of OPG and RANKL in primary OBs compared with non-treated (Ct) while PC3c conditioned media only inhibits the expression of OPG compared with Ct leading to an higher RANKL/OPG ratio in PC3c conditions. (D) Detection by real-time PCR of SOST, DMP1, OPG and RANKL mRNA expression in MLO-Y4 cells treated with PC3 and PC3c conditioned medium. Results are plotted as the mean number of OC ± SD and OB nodules ± SD of three wells for controls and each condition and are representative of two independent experiments. Genes expression was assessed by real-time PCR on triplicate samples and normalized against that of the ribosomal protein gene L32 p<0.05; p<0,001, p<0,0001. doi: 10.1371/journal.pone.0075092.g003 September 2013 \| Volume 8 \| Issue 9 \| e75092 7 7 PLOS ONE \| www.plosone.org New Androgen-Resistant Bone Metastasis Model Figure 4. PC3c cells osteomimicry properties. (A) Immunodetection of OPN in OB in bone metastases induced by PC3c cells (see OBs a and b magnifications of a; see black arrows) (A) and in tumors cells (B a). Similarly to OPN, BSP expression is detected in tumors cells in vivo (B, b). (C) Similarly to primary mouse calvaria cells, PC3 and PC3c were cultured in osteogenic conditions for 21 days. ALP (a, c) and von Kossa staining (b, d) show high expression of ALP (c) and mineralization (white arrow) (d) in PC3c cells while no ALP expression (a) and mineralization were detected in PC3 cells (b). (D) Detection by real-time PCR of OPN, ALP and OCN mRNA expression in PC3c and PC3 cells cultured in osteogenic conditions for 21 days. Gene expression was assessed by real-time PCR on triplicate samples and normalized against that of the ribosomal protein gene L32 p<0,001. Bar=200µm T: Figure 4. PC3c cells osteomimicry properties. (A) Immunodetection of OPN in OB in bone metastases induced by PC3c cells (see OBs a and b magnifications of a; see black arrows) (A) and in tumors cells (B a). Discussion (A) Detection by real-time PCR of type I collagen mRNA expression in PC3c and PC3 cells cultured in normal conditions. Gene expression was assessed by real-time PCR on triplicate samples and normalized against that of the ribosomal protein gene L32 p<0,05. (B) Visualization of type I collagen in PC3c cells cultured on glass coverslip by electron microscopy (see black arrows). doi: 10.1371/journal.pone.0075092.g005 Figure 5. PC3c cells highly expressed type I collagen. (A) Detection by real-time PCR of type I collagen mRNA expression in PC3c and PC3 cells cultured in normal conditions. Gene expression was assessed by real-time PCR on triplicate samples and normalized against that of the ribosomal protein gene L32 p<0,05. (B) Visualization of type I collagen in PC3c cells cultured on glass coverslip by electron microscopy (see black arrows). doi: 10 1371/journal pone 0075092 g005 after intra-tibial tumor cells injection. This model may be useful to study cellular and molecular mechanisms that differ between androgen-dependent and androgen-independent prostate carcinomas when they metastasize to bone. In prostate cancer, androgen blockade strategies are usually used to treat osteoblastic bone metastases. However, responses to these therapies are often brief due to post-traductional modifications or mutations of AR that reduce ligand binding and inevitably lead to CRPC [26,27]. The role of AR pathway in the osteoblastic progression of prostate cancer is poorly understood because available models of mixed and pure osteoblastic lesions (C4-2B, PCa2B, VCaP) are mainly androgen responsive like [28,29]. Consequently, an active AR pathway is believed to be implicated in the osteoblastic progression of prostate cancer. However, concerning our PC3c model, this assumption does not occur as both cells lines PC3 and PC3c cell lines do not express AR. On the other side, AR negative models commonly used as PC3 and DU145 cells that derived from CRPC patients induced pure osteolytic lesions and do not reproduce what it is observed in clinic [30,31]. Then, in order to meet the need for clinically relevant models of prostate cancer-associated bone lesions, more recent hormone-independent models have been developed like MDA PCa 118 and Ace-1 that similarly to the PC3c model do not express AR and induce mixed lesions [22,32]. Nevertheless, osteogenesis that was induced by FGF-9 in the MDA PCa 118 model was not implicated in the osteoblastic response induced by PC3c as FGF-9 expression was not statistically significantly modulated between PC3 and PC3c cells. Discussion Similarly to OPN, BSP expression is detected in tumors cells in vivo (B, b). (C) Similarly to primary mouse calvaria cells, PC3 and PC3c were cultured in osteogenic conditions for 21 days. ALP (a, c) and von Kossa staining (b, d) show high expression of ALP (c) and mineralization (white arrow) (d) in PC3c cells while no ALP expression (a) and mineralization were detected in PC3 cells (b). (D) Detection by real-time PCR of OPN, ALP and OCN mRNA expression in PC3c and PC3 cells cultured in osteogenic conditions for 21 days. Gene expression was assessed by real-time PCR on triplicate samples and normalized against that of the ribosomal protein gene L32 p<0,001. Bar=200µm T: Tumor; OB: osteoblasts; NB: New Bone. doi: 10.1371/journal.pone.0075092.g004 Figure 4. PC3c cells osteomimicry properties. (A) Immunodetection of OPN in OB in bone metastases induced by PC3c cells (see OBs a and b magnifications of a; see black arrows) (A) and in tumors cells (B a). Similarly to OPN, BSP expression is detected in tumors cells in vivo (B, b). (C) Similarly to primary mouse calvaria cells, PC3 and PC3c were cultured in osteogenic conditions for 21 days. ALP (a, c) and von Kossa staining (b, d) show high expression of ALP (c) and mineralization (white arrow) (d) in PC3c cells while no ALP expression (a) and mineralization were detected in PC3 cells (b). (D) Detection by real-time PCR of OPN, ALP and OCN mRNA expression in PC3c and PC3 cells cultured in osteogenic conditions for 21 days. Gene expression was assessed by real-time PCR on triplicate samples and normalized against that of the ribosomal protein gene L32 p<0,001. Bar=200µm T: Tumor; OB: osteoblasts; NB: New Bone. doi: 10.1371/journal.pone.0075092.g004 September 2013 \| Volume 8 \| Issue 9 \| e75092 PLOS ONE \| www.plosone.org 8 Figure 5. PC3c cells highly expressed type I collagen. (A) Detection by real-time PCR of type I collagen mRNA expression in PC3c and PC3 cells cultured in normal conditions. Gene expression was assessed by real-time PCR on triplicate samples and normalized against that of the ribosomal protein gene L32 p<0,05. (B) Visualization of type I collagen in PC3c cells cultured on glass coverslip by electron microscopy (see black arrows). doi: 10.1371/journal.pone.0075092.g005 New Androgen-Resistant Bone Metastasis Model New Androgen-Resistant Bone Metastasis Model Figure 5. PC3c cells highly expressed type I collagen. New Androgen-Resistant Bone Metastasis Model Supporting Information OC differentiation. Based on OC and OB in vitro assay and TRAP staining in vivo, it appears that PC3c cells can also over stimulate osteoclastogenesis directly by acting on OC precursors and indirectly by increasing the RANKL/OPG ratio by OBs conducting to a general bone remodeling stimulation. On the other side, the RANKL/OPG ratio was not modulated by osteocytes like MLO-Y4 cells. Concerning osteocyte involvement into prostate cancer progression in bone, we found that the expression of SOST, an inhibitor of Wnt signaling, was induced by PC3 cells conditioned medium, suggesting a direct effect of osteocytes on OB differentiation during tumor bone progression. Meanwhile the expression of DMP1, a member of the small integrin-binding ligand N-linked glycoproteins (SIBLINGs) that is involved in phosphatemia regulation, was only stimulated by PC3c cells conditioned media which, combined with the expression of type I collagen by the tumor cells, may contribute to the formation of the new bone mineralized matrix observed in vivo and by infrared microspectroscopy [36,37]. Figure S1. Epithelial phenotype of PC3c cells. Detection by real-time PCR of E-Cadherin, N-Cadherin and vimentin mRNA expression in PC3 and PC3c cancer cells lines. Genes expression was assessed by real-time PCR on triplicate samples and normalized against that of the ribosomal protein gene L32 p<0,001, *p<0,0001. Figure S2. ET-1 and OPG expression by PC3c cells in vivo. Immunostaining for ET-1 (A, B) and OPG (C, D) is higher in bone metastases induced by PC3c cells (B, D) compared to PC3 cells (A, C). Bar=200µm T: Tumor. (TIF) Figure S3. Visualization of TRAP positive OC in bone metastases induced by PC3 and PC3c cells. TRAP (red) staining of OCs (black arrow) realized in sections of tibiae taken from mice injected with Sham (A), PC3 (B) and PC3c cells (C). Bar=200µm T: Tumor. (TIF) Finally, PC3c cells had strong osteomimicry properties compared to PC3 cells, as judged by the expression of bone- associated markers such as OPN, BSP, Runx2, and type I collagen and by the ability of this cell line to form calcified matrix under osteogenic conditions, while PC3 are negative for both. Bone matrix proteins are also known to influence tumor localization in the skeleton and the phenotype of skeletal lesions. For example, OPN and BSP have been associated with breast and prostate cancer cells affinity for bone, migration and survival [38]. Supporting Information Moreover, differential expression of BSP and OPN has been shown to be implicated in the switch between osteolytic versus osteoblastic bone lesions. Indeed, strong BSP expression has been found mainly in prostate when compared with breast cancer lesions while high level of OPN was essentially shown in breast versus prostate cancer. Thus, associating BSP to sclerotic lesions as a stimulator of bone mineralization and OPN to osteolytic bone lesions as an activator of OC [39]. Differences into OPN expression level between PC3 and PC3c cells under normal conditions cannot explain the osteoblastic pattern observed in PC3c model and it is likely that OPN and BSP contribute to the osteomimicry properties of PC3c cells by promoting cell attachment and bone matrix mineralization, respectively. Additionally, type I collagen and ALP may also be part of this osteosclerotic pattern induced by PC3c cells by being involved in the new bone matrix formation and mineralization respectively. Figure S4. Identification of amide groups in PC3c subcutaneous tumors. (A) IR spectra obtained on PC3c tumors (red curve), on cortical tibial bone matrix bone from Sham (Black curve) or PC3c mice (Blue curve) illustrates the presence of amide I and II and III groups usually corresponding to organic matrix (mainly to type I collagen) observed in bone matrix. As expected, mineral was shown by IR on bone matrix of tibia injected by PBS (Sham) (Black curve) or PC3c cells (Blue curve) (see peaks corresponding to ν3PO4, ν2CO3 and ν4PO4 groups) while it was not present in PC3c tumors (red curve; see ν3PO4, ν2CO3 and ν4PO4 groups). (B) Bone quality analysis performed by infrared microspectroscopy. No differences were observed between normal cortical bone and the new bone (NB) induced by PC3c tumor cells (n=2). (TIF) Acknowledgements In summary, we have established a new AR-negative prostate cancer cell line that is derived from PC3 cells and that recapitulates the osteoblastic phenotype of prostate cancer in bone. This new model may be helpful for the identification of new signaling pathways that are involved in the progression of prostate cancer in bone and may provide a valuable tool for investigating the mechanisms of androgen-independent prostate cancer cells osteomimicry in osteoblastic bone lesions. Finally, it can also provide a clinically relevant experimental model for testing the effects of new treatments for bone metastases. The authors thank the CeCIL platform (Faculté de Médecine Laennec, Lyon, France) for technical assistance. They also thank Marianne Till and Blandine Deux. Discussion Interestingly, when compared with PC3, PC3c cells highly expressed ET-1, a mitogenic factor for OB [33,34]. ET-1 is known to contribute to osteoblastic progression of prostate cancer cells by stimulating OB proliferation through the negative regulation of the inhibitor of the Wnt signaling, DKK1 which may explain, at least in part, the decrease of DKK1 observed in PC3c cells [35]. Moreover, a decrease of DKK1 serum level in patients with advanced prostate cancer has been reported to be associated with occurrence of osteoblastic lesions which suggests that DKK1 inhibition in PC3c cells may contribute to the osteoblastic phenotype induced by these cells [12]. BMPs have also been implicated in the formation of new bone induced by prostate cancer and the inhibition of BMPs by their inhibitor Noggin in C4-2B cells induces a decrease in the osteoblastic response, suggesting that the low expression level of Noggin in PC3c cells compared with PC3 cells may also contribute to the osteoblastic lesions induced by PC3c cells [4,10]. Finally the over-expression of OPG may also contribute to the osteoblastic response induced by PC3c cells by limiting PLOS ONE \| www.plosone.org September 2013 \| Volume 8 \| Issue 9 \| e75092 9 New Androgen-Resistant Bone Metastasis Model References Bagnato A, Loizidou M, Pflug BR, Curwen J, Growcott J (2011) Role of the endothelin axis and its antagonists in the treatment of cancer. Br J Pharmacol 163: 220-233. doi:10.1111/j.1476-5381.2011.01217.x. PubMed: 21232046. 14. Flajollet S, Tian T, Flourens A, Tomavo N, Villers A et al. (2011) Abnormal expression of the ERG transcription factor in prostate cancer cells activates osteopontin. Mol Cancer Res 9: 914-924. doi: 10.1158/1541-7786.MCR-10-0537. PubMed: 21669963. 34. von Schroeder HP, Veillette CJ, Payandeh J, Qureshi A, Heersche JN (2003) Endothelin-1 promotes osteoprogenitor proliferation and differentiation in fetal rat calvarial cell cultures. Bone 33: 673-684. doi: 10.1016/S8756-3282(03)00215-1. PubMed: 14555273. 15. Tian TV, Tomavo N, Huot L, Flourens A, Bonnelye E et al. (2013) Identification of novel TMPRSS2:ERG mechanisms in prostate cancer metastasis: involvement of MMP9 and PLXNA2. Oncogene [Epub ahead of print]. doi:10.1038/onc.2013.176. PubMed: 23708657. 35. Clines GA, Mohammad KS, Bao Y, Stephens OW, Suva LJ et al. (2007) Dickkopf homologue 1 mediates endothelin-1-stimulated new bone formation. Mol Endocrinol 21: 486-498. PubMed: 17068196. 16. Bonnelye E, Chabadel A, Saltel F, Jurdic P (2008) Dual effect of strontium ranelate: stimulation of osteoblast differentiation and inhibition of osteoclast formation and resorption in vitro. Bone 42: 129-138. doi:10.1016/j.bone.2007.08.043. PubMed: 17945546. 36. Feng JQ, Ward LM, Liu S, Lu Y, Xie Y et al. (2006) Loss of DMP1 causes rickets and osteomalacia and identifies a role for osteocytes in mineral metabolism. Nat Genet 38: 1310-1315. doi:10.1038/ng1905. PubMed: 17033621. 17. Fradet A, Sorel H, Bouazza L, Goehrig D, Dépalle B et al. (2011) Dual function of ERRα in breast cancer and bone metastasis formation: implication of VEGF and osteoprotegerin. Cancer Res 71: 5728-5738. doi:10.1158/0008-5472.CAN-11-1431. PubMed: 21734015. 37. Harris SE, Gluhak-Heinrich J, Harris MA, Yang W, Bonewald LF et al. (2007) DMP1 and MEPE expression are elevated in osteocytes after mechanical loading in vivo: theoretical role in controlling mineral quality in the perilacunar matrix. J Musculoskelet Neuronal Interact 7: 313-315. PubMed: 18094489. 18. Bonnelye E, Merdad L, Kung V, Aubin JE (2001) The orphan nuclear estrogen receptor-related receptor alpha (ERRalpha) is expressed throughout osteoblast differentiation and regulates bone formation in vitro. J Cell Biol 153: 971-984. doi:10.1083/jcb.153.5.971. PubMed: 11381083. 38. Bellahcène A, Castronovo V, Ogbureke KU, Fisher LW, Fedarko NS (2008) Small integrin-binding ligand N-linked glycoproteins (SIBLINGs): multifunctional proteins in cancer. Nat Rev Cancer 8: 212-226. doi: 10.1038/nrc2345. PubMed: 18292776. 19. References (1995) Identification of endothelin-1 in the pathophysiology of metastatic adenocarcinoma of the prostate. Nat Med 1: 944-949. doi: 10.1038/nm0995-944. PubMed: 7585222. 27. Saraon P, Jarvi K, Diamandis EP (2011) Molecular alterations during progression of prostate cancer to androgen independence. Clin Chem 57: 1366-1375. doi:10.1373/clinchem.2011.165977. PubMed: 21956922. 9. Dai J, Hall CL, Escara-Wilke J, Mizokami A, Keller JM et al. (2008) Prostate cancer induces bone metastasis through Wnt-induced bone morphogenetic protein-dependent and independent mechanisms. Cancer Res 68: 5785-5794. doi:10.1158/0008-5472.CAN-07-6541. PubMed: 18632632. 28. Navone NM, Olive M, Ozen M, Davis R, Troncoso P et al. (1997) Establishment of two human prostate cancer cell lines derived from a single bone metastasis. Clin Cancer Res 3: 2493-2500. PubMed: 9815652. 10. Logothetis CJ, Lin SH (2005) Osteoblasts in prostate cancer metastasis to bone. Nat Rev Cancer 5: 21-28. doi:10.1038/nri1529. PubMed: 15630412. 29. Korenchuk S, Lehr JE, MClean L, Lee YG, Whitney S et al. (2001) VCaP, a cell-based model system of human prostate cancer. In Vivo 15: 163-168. PubMed: 11317522. 11. Hall JM, Korach KS (2003) Stromal cell-derived factor 1, a novel target of estrogen receptor action, mediates the mitogenic effects of estradiol in ovarian and breast cancer cells. Mol Endocrinol 17: 792-803. doi: 10.1210/me.2002-0438. PubMed: 12586845. 30. Stone KR, Mickey DD, Wunderli H, Mickey GH, Paulson DF (1978) Isolation of a human prostate carcinoma cell line (DU 145). Int J Cancer 21: 274-281. doi:10.1002/ijc.2910210305. PubMed: 631930. 12. Chen G, Shukeir N, Potti A, Sircar K, Aprikian A et al. (2004) Up- regulation of Wnt-1 and beta-catenin production in patients with advanced metastatic prostate carcinoma: potential pathogenetic and prognostic implications. Cancer 101: 1345-1356. doi:10.1002/cncr. 20518. PubMed: 15316903. 31. Kaighn ME, Narayan KS, Ohnuki Y, Lechner JF, Jones LW (1979) Establishment and characterization of a human prostatic carcinoma cell line (PC-3). Invest Urol 17: 16-23. PubMed: 447482. 32. LeRoy BE, Thudi NK, Nadella MV, Toribio RE, Tannehill-Gregg SH et al. (2006) New bone formation and osteolysis by a metastatic, highly invasive canine prostate carcinoma xenograft. Prostate 66: 1213-1222. doi:10.1002/pros.20408. PubMed: 16683269. 13. Huang WC, Xie Z, Konaka H, Sodek J, Zhau HE et al. (2005) Human osteocalcin and bone sialoprotein mediating osteomimicry of prostate cancer cells: role of cAMP-dependent protein kinase A signaling pathway. Cancer Res 65: 2303-2313. doi: 10.1158/0008-5472.CAN-04-3448. PubMed: 15781644. 33. References 1. Vela I, Gregory L, Gardiner EM, Clements JA, Nicol DL (2007) Bone and prostate cancer cell interactions in metastatic prostate cancer. BJU Int 99: 735-742. doi:10.1111/j.1464-410X.2006.06670.x. PubMed: 17166237. 21. Chauchereau A, Al Nakouzi N, Gaudin C, Le Moulec S, Compagno D et al. (2011) Stemness markers characterize IGR-CaP1, a new cell line derived from primary epithelial prostate cancer. Exp Cell Res 317: 262-275. doi:10.1016/j.yexcr.2010.10.012. PubMed: 20974126. 2. Roudier MP, Morrissey C, True LD, Higano CS, Vessella RL et al. (2008) Histopathological assessment of prostate cancer bone osteoblastic metastases. J Urol 180: 1154-1160. doi:10.1016/j.juro. 2008.04.140. PubMed: 18639279. 22. Li ZG, Mathew P, Yang J, Starbuck MW, Zurita AJ et al. (2008) Androgen receptor-negative human prostate cancer cells induce osteogenesis in mice through FGF9-mediated mechanisms. J Clin Invest 118: 2697-2710. PubMed: 18618013. 3. Keller ET, Brown J (2004) Prostate cancer bone metastases promote both osteolytic and osteoblastic activity. J Cell Biochem 91: 718-729. doi:10.1002/jcb.10662. PubMed: 14991763. 23. Kamiya N, Suzuki H, Endo T, Takano M, Yano M et al. (2011) Significance of serum osteoprotegerin and receptor activator of nuclear factor κB ligand in Japanese prostate cancer patients with bone metastasis. Int J Clin Oncol 16: 366-372. doi:10.1007/ s10147-011-0193-7. PubMed: 21327451. j 4. Weilbaecher KN, Guise TA, McCauley LK (2011) Cancer to bone: a fatal attraction. Nat Rev Cancer 11: 411-425. doi:10.1038/nrc3055. PubMed: 21593787. 24. Secondini C, Wetterwald A, Schwaninger R, Thalmann GN, Cecchini MG et al. (2011) The role of the BMP signaling antagonist noggin in the development of prostate cancer osteolytic bone metastasis. PLOS ONE 6: e16078. doi:10.1371/journal.pone.0016078. PubMed: 21249149. 5. Theriault RL (2012) Biology of bone metastases. Cancer Control 19: 92-101. PubMed: 22487971. 6. Coleman R, Gnant M, Morgan G, Clezardin P (2012) Effects of bone- targeted agents on cancer progression and mortality. J Natl Cancer Inst 104: 1059-1067. doi:10.1093/jnci/djs263. PubMed: 22752060. 6: e16078. doi:10.1371/journal.pone.0016078. PubMed: 2124914 25. Hall CL, Bafico A, Dai J, Aaronson SA, Keller ET (2005) Prostate cancer cells promote osteoblastic bone metastases through Wnts. Cancer Res 65: 7554-7560. PubMed: 16140917. j j 7. Jimeno A (2004) Atrasentan: targeting the endothelin axis in prostate cancer. Expert Opin Investig Drugs 13: 1631-1640. doi: 10.1517/13543784.13.12.1631. PubMed: 15566319. 26. Anbalagan M, Huderson B, Murphy L, Rowan BG (2012) Post- translational modifications of nuclear receptors and human disease. Nucl Recept Signal 10: e001. PubMed: 22438791. 8. Nelson JB, Hedican SP, George DJ, Reddi AH, Piantadosi S et al. Author Contributions Conceived and designed the experiments: EB PC. Performed the experiments: AF HS BD CMS DF EB. Analyzed the data: EB AF DF. Contributed reagents/materials/analysis tools: HF AT AB VC. Wrote the manuscript: EB AF. PLOS ONE \| www.plosone.org 10 September 2013 \| Volume 8 \| Issue 9 \| e75092 New Androgen-Resistant Bone Metastasis Model New Androgen-Resistant Bone Metastasis Model References Bellows CG, Aubin JE, Heersche JN, Antosz ME (1986) Mineralized bone nodules formed in vitro from enzymatically released rat calvaria cell populations. Calcif Tissue Int 38: 143-154. doi:10.1007/ BF02556874. PubMed: 3085892. 39. Carlinfante G, Vassiliou D, Svensson O, Wendel M, Heinegård D et al. (2003) Differential expression of osteopontin and bone sialoprotein in bone metastasis of breast and prostate carcinoma. Clin Exp Metastasis 20: 437-444. doi:10.1023/A:1025419708343. PubMed: 14524533. 20. Kato Y, Windle JJ, Koop BA, Mundy GR, Bonewald LF (1997) Establishment of an osteocyte-like cell line, MLO-Y4. J Bone Miner Res 12: 2014-2023. PubMed: 9421234. 11 PLOS ONE \| www.plosone.org September 2013 \| Volume 8 \| Issue 9 \| e75092
https://openalex.org/W4383994480	https://zenodo.org/records/8138692/files/TYPES%20AND%20CAUSES%20OF%20CITRULLINEMIA%20TYPE.pdf	English	null	Types and Causes of Citrullinemia Type I, Type II, Symptoms of Citrullinemia Type I, Type II, Diagnosis of Citrullinemia, Clinical Evaluation of Citrullinemia and Treatment of Cutrullinemia	Zenodo (CERN European Organization for Nuclear Research)	2,023	cc-by	3,317	ABSTRACT Citrullinemia is a rare genetic disorder that affects the metabolism of the amino acid arginine. It is manifested by the collection of citrulline, a toxic substance, in the blood and tissues due to the deficiency or malfunctioning of an enzyme called argininosuccinate synthetase. This article gives information about the understanding of citrullinemia, along with its causes, symptoms, diagnosis, treatment, and ongoing research efforts. Keywords: Arginine, arginosuccinate synthase, genetic disorders, urea cycle, ammonia, autosomal recessive pattern, hyperammonemia, neurological symptoms, poor feeding, vomiting, lethargy, irritability, seizures, hypotonia, respiratory distress, apnea, jaundice, citrin deficiency, delayed motor skills, cognitive impairement, speech difficulties, hepatomegaly, hypoglycemia, fatigue, weakness, dizziness, difficult concentration, confusion, loss of consciousness, fasting, disorientation, coma, alcohol consumption, low-protein diet, personality changes, dietary modifications, avoiding known triggers, poor appetite, neurological problems, metabolic markers, genetic testing, computed tomography, magnetic resonance imaging, liver damage, electrolyte imbalance, blood gas analysis, ornithine transcarbamylase deficiency, argino succinic aciduria, homocitrulline, N-acetyl glutamate synthase, organic acidemia, fatty acid oxidation disorders, mitoichondrial disorders, essential amino acids, vitamins minerals, ammonia scavenging drugs, sodium phenylacetate sodium benzoate, ammonia detoxification, liver transplantation and enzyme replacement therapy Journal of Advances in Clinical Pharmacology Volume 1 Issue 2 Journal of Advances in Clinical Pharmacology Volume 1 Issue 2 Types and Causes of Citrullinemia Type I, Type II, Symptoms of Citrullinemia Type I, Type II, Diagnosis of Citrullinemia, Clinical Evaluation of Citrullinemia and Treatment of Cutrullinemia Vinayasree.C1, Mohan Naidu.K2, Muralinath.E3*, Amrutham Sandeep4, Venkat Naveen .A5, Guruprasad.M6, Sravani Pragna.K7 1Assistant Professor, College of Veterinary Science, Korutla, PVNRTVU, Telengana, Hyderabad, India 2,5Veterinay Doctor, College of veterinary science, Proddatur, Andhra Pradesh, India 3Associate Professor, College of veterinary science, Proddatur, Andhra Pradesh, India 4Staff, Indian Veterinary Research Institute (IVRI), Izathnagar, Barielly district, Uttar Pradesh, India 6Assistant General Manager, Vaishnavi Bio pharma pvt. Ltd., Telanagana, India 7Assistant Professor, Contract Teaching Faculty, College of veterinary, science, Proddatur, Andhra Pradesh, India 6Assistant General Manager, Vaishnavi Bio pharma pvt. Ltd., Telanagana, India 7Assistant Professor, Contract Teaching Faculty, College of veterinary, science, Proddatur, Andhra Pradesh, India Hyperammonemia Citrullinemia is categorized into two types: Citrullinemia is categorized into two types: Elevated levels of ammonia in the blood is a hallmark feature of citrullinemia. High ammonia levels act as toxic to the brain and cause various neurological symptoms.  Citrullinemia type I  Citrullinemia type II Respiratory distress Some infants may show difficulty breathing, rapid breathing, or episodes of apnea (brief pauses in breathing). Seizures High levels of ammonia lead to the occurrence of seizures, which may present as jerking movements, loss of consciousness, or staring spells. Jaundice Yellowing of the skin and eyes (jaundice) may be present because of the liver dysfunction. and symptoms associated with type I citrullinemia: accumulation and subsequent toxicity.[1- 4] Hypotonia Babies with citrullinemia often show poor muscle tone (hypotonia), which can make them feel floppy or “limp.” The specific genetic mutations essential for type I citrullinemia can vary among affected individuals. However, most cases are affected by “loss-of-function” mutations, which severely impair or entirely abolish the production or activity of the ASS enzyme. As a result, the disruption of urea cycle happens and ammonia cannot be efficiently changed to urea for elimination from the body. INTRODUCTION enzyme critical for the breakdown of nitrogen-containing compounds in the body. Without functional ASS, the body cannot effectively change the amino acid citrulline into arginine, leading to its Citrullinemia is an autosomal recessive disorder, meaning that it occurs when an individual inherits two mutated copies of the gene responsible for producing argininosuccinate synthetase (ASS), an HBRP Publication Page 8-15 2023. All Rights Reserved Page 8 Journal of Advances in Clinical Pharmacology Volume 1 Issue 2 Journal of Advances in Clinical Pharmacology Volume 1 Issue 2 Journal of Advances in Clinical Pharmacology Volume 1 Issue 2 Poor feeding and vomiting yp Type I citrullinemia is a rare genetic disorder manifested by a deficiency of the enzyme argininosuccinate synthetase (ASS). This enzyme is responsible for the urea cycle, which assists in removing ammonia, a toxic waste product, from the body. Without sufficient ASS activity, ammonia retaines in the blood, leading to hyperammonemia and subsequent neurological symptoms. Infants with citrullinemia exhibit difficulty feeding, and they may vomit frequently. These symptoms result in poor weight gain and failure to thrive. Lethargy and irritability Babies with citrullinemia may look normally tired, sluggish, and irritable. They may have a weak cry and reduced muscle tone. Type I citrullinemia occurs by mutations in the ASS1 gene, which provides instructions for producing the ASS enzyme. These mutations result in the disturbance of the normal function of the enzyme, resulting in its reduced or absent activity. Because the ailment is autosomal recessive, a person must inherit two copies of the defective gene (one from each parent) in order to become afflicted. Symptoms Without proper treatment, citrullinemia result in developmental delays and intellectual disability. The symptoms of type I citrullinemia usually observe within the first few days of life, but in some cases, they may develop later in infancy or childhood. The severity of symptoms can vary among affected individuals. Here are some common signs Failure to thrive Infants with type 2 citrullinemia may have difficulties gaining weight and growing at a normal rate, leading to failure to thrive. Clinical Evaluation The doctor will review the patient’s medical history and perform a thorough physical examination. They will inquire about symptoms such as poor appetite, vomiting, lethargy, developmental delay, Adult-Onset Citrullinemia Unlike the early-onset forms, adult-onset citrullinemia appears later in life, specifically during adolescence or adulthood. It is happened by mutations in the SLC25A13 gene, similar to Citrullinemia Type II. The onset of symptoms is often triggered by factors namely alcohol consumption, fasting, or a high-protein diet. One of the essential symptoms is more levels of ammonia in the blood (hyperammonemia). This happens because the liver is not able to process ammonia in an effective manner, leading to its retention in the bloodstream. Developmental delay Infants with type 2 citrullinemia may show developmental delays, namely delayed motor skills, cognitive impairment, and speech difficulties. The symptoms of adult-onset citrullinemia such as episodes of hyperammonemia, confusion, disorientation, personality changes, and coma. These symptoms may be mistaken for other conditions namely psychiatric disorders or liver disease. Treatment is related to the management of the symptoms with the help of a combination of dietary modifications, medications, and avoiding known triggers. DIAGNOSIS[20,21] Low blood sugar levels (hypoglycemia) can happen in affected individuals. This may result in symptoms namely fatigue, weakness, dizziness, and difficulty concentrating. The diagnosis of citrullinemia typically involves several steps and may need input from different doctors. Here is an overview of the diagnostic process for citrullinemia: Citrullinemia Type II[13-19] Citrullinemia Type II[13-19] HBRP Publication Page 8-15 2023. All Rights Reserved Page 9 Page 9 Journal of Advances in Clinical Pharmacology Volume 1 Issue 2 Journal of Advances in Clinical Pharmacology Volume 1 Issue 2 Some individuals with citrin deficiency may show neuropsychiatric symptoms, including irritability, behavioral changes, confusion, seizures, and loss of consciousness. Citrullinemia Type II, also termed as citrin deficiency, is happened by a mutation in the SLC25A13 gene, which encodes the citrin protein. Citrin plays a major role in transporting certain amino acids, including citrulline, across the cell membrane of the liver and other tissues. When citrin is absent, citrulline accumulates, leading to hyperammonemia. Liver problems Individuals with citrin deficiency may exhibit liver-related symptoms, along with hepatomegaly (enlarged liver), jaundice (yellowing of the skin and eyes), and liver dysfunction. Metabolic crises Stressors such as infections, fasting, or certain medications can trigger metabolic crises in individuals with type 2 citrullinemia. During these episodes, ammonia levels can rise rapidly, leading to severe symptoms such as vomiting, lethargy, disorientation, and even coma. Symptoms Some common symptoms associated with type 2 citrullinemia: Hyperammonemia Adult-Onset Citrullinemia and neurological problems, which are commonly linked to the citrullinemia. Argininosuccinic Aciduria (ASA) ASA is another urea cycle disorder occurs by a deficiency of the enzyme argininosuccinate lyase. It can present with symptoms similar to citrullinemia, along with hyperammonemia and neurological abnormalities. Other Urea Cycle Disorders There are several other genetic disorders affecting different enzymes in the urea cycle, namely carbamoyl phosphate synthetase I (CPS I) deficiency, arginase deficiency, and N-acetylglutamate synthase (NAGS) deficiency. These disorders can present with hyperammonemia and related symptoms. Blood Tests Blood tests play a crucial role in diagnosing citrullinemia. The primary test is the measurement of ammonia levels in the blood. Elevated ammonia levels are a key indicator of urea cycle disorders like citrullinemia. In addition to ammonia, other metabolic markers namely amino acids and organic acids may be assessed. Additional Tests Depending on the clinical presentation and suspected complications, additional tests may be required. These can include tests for electrolyte imbalances, blood gas analysis, and neurological assessments. DIFFERENTIAL DIAGNOSIS FOR CITRULLINEMIA[22,23] When considering a differential diagnosis for citrullinemia, it is very essential to rule out other conditions that can present with similar symptoms. Here are some disorders that play a role in the differential diagnosis: Hyperornithinemia-Hyperammonemia Hyperornithinemia-Hyperammonemia Homocitrullinuria (HHH) syndrome: HHH syndrome is a disorder manifested by enhanced levels of ornithine, ammonia, and Homocitrulline. It leads to the occurrence of symptoms similar to citrullinemia, including hyperammonemia and neurological deficits. Citrullinemia can show an impact on liver function. Liver function tests, measuring liver enzyme levels, may be conducted to evaluate the liver’s health and identify any abnormalities. Ornithine Transcarbamylase Deficiency (OTCD) OTCD is another urea cycle disorder that can lead to the occurrence of hyperammonemia and similar neurological symptoms. However, it typically is seen in the neonatal period with severe symptoms, while citrullinemia may have a later onset. Imaging Studies g g In some cases, imaging studies namely ultrasound, computed tomography (CT), or magnetic resonance imaging (MRI) may be conducted to estimate the liver and other organs influenced by citrullinemia. These imaging techniques can help identify any structural abnormalities or liver damage. Neuropsychiatric symptoms HBRP Publication Page 8-15 2023. All Rights Reserved Page 10 Journal of Advances in Clinical Pharmacology Volume 1 Issue 2 Genetic Testing Genetic testing is typically performed to confirm the diagnosis of citrullinemia and identify the specific gene mutation responsible for the condition. This involves obtaining a blood or saliva sample especially from the individual and analyzing their DNA. Genetic testing can assist in differentiating between the different types of citrullinemia, such as citrullinemia type I and citrullinemia type II. TREATMENT The treatment approach for citrullinemia primarily focuses on managing the symptoms and stopping complications regarding elevated ammonia levels in the body. The treatment is based on the following strategies. Other Metabolic Disorders There are various metabolic disorders that can present with hyperammonemia and neurological symptoms, namely organic HBRP Publication Page 8-15 2023. All Rights Reserved Page 11 Journal of Advances in Clinical Pharmacology Volume 1 Issue 2 Journal of Advances in Clinical Pharmacology Volume 1 Issue 2 specialists, can help assess the effectiveness of the treatment plan and make any necessary adjustments. acidemias, fatty acid oxidation disorders, and mitochondrial disorders. These conditions may require specific metabolic testing regarding confirmation of the diagnosis Medications Medications may be prescribed to help manage symptoms and lower ammonia levels. Ammonia scavenging drugs, namely sodium phenylacetate and sodium benzoate, can help remove excess ammonia from the body. Other medications, such as arginine or citrulline supplements, may be helpful in enhancing the urea cycle and promote ammonia detoxification. Diagnostic Methods g Efforts are being made to evolve improved diagnostic techniques for citrullinemia. These include the use of newborn screening programs to detect the disorder early, as well as the development of more sensitive and specific diagnostic tests. Genetic Studies Researchers continue to investigate the underlying genetic mutations linked to citrullinemia. Identifying specific mutations can contribute to early diagnosis, genetic counseling, and potential gene therapy approaches. Dietary Management A low-protein diet is typically recommended to restrict the intake of citrulline and other amino acids. Protein restriction assists in the the production of ammonia in the body. A dietitian experienced in metabolic disorders can help create an appropriate meal plan and monitor the individual’s nutritional needs. Current research efforts focus on improving early detection methods, developing new therapeutic approaches, and obtaining a deeper understanding of the molecular mechanisms underlying citrullinemia. Liver Transplantation p In severe cases of citrullinemia type I that do not respond to other treatments, liver transplantation plays a major role. A liver transplant can provide a new source of the enzyme that is lacking in individuals with citrullinemia type I, thereby enhancing the body’s ability to break down citrulline. Supplementation Certain supplements may be required to meet the nutritional requirements while maintaining a low-protein diet. These may include essential amino acids, vitamins, and minerals. Specialized formulas may be helpful in infants or young children. Novel Treatments Current treatment options for citrullinemia primarily include dietary management and medications to reduce ammonia levels. However, researchers are exploring new therapeutic approaches. Some potential strategies namely enzyme replacement therapy, gene therapy, and small molecule drugs to enhance the urea cycle function. 1. Albayram, S., Murphy, K. J., Gailloud, P., Moghekar, A., & Brunberg, J. A. (2002). CT findings in the infantile form of citrullinemia. American journal of neuroradiology, 23(2), 334- 336. Patient Support and Advocacy The citrullinemia community, along with patient support groups and advocacy organizations, play an essential role in raising awareness, supporting affected individuals and their families, and facilitating research collaborations. 3. Ando, T., Fuchinoue, S., & Shiraga, H. (2003). Living-related liver transplantation for citrullinemia: different features and clinical problems between classical types (CTLN1) and adult-onset type (CTLN2) citrullinemia. Jap J Transpl, 38, 143-7. Gene therapy, enzyme replacement therapy, and novel drugs targeting specific metabolic pathways are among the potential future treatments being explored. Monitoring and Regular Follow-Up Close monitoring of ammonia levels in the blood is important in obstructing hyperammonemic crises. Regular check- ups with doctors, including metabolic HBRP Publication Page 8-15 2023. All Rights Reserved Page 12 Journal of Advances in Clinical Pharmacology Volume 1 Issue 2 Journal of Advances in Clinical Pharmacology Volume 1 Issue 2 Biomarkers and Monitoring Biomarkers that can predict disease severity and monitor treatment effectiveness are still under investigation. These biomarkers may assist in personalized treatment plans and provide insights into disease progression. 2. Ames, E. G., Powell, C., Engen, R. M., Weaver Jr, D. J., Mansuri, A., Rheault, M. N., ... & Luckritz, K. E. (2022). Multisite retrospective review of outcomes in renal replacement therapy for neonates with inborn errors of metabolism. The Journal of pediatrics, 246, 116-122. Harting, I. (2017). Extrastriatal changes in patients with late-onset glutaric aciduria type I highlight the risk of long-term neurotoxicity. Orphanet journal of rare diseases, 12(1), 1-13. Harting, I. (2017). Extrastriatal changes in patients with late-onset glutaric aciduria type I highlight the risk of long-term neurotoxicity. Orphanet journal of rare diseases, 12(1), 1-13. 16. Gunz, A. C., Choong, K., Potter, M., & Miller, E. (2013). Magnetic resonance imaging findings and neurodevelopmental outcomes in neonates with urea-cycle defects. International Medical Case Reports Journal, 41-48. 9. Brunetti-Pierri, N., Lamance, K. M., Lewis, R. A., & Craigen, W. J. (2012). 30-year follow-up of a patient with classic citrullinemia. Molecular Genetics and Metabolism, 106(2), 248-250. 17. Häberle, J., Boddaert, N., Burlina, A., Chakrapani, A., Dixon, M., Huemer, M., ... & Dionisi-Vici, C. (2012). Suggested guidelines for the diagnosis and management of urea cycle disorders. Orphanet journal of rare diseases, 7, 1-30. 10. Gravel, R. A. (2001). The metabolic and molecular basis of inherited disease. (No Title), 3827. 18. Häberle, J., Boddaert, N., Burlina, A., Chakrapani, A., Dixon, M., Huemer, M., ... & Dionisi-Vici, C. (2012). Suggested guidelines for the diagnosis and management of urea cycle disorders. Orphanet journal of rare diseases, 7, 1-30. 11. Diez‐ Fernandez, C., Rüfenacht, V., & Häberle, J. (2017). Mutations in the human Argininosuccinate Synthetase (ASS1) gene, impact on patients, common changes, and structural considerations. Human Mutation, 38(5), 471-484. 19. Häberle, J., Vilaseca, M. A., Meli, C., Rigoldi, M., Jara, F., Vecchio, I., ... & Parini, R. (2010). First manifestation of citrullinemia type I as differential diagnosis to postpartum psychosis in the puerperal period. European Journal of Obstetrics and Gynecology and Reproductive Biology, 149(2), 228-229. 12. Engel, K., Höhne, W., & Häberle, J. (2009). Mutations and polymorphisms in the human argininosuccinate synthetase (ASS1) gene. Human mutation, 30(3), 300-307. 13. Enns, G. M., O'Brien, W. E., Kobayashi, K., Shinzawa, H., & Pellegrino, J. E. (2005). Postpartum “psychosis” in mild argininosuccinate synthetase deficiency. Obstetrics & Gynecology, 105(5 Part 2), 1244-1246. 20. Häberle, J., Pauli, S., Linnebank, M., Kleijer, W., Bakker, H., Wanders, R., ... & Koch, H. (2002). Structure of the human argininosuccinate synthetase gene and an improved system for molecular diagnostics in patients with classical and mild citrullinemia. Human genetics, 110, 327-333. 14. Faghfoury, H., Baruteau, J., de Baulny, H. O., Häberle, J., & Schulze, A. (2011). Transient fulminant liver failure as an initial presentation in citrullinemia type I. CONCLUSION 4. Bachmann, C. (2003). Outcome and survival of 88 patients with urea cycle disorders: a retrospective evaluation. European journal of pediatrics, 162, 410-416. 4. Bachmann, C. (2003). Outcome and survival of 88 patients with urea cycle disorders: a retrospective evaluation. European journal of pediatrics, 162, 410-416. Citrullinemia is a rare metabolic disorder that can lead to the occurrence of severe implications if not diagnosed and managed promptly. With developments in screening techniques and treatment modalities, the prognosis for individuals with citrullinemia has improved significantly. However, further research is required to refine diagnostic methods, expand treatment options, and enhance long-term outcomes for patients with this complex disorder. 5. Batshaw, M. L., MacArthur, R. B., & Tuchman, M. (2001). Alternative pathway therapy for urea cycle disorders: twenty years later. The Journal of pediatrics, 138(1), S46-S55. 6. Bireley, W. R., Van Hove, J. L., Gallagher, R. C., & Fenton, L. Z. (2012). Urea cycle disorders: brain MRI and neurological outcome. Pediatric radiology, 42, 455-462. 6. Bireley, W. R., Van Hove, J. L., Gallagher, R. C., & Fenton, L. Z. (2012). Urea cycle disorders: brain MRI and neurological outcome. Pediatric radiology, 42, 455-462. In conclusion, citrullinemia remains a challenging condition, but with enhanced awareness, early diagnosis, and effective management strategies, individuals affected by this rare disorder can lead fulfilling lives. Continued research endeavors hold promise for further enhancing the understanding and treatment of citrullinemia, offering hope to affected individuals and their families. 7. Bourdeaux, C., Darwish, A., Jamart, J., Tri, T. T., Janssen, M., Lerut, J., ... & Reding, R. (2007). Living-related versus deceased donor pediatric liver transplantation: a multivariate analysis of technical and immunological complications in 235 recipients. American journal of transplantation, 7(2), 440-447. REFERENCES OR FURTHER READING REFERENCES OR FURTHER READING 8. Boy, N., Heringer, J., Brackmann, R., Bodamer, O., Seitz, A., Kölker, S., & HBRP Publication Page 8-15 2023. All Rights Reserved Page 13 Journal of Advances in Clinical Pharmacology Volume 1 Issue 2 Journal of Advances in Clinical Pharmacology Volume 1 Issue 2 Molecular Genetics and Metabolism, 102(4), 413-417. 21. Häberle, J., Pauli, S., Schmidt, E., Schulze-Eilfing, B., Berning, C., & Koch, H. G. (2003). Mild citrullinemia in Caucasians is an allelic variant of argininosuccinate synthetase deficiency (citrullinemia type 1). Molecular genetics and metabolism, 80(3), 302-306. 15. Gao, H. Z., Kobayashi, K., Tabata, A., Tsuge, H., Iijima, M., Yasuda, T., ... & Saheki, T. (2003). Identification of 16 novel mutations in the argininosuccinate synthetase gene and genotype–phenotype correlation in 38 classical citrullinemia patients. Human mutation, 22(1), 24- 34. 22. Häberle, J., & Rubio, V. (2016). Disorders of the urea cycle and related HBRP Publication Page 8-15 2023. All Rights Reserved Page 14 Journal of Advances in Clinical Pharmacology Volume 1 Issue 2 Journal of Advances in Clinical Pharmacology Volume 1 Issue 2 enzymes. Inborn metabolic diseases: diagnosis and treatment, 295-308. 23. Häberle, J., Vilaseca, M. A., Meli, C., Rigoldi, M., Jara, F., Vecchio, I., ... & Parini, R. (2010). First manifestation of citrullinemia type I as differential diagnosis to postpartum psychosis in the puerperal period. European Journal of Obstetrics and Gynecology and Reproductive Biology, 149(2), 228-229. HBRP Publication Page 8-15 2023. All Rights Reserved Page 15
https://openalex.org/W2912830178	https://ejournal.bsi.ac.id/ejurnal/index.php/swabumi/article/download/4868/2837	Indonesian	null	PERANCANGAN SISTEM ABSENSI PEGAWAI BERBASIS WEB Studi Kasus : Kantor Kecamatan Purwodadi	Ilmu Komputer, Manajemen dan Sosial Swabumi : Suara Wawasan Sukabumi/Swabumi (Suara Wawasan Sukabumi)	2,018	cc-by	2,520	Abstrak Absensi kehadiran pegawai merupakan faktor penting bagi sebuah instansi atau perusahaan untuk mencapai tujuan, hal ini berkaitan pada kedisiplinan dan berdampak pada kinerja dari masing-masing pegawai. Oleh karena itu, perlu adanya pendataan khusus untuk mencatat absensi kehadiran dan ketidakhadiran agar aktifitas kerja dapat tercatat secara realtime dan baik. Banyak cara yang dapat dilakukan untuk mencapai sistem informasi absensi yang baik, salah satunya menggunakan teknologi komputer dimana penerapannya dengan aplikasi absensi berbasis website. Pada Kantor Kecamatan Purwodadi sistem yang digunakan dalam proses absensi masih manual menggunakan buku absensi harian yang berdampak pada efisiensi dan efektifitas pendataan, pencarian data sekaligus perhitungan rekap data yang membutuhkan waktu yang relatif lama. Disamping itu resiko kesalahan dan kehilangan data absensi semakin besar. Berdasarkan permasalahan diatas dibuatlah Sistem Informasi Absensi Kepegawaian Pada Kantor Kecamatan Purwodadi. Metode penelitian merupakan metode yang digunakan dalam pengumpulan data yang meliputi: metode riset, wawancara dan pustaka. Sedangkan pengembangan perangkat lunak menggunakan waterfall yang meliputi : analisa kebutuhan, desain, pengkodean, pengujian dan implementasi. Dengan dihasilkannya aplikasi absensi kepegawaian berbasis web dapat memberikan kemudahan dalam proses absensi, pencarian data dan perhitungan rekap absensi, serta meminimalisir kehilangan dan kesalahan pencatatan data absensi pada Kantor Kecamatan Purwodadi. p Kata Kunci: Sistem Informasi, Absensi Karyawan, Berbasis Web Abstract Attendance data is an important factor for the institution or to achieve its objectives, this is related to discipline and affects every employee. Therefore, there is a need to issue data specifically to report and find activities that can be done in real time and good. There are many ways that can be done to achieve a good information system, one of which uses computer technology where the application is with an application-based attendance website. The Purwodadi District Office system which is still in manual attendance using books, data and data recapitulation calculations takes a relatively long time. In addition, data and data shows are larger. Based on the context, an employee attendance information system is created in the Purwodadi Regional Office. The research method is the method used in the data which includes: research methods, interviews and literature. While the development of software using waterfall which includes: needs analysis, design, coding, testing and implementation. With the attendance results employees can provide assistance in the attendance process, search data and recapitulation, and explore and record attendance data at Purwodadi District Office. K d I f i S E l A d W b B d Keywords: Information Systems, Employee Attendance, Web-Based PERANCANGAN SISTEM ABSENSI PEGAWAI BERBASIS WEB Studi Kasus : Kantor Kecamatan Purwodadi 1)Subiantoro, 2)Sardiarinto 1) Manajemen Informatika, AMIK BSI Yogyakarta subiantoro558@gmail.com 2) Teknik Informatika, AMIK BSI Purwokerto sardiarinto.sdo@bsi.ac.id 2) Teknik Informatika, AMIK BSI Purwokerto sardiarinto.sdo@bsi.ac.id JURNAL SWABUMI, Vol.6 No.2 September 2018, pp. 184~189 ISSN: 2355-990X E-ISSN: 2549-5178 JURNAL SWABUMI, Vol.6 No.2 September 2018, pp. 184~189 ISSN: 2355-990X E-ISSN: 2549-5178 184 2. Metode Penelitian Kantor Kecamatan Purwodadi Kabupaten Purworejo merupakan salah satu instansi Pemerintah yang mempunyai tanggung jawab terhadap pelayanan masyakarakat dalam bidang pelayanan sosial. Jumlah pegawai yang dimiliki saat ini 48 personel dan dapat bertambah sesuai kebutuhan instansi. Semakin besar instansi pemerintah maka akan berdampak semakin besar pula dengan Sumber Daya Manusia (SDM) yang dibutuhkan. Berdasarkan hal itu maka instansi pemerintah dituntut untuk melakukan manajemen pengolahan data absensi yang baik, cepat dan dan efisien. Karena pada dasarnya sistem informasi absensi digunakan sebagai salah satu tolak ukur metode pengembangan pegawai, jika dalam absensi pegawai setelah mengikuti pengembangan menurun, maka metode pengembangan yang diterapkan berjalan dengan baik, sebaliknya jika absensi pegawai tetap berarti metode pengambangan yang diterapkan kurang baik (Hasibuan, 2008). Metode penelitian yang penulis gunakan untuk pengembangan perangkat lunaknya yaitu menggunakan model air terjun (waterfall). Model ini merupakan pendekatan perangkat lunak secara terurut yang dimulai dari analisis, desain, pengkodean, pengujian dan tahap pendukung(Rosa dan Shalahudin, 2013). Bentuk kerangka penelitian ini dapat dilihat pada gambar 1 berikut. Gambar 1. Bentuk Kerangka Penelitian Analisa Kebutuhan Design Database dan Web Pengkodean/Implementasi Web Pengujian Web Gambar 1. Bentuk Kerangka Penelitian Prosedur absensi di Kantor Kecamatan Purwodadi yang diterapkan sekarang ini dapat dikatakan masih kurang efisien dan efektif, dimana semua masih dilakukan secara manual mulai dari pendataan dan perhitungan jam hadir, jam keluar, sampai dengan keterangan tidak masuk, hal ini berdampak pada waktu yang relatif lama dalam proses perhitungan rekapitulasi absensi dari pegawai, serta bentuk laporan absensi yang dibuat berupa hardcopy yang dapat menyebabkan kesalahan dalam pencatatan data, menyulitkan dalam proses pencarian data dan dapat dikhawatirkan terjadi kehilangan data absensi pegawai. 1. Pendahuluan informasi absensi yang baik maka diperlukan teknologi informasi yang meliputi teknologi komputer, teknologi telekomunikasi dan teknologi apapun yang dapat memberikan nilai tambah untuk mengelola sistem tersebut(Jogiyanto, 2009). Absensi atau kartu jam hadir adalah informasi absensi yang baik maka diperlukan teknologi informasi yang meliputi teknologi komputer, teknologi informasi absensi yang baik maka diperlukan teknologi informasi yang meliputi teknologi komputer, teknologi telekomunikasi dan teknologi apapun yang dapat memberikan nilai tambah untuk mengelola sistem tersebut(Jogiyanto, 2009). Absensi atau kartu jam hadir adalah dokumen yang mencatat jam hadir setiap pegawai di suatu perusahaan yang dapat Absensi merupakan suatu hal yang penting dalam sebuah instansi pemerintah. Dengan sistem absensi yang baik maka diharapkan dapat membantu dalam mengendalikan proses penyelesaian pekerjaan sehingga didapatkan hasil yang maksimal dan sesuai dengan tujuan yang ditetapkan. Untuk mencapai sistem g p g telekomunikasi dan teknologi apapun yang dapat memberikan nilai tambah untuk mengelola sistem tersebut(Jogiyanto, 2009). g p g telekomunikasi dan teknologi apapun yang dapat memberikan nilai tambah untuk mengelola sistem tersebut(Jogiyanto, 2009). Absensi atau kartu jam hadir adalah dokumen yang mencatat jam hadir setiap pegawai di suatu perusahaan yang dapat Diterima April 05, 2018; Revisi Mei 16 , 2018; Disetujui Agustus 15, 2018 185 permasalahan-permasalahan absensi yang ada di Kantor Kecamatan Purwodadi. permasalahan-permasalahan absensi yang ada di Kantor Kecamatan Purwodadi. berupa daftar hadir biasa atau kartu hadir yang diisi dengan mesin pencatat waktu(Setiawan, 2015). 3.1. Analisa Kebutuhan Dalam perancangan website Sistem Informasi Absensi Kantor Kecamatan Purwodadi, memiliki kebutuhan fungsional dari masing-masing pengguna adalah sebagai berikut: g 1. Kebutuhan Pengguna JURNAL SWABUMI Vol.6 No.2, September 2018: 184-189 g 1. Kebutuhan Pengguna Kebutuhan pengguna yang digunakan dalam pembanguna website Sistem Informasi Absensi Kepegawaian adalah sebagai berikut: a. Kebutuhan Admin 1) Admin melakukan login sebelum masuk kehalaman administrator dengan menggunakan id_user dan password. p g Untuk mengatasi hal tersebut diperlukan adanya sistem absensi berbasis web dengan intranet. Website adalah halaman informasi yang disediakan melalui jalur internet sehingga bisa diakses di seluruh dunia, selama terkoneksi dengan jaringan internet(Puspitosari, 2010). Sedangkan Intranet merupakan jaringan komputer didalam suatu organisasi yang menggunakan teknologi internet, menyediakan lingkungan yang mirip dengan internet, sehingga memungkinkan saling berbagai informasi(Mulyanto dalam O’Brien, 2009). Dengan adanya aplikasi tersebut diharapkan mampu meminimalisir p 2) Admin dapat menambah, mengubah password, dan menghapus data administrator . 3) Admin dapat menambah data pegawai pada halaman login . 4) Admin dapat melakukan proses absensi kepegawaian. p g 5) Admin dapat mengubah dan menghapus data pegawai. 6) Admin dapat menambah, mengubah dan menghapus data pegawai, bagian, jabatan ,pelatihan, kenaikan jabatan. JURNAL SWABUMI Vol.6 No.2, September 2018: 184-189 186 Absensi Kepegawaian Kantor Kecamatan Purwodadi. 7) Admin dapat melihat dan mencetak laporan data pegawai, data pelatihan, data absensi. 5) Sistem akan menghentikan akses Admin pada halaman administrator apabila telah melakukan proses logout. b. Kebutuhan Pegawai g 1) Pegawai dapat menambahkan data pegawai pada halaman login pegawai. p g Analisa Kebutuhan Sistem untuk Pegawai p g 2) Pegawai melakukan login sebelum masuk kehalaman index pegawai dengan menggunakan NIP sebagai id_petugas dan password yang telah terdaftar. g 1) Sistem melakukan validasi username dan password untuk masuk ke halaman user 2) Sistem akan memberikan hak akses kepada user untuk dapat masuk ke halaman user jika username dan password sesuai. y g 3) Pegawai dapat melakukan proses absensi masuk dan proses absensi keluar berdasarkan waktu pada halaman absensi 3) Sistem akan menolak hak akses kepada user jika username dan password tidak sesuai. p 4) Pegawai dapat mengubah data pegawai, password. p 4) Sistem menyediakan fasilitas untuk dapat menambah, mengubah, dan menghapus beberapa data yang digunakan untuk menampilkan data absensi. p g 5) Pegawai dapat menambah, mengubah, dan menghapus data riwayat pendidikan data pengalaman kerja. 4.2. Design Sistem 2) Sistem akan memberikan hak akses kepada Admin untuk dapat masuk ke halaman administrator jika username dan password sesuai. p g 2. Kebutuhan Sistem 2. Kebutuhan Sistem p 5) Sistem melakukan penyimpanan data absensi yang dilakukan oleh user. Kebutuhan sistem dalam perancangan website Sistem Informasi Absensi Kepegawaian Kantor Kecamatan Purwodadi adalah sebagai berikut: 6) Sistem akan menghentikan akses user pada halaman user apabila telah melakukan proses logout. a. Analisa Kebutuhan Sistem untuk Admin 1) Sistem melakukan validasi username dan password. ERD(Entity Relationship Diagram) Entity Relationship Diagram (ERD) ERD(Entity Relationship Diagram) Entity Relationship Diagram (ERD) y p g ( ) adalah suatu diagram untuk menggambarkan desain konseptual dari model konseptual suatu basis data relasional. Entity Relationship Diagram (ERD) juga merupakan gambaran yang menghubungkan antara objek satu dengan objek yang lain dalam dunia nyata(Utami dan Hartanto, 2012). Bentuk rancangan ERD dalam pengembangan sistem ini dapat dilihat pada gambar 2. 3) Sistem akan menolak hak akses kepada Admin jika username dan password tidak sesuai. p 4) Sistem menyediakan fasilitas untuk dapat menambah, mengubah, dan menghapus beberapa data yang digunakan untuk menampilkan secara keseluruhan pada website JURNAL SWABUMI Vol.6 No.2, September 2018: 184-189 187 Gambar 2. ERD Sistem admin pegawai bagian pelatihan petugas mengelola id_admin nama_adm password jenis_kelamin tgl_lahir tempat_lahir tgl_masuk foto id_jabatan nip nama id_bagian password nama_bagian id_bagian jabatan nama_jabatan id_jabatan tgl_pelatihan nip id_pelatihan hasil penyelengara topik_pelatihan naik_ jabatan nip id_kjabatan keterangan masa_kerja mengelola mengelola mengelola mengelola memiliki memiliki memiliki nama_petugas id_petugas password pendidikan nip id_rpendidikan detail_pendidikan tahun_pendidikan mengelola pengalaman _kerja nip id_pekerjaan detail_pekerjaan nama_pekerjaan mengelola absensi tanggal_absen jam_masuk id_absensi status_keluar jam_keluar terlambat nip ket status_masuk melakukan 1 1 1 1 1 1 1 1 1 M M M M M M M M 1 1 1 riwayat _jabatan jabatan_lama id_kjabatan id_rjabatan tgl_njabatan jabatan_baru tgl_ajabatan memiliki 1 1 mengelola 1 M level id_admin id_admin id_admin id_admin 1 memiliki 1 id_petugas mengelola 1 M id_petugas tempat_lahir pegawai password status_keluar pengalaman _kerja mengelola absensi terlambat status_masuk mengelola Gambar 2. ERD Sistem 3.3. Implementasi a. Halaman Index Website Menampilkan pilihan untuk login sebagai administrator website atau login sebagai pegawai dan menampilkan tampilan form pendaftaran pegawai baru. Berikut tampilan halamannya: a. Halaman Index Website Menampilkan pilihan untuk login sebagai administrator website atau login sebagai pegawai dan menampilkan tampilan form pendaftaran pegawai baru. Berikut tampilan halamannya: JURNAL SWABUMI Vol.6 No.2, September 2018: 184-189 188 Gambar 3. Halaman Index Website b. Halaman Login Administrator Admin harus melakukan login terlebih dahulu sebelum masuk halaman administrator. Berikut tampilan halamannya: Gambar 4. Login Administrator c. Halaman Data Pegawai Menampilkan data pegawai secara keseluruhan dimana admin dapat melakukan proses melihat, menambah, mengubah, menghapus data pegawai baru. Berikut tampilan halaman data pegawai: Gambar 5. Halaman Data Pegawai Gambar 3. Halaman Index Website e. Halaman Absensi Pegawai Halaman baru digunakan untuk menampilkan dan mengelola absensi pegawai. Berikut Halaman absensi Pegawai: Gambar 7. Absensi Pegawai e. Halaman Absensi Pegawai Halaman baru digunakan untuk menampilkan dan mengelola absensi pegawai. Berikut Halaman absensi Pegawai: Gambar 7. Absensi Pegawai Gambar 3. Halaman Index Website b. Halaman Login Administrator Admin harus melakukan login terlebih dahulu sebelum masuk halaman administrator. Berikut tampilan halamannya: b. Halaman Login Administrator Admin harus melakukan login terlebih dahulu sebelum masuk halaman administrator. Berikut tampilan halamannya: Gambar 4. Login Administrator Gambar 7. Absensi Pegawai f. Halaman Cetak Data Absensi Pegawai Halaman baru yang digunakan untuk mencetak laporan data absensi pegawai sekaligus bentuk tampilan laporannya. Berikut tampilan halaman cetak data absensi pegawai: Gambar 4. Login Administrator Gambar 8. Cetak Data Absensi Pegawai c. Halaman Data Pegawai Menampilkan data pegawai secara keseluruhan dimana admin dapat melakukan proses melihat, menambah, mengubah, menghapus data pegawai baru. Berikut tampilan halaman data pegawai: p p g Gambar 5. Halaman Data Pegawai p p g Gambar 5. Halaman Data Pegawai Gambar 8. Cetak Data Absensi Pegawai g. Halaman Cetak Laporan Data Pelatihan Halaman baru yang digunakan untuk mencetak laporan data pelatihan pegawai sekaligus bentuk tampilannya. Berikut tampilan halaman cetak laporan data pelatihan pegawai: Gambar 5. Halaman Data Pegawai Gambar 9. Cetak Laporan Data Pelatihan Gambar 9. Cetak Laporan Data Pelatihan d. Halaman Data Pelatihan Halaman baru yang digunakan untuk menampilkan data pelatihan pegawai dimana admin dapat mengelola data di dalamnya yang terdiri atas menambah, mengubah dan menghapus data pelatihan. Gambar 9. Cetak Laporan Data Pelatihan Gambar 9. Cetak Laporan Data Pelatihan Berikut tampilan halaman data pelatihan: Berikut tampilan halaman data pelatihan: Gambar 6.Halaman Data Naik Jabatan 3.4. Pengujian e ut ta p a a a a data pe at a Gambar 6.Halaman Data Naik Jabatan Hasil pengujian unit yang ditunjukkan pada tabel 1 dapat simpulkan bahwa hasil pengujian sistem menggunakan metode Black Box Testing, semua kebutuhan fungsionalitas yang diinginkan telah memenuhi syarat kriteria atau berjalan dengan lancar. Gambar 6.Halaman Data Naik Jabatan JURNAL SWABUMI Vol.6 No.2, September 2018: 184-189 189 Web Pada Program Studi Teknik Informatika Universitas Diponegoro. Journal of Informatics and Technology: 72-84. Vol 1 No 1, Tahun 2012. Diambil dari: http://ejournal- s1.undip.ac.id/index.php/joint/article/v iew/434/434 (27 Juli 2016). Web Pada Program Studi Teknik Informatika Universitas Diponegoro. Journal of Informatics and Technology: 72-84. Vol 1 No 1, Tahun 2012. Diambil dari: http://ejournal- s1.undip.ac.id/index.php/joint/article/v iew/434/434 (27 Juli 2016). Tabel 1. Kesimpulan hasil pengujian unit No Jenis pengujian Validasi data Kosong Valida si kesala han data Fungsi- fungsi tombol 1 Login Admin V V V 2 Login Pegawai V V V 3 Halaman Admin V V V 4 Halaman Pegawai V V V Frieyadie. 2007. Belajar Sendiri Pemograman Database Menggunakan FoxPro 9.0. Jakarta: Elex Media Komputindo. Hasibuan, Malayu S.P. 2008. Manajemen Sumber Daya Manusia. Jakarta: PT. Bumi Aksara. Jogiyanto. 2009. Sistem Teknologi Informasi. Yogyakarta : Andi Offset. p Saran Terdapat saran yang dapat membantu pengembangan aplikasi absensi pada Kantor Kecamatan Purwodadi di masa yang akan datang, antara lain: g 1. Bagian Administrator harus melakukan back-up data serta pemeliharaan sistem yang baik dan efektif secara rutin guna mengantisipasi kerusakan pada sistem dan human error. 2. Sistem absensi kepegawaian pada Kantor Kecamatan Purwodadi dapat dikembangkan lagi dengan menggabungkan aplikasi lain seperti finger print. 5. Kesimpulan dan Saran Mulyanto, Agus. 2009. Sistem Informasi Konsep dan Aplikasi. Yogyakarta : Pustaka Pelajar. Berdasarkan pembahasan diatas, maka dapat diambil kesimpulan sebagai berikut : 1. Telah dihasilkan aplikasi absensi berbasis website Kantor Kecamatan Purwodadi yang memberikan kemudahan bagi pegawai dalam mengelola dan mengakses informasi absensi baik dari segi penggunaanya maupun pada proses pembuatan laporan. Peranginangin, Kasiman. 2006. Aplikasi WEB dengan PHP dan MySQL. Yogyakarta: Andi Offset. Puspitosari, Heni A. 2010. Membangun Website Interaktif dengan Adobe Creative Suite 5. Yogyakarta:PT. Skripta Media Creative. p Rosa, A.S dan M. Shalahuddin. 2013. Rekayasa Perangkat Lunak Terstuktur dan Berorientasi Objek. Bandung: Informatika Bandung. p 2. Dengan menggunakan aplikasi absensi berbasis web pegawai dapat meminimalisir kehilangan dan kesalahan pencatatan data baik dalam proses absensi itu sendiri maupun pembuatan laporan absensi. Setiawan, Parta. 2015. Pengertian Absensi dan Jenis-Jenis Absensi. Diambil dari :http://www.gurupendidikan.com/pen gertian-dan-jenis-jenis-absensi/. (18 Mei 2016). 3. Data absensi dapat diolah secara terstruktur yang dapat memberikan kemudahan kepada pegawai dalam proses pencarian data absensi. Utami, Erna dan Anggit Dwi Hartanto. 2012. Sistem Basis Data Menggunakan Microsoft SQL Server 2005. Yogyakarta : Andi Offset. Referensi Binarso, Yusi Ardi, Eko Adi Sarwoko dan Nurdin Bahtiar. 2012. Pembangunan Sistem Informasi Alumni Berbasisi JURNAL SWABUMI Vol.6 No.2, September 2018: 184-189
https://openalex.org/W2885292374	https://repository.arizona.edu/bitstream/10150/631118/1/s12881-018-0656-z.pdf	English	null	Genome-wide association study of lung function and clinical implication in heavy smokers	BMC medical genetics	2,018	cc-by	9,225	Genome-wide association study of lung function and clinical implication in heavy smokers Item Type Article Authors Li, Xingnan; Ortega, Victor E.; Ampleford, Elizabeth J.; Graham Barr, R.; Christenson, Stephanie A.; Cooper, Christopher B.; Couper, David; Dransfield, Mark T.; Han, Mei Lan K.; Hansel, Nadia N.; Hoffman, Eric A.; Kanner, Richard E.; Kleerup, Eric C.; Martinez, Fernando J.; Paine, Robert; Woodruff, Prescott G.; Hawkins, Gregory A.; Bleecker, Eugene R.; Meyers, Deborah A. Citation Li et al. BMC Medical Genetics (2018) 19:134 DOI 10.1186/s12881-018-0656-z Publisher BMC Journal BMC MEDICAL GENETICS Rights © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License. Download date 24/10/2024 04:06:53 Item License https://creativecommons.org/licenses/by/4.0/ Version Final published version Link to Item http://hdl.handle.net/10150/631118 Genome-wide association study of lung function and clinical implication in heavy smokers Item Type Article Authors Li, Xingnan; Ortega, Victor E.; Ampleford, Elizabeth J.; Graham Barr, R.; Christenson, Stephanie A.; Cooper, Christopher B.; Couper, David; Dransfield, Mark T.; Han, Mei Lan K.; Hansel, Nadia N.; Hoffman, Eric A.; Kanner, Richard E.; Kleerup, Eric C.; Martinez, Fernando J.; Paine, Robert; Woodruff, Prescott G.; Hawkins, Gregory A.; Bleecker, Eugene R.; Meyers, Deborah A. Citation Li et al. BMC Medical Genetics (2018) 19:134 DOI 10.1186/s12881-018-0656-z Publisher BMC Journal BMC MEDICAL GENETICS Rights © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License. Download date 24/10/2024 04:06:53 Item License https://creativecommons.org/licenses/by/4.0/ Version Final published version Link to Item http://hdl.handle.net/10150/631118 © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Li et al. BMC Medical Genetics (2018) 19:134 https://doi.org/10.1186/s12881-018-0656-z Open Access Genome-wide association study of lung function and clinical implication in heavy smokers Xingnan Li1* , Victor E. Ortega2, Elizabeth J. Ampleford2, R. Graham Barr3, Stephanie A. Christenson4, Christopher B. Cooper5, David Couper6, Mark T. Dransfield7, Mei Lan K. Han8, Nadia N. Hansel9, Eric A. Hoffman10, Richard E. Kanner11, Eric C. Kleerup5, Fernando J. Martinez12, Robert Paine III11, Prescott G. Woodruff4, Gregory A. Hawkins2, Eugene R. Bleecker1, Deborah A. Meyers1 and for the SPIROMICS Research Group Abstract Background: The aim of this study is to identify genetic loci associated with post-bronchodilator FEV1/FVC and FEV1, and develop a multi-gene predictive model for lung function in COPD. Methods: Genome-wide association study (GWAS) of post-bronchodilator FEV1/FVC and FEV1 was performed in 1645 non-Hispanic White European descent smokers. Results: A functional rare variant in SERPINA1 (rs28929474: Glu342Lys) was significantly associated with post- bronchodilator FEV1/FVC (p = 1.2 × 10−8) and FEV1 (p = 2.1 × 10−9). In addition, this variant was associated with COPD (OR = 2.3; p = 7.8 × 10−4) and severity (OR = 4.1; p = 0.0036). Heterozygous subjects (CT genotype) had significantly lower lung function and higher percentage of COPD and more severe COPD than subjects with the CC genotype. 8.6% of the variance of post-bronchodilator FEV1/FVC can be explained by SNPs in 10 genes with age, sex, and pack-years of cigarette smoking (P < 2.2 × 10−16). Conclusions: This study is the first to show genome-wide significant association of rs28929474 in SERPINA1 with lung function. Of clinical importance, heterozygotes of rs28929474 (4.7% of subjects) have significantly reduced pulmonary function, demonstrating a major impact in smokers. The multi-gene model is significantly associated with CT-based emphysema and clinical outcome measures of severity. Combining genetic information with demographic and environmental factors will further increase the predictive power for assessing reduced lung function and COPD severity. Keywords: COPD, GWAS, Lung function, rs28929474, SERPINA1, SPIROMICS Keywords: COPD, GWAS, Lung function, rs28929474, SERPINA1, SPIROMICS * Correspondence: lixingnan1@deptofmed.arizona.edu 1Division of Genetics, Genomics and Precision Medicine, Department of Medicine, University of Arizona, BioScience Research Lab, Room 253, 1230 N. Cherry Avenue, PO Box 210242, Tucson, AZ 85721, USA Full list of author information is available at the end of the article Background reduction in post-bronchodilator percent predicted FEV1, i.e., GOLD stages 1–4 (mild, moderate, severe, and very severe COPD) have post-bronchodilator percent predicted FEV1 ≥80%, ≥50%, ≥30%, or < 30%, respectively [1]. Chronic obstructive pulmonary disease (COPD) is a com- mon respiratory disease caused by the interaction of gen- etic susceptibility with environmental influences, primarily tobacco exposure. COPD is defined as a reduced ratio of post-bronchodilator forced expiratory volume in 1 s (FEV1) to forced vital capacity (FVC) (post-bronchodilator FEV1/FVC < 0.70) [1]. COPD severity is measured by the Twenty-eight genomic loci associated with baseline FEV1/FVC or FEV1 have been identified by meta-analyses of genome-wide association studies (GWAS) in general populations of European descent [2–4]. A recent GWAS comparing extremes of high and low baseline FEV1 in subjects of European ancestry from the UK Biobank has identified five loci (KANSL1, HLA-DQ, NPNT, TET2, and TSEN54) in never smokers and RBM19-TBX5 in heavy smokers [5]. HHIP, FAM13A1, CHRNA3, RIN3, MMP12, * Correspondence: lixingnan1@deptofmed.arizona.edu 1Division of Genetics, Genomics and Precision Medicine, Department of Medicine, University of Arizona, BioScience Research Lab, Room 253, 1230 N. Cherry Avenue, PO Box 210242, Tucson, AZ 85721, USA Full list of author information is available at the end of the article GWAS of post-bronchodilator pulmonary function GWAS of post-bronchodilator pulmonary function After quality control analysis, 1645 non-Hispanic White subjects (1086 subjects with COPD and 559 current and former smokers with preserved lung function [8]) were included in the analysis (Table 1). GWAS of post-bron- chodilator FEV1/FVC and percent predicted FEV1 were performed for 635,970 single nucleotide polymorphisms (SNPs) with MAF ≥0.01 in 1645 non-Hispanic White smokers with age, sex, current smoking status, pack-years of cigarette smoking, and the first two principal compo- nents as covariates in the linear additive model. Genomic inflation factors are 1.013 and 1.017 for GWAS of Page 2 of 10 Li et al. BMC Medical Genetics (2018) 19:134 criteria were excluded, SNPs were removed if 1) call rates< 95%, 2) inconsistent with Hardy-Weinberg Equi- librium (HWE) (p < 10−6), or 3) minor allele frequency (MAF) < 0.01. and TGFB2 have been associated with COPD at genome-wide significant levels [6]. To our knowledge, no GWAS study has been performed on post-bronchodilator FEV1/FVC and FEV1 in smokers, which defines a diagno- sis of COPD and determines COPD severity, respectively. A linear additive model was used for analysis of pre −/post-bronchodilator FEV1/FVC, percent predicted FEV1, FVC, and % change in FEV1 bronchodilator re- sponse using PLINK software (URL: zzz.bwh.harvar- d.edu/plink/) [9], adjusted for age, sex, current smoking status, pack-years of cigarette smoking, and the first two principal components from the multidimensional scaling analysis of genotypes on the chip. Association analyses of Pre-/Post-bronchodilator FEV1 and FVC in ml were performed using linear regression adjusted for sex, age, age2, height, height2, weight, current smoking status, pack-years of cigarette smoking, and the first two princi- pal components. Association analyses of COPD and COPD severity were performed using logistic regression adjusted for age, sex, current smoking status, pack-years of cigarette smoking, and the first two principal compo- nents. P values≤5 × 10−8 were considered genome-wide significant. P values ≤0.05 were considered significant for SNP-level evaluation of previously reported candi- date SNPs associated with baseline lung function. SNAP software (URL: http://www.broad.mit.edu/mpg/snap/) was used to generate the association plots [10]. GWAS of post-bronchodilator FEV1/FVC and percent predicted FEV1 were performed in non-Hispanic White smokers (n = 1645, GOLD stage 0–4, smoking≥20 packs/ year) from the NHLBI-sponsored SubPopulations and InteRmediate Outcome Measures In COPD Study (SPIROMICS). In addition to evaluating previously re- ported loci associated with baseline lung function in general populations, we aimed to identify novel genes associated with abnormal post-bronchodilator lung func- tion in smokers enriched for COPD and develop a model to predict lung function using multiple genes and demographic/environmental factors. Study subjects y j SPIROMICS is a prospective cohort study that enrolled 2981 participants with the goals of identifying new COPD subgroups and intermediate markers of disease progression [7, 8]. SPIROMICS is a well-characterized longitudinal cohort with comprehensive phenotyping in- cluding measurements of lung function and quantitative CT scan. Spirometry was performed before and after four inhalations with 90 μg albuterol and 18 μg ipratro- pium per inhalation according to ATS recommenda- tions. Non-Hispanic White smokers (ever or current smoking≥20 packs/year) with genotyping information available were included in this analysis. Smokers with COPD were defined as smokers (smoking≥20 packs/ year) with post-bronchodilator FEV1/FVC < 0.7 (GOLD stage 1–4) and ‘healthy’ smoking controls were defined as smokers (smoking≥20 packs/year) with post-broncho- dilator FEV1/FVC ≥0.7 (GOLD stage 0). DNA was iso- lated using standard protocols, and SNP genotyping performed using Illumina HumanOmniExpressExome BeadChip and BeadStudio (Illumina, Inc., San Diego, CA). Joint analysis of 10 confirmed candidate SNPs was per- formed, in which eight subjects with homozygous TT genotype of rs28929474 in SERPINA1 (PiZZ genotype) were not included in joint analysis to avoid bias. Genetic scores were defined by the number of risk alleles pre- sented in these 10 SNPs. A linear model was used for analysis of post-bronchodilator FEV1/FVC and percent predicted FEV1 with genetic scores in 1632 current or former smokers. Joint analysis of these 10 candidate SNPs was also performed for post-bronchodilator per- cent predicted FEV1 and percentage of subjects with se- vere COPD (GOLD stage 3–4) in 1077 smokers with COPD. Results Participants were recruited at each center through physician referral, advertisement in clinical areas or self-referral using the SPIROMICS study website (www.spiromics.com). The research protocol was ap- proved by the institutional review boards of all partici- pating institutions with written informed consent from all participants. Statistical analysis For quality control, subjects were removed if they 1) had genotyping call rates< 95%, 2) were discrepant for gen- etic sex, 3) failed the check for family relatedness, or 4) were detected as an outlier. After subjects meeting these Li et al. BMC Medical Genetics (2018) 19:134 Page 3 of 10 Table 1 Demographics (Mean ± SD) of Non-Hispanic White Subjects in SPIROMICS Cases Controls P value All Current smokers Former smokers All Current smokers Former smokers n 1086 325 761 559 210 349 NA Age at enrollment, years 66.2 ± 7.6 62.9 ± 8.1 67.7 ± 6.9 63.6 ± 9.0 58.0 ± 9.1 66.9 ± 7.0 < 0.0001 Female, n (%) 437 (40) 137 (42) 300 (39) 294 (53) 122 (58) 172 (49) < 0.0001 Body mass index 27.4 ± 5.1 25.7 ± 4.9 28.1 ± 5.0 28.6 ± 5.0 27.5 ± 5.1 29.2 ± 4.8 < 0.0001 Current smokers, n (%) 325 (30) 325 (100) 0 (0) 210 (38) 210 (100) 0 (0) 0.0019 Pack-years of cigarette smoking 55.0 ± 25.7 52.6 ± 24.7 56.0 ± 26.1 46.3 ± 27.3 45.3 ± 31.1 46.9 ± 24.8 < 0.0001 Post-bronchodilator FEV1/FVC 0.52 ± 0.13 0.55 ± 0.11 0.50 ± 0.13 0.77 ± 0.05 0.78 ± 0.05 0.77 ± 0.05 < 0.0001 Post-bronchodilator FEV1, % predicted 60.1 ± 22.5 63.6 ± 19.7 58.6 ± 23.5 94.4 ± 13.9 93.4 ± 13.2 95.0 ± 14.3 < 0.0001 Subjects with available GWAS genotyping information available at current stage were included; Cases: GOLD stage 1–4; Controls: GOLD stage 0 Table 1 Demographics (Mean ± SD) of Non-Hispanic White Subjects in SPIROMICS significant level (Table 2 and Additional file 1: Tables S1-S2). No other SNPs in the SERPINA1 region were in strong linkage disequilibrium (LD) with rs28929474 or strongly associated with post-bronchodilator lung function (Figs. 1 and 2). rs4537555 in hedgehog acyltransferase (HHAT) and rs8079868 in myosin heavy chain 3 (MYH3) were strongly associated with post-bronchodilator FEV1/ FVC (p = 2.1 × 10−7) and percent predicted FEV1 (p = 5.9 × 10−7), respectively (Table 2 and Additional file 1: Tables S1-S2). post-bronchodilator FEV1/FVC and percent predicted FEV1, respectively, indicating limited genomic inflation. SNPs in nine genes previously identified for baseline FEV1/FVC or FEV1 in general populations [2–4], extremes of high and low baseline FEV1 [5] or COPD [6] were also associated (p < 0.05) with post-bronchodilator FEV1/FVC or FEV1 (Table 2). Statistical analysis SNPs in RARB, HDAC4, CHRNA3, and RIN3 were associated with post-bronchodilator FEV1/FVC and FEV1. SNPs in HHIP, AGER, FAM13A1, and PID1 were only associated with post-bronchodilator FEV1/FVC. A SNP in MMP12 was only associated with post-bronchodilator FEV1. The associations were signifi- cant at the SNP level with same effect direction as previ- ous findings [2–4, 6]. Association analyses of Post-bronchodilator % predicted FEV1 and FEV1/FVC were performed using linear regression adjusted for age, pack-years of cigarette smoking, and the first two principal components ost-bronchodilator % predicted FEV1 and FEV1/FVC were performed using linear regression adjusted for age, sex, current smoking statu moking, and the first two principal components Association of SERPINA1 with lung function, COPD, and COPD severity Red color scale represents the strength of linkage disequilibrium of SNPs with rs28929474 Prediction of post-bronchodilator pulmonary function Joint analysis of the most consistently associated 10 SNPs, based on our analyses and previous findings was per- formed. Genetic scores (the number of risk alleles) and pack-years of cigarette smoking were significantly associ- ated with post-bronchodilator FEV1/FVC and percent pre- dicted FEV1 (Table 4). Age at enrollment and sex were significantly associated with post-bronchodilator FEV1/ FVC but not associated with percent predicted FEV1. In 1632 SPIROMICS non-Hispanic White smokers (GOLD stage 0–4), genetic score, age, sex, and pack-years of cigarette smoking explained 3.6, 1.5, 1.9, 3.0%, and to- gether 8.6% of the variance of post-bronchodilator FEV1/ FVC (Table 4). Genetic score and pack-years of cigarette smoking explained 3.0, 2.9%, and together 5.8% of the variance of post-bronchodilator percent predicted FEV1 (Table 4). In 1077 SPIROMICS non-Hispanic White smokers with COPD (GOLD stage 1–4), post-broncho- dilator FEV1 decreased significantly with the increase in the number of risk alleles, from 65.4 to 54.0 (p = 1.2 × 10−5) and the percentage of subjects with severe COPD (GOLD stage 3–4) increased significantly from 25.6 to 48.3% (p = 5.5 × 10−5) (Fig. 3). associated with pre−/post-bronchodilator FEV1 (33.5, 61.3, and 72.5 or 1210, 1841, and 2115 ml for genotype TT, CT, and CC, respectively; p = 2.1 × 10−9). Pre−/post- bronchodilator lung function was significantly different between CT and CC or TT and CC genotype groups, how- ever differences between TT and CT genotype groups were not as marked. rs28929474 was associated with post-bronchodilator FEV1/FVC (β = −0.060, p = 1.1 × 10−5) and percent predicted FEV1 (β = −8.73; p = 2.6 × 10−4) in subjects with COPD (GOLD stage 1–4), but not in subjects without COPD (GOLD stage 0; data not shown). Thus, the association of rs28929474 with lung function was driven by subjects with COPD. Additional COPD-related phenotypes were analyzed for association with rs28929474 (Table 3). rs28929474 was also associated with COPD status (odds ratio = 2.3, p = 7.8 × 10−4) and COPD severity (odds ratio = 4.1, p = 0.0036) (Table 3). The percentage of subjects with COPD or severe COPD was significantly higher in subjects with CT genotype than CC genotype. rs28929474 was a less common SNP with minor allele frequency (MAF) of 0.029 in SPIROMICS (Additional file 1: Table S3). Association of SERPINA1 with lung function, COPD, and COPD severity Pre−/post-bronchodilator lung function was stratified by genotypes of rs28929474 (Table 3). rs28929474 was associ- ated in a stepwise fashion with pre−/post-bronchodilator FEV1/FVC ratio (0.39, 0.54, and 0.61 for genotype TT, CT, and CC, respectively; p = 1.2 × 10−8). rs28929474 was rs28929474 (Glu342Lys) in alpha-1 antitrypsin member 1 (SERPINA1) was associated with post-bronchodilator FEV1/FVC (β = −0.087, p = 1.2 × 10−8) and percent pre- dicted FEV1 (β = −13.6, p = 3.5 × 10−8) at a genome-wide ults of the Top SNPs (P < 10−6) and Candidate Lung Function and COPD SNPs (P < 0.05) ble 2 Association Results of the Top SNPs (P < 10−6) and Candidate Lung Function and COPD SNPs (P < 0.05) Table 2 Association Results of the Top SNPs (P < 10−6) and Candidate Lung Function and COPD SNPs (P < 0.05) SNP Gene Chr Location Minor (Effect)/ Major Allele Minor AlleleFrequency Post-bronchodilator FEV1/FVC Post-bronchodilator % predicted FEV1 β P value β P value rs28929474 SERPINA1 14 coding T/C 0.029 −0.087 1.2 × 10−8 −13.6 3.5 × 10−8 rs4537555 HHAT 1 intron G/A 0.11 −0.044 2.1 × 10−7 −6.3 4.1 × 10−6 rs8079868 MYH3 17 3’ C/T 0.12 −0.034 3.5 × 10−5 −6.7 5.9 × 10−7 rs1980057 HHIP 4 5’ T/C 0.37 0.011 0.049 1.4 0.13 rs2070600 AGER 6 coding A/G 0.047 0.026 0.047 3.4 0.10 rs2869967 FAM13A1 4 intron C/T 0.41 −0.014 0.016 −1.4 0.12 rs1435867 PID1 2 3’ C/T 0.075 0.021 0.043 1.9 0.26 rs12477314 HDAC4 2 3’ T/C 0.21 0.014 0.033 2.3 0.035 rs1529672 RARB 3 intron A/C 0.15 0.026 5.1 × 10−4 3.4 5.0 × 10−3 rs12914385 CHRNA3 15 intron T/C 0.43 −0.014 0.014 −2.2 0.014 rs10498635 RIN3 14 intron T/C 0.18 0.021 2.6 × 10−3 3.7 1.5 × 10−3 rs615098 MMP12 11 3’ A/C 0.18 0.013 0.056 2.4 0.034 Association analyses of Post-bronchodilator % predicted FEV1 and FEV1/FVC were performed using linear regression adjusted for age, sex, current smoking status, pack-years of cigarette smoking, and the first two principal components Li et al. BMC Medical Genetics (2018) 19:134 Page 4 of 10 Fig. 1 Association plot of SERPINA1 region with post-bronchodilator FEV1/FVC. Negative logarithm-transformed P value (left) and recombination rate (right). Red color scale represents the strength of linkage disequilibrium of SNPs with rs28929474 Fig. 1 Association plot of SERPINA1 region with post-bronchodilator FEV1/FVC. Negative logarithm-trans rate (right). Association of SERPINA1 with lung function, COPD, and COPD severity Homozygous risk genotype TT was present only in subjects (n = 8) with severe COPD (GOLD stage 3–4). Joint analysis of the top 10 SNPs associated with post-bronchodilator % predicted FEV1 in this study was Li et al. BMC Medical Genetics (2018) 19:134 Page 5 of 10 Fig. 2 Association plot of SERPINA1 region with post-bronchodilator percent predicted FEV1. Negative logarithm-transformed P value (left) and recombination rate (right). Red color scale represents the strength of linkage disequilibrium of SNPs with rs28929474 Fig. 2 Association plot of SERPINA1 region with post-bronchodilator percent predicted FEV1. Negative logarithm-transformed P value (left) an recombination rate (right). Red color scale represents the strength of linkage disequilibrium of SNPs with rs28929474 Fig. 2 Association plot of SERPINA1 region with post-bronchodilator percent predicted FEV1. Negative logarithm-transf recombination rate (right). Red color scale represents the strength of linkage disequilibrium of SNPs with rs28929474 also performed (Additional file 1: Table S2). In 1634 SPIROMICS non-Hispanic White smokers (GOLD stage 0–4), genetic score, age, sex, and pack-years of cigarette smoking explained 7.5, 1.4, 1.9, 3.1%, and together 12.8% of the variance of post-bronchodilator FEV1/FVC (Additional file 1: Table S4). Genetic score and pack-years of cigarette smoking explained 9.9, 3.0%, and together 12.9% of the variance of post-bronchodilator percent predicted FEV1 (Additional file 1: Table S4). Increase in the number of risk alleles from 4 to 6 to 11– 13 was associated with significant decrease in post-bron- chodilator FEV1 from 69.4 to 45.6 (p < 2.2 × 10−16) and with a significant increase in the percentage of subjects with severe COPD (GOLD stage 3–4) from 21.4 to 57.9% (p = 2.2 × 10−12) (Fig. S1). ED visit or hospitalization in last 12 month (Table 5). In general, with the increase of genetic scores, emphysema (p < 0.0001) and airtrapping (p < 0.0001) increased, BODE index (p < 0.0001) and SGRQ total score (p = 0.0044) in- creased, 6MWD (p = 0.0086) decreased, and the percent- age of subjects with exacerbations (p = 0.001) increased. Two extreme genetic score groups (8 to 11 risk alleles vs. 16 to 18 risk alleles) showed statistical and clinical differ- ence for emphysema (5.54 vs. 12.5 of TLC % area < −950 HU), airtrapping (21.8 vs. 33.9 RV % area < −856 HU), BODE index (1.15 vs. 2.21), SGRQ total score (30.4 vs. 35.5), 6MWD (416 m vs. Discussion In this study, we performed GWAS of post-bronchodilator FEV1/FVC and percent predicted FEV1, and identified rs28929474 in SERPINA1. In 1963, Laurell and Eriksson identified the connection between alpha 1-antitrypsin (A1AT) deficiency and degenerative pulmonary disease [11]. The SERPINA1 gene on chr14q32 encodes A1AT pro- tein. The most common variant of SERPINA1 causing A1AT deficiency is the Z allele (rs28929474: Glu342Lys), which is a missense mutation of glutamic acid to lysine at position 342 of A1AT protein. The homozygous TT Association of SERPINA1 with lung function, COPD, and COPD severity 390 m), and percentage of exac- erbations (7.5% vs. 14%). Joint analysis of 10 SNPs with emphysema, clinical symptoms, and exacerbation Association analyses of Pre-/Post-bronchodilator FEV1 and FVC in ml were performed using linear regression adjusted for sex, age, age2, height, height2, weight, current smoking status, pack-years of cigarette smoking, and the first two principal components. Association analyses of Pre-/Post-bronchodilator FEV1, FVC, and FEV1/FVC, and % change in FEV1 were performed using linear regression adjusted for age, sex, current smoking status, pack-years of cigarette smoking, and the first two principal components. Association analyses of COPD and COPD severity were performed using logistic regression adjusted for age, sex, current smoking status, pack-years of cigarette smoking, and the first two principal components Association analyses of age or sex were performed using linear or logistic regression without adjustment. Association analyses of Pre-/Post-bronchodilator FEV1 and FVC in ml were performed using linear regression adjusted for sex, age, age2, height, height2, weight, current smoking status, pack-years of cigarette smoking, and the first two principal components. Association analyses of Pre-/Post-bronchodilator FEV1, FVC, and FEV1/FVC, and % change in FEV1 were performed using linear regression adjusted for age, sex, current smoking status, pack-years of cigarette smoking, and the first two principal components. Association analyses of COPD and COPD severity were performed using logistic regression adjusted for age, sex, current smoking status, pack-years of cigarette smoking, and the first two principal components Association analyses of age or sex were performed using linear or logistic regression without adjustment. Association analyses of Pre-/Post-bronchodilator FEV1 and FVC in ml were performed using linear regression adjusted for sex, age, age2, height, height2, weight, current smoking status, pack-years of cigarette smoking, and the first two principal components. Association analyses of Pre-/Post-bronchodilator FEV1, FVC, and FEV1/FVC, and % change in FEV1 were performed using linear regression adjusted for age, sex, current smoking status, pack-years of cigarette smoking, and the first two principal components. Joint analysis of 10 SNPs with emphysema, clinical symptoms, and exacerbation Joint analysis of 10 candidate SNPs was further per- formed on quantitative Computed Tomography (CT) evidence of emphysema (TLC % area < −950 HU) and airtrapping (RV % area < −856 HU), BODE index, COPD Assessment Test (CAT) score, St. George’s Re- spiratory Questionnaire (SGRQ) total score, 6-Minute Walk Distance (6MWD), and exacerbations requiring Li et al. BMC Medical Genetics (2018) 19:134 Page 6 of 10 Table 3 Association Results of rs28929474 in SERPINA1 with Lung Function, COPD, and COPD Severity Phenotype CC (n = 1559) CT (n = 78) TT (n = 8) TT vs. CT vs. CC CT vs. CC TT vs. CT TT vs. CC β or OR P value β or OR P value β or OR P value β or OR P value Age at enrollment, years 65.4 ± 8.2 64.5 ± 8.1 53.7 ± 3.9 −2.38 0.0029 −0.92 0.33 −10.8 3.6 × 10−4 −11.7 5.7 × 10−5 Sex (Female vs. Male), n 696 vs. 863 33 vs. 45 2 vs. Joint analysis of 10 SNPs with emphysema, clinical symptoms, and exacerbation 6 0.83 0.35 0.91 0.69 0.46 0.35 0.41 0.28 Pack-years of cigarette smoking 52.3 ± 26.6 48.9 ± 26.4 35.2 ± 13.4 −3.97 0.12 −3.10 0.3 −6.62 0.52 −12.3 0.19 Post-bronchodilator FEV1/FVC 0.61 ± 0.16 0.54 ± 0.18 0.39 ± 0.09 −0.087 1.2 × 10−8 −0.077 2.3 × 10−5 −0.090 0.21 −0.23 3.2 × 10−5 Pre-bronchodilator FEV1/FVC 0.59 ± 0.15 0.52 ± 0.17 0.37 ± 0.09 −0.081 9.9 × 10−8 −0.069 1.7 × 10−4 −0.086 0.22 −0.22 3.5 × 10−5 Post-bronchodilator % predicted FEV1 72.5 ± 25.6 61.3 ± 26.4 33.5 ± 7.89 −13.6 3.5 × 10−8 −11.4 9.1 × 10−5 −22.5 0.037 −38.1 2.2 × 10−5 Post-bronchodilator FEV1, ml 2115 ± 888 1841 ± 860 1210 ± 240 −439 2.1 × 10−9 −329 1.4 × 10−4 −1066 1.5 × 10−4 −1395 1.4 × 10−7 Pre-bronchodilator % predicted FEV1 65.6 ± 26.1 54.8 ± 26.9 30.2 ± 7.57 −12.9 8.3 × 10−7 −10.9 5.0 × 10−4 −17.5 0.12 −34.7 1.6 × 10−4 Pre-bronchodilator FEV1, ml 1916 ± 883 1629 ± 833 1091 ± 240 −426 1.1 × 10−8 −339 1.2 × 10−4 −924 8.1 × 10−4 −1263 2.9 × 10−6 % change in FEV1 (BDR) 13.6 ± 13.5 17.1 ± 18.1 11.6 ± 9.75 2.08 0.13 3.49 0.030 −3.56 0.65 −2.77 0.57 Post-bronchodilator % predicted FVC 90.7 ± 17.7 86.9 ± 17.2 71.5 ± 21.1 −5.23 0.0024 −3.83 0.060 −18.0 0.019 −17.0 0.0070 Post-bronchodilator FVC, ml 3518 ± 1011 3482 ± 1036 3365 ± 1015 −236 4.6 × 10−4 −162 0.042 −661 0.023 −822 7.1 × 10−4 Pre-bronchodilator % predicted FVC 84.4 ± 19.4 78.9 ± 18.6 68.7 ± 21.7 −6.07 0.0018 −5.55 0.017 −12.6 0.13 −14.2 0.038 Pre-bronchodilator FVC, ml 3274 ± 1032 3145 ± 1015 3240 ± 1071 −277 2.0 × 10−4 −249 0.0046 −434 0.16 −683 0.011 COPD (GOLD stage 2–4 vs. 0), n 803 vs. 539 53 vs. 20 8 vs. 0 2.31 7.8 × 10−4 1.91 0.019 NA NA NA NA COPD severity (GOLD stage 3–4 vs. 1), n 331 vs. 217 32 vs. 5 8 vs. 0 4.08 0.0036 3.79 0.0081 NA NA NA NA Association analyses of age or sex were performed using linear or logistic regression without adjustment. Joint analysis of 10 SNPs with emphysema, clinical symptoms, and exacerbation Blue bars represent post-bronchodilator percent predicted FEV1, and red bars represent percentages of subjects with severe COPD (GOLD stage 3–4) subjects with a significant history of cigarette smoking, a necessary environmental exposure. Thus, these studies in general populations have limited power to identify the association between rs28929474 and lung function. In this study, we performed GWAS of post-bronchodilator lung function in heavy smokers enriched for COPD. As expected the number of subjects with homozygous TT genotype was rare (n = 8 in 1645 or 0.49%) but the het- erozygous CT genotype was more common (n = 78 or 4.74%). In addition, rs28929474 is not included in the previously designed GWAS chips nor are there other SNPs in strong LD (r2 > 0.5) with rs28929474, preventing the identification of association with COPD and emphysema [6, 14]. The Illumina OmniExpressExome BeadChip used in this study includes exonic markers genotype of rs28929474 (PiZZ genotype) is consistently as- sociated with emphysema, decreased lung function, and COPD [12, 13]. Previous GWAS of COPD, emphysema, and lung function did not identify rs28929474 in SERPINA1 [2–6, 14]. There are several potential reasons for missing this association. rs28929474 is relatively rare in the general population, for example, approximately 2 and 0.01% of the population in the United States are heterozygous or homozygous for the T allele, respectively [15]. The lar- gest meta-analyses of GWAS of baseline lung function in general populations of European descent [2–5] have included tens of thousands subjects, however very few subjects may have been homozygous for the T allele and more importantly these studies did not ascertain Table 5 Joint analysis of 10 SNPs with emphysema, clinical symptoms, and exacerbation Genetic Scorea 8–11 (n = 228) 12–13 (n = 643) 14–15 (n = 612) 16–18 (n = 149) P valueǂ CT Evidence of Emphysema (TLC % Area < −950 HUb) 5.54 ± 6.59 7.60 ± 9.84 8.67 ± 10.4 12.5 ± 12.4 < 0.0001 CT Evidence of Airtrapping (RV % Area < −856 HUb) 21.8 ± 18.1 24.0 ± 20.6 27.6 ± 21.0 33.9 ± 23.2 < 0.0001 BODE Index 1.15 ± 1.62 1.33 ± 1.75 1.61 ± 1.98 2.21 ± 2.41 < 0.0001 COPD Assessment Test (CAT) 12.7 ± 8.0 13.0 ± 8.0 14.0 ± 8.2 13.5 ± 7.9 0.07 St. Joint analysis of 10 SNPs with emphysema, clinical symptoms, and exacerbation Association analyses of COPD and COPD severity were performed using logistic regression adjusted for age, sex, current smoking status, pack-years of cigarette smoking, and the first two principal components Table 4 Prediction Models for Post-bronchodilator Lung Function Post-bronchodilator FEV1/FVC Post-bronchodilator % predicted FEV1 β R2 P value β R2 P value Genetic Score (8–18)a −0.018 0.0363 8.6 × 10−15 −2.6 0.0296 2.7 × 10−12 Age at enrollment, years −0.0024 0.0145 1.1 × 10−6 −0.042 0.000176 0.59 Sex (Male = 0, Female = 1) 0.044 0.0186 3.1 × 10−8 1.88 0.000132 0.14 Pack-years of cigarette smoking −0.0011 0.0304 1.3 × 10−12 −0.017 0.0294 3.2 × 10−12 All NA 0.0859 < 2.2 × 10−16 NA 0.0583 < 2.2 × 10−16 aGenetic scores (the number of risk alleles) of 10 candidate SNPs (rs28929474 in SERPINA1, rs1980057 in HHIP, rs2869967 in FAM13A1, rs2070600 in AGER, rs1435867 in PID1, rs12477314 in HDAC4, rs1529672 in RARB, rs12914385 in CHRNA3, rs10498635 in RIN3, and rs615098 in MMP12). 1632 SPIROMICS non-Hispanic White smokers (GOLD stage 0–4) were included. Eight subjects with TT genotype of rs28929474 in SERPINA1 (PiZZ genotype) were excluded Table 4 Prediction Models for Post-bronchodilator Lung Function Genetic scores (the number of risk alleles) of 10 candidate SNPs (rs28929474 in SERPINA1, rs1980057 in HHIP, rs2869967 in FAM13A1, rs2070600 in AGER, rs1435867 in PID1, rs12477314 in HDAC4, rs1529672 in RARB, rs12914385 in CHRNA3, rs10498635 in RIN3, and rs615098 in MMP12). 1632 SPIROMICS non-Hispanic White smokers (GOLD stage 0–4) were included. Eight subjects with TT genotype of rs28929474 in SERPINA1 (PiZZ genotype) were excluded Li et al. BMC Medical Genetics (2018) 19:134 Page 7 of 10 Fig. 3 Joint analysis of 10 candidate SNPs in 1077 SPIROMICS non-Hispanic White smokers with COPD. 10 SNPs include rs28929474 in SERPINA1, rs1980057 in HHIP, rs2869967 in FAM13A1, rs2070600 in AGER, rs1435867 in PID1, rs12477314 in HDAC4, rs1529672 in RARB, rs12914385 in CHRNA3, rs10498635 in RIN3, and rs615098 in MMP12. Blue bars represent post-bronchodilator percent predicted FEV1, and red bars represent percentages of subjects with severe COPD (GOLD stage 3–4) Fig. 3 Joint analysis of 10 candidate SNPs in 1077 SPIROMICS non-Hispanic White smokers with COPD. 10 SNPs include rs28929474 in SERPINA1, rs1980057 in HHIP, rs2869967 in FAM13A1, rs2070600 in AGER, rs1435867 in PID1, rs12477314 in HDAC4, rs1529672 in RARB, rs12914385 in CHRNA3, rs10498635 in RIN3, and rs615098 in MMP12. aGenetic scores (the number of risk alleles) of 10 candidate SNPs (rs28929474 in SERPINA1, rs1980057 in HHIP, rs2869967 in FAM13A1, rs2070600 in AGER, rs1435867 in PID1, rs12477314 in HDAC4, rs1529672 in RARB, rs12914385 in CHRNA3, rs10498635 in RIN3, and rs615098 in MMP12). 1632 SPIROMICS non-Hispanic White smokers (GOLD stage 0–4) were included. Eight subjects with TT genotype of rs28929474 in SERPINA1 (PiZZ genotype) were excluded. b CT scan-based measures of emphysema (−950 Hounsfield Units or less [%Bilateral Lung Area ≤−950]) and airtrapping (−856 Hounsfield Units or less [%Bilateral Lung Area ≤−856]) measures log-transformed for analysis and adjusted by study site, age, sex, height, BMI, and pack-year smoking history. ǂGeneralized linear model was used with adjusted of age, sex, current smoking status, and pack-years of cigarette smoking. In generalized linear models, CT evidence of emphysema and airtrapping were natural logarithm transformed; 6MWD was logarithm (base 10) transformed Joint analysis of 10 SNPs with emphysema, clinical symptoms, and exacerbation In a previous study, we have showed that HHIP, FAM13A1, AGER and RARB associated with pre-bronchodilator lung function in subjects with asthma [21]. The lung expression quantitative trait locus (eQTL) analysis has identified cis-eQTL SNPs in HHIP, FAM13A1, and AGER [22]. All the evidence indicates rs1980057 in HHIP, rs2869967 in FAM13A1, and rs2070600 in AGER are functionally relevant SNPs im- portant for lung function in the general population and in subjects with COPD or asthma. rs4537555 in HHAT was strongly associated with post-bronchodilator FEV1/ FVC (Table 2). HHAT is a hedgehog acyltransferase which catalyzes N-terminal palmitoylation of sonic hedgehog (SHH). Hedgehog interacting protein (HHIP) and patched homolog 1 (PTCH1) are the other two genes involved in hedgehog signaling pathway and asso- ciated with lung function [2–4, 21], indicating the im- portance of this pathway in lung development and function. Independent replication and functional study of HHAT are warranted. identified from exome and whole genome sequencing projects. rs28919474 (exm1124179) was directly geno- typed. This study found rs28929474 in SERPINA1 to be associated with pre- and post-bronchodilator FEV1/FVC and FEV1 at a genome-wide significant level (Table 3). Although the function of homozygous TT has been known for a long while, the effect of heterozygous CT is more controversial and has been questioned in candidate-gene studies in the past [16–18]. For example, in a general population (n = 4600), baseline FEV1/FVC and FEV1 were not significantly different between PiMM and PiMZ [17]. In a case-control study (834 COPD cases and 835 controls), post-bronchodilator FEV1/FVC and FEV1 were not significantly different between PiMM and PiMZ [16]. In a small study composed of mainly healthy subjects, post-bronchodilator FEV1/FVC (0.77 or 0.71 for PiMM or PiMZ) and percent predicted FEV1 (96.4 or 84.6 for PiMM or PiMZ) were significantly different in ever-smokers but not in never-smokers [18]. In a recent candidate-gene study (5518 non-Hispanic Whites and 2753 African Americans with ≥10 pack-years of smok- ing), subjects with PiMZ had significant lower lung func- tion than subjects with PiMM in both Whites and African Americans [19]. In the current study, subjects with CT genotype had intermediate values for lung function between subjects with TT and CC genotype (Table 3). Subjects with CT genotypes had significantly lower post-bronchodilator FEV1/FVC and percent pre- dicted FEV1, and higher percentage of COPD and more severe COPD than subjects with CC genotype. Joint analysis of 10 SNPs with emphysema, clinical symptoms, and exacerbation Thus, SERPINA1 CT heterozygosity has important functional effects on COPD and lung function. All subjects in- cluded in our study had a history of tobacco smoking with at least a 20-pack-years. Association results were unaffected by the number of pack-years of cigarette smoking in our study. Compared with results from COPDGene study [19], this study included heavier smokers, and thus had lower lung function. More im- portantly, this study is a hypothesis-free GWAS study, which identified association of rs28929474 with lung function at genome-wide significant level for the first time. More than a hundred common and rare variants exist in the SERPINA1 gene. Thun et al. have identified synthetic association between common variants in SER- PINA1 and serum A1AT levels, suggesting A1AT levels are causally determined by rare variants such as Z allele and S allele (rs17580) [20]. Cho et al. have identified rs45505795 in SERPINA10 with MAF of 0.04 (not in strong LD with rs28929474: r2 = 0.295) associated with emphysema [14]. We found no SNP other than rs28929474 in SERPINA1 region to be strongly associ- ated with lung function (Figs. 1 and 2). To develop a multi gene predictive model for lung Since each of these variants alone had smaller effects, we performed a joint analysis of 10 confirmed candidate SNPs. This analysis explained 3.63 and 2.96% variance of post-bronchodilator FEV1/FVC and percent predicted FEV1, respectively (Table 4). In contrast, pack-years of cigarette smoking explained 3.04 and 2.94% variance of post-bronchodilator FEV1/FVC and percent predicted FEV1. A genetic score using these 10 candidate SNPs, age, sex, and pack-years of cigarette smoking together ex- plained 8.59 and 5.83% variance of post-bronchodilator FEV1/FVC and percent predicted FEV1. In addition, joint analysis of 10 confirmed candidate SNPs (with Z allele ho- mozygotes removed) was performed on CT evidence of emphysema and airtrapping, BODE index, COPD, CAT score, SGRQ total score, 6MWD, and exacerbations (Table 5) in all heavy smokers (Gold stage 0–4). Statistical and clinical significant difference was shown between two extreme genetic score groups (8–11 vs. 16–18) for emphy- sema, airtrapping, BODE index, SGRQ total score, 6MWD, and exacerbations, indicating the potential useful- ness of genetic information to distinguish clinical sub- groups of heavy smokers. It will be important to evaluate the power of this model to predict decline in lung function and progression of COPD severity longitudinally in clin- ical settings. Joint analysis of 10 SNPs with emphysema, clinical symptoms, and exacerbation George’s Respiratory Questionnaire (SGRQ) Total Score 30.4 ± 19.7 30.3 ± 19.6 32.9 ± 20.0 35.5 ± 20.6 0.0044 6-Minute Walk Distance (6MWD, meters) 416 ± 114 415 ± 111 418 ± 124 390 ± 126 0.0086 Exacerbations requiring ED Visit or Hospitalization in last 12 months (% Yes) 7.5% 6.5% 11% 14% 0.001 aGenetic scores (the number of risk alleles) of 10 candidate SNPs (rs28929474 in SERPINA1, rs1980057 in HHIP, rs2869967 in FAM13A1, rs2070600 in AGER, rs1435867 in PID1, rs12477314 in HDAC4, rs1529672 in RARB, rs12914385 in CHRNA3, rs10498635 in RIN3, and rs615098 in MMP12). 1632 SPIROMICS non-Hispanic White smokers (GOLD stage 0–4) were included. Eight subjects with TT genotype of rs28929474 in SERPINA1 (PiZZ genotype) were excluded. b CT scan-based measures of emphysema (−950 Hounsfield Units or less [%Bilateral Lung Area ≤−950]) and airtrapping (−856 Hounsfield Units or less [%Bilateral Lung Area ≤−856]) measures log-transformed for analysis and adjusted by study site, age, sex, height, BMI, and pack-year smoking history. ǂGeneralized linear model was used with adjusted of age, sex, current smoking status, and pack-years of cigarette smoking. In generalized linear models, CT evidence of emphysema and airtrapping were natural logarithm transformed; 6MWD was logarithm (base 10) transformed aGenetic scores (the number of risk alleles) of 10 candidate SNPs (rs28929474 in SERPINA1, rs1980057 in HHIP, rs2869967 in FAM13A1, rs2070600 in AGER, rs1435867 in PID1, rs12477314 in HDAC4, rs1529672 in RARB, rs12914385 in CHRNA3, rs10498635 in RIN3, and rs615098 in MMP12). 1632 SPIROMICS non-Hispanic White smokers (GOLD stage 0–4) were included. Eight subjects with TT genotype of rs28929474 in SERPINA1 (PiZZ genotype) were excluded. b CT scan-based measures of emphysema (−950 Hounsfield Units or less [%Bilateral Lung Area ≤−950]) and airtrapping (−856 Hounsfield Units or less [%Bilateral Lung Area ≤−856]) measures log-transformed for analysis and adjusted by study site, age, sex, height, BMI, and pack-year smoking history. ǂGeneralized linear model was used with adjusted of age, sex, current smoking status, and pack-years of cigarette smoking. In generalized linear models, CT evidence of emphysema and airtrapping were natural logarithm transformed; 6MWD was logarithm (base 10) transformed Li et al. BMC Medical Genetics (2018) 19:134 Page 8 of 10 in previous published studies were evaluated. We identi- fied the association of HHIP, FAM13A1, AGER, PID1, HDAC4, RARB, CHRNA3, RIN3, and MMP12 with post-bronchodilator lung function at the SNP level (Table 2). Ethics approval and consent to participate Participants were recruited at each center through physician referral, advertisement in clinical areas or self-referral using the SPIROMICS study website (www.spiromics.com). The research protocol was approved by the institutional review boards of all participating institutions (Wake Forest School of Medicine, Columbia University, University of California at San Francisco, University of California at Los Angeles, University of North Carolina at Chapel Hill, University of Alabama at Birmingham, University of Michigan, Johns Hopkins University School of Medicine, University of Iowa, University of Utah, Weill Cornell Medical College of Cornell University) with written informed consent from all participants. Consent for publication Not applicable. Consent for publication Not applicable. Competing interests X.L.: Associate Editor of BMC Medical Genetics. V.E.O.: funding from the Foundation for the NIH NHLBI in the form of a K08 training award; consultancy fees from CSL Behring. X.L.: Associate Editor of BMC Medical Genetics. V.E.O.: funding from the Foundation for the NIH NHLBI in the form of a K08 training award; consultancy fees from CSL Behring. E.J.A.: no conflicts of interest to disclose. R.G.B.: grants from the Foundation for the NIH, Alpha1 Foundation and personal fees from UpToDate and the COPD Foundation all outside of the submitted work. X.L.: Associate Editor of BMC Medical Genetics. V.E.O.: funding from the Foundation for the NIH NHLBI in the form of a K08 training award; consultancy fees from CSL Behring. Abbreviations COPD Ch E.J.A.: no conflicts of interest to disclose. E.J.A.: no conflicts of interest to disclose. COPD: Chronic obstructive pulmonary disease; FEV1: Forced expiratory volume in 1 s; GWAS: Genome-wide association studies; SPIROMICS: The SubPopulations and InteRmediate Outcome Measures In COPD Study R.G.B.: grants from the Foundation for the NIH, Alpha1 Foundation and personal fees from UpToDate and the COPD Foundation all outside of the submitted work. R.G.B.: grants from the Foundation for the NIH, Alpha1 Foundation and personal fees from UpToDate and the COPD Foundation all outside of the submitted work. S.A.C.: no conflicts of interest to disclose. S.A.C.: no conflicts of interest to disclose. C.B.C.: grants from the Foundation for the NIH and NIH NHLBI; part-time employment by the Global Respiratory Franchise in the GlaxoSmithKline. D.C.: grants from the Foundation for the NIH and NIH NHLBI. M.T.D.: grants from the NIH, the Department of Defense, and the American Heart. C.B.C.: grants from the Foundation for the NIH and NIH NHLBI; part-time employment by the Global Respiratory Franchise in the GlaxoSmithKline. D.C.: grants from the Foundation for the NIH and NIH NHLBI. M T D : grants from the NIH the Department of Defense and the American C.B.C.: grants from the Foundation for the NIH and NIH NHLBI; part-time employment by the Global Respiratory Franchise in the GlaxoSmithKline. D.C.: grants from the Foundation for the NIH and NIH NHLBI. M.T.D.: grants from the NIH, the Department of Defense, and the American H t Authors’ contributions Study design: XL, RGB, DC, MH, EAH, RK, EK, FJM, PGW, ERB, and DAM. Phenotype acquisition: VEO, RGB, SAC, CBC, DC, MTD, MH, NNH, EAH, RK, EK, FJM, RP, PGW, and ERB. Genotype acquisition: XL, VEO, EJA, GAH, ERB, and DAM. Statistical Analysis: XL, EJA, and DAM. All authors have read and approved the manuscript. Availability of data and materials All data generated or analyzed during this study are included in this published article and its supplementary information files. Raw genotype data may be accessible by contacting SPIROMICS (https://www.spiromics.org). Acknowledgments R.E.K.: grants from the Foundation for the NIH and NIH NHLBI. E.K.: grants from the Foundation for the NIH and NIH NHLBI; grants from Boehringer-Ingelheim, Novartis, Pearl, AstraZeneca, and Sunovion outside of the submitted work. E.A.H.: grants from the Foundation for the NIH and NIH NHLBI; founder and shareholder of VIDA Diagnostics. E.A.H.: grants from the Foundation for the NIH and NIH NHLBI; founder and shareholder of VIDA Diagnostics. R.E.K.: grants from the Foundation for the NIH and NIH NHLBI. E.K.: grants from the Foundation for the NIH and NIH NHLBI; grants from Boehringer-Ingelheim, Novartis, Pearl, AstraZeneca, and Sunovion outside of the submitted work. F.J.M.: grants from National Institutes of Health, Clarion, Continuing Education, Potomac, Afferent, and Adept; personal fees from Forest, Janssen, GlaxoSmithKline, Nycomed/Takeda, Amgen, Astra Zeneca, Boehringer- Ingelheim, Ikaria/Bellerophon, Genentech, Janssen, Johnson & Johnson, Novartis, Pearl, Pfizer, Roche, Sunovion, Theravance, Axon Communication, CME Incite, California Society for Allergy and Immunology, Annenberg, Integritas, InThought, Miller Medical, National Association for Continuing Education, Paradigm, Peer Voice, UpToDate, Haymarket Communications, Western Society of Allergy and Immunology, Bioscale, Unity Biotechnology, ConCert, Lucid, Methodist Hospital, Prime, WebMD, Mereo, Kadmon, Pfizer, Veracyte, American Thoracic Society, Academic CME, Falco, and the National Association for Continuing Education. Joint analysis of 10 SNPs with emphysema, clinical symptoms, and exacerbation In summary, rs28929474 in SERPINA1 is clearly asso- ciated with post-bronchodilator FEV1/FVC and FEV1 among heavy smokers. This study is the first to show genome-wide significant association of rs28929474 with lung function. In addition, rs28929474 is associated with To develop a multi-gene predictive model for lung function, genes associated with lung function and COPD Li et al. BMC Medical Genetics (2018) 19:134 Page 9 of 10 Page 9 of 10 COPD and COPD severity. While well-established rare ZZ homozygotes have severe COPD and emphysema, this study establishes that more common heterozygotes (4.7% of subjects) at this locus lead to pulmonary abnor- mality in smokers and COPD. Thus, in future clinical studies, this largely ignored heterozygotes group should be carefully examined. A joint genetic model combined with environmental factors is associated with reduced lung function, emphysema, exacerbation, and clinical symptoms. The models should be tested in other popu- lations as well as longitudinally to evaluate potential value of predicting COPD progression and severity. Additional file Additional file 1: Table S1. Association Results of the Top SNPs (P < 10−4) with Post-bronchodilator FEV1/FVC. Table S2. Association Results of the Top SNPs (P < 10−4) with Post-bronchodilator % Predicted FEV1.Table S3. Genotype Frequency of rs28929474 in SERPINA1 Stratified by GOLD Stages. Table S4. Prediction Models for Post-bronchodilator Lung Function Using Top 10 SNPs for Post-bronchodilator % Predicted FEV1.Figure S1. Joint analysis of the top10 SNPs for post-bronchodilator % predicted FEV1 in 1075 SPIROMICS non-Hispanic White smokers with COPD. (DOCX 141 kb) Additional file 1: Table S1. Association Results of the Top SNPs (P < 10−4) with Post-bronchodilator FEV1/FVC. Table S2. Association Results of the Top SNPs (P < 10−4) with Post-bronchodilator % Predicted FEV1.Table S3. Genotype Frequency of rs28929474 in SERPINA1 Stratified by GOLD Stages. Table S4. Prediction Models for Post-bronchodilator Lung Function Using Top 10 SNPs for Post-bronchodilator % Predicted FEV1.Figure S1. Joint analysis of the top10 SNPs for post-bronchodilator % predicted FEV1 in 1075 SPIROMICS non-Hispanic White smokers with COPD. (DOCX 141 kb) Acknowledgments The authors thank the SPIROMICS participants and participating physicians, investigators and staff for making this research possible. More information about the study and how to access SPIROMICS data is at www.spiromics.org. We would like to acknowledge the following current and former investigators of the SPIROMICS sites and reading centers: Neil E Alexis, PhD; Wayne H Anderson, PhD; R Graham Barr, MD, DrPH; Eugene R Bleecker, MD; Richard C Boucher, MD; Russell P Bowler, MD, PhD; Elizabeth E Carretta, MPH; Stephanie A Christenson, MD; Alejandro P Comellas, MD; Christopher B Cooper, MD, PhD; David J Couper, PhD; Gerard J Criner, MD; Ronald G Crystal, MD; Jeffrey L Curtis, MD; Claire M Doerschuk, MD; Mark T Dransfield, MD; Christine M Freeman, PhD; MeiLan K Han, MD, MS; Nadia N Hansel, MD, MPH; Annette T Hastie, PhD; Eric A Hoffman, PhD; Robert J Kaner, MD; Richard E Kanner, MD; Eric C Kleerup, MD; Jerry A Krishnan, MD, PhD; Lisa M LaVange, PhD; Stephen C Lazarus, MD; Fernando J Martinez, MD, MS; Deborah A Meyers, PhD; John D Newell Jr., MD; Elizabeth C Oelsner, MD, MPH; Wanda K O’Neal, PhD; Robert Paine, III, MD; Nirupama Putcha, MD, MHS; Stephen I. Rennard, MD; Donald P Tashkin, MD; Mary Beth Scholand, MD; J Michael Wells, MD; Robert A Wise, MD; and Prescott G Woodruff, MD, MPH. The project officers from the Lung Division of the National Heart, Lung, and Blood Institute were Lisa Postow, PhD, and Thomas Croxton, PhD, MD. M.T.D.: grants from the NIH, the Department of Defense, and the American Heart. Association; consultancy fees from Boehringer Ingelheim, Boston Scientific, and GlaxoSmithKline and contracted clinical trials from Boehringer Ingelheim, Boston Scientific, GlaxoSmithKline, Pearl, Pulmonx, PneumRx, AstraZeneca, Novartis, and Yungjin. M.H.: grants from the NIH NHLBI and the Foundation for the NIH; consultancy fees from GlaxoSmithKline, Boehringer-Ingelheim, Novartis, and AstraZeneca. Association; consultancy fees from Boehringer Ingelheim, Boston Scientific, and GlaxoSmithKline and contracted clinical trials from Boehringer Ingelheim, Boston Scientific, GlaxoSmithKline, Pearl, Pulmonx, PneumRx, AstraZeneca, Novartis, and Yungjin. M.H.: grants from the NIH NHLBI and the Foundation for the NIH; consultancy fees from GlaxoSmithKline, Boehringer-Ingelheim, Novartis, and AstraZeneca. N.N.H.: grants from the Foundation for the NIH and NIH NHLBI. N.N.H.: grants from the Foundation for the NIH and NIH NHLBI. E.A.H.: grants from the Foundation for the NIH and NIH NHLBI; founder and shareholder of VIDA Diagnostics. Author details 1 17. Thun GA, Ferrarotti I, Imboden M, et al. SERPINA1 PiZ and PiS heterozygotes and lung function decline in the SAPALDIA cohort. PLoS One. 2012;7:e42728. 1Division of Genetics, Genomics and Precision Medicine, Department of Medicine, University of Arizona, BioScience Research Lab, Room 253, 1230 N. Cherry Avenue, PO Box 210242, Tucson, AZ 85721, USA. 2Center for Genomics and Personalized Medicine Research, Wake Forest School of Medicine, Winston-Salem, North Carolina, USA. 3Department of Medicine, Columbia University, New York, NY, USA. 4Division of Pulmonary, Critical Care, Sleep & Allergy, Department of Medicine and Cardiovascular Research Institute, University of California at San Francisco, San Francisco, California, USA. 5Department of Medicine, University of California at Los Angeles, Los Angeles, California, USA. 6Department of Biostatistics, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA. 7Division of Pulmonary, Allergy & Critical Care Medicine, Lung Health Center, University of Alabama at Birmingham, Birmingham, AL, USA. 8Division of Pulmonary & Critical Care, University of Michigan, Ann Arbor, MI, USA. 9Division of Pulmonary and Critical Care Medicine, Johns Hopkins University School of Medicine, Baltimore, MD, USA. 10Department of Radiology, University of Iowa, Iowa City, Iowa, USA. 11Department of Internal Medicine/Pulmonary and Critical Care Medicine, University of Utah, Salt Lake City, UT, USA. 12Department of Medicine Weill Cornell Medical College of Cornell 18. Molloy K, Hersh CP, Morris VB, et al. Clarification of the risk of chronic obstructive pulmonary disease in α1-antitrypsin deficiency PiMZ heterozygotes. Am J Respir Crit Care Med. 2014;189:419–27. 19. Foreman MG, Wilson C, DeMeo DL, et al. Alpha-1 antitrypsin PI MZ genotype is associated with COPD in two racial groups. Ann Am Thorac Soc 2017. https://doi.org/10.1513/AnnalsATS.201611-838OC. Epub ahead of print. 20. Thun GA, Imboden M, Ferrarotti I, et al. Causal and synthetic associations of variants in the SERPINA gene cluster with alpha1-antitrypsin serum levels. PLoS Genet. 2013;9:e1003585. 21. Li X, Hawkins GA, Ampleford EJ, et al. Genome-wide association study identifies TH1 pathway genes associated with lung function in asthmatic patients. J Allergy Clin Immunol. 2013;132:313–20. 22. Obeidat M, Hao K, Bossé Y, et al. Molecular mechanisms underlying variations in lung function: a systems genetics analysis. Lancet Respir Med. 2015;3:782–95. 12Department of Medicine, Weill Cornell Medical College of Cornell University, New York, NY, USA. 12Department of Medicine, Weill Cornell Medical College of Cornell University, New York, NY, USA. Funding SPIROMICS g SPIROMICS was supported by contracts from the NIH/NHLBI (HHSN268200900013C, HHSN268200900014C, HHSN268200900015C, HHSN268200900016C, HHSN268200900017C, HHSN268200900018C, HHSN268200900019C, HHSN268200900020C), which were supplemented by contributions made through the Foundation for the NIH from AstraZeneca; Bellerophon Pharmaceuticals; Boehringer-Ingelheim Pharmaceuticals, Inc.; Chiesi Farm- aceutici SpA; Forest Research Institute, Inc.; GSK; Grifols Therapeutics, Inc.; Ikaria, Inc.; Nycomed GmbH; Takeda Pharmaceutical Company; Novartis Pharmaceuticals Corporation; Regeneron Pharmaceuticals, Inc.; and Sanofi. The funders had no role in the study design, data collection, data analysis, decision to publish, or preparation of the manuscript. R.P.: grants from the Foundation for the NIH and NIH NHLBI. P.G.W.: grants from Medimmune and consultancy fees from Genentech/ Roche, Astra Zeneca, Novartis, Neostem, Janssen outside the submitted work; a patent with Asthma diagnostics pending. Page 10 of 10 Page 10 of 10 Page 10 of 10 Li et al. BMC Medical Genetics (2018) 19:134 G.A.H.: no conflicts of interest to disclose. 11. Laurell CB, Eriksson S. The electrophoretic alpha 1 globulin pattern of serum in alpha 1 antitrypsin deficiency. Scand J Clin Lab Invest. 1963;15:132–40. G.A.H.: no conflicts of interest to disclose. E.R.B.: grants from the NIH NHLBI for the Severe Asthma Research Program, AsthmaNet, SPIROMICS, and the Foundation for the NIH; consultancy fees from Amgen, AstraZeneca-MedImmune, Boehringer-Ingelheim, Genentech/ Roche, GlaxoSmithKline, Knopp, Novartis, and Sanofi/Regeneron; funds for clinical trials administered through the Wake Forest School of Medicine from Amgen, AstraZeneca-MedImmune, Boehringer-Ingelheim, Genentech/Roche, GlaxoSmithKline, Janssen/Johnson & Johnson, Novartis, Pzifer, Sanofi- Regeneron, and Teva. 12. DeMeo DL, Silverman EK. Alpha1-antitrypsin deficiency. 2: genetic aspects of alpha(1)-antitrypsin deficiency: phenotypes and genetic modifiers of emphysema risk. Thorax. 2004;59:259–64. 13. Brebner JA, Stockley RA. Recent advances in α-1-antitrypsin deficiency- related lung disease. Expert Rev Respir Med. 2013;7:213–29. 14. Cho MH, Castaldi PJ, Hersh CP, et al. A genome-wide association study of emphysema and airway quantitative imaging phenotypes. Am J Respir Crit Care Med. 2015;192:559–69. D.A.M.: no conflicts of interest to disclose. 15. de Serres FJ, Blanco I. Prevalence of α1-antitrypsin deficiency alleles PIS and PIZ worldwide and effective screening for each of the five phenotypic classes PIMS, PIMZ, PISS, PISZ, and PI*ZZ: a comprehensive review. Ther Adv Respir Dis. 2012;6:277–95. Publisher’s Note S Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. 16. Sørheim IC, Bakke P, Gulsvik A, et al. α1-Antitrypsin protease inhibitor MZ heterozygosity is associated with airflow obstruction in two large cohorts. Chest. 2010;138:1125–32. References 1. Vestbo J, Hurd SS, Agustí AG, et al. Global strategy for the diagnosis, management, and prevention of chronic obstructive pulmonary disease: GOLD executive summary. Am J Respir Crit Care Med. 2013;187:347–65. 1. Vestbo J, Hurd SS, Agustí AG, et al. Global strategy for the diagnosis, management, and prevention of chronic obstructive pulmonary disease: GOLD executive summary. Am J Respir Crit Care Med. 2013;187:347–65. 2. Soler Artigas M, Loth DW, Wain LV, et al. Genome-wide association and large-scale follow up identifies 16 new loci influencing lung function. Nat Genet. 2011;43:1082–90. 2. Soler Artigas M, Loth DW, Wain LV, et al. Genome-wide association and large-scale follow up identifies 16 new loci influencing lung function. Nat Genet. 2011;43:1082–90. 3. Hancock DB, Eijgelsheim M, Wilk JB, et al. Meta-analyses of genome-wide association studies identify multiple loci associated with pulmonary function. Nat Genet. 2010;42:45–52. 3. Hancock DB, Eijgelsheim M, Wilk JB, et al. Meta-analyses of genome-wide association studies identify multiple loci associated with pulmonary function. Nat Genet. 2010;42:45–52. 4. Repapi E, Sayers I, Wain LV, et al. Genome-wide association study identifies five loci associated with lung function. Nat Genet. 2010;42:36–44. 4. Repapi E, Sayers I, Wain LV, et al. Genome-wide association study identifies five loci associated with lung function. Nat Genet. 2010;42:36–44. 5. Wain LV, Shrine N, Miller S, et al. Novel insights into the genetics of smoking behaviour, lung function, and chronic obstructive pulmonary disease (UK BiLEVE): a genetic association study in UK biobank. Lancet Respir Med. 2015;3:769–81. 6. Cho MH, McDonald ML, Zhou X, et al. Risk loci for chronic obstructive pulmonary disease: a genome-wide association study and meta-analysis. Lancet Respir Med. 2014;2:214–25. 6. Cho MH, McDonald ML, Zhou X, et al. Risk loci for chronic obstructive pulmonary disease: a genome-wide association study and meta-analysis. Lancet Respir Med. 2014;2:214–25. 7. Couper D, LaVange LM, Han M, et al. Design of the Subpopulations and Intermediate Outcomes in COPD study (SPIROMICS). Thorax. 2014;69:491–4. 7. Couper D, LaVange LM, Han M, et al. Design of the Subpopulations and Intermediate Outcomes in COPD study (SPIROMICS). Thorax. 2014;69:491–4 8. Woodruff PG, Barr RG, Bleecker E, et al. Clinical significance of symptoms in smokers with preserved pulmonary function. N Engl J Med. 2016;374:1811–21. 8. Woodruff PG, Barr RG, Bleecker E, et al. Clinical significance of symptoms in smokers with preserved pulmonary function. N Engl J Med. 2016;374:1811–21. 9. Purcell S, Neale B, Todd-Brown K, et al. 20. Thun GA, Imboden M, Ferrarotti I, et al. Causal and synthetic associations of variants in the SERPINA gene cluster with alpha1-antitrypsin serum levels. PLoS Genet. 2013;9:e1003585. 14. Cho MH, Castaldi PJ, Hersh CP, et al. A genome-wide association study of emphysema and airway quantitative imaging phenotypes. Am J Respir Crit Care Med. 2015;192:559–69. References PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet. 2007; 81:559–75. 10. Johnson AD, Handsaker RE, Pulit SL, et al. SNAP: a web-based tool for identification and annotation of proxy SNPs using HapMap. Bioinformatics. 2008;24:2938–9.
https://openalex.org/W2976781807	https://europepmc.org/articles/pmc6776974?pdf=render	English	null	Neural Correlates of Facial Expression Recognition in Earthquake Witnesses	Frontiers in neuroscience	2,019	cc-by	9,479	ORIGINAL RESEARCH published: 27 September 2019 doi: 10.3389/fnins.2019.01038 ORIGINAL RESEARCH published: 27 September 2019 doi: 10.3389/fnins.2019.01038 Neural Correlates of Facial Expression Recognition in Earthquake Witnesses Francesca Pistoia1, Massimiliano Conson2, Mario Quarantelli3, Luca Panebianco4, Antonio Carolei1, Giuseppe Curcio4, Simona Sacco1, Gennaro Saporito4, Ernesto Di Cesare4, Antonio Barile4, Carlo Masciocchi4 and Alessandra Splendiani4* Francesca Pistoia1, Massimiliano Conson2, Mario Quarantelli3, Luca Panebianco4, Antonio Carolei1, Giuseppe Curcio4, Simona Sacco1, Gennaro Saporito4, Ernesto Di Cesare4, Antonio Barile4, Carlo Masciocchi4 and Alessandra Splendiani4* 1 Department of Biotechnological and Applied Clinical Sciences, Neurological Institute, University of L’Aquila, L’Aquila, Italy, 2 Developmental Neuropsychology Laboratory, Department of Psychology, University of Campania Luigi Vanvitelli, Campania, Italy, 3 Institute of Biostructure and Bioimaging, National Research Council, Naples, Italy, 4 Department of Biotechnological and Applied Clinical Sciences, University of L’Aquila, L’Aquila, Italy Keywords: earthquake, emotional, fMRI, visual network, default-mode network Reviewed by: Reviewed by: Wenbin Guo, The Second Xiangya Hospital, Central South University, China Antonio Ferretti, Università degli Studi “G. d’Annunzio” Chieti–Pescara, Italy *Correspondence: Alessandra Splendiani alessandra.splendiani@univaq.it Specialty section: This article was submitted to Brain Imaging Methods, a section of the journal Frontiers in Neuroscience Received: 03 April 2019 Accepted: 13 September 2019 Published: 27 September 2019 Edited by: Edited by: Roberto Esposito, Azienda Ospedaliera Ospedali Riuniti Marche Nord, Italy Major adverse events, like an earthquake, trigger different kinds of emotional dysfunctions or psychiatric disorders in the exposed subjects. Recent literature has also shown that exposure to natural disasters can increase threat detection. In particular, we previously found a selective enhancement in the ability to read emotional facial expressions in L’Aquila earthquake witnesses, suggesting hypervigilance to stimuli signaling a threat. In light of previous neuroimaging data showing that trauma exposure is related to derangement of resting-state brain activity, in the present study we investigated the neurofunctional changes related to the recognition of emotional faces in L’Aquila earthquake witnesses. Speciﬁcally, we tested the relationships between accuracy in recognizing facial expressions and activity of the visual network (VN) and of the default-mode network (DMN). Resting-state functional connectivity (FC) with the main hub of the VN (primary, ventral, right-dorsal, and left-dorsal visual cortices) and DMN (posterior cingulate/precuneus, medial prefrontal, and right and left inferior parietal cortices) was investigated through a seed-based functional magnetic resonance imaging (fMRI) analysis in both earthquake-exposed subjects and non-exposed persons who did not live in an earthquake-affected area. The results showed that, in earthquake- exposed subjects, there is a signiﬁcant reduction in the correlation between accuracy in recognizing facial expressions and the FC of the dorsal seed of the VN with the right inferior occipito-temporal cortex and the left lateral temporal cortex, and of two parietal seeds of DMN, i.e., lower parietal and medial prefrontal cortex, with the precuneus bilaterally. These ﬁndings suggest that a functional modiﬁcation of brain systems involved in detecting and interpreting emotional faces may represent the neurophysiological basis of the speciﬁc “emotional expertise” observed in the earthquake witnesses. INTRODUCTION Results actually showed that, compared to no-trauma controls, trauma-exposed women showed greater connectivity between the amygdala and the cingulate cortex not only during aﬀective processing but also at rest. Long et al. (2014) also demonstrated that testing functional brain changes using resting-state functional magnetic resonance imaging (rs- fMRI) is a useful approach in classifying people who have survived an earthquake who develop psychological responses to trauma exposure. Furthermore, although earthquake witnesses, especially those without PTSD, may not have structural brain changes shortly after the trauma, functional brain changes have been demonstrated as occurring in as little as 1 month after an earthquake (Lui et al., 2009). Against this background, in the present study, we used rs-fMRI to investigate the neurofunctional basis of enhanced recognition of facial expressions in earthquake witnesses. Data from diﬀerent neurofunctional approaches demonstrate that the processing of facial expressions crucially involves central nodes of the visual network (VN), like the inferior occipito- temporal cortex, the fusiform and the lingual gyrus, and the lateral temporal cortex (Haxby et al., 2000; Gorno-Tempini et al., 2001; Ganel et al., 2005; Said et al., 2011). Several studies also support the role of the default-mode network (DMN) in interpreting emotional faces, with structures including the medial prefrontal and parietal cortices (Phillips et al., 1998; Harmer et al., 2001; Mattavelli et al., 2011, 2016); the DMN is especially relevant in conditions where explicit expression processing is required (Mattavelli et al., 2016; Zhao et al., 2017). Importantly, the involvement of the VN and the DMN has been demonstrated in trauma-exposed persons during processing of arousing stimuli (Fani et al., 2012; Todd et al., 2015). In this context, we recently reported on a speciﬁc emotional expertise developed by earthquake-exposed subjects without PTSD (Pistoia et al., 2018). In detail, we compared two groups of students, one with a permanent residence in the earthquake epicenter of L’Aquila (central Italy) on April 6, 2009, and one not living in an earthquake-aﬀected area. Participants took part in two behavioral experiments aimed at evaluating their ability to recognize facial expressions and to evaluate emotionally evocative scenes. The results demonstrated that students living in the earthquake-aﬀected areas were signiﬁcantly more accurate than controls in recognizing facial expressions, whereas the two groups did not diﬀer in the evaluation of emotionally evocative scenes. INTRODUCTION maintain their safety by systematically paying attention to potential signs of approaching threat, such as emotional facial expressions (Bell et al., 2017). This interpretation is even more convincing if we consider that the earthquake-exposed students living in L’Aquila experienced a long-lasting exposure to earthquakes with massive psychological distress, as the main event in 2009 was followed by hundreds of thousands of aftershocks in the months afterward and by additional earthquakes in 2016 and 2017. After a natural disaster, like an earthquake, people usually experience diﬀerent kinds of emotional dysfunctions or disorders. Among post-earthquake psychiatric complications, the most frequently reported include post-traumatic stress disorder (PTSD), depression, anxiety, obsessive-compulsive disorders, and social phobia (Farooqui et al., 2017; Dube et al., 2018; Geng et al., 2018; Raﬁey et al., 2019). The risk of sleep disorders and of prolonged grief symptoms also seems to be increased in earthquake-exposed subjects (Tang et al., 2018; Eisma et al., 2019). On the other hand, experiencing a natural disaster may trigger post-traumatic growth, which refers to positive personality changes following threatening life events and a higher level of functioning (Calhoun and Tedeschi, 2006). Therefore, the spectrum of trauma-related changes may encompass a wide range of manifestations, with subjects who have a poor coping ability being more likely to develop psychiatric disorders and subjects with greater adaptability being able to obtain advantages from adverse events. While the former experience a decreased quality of life after a natural disaster and often need speciﬁc psychological or pharmacological support, the latter may thrive (O’Leary and Ickovics, 1995). Independent of developing a clear psychopathological condition, such as PTSD, convergent evidence suggests that trauma exposure increases threat detection (Hayes et al., 2012; Zhang et al., 2014; Bell et al., 2017; Pistoia et al., 2018). In a relevant study on earthquake- exposed persons, Bell et al. (2017) demonstrated that both individuals who develop PTSD and individuals without PTSD are signiﬁcantly more accurate than non-exposed controls in recognizing emotional facial expressions. The authors interpreted their results in terms of increased sensitivity to threat due to the prolonged exposure to aftershocks in the earthquake-exposed groups. Recently, Kleshchova et al. (2019) tested the hypothesis that since chronic hypervigilance is a persistent rather than a reactive state, brain correlates can be directly observable under resting- state conditions without the need for exposure to aﬀectively charged stimuli. Citation: Citation: Pistoia F, Conson M, Quarantelli M, Panebianco L, Carolei A, Curcio G, Sacco S, Saporito G, Di Cesare E, Barile A, Masciocchi C and Splendiani A (2019) Neural Correlates of Facial Expression Recognition in Earthquake Witnesses. Front. Neurosci. 13:1038. doi: 10.3389/fnins.2019.01038 September 2019 \| Volume 13 \| Article 1038 1 Frontiers in Neuroscience \| www.frontiersin.org rs-fMRI in Earthquake Witnesses Pistoia et al. Frontiers in Neuroscience \| www.frontiersin.org MATERIALS AND METHODS as a dependent variable. The Tolerance of Uncertainty Scale Short Form (IUS-12; Freeston et al., 1994) measures responses to uncertainty, ambiguous situations, and the future. It provides a measure of both prospective anxiety and inhibitory anxiety, as well as a total measure of uncertainty (by summing the scores to all the 12 items). We considered the total score as a dependent variable. The Uncertainty Response Scale (URS; Greco and Roger, 2001) is a scale for the evaluation of styles of coping with uncertainty and can provide a measure of three subscales (emotional uncertainty, desire for control, and cognitive uncertainty). We considered the three subscale scores and the total score as dependent variables. The Anxiety Sensitivity Index 3 (ASI-3; Taylor et al., 2007; Petrocchi et al., 2015) measures vulnerability to anxiety. Higher scores reﬂect higher levels of anxiety. We considered the physical concerns, social concerns, and cognitive concerns subscales as well as the total score (sum of all the three subscales) as dependent variables. Finally, the Eysenck Personality Questionnaire-Revised Short Form (EPQ- RS; Eysenck et al., 1985; Picconi et al., 2018) was used to assess the personality characteristics of participants. In particular, here we used the scores for neuroticism, extraversion/introversion, and psychoticism scales as dependent variables. INTRODUCTION This enhanced recognition ability was not selective for speciﬁc emotions, at variance to what has previously been found in neurological patients (Mather and Carstensen, 2005; Pistoia et al., 2010). However, both positive (happiness, surprise) and negative (disgust, fear, anger, sadness) emotions were involved. These results were interpreted in terms of hypervigilance in respect of threats in earthquake witnesses, since trauma exposure, especially in an environment of ongoing threat, requires individuals to By capitalizing on the above evidence, here we used a subsample of Pistoia et al.’s (2018) group of L’Aquila earthquake witnesses to test resting-state functional connectivity (FC) within the major hubs of the VN and the DMN and, crucially, to test FC as it relates to behavioral performance in facial expression recognition task (Pistoia et al., 2018). We hypothesized that enhanced recognition of facial expressions in earthquake- exposed persons could imply an altered pattern of FC between the seeds of the VN and DMN and brain regions related to detection and interpretation of emotional facial expressions, such as the inferior occipito-temporal cortex, the lateral temporal cortex, and the medial parietal cortex. September 2019 \| Volume 13 \| Article 1038 2 rs-fMRI in Earthquake Witnesses Pistoia et al. Recognition of Facial Expressions Task g p In Pistoia et al.’s (2018) study, participants also took part in behavioral experiments aimed at evaluating their ability to recognize facial expressions (using the Ekman and Friesen Pictures of Facial Aﬀect) and to evaluate emotionally evocative scenes (using the International Aﬀective Picture System). In the present study, we speciﬁcally focused on the participants’ accuracy in recognizing emotional facial expression, that is the ability to correctly identify actors from the Ekman and Friesen (1976) set of Pictures of Facial Aﬀect (Ekman, 1993) displaying the six basic emotions: happiness, sadness, anger, fear, disgust, and surprise [see Pistoia et al. (2018) for a detailed description of the experimental procedure]. The research protocol was approved by the Internal Review Board of the University of L’Aquila (January 2017). The study was conducted in accordance with the ethical standards of the Helsinki Declaration and signed informed consent was obtained from all the participants. rs-fMRI All participants were assessed by means of a series of formalized self-report measures. The Beck Depression Inventory (BDI; Beck, 1967) is one of the most widely used self-report measures for the assessment of depression severity. The score can range from 0 to 63, with higher scores indicating an increasing level of depressive symptoms. The score is usually taken as a dependent variable. The State–Trait Anxiety Inventory (STAI; Spielberger et al., 1983; Pedrabissi and Santiniello, 1989) is a commonly used measure of trait and state anxiety: here, only the 20 items for the assessment of trait anxiety were used. The score can range from 20 to 60; a high score reﬂects a high level of anxiety. This score was used as a dependent variable. The Insomnia Severity Index (ISI; Bastien et al., 2001; Castronovo et al., 2016) is a self-report questionnaire evaluating diﬀerent dimensions of insomnia (sleep onset, sleep maintenance and early morning awakening problems, sleep dissatisfaction, interference with daytime functioning, noticeability of sleep problems by others, and distress caused by the sleep diﬃculties). The score ranges from 0 to 28, with higher scores indicating higher severity of insomnia symptoms; the score was taken into consideration Frontiers in Neuroscience \| www.frontiersin.org Participants The original sample of the main neuropsychological study by Pistoia et al. (2018) included 107 students, 48 belonging to the experimental earthquake-exposed group with a permanent residence in L’Aquila at the time of the 2009 earthquake (20 males and 28 females, mean age = 22.6, SD = 2.3 years) and 59 belonging to the control group not living in an earthquake-aﬀected area (30 males and 29 females, mean age = 23.1, SD = 1.6 years). The original inclusion criteria were: (i) no history of previous or coexistent neurological or psychiatric diseases including PTSD, as revealed by a psychiatric examination; (ii) no assumption of drugs or substances acting on the central nervous system; and (iii) signed informed consent to participate in the study. Here, a subsample was selected from both the earthquake- exposed and the non-exposed group to perform the rs-fMRI study. MRI assessment was restricted to a subsample of subjects because neuroimaging assessment requires a longer time to organize and complete, and not all the subjects originally included were available at the time of the neurofunctional assessment. Following selection, 41 (38%) subjects were included, 18 students belonging to the earthquake-exposed group (8 males and 10 females, mean age = 24.5, SD = 1.8 years) and 23 students belonging to the non-exposed control group (14 males and 9 females, mean age = 23.7, SD = 2.0 years); non-parametric between-group comparisons showed that the two groups did not diﬀer with respect to both sex and age (both p > 0.05). Statistical Analysis For each seed, FC maps were then entered in a second-level analysis. To identify diﬀerences between the two groups in the strength of the correlation with the VN, or signiﬁcant interactions between the group and the correlations of the FC of the seeds of both the VN and the DMN with the selected behavioral scores, FC maps were statistically analyzed using a multiple regression analysis within the general linear model framework. Both contrasts (exposed > non-exposed; non-exposed > exposed) were probed when comparing the two subject groups for both the between-group diﬀerences and the interaction analyses. Seed- based fMRI analysis was restricted to voxels falling in a GM mask, obtained thresholding at 0.2 the mean of the GM maps obtained in the segmentation step. For all the analyses, age and sex were included as nuisance covariates in the model, along with the mean framewise displacement derived from the motion correction procedure. Results, corrected for family-wise error (FWE) at cluster level, following a cluster-deﬁning threshold of 0.001, were considered signiﬁcant when surviving an alpha level of 0.05, corrected according to Bonferroni for the number of tests Normalized EPI volumes were then resampled to a voxel size of 3 × 3 × 3 mm3. A rigorous removal of signal contributions from head movements and from physiological variations unrelated to neuronal activity was implemented by regressing out the mean white matter and cerebro-spinal ﬂuid signals (Whitﬁeld-Gabrieli and Nieto-Castanon, 2012), along with six framewise motion parameters derived from the motion correction routine (i.e., rotations and shifts along the three orthogonal main axes). 2http://www.humanconnectomeproject.org p p j g TABLE 1 \| XYZ coordinates in the MNI space of the center of mass and size of the eight seeds used for the analysis. Network Seed Center of mass MNI coordinates (mm) Size (mm3) Visual Primary 2, −79, −12 79,224 Ventral 0, −93, −4 48,712 Right dorsal 38, −72, 13 33,968 Left dorsal −37, −79, 10 24,832 Default-mode Posterior cingulate/precuneus 1, −61, 38 38,664 Medial prefrontal 1, 55, −3 10,768 Right inferior parietal 47, −67, 29 10,608 Left inferior parietal −39, −77, 33 8,328 September 2019 \| Volume 13 \| Article 1038 In addition, a “scrubbing” procedure (Power et al., 2012) was applied, consisting of the introduction of dummy regressors to censor the eﬀect of frames with excessive movements and/or signal changes. Data Acquisition Magnetic resonance imaging studies were carried out at three Tesla (Discovery MR, General Electric Medical Systems, Erlangen, Germany), using a 32-channel head coil. Structural T1w volumes were acquired using a three- dimensional magnetization-prepared fast spoiled gradient echo sequence (144 sagittal partitions; TR 6.6 ms; TE 2.3 ms; TI 1100 ms; ﬂip angle 7◦; voxel size 1 × 1 × 1 mm3). Resting-state functional magnetic resonance imaging data were acquired using an EPI sequence (50 contiguous axial slices, TR = 3000 ms, TE = 33 ms, FOV = 240 mm, 64 × 64 matrix, slice thickness 3.6 mm, 120 time-points). In addition, turbo-spin-echo FLAIR axial images were acquired (144 sagittal partitions; TR 8000 ms; TE 119 ms; TI 2032 ms; ﬂip angle 90◦; voxel size 1 × 1 × 1 mm3) to help rule out the presence of chronic cerebrovascular disease or other CNS pathologies. During the MRI study, the subjects were lying in a supine position with the head lightly ﬁxed by straps and foam pads to September 2019 \| Volume 13 \| Article 1038 3 rs-fMRI in Earthquake Witnesses Pistoia et al. minimize head movement. They were asked to relax with their eyes closed but not to fall asleep during the examination. change Z-values (1.1 ± 0.5 vs. 9 ± 0.2) were signiﬁcantly diﬀerent between the two groups. Functional magnetic resonance imaging volumes were ﬁnally smoothed with an isotropic Gaussian ﬁlter of 8 mm (FWHM). Voxel-wise maps of FC were then generated by calculating the Fisher-transformed Pearson correlation coeﬃcients between the time course of each voxel and the time course averaged over each of eight seeds, sampling the major hubs of the VN (primary, ventral, right-dorsal, and left-dorsal visual cortices) and of the DMN (posterior cingulate/precuneus, medial prefrontal, right and left inferior parietal cortices). To this end, the seeds provided in CONN were used, which were obtained by independent component analysis of 497 normal subjects from the human connectome project dataset2, after masking by the GM map of each patient. The size and the MNI coordinates of the centers of mass of these eight seeds are reported in Table 1. 1http://www.nitrc.org/projects/conn Data Processing Data Processing Magnetic resonance imaging data were pre-processed and analyzed using a toolbox for FC data analysis (CONN – FC toolbox v18b, Gabrieli Lab., McGovern Institute for Brain Research, Massachusetts Institute of Technology1; Whitﬁeld-Gabrieli and Nieto-Castanon, 2012) running in Matlab (MathWorks Inc.). CONN is a toolbox for fMRI analysis based on libraries from the Statistical Parametric Mapping package (SPM12, the Wellcome Department of Neurology, London, United Kingdom). Brain tissue probability maps were used to derive the white matter and Cerebro Spinal Fluid (CSF) mean signal time-courses for fMRI pre-processing (see the section “Data Processing”) and to restrict the deﬁnition of seeds to each subject’s GM voxels. To this end, for each subject, GM, WM, and CSF probability maps, normalized to the Montreal Neurological Institute (MNI) space, were obtained using the uniﬁed segmentation (Ashburner and Friston, 2005), implemented in SPM12. For all the segmentation preprocessing steps, the default SPM12 parameters were used. rs-fMRI pre- processing steps included the following: exclusion of the ﬁrst ﬁve time-points to avoid the eﬀects of the possible instability of the initial MRI signal; correction for diﬀerences in acquisition time across slices; motion correction by rigid-body co-registration of all the time-points to the ﬁrst EPI volume (Friston et al., 1996); and band-pass ﬁltering (0.008–0.09 Hz, to remove low-frequency signal drifts related to scanner instability and high frequency noise). rs-fMRI data were then normalized to the standard MNI space by ﬁrst coregistering to the corresponding T1-weighted volumes (to avoid misregistration related to inter-sequence movements) and then applying the normalization parameters calculated for the T1 volumes in the segmentation step to the co-registered EPI volumes. For all the pre-processing steps, an experienced operator, blind to participants’ clinical conditions, visually assessed accuracy of the segmentation and spatial normalization. Frontiers in Neuroscience \| www.frontiersin.org September 2019 \| Volume 13 \| Article 1038 RESULTS Behavioral and rs-fMRI raw data are available upon request to the corresponding author. Self-Report Measures and Facial Expression Recognition Task Participants’ scores on the self-report measures strongly overlapped with Pistoia et al.’s (2018) data (Table 2) in revealing higher scores in earthquake witnesses than in controls on several of the selected measures, although the multivariate results did not show signiﬁcant eﬀects for group [Pillai’s Trace = 0.363; Wilks’ Lambda = 0.637; F(14,24) = 0.978; p = 0.501, η2 p = 0.363] and for sex [Pillai’s Trace = 0.353; Wilks’ Lambda = 0.647; F(14,24) = 0.935; p = 0.539, η2 p = 0.353]. The group by sex interaction was also not signiﬁcant (p > 0.05). TABLE 2 \| Scores (mean and SD) of the two groups on the self-reported measures and on the facial expression recognition task. Controls Earthquake witnesses Mean SD Mean SD Self-report measures ISI 4.52 3.1 5.06 3.15 IUS-12 32.52 10.4 44.22 11.7 URS-total score 121.61 12.9 131.06 14.9 URS-emotional uncertainty 27.43 6.5 34.17 8.8 URS-desire for control 45.52 6.2 44.67 8.7 URS-cognitive uncertainty 48.43 8.3 52.5 9.6 STAI-2 38.52 8.2 40.06 6.4 BDI 5.78 4.5 6.89 4.9 ASI-total score 9.35 6.2 16.56 15.9 ASI-physical concerns 2.35 2.3 4.94 6.9 ASI-cognitive concerns 4.13 3.7 5.33 5.8 ASI-social concerns 2.87 3.3 6.28 5.8 EPQ-R-extraversion/introversion 9.22 2.9 7.78 4.1 EPQ-R-neuroticism 4.74 2.8 6.17 3.6 EPQ-R-psychoticism 3.52 1.8 2.5 1.2 Recognition of facial expressions Disgust 0.76 0.1 0.91 0.1 Happiness 0.98 0.1 1 0.2 Fear 0.64 0.1 0.76 0.1 Anger 0.77 0.2 0.97 0.2 Surprise 0.91 0.2 0.99 0.2 Sadness 0.71 0.1 0.77 0.2 ISI, Insomnia Severity Index (Bastien et al., 2001; Castronovo et al., 2016); IUS- 12, Tolerance of Uncertainty Scale Short Form (Freeston et al., 1994); URS, Uncertainty Response Scale (Greco and Roger, 2001); STAI, State–Trait Anxiety Inventory (Spielberger et al., 1983; Pedrabissi and Santiniello, 1989); BDI, Beck Depression Inventory (Beck, 1967); ASI-3, Anxiety Sensitivity Index (Taylor et al., 2007; Petrocchi et al., 2015); EPQ-R, Eysenck Personality Questionnaire-Revised Short Form (Eysenck et al., 1985; Picconi et al., 2018). TABLE 2 \| Scores (mean and SD) of the two groups on the self-reported measures and on the facial expression recognition task. Statistical Analysis Accordingly, for each EPI sequence, volumes were identiﬁed that had, compared to the previous one, a mean signal diﬀerence exceeding three Z-values, and/or a mean framewise displacement exceeding 0.5 mm. TABLE 1 \| XYZ coordinates in the MNI space of the center of mass and size of the eight seeds used for the analysis. On average, 10.1 ± 9.7 (mean ± SD) time-points were removed due to the scrubbing procedure (10.3 ± 10.2 in the controls, 9.7 ± 9.2 in the exposed; p = not signiﬁcant). Analogously, neither mean framewise displacement (0.15 ± 0.11 mm vs. 14 ± 0.07 mm) nor mean global signal 4 rs-fMRI in Earthquake Witnesses Pistoia et al. performed (p = 0.01 for the comparison between the two groups and 0.006 for the imaging/clinical correlations). The performance on the facial expression recognition task largely conﬁrmed our previous data for the whole sample, demonstrating a higher accuracy by earthquake witnesses than non-exposed persons in recognizing all the six emotional categories (happiness, sadness, anger, fear, disgust, and surprise; percentages of correct responses are shown in Table 2). Indeed, the three-way mixed ANOVA on recognition accuracy, with emotion (disgust, happiness, fear, anger, surprise, and sadness) as a within-subject factor, and with group and sex as between- subject factors, showed a signiﬁcant main eﬀect of emotion [F(5,185) = 11.547, p = 0.0001, η2 p = 0.238], with recognition of fear (0.70) being less accurate than all other emotions (disgust = 0.83; happiness = 0.99; anger = 0.87; surprise = 0.95; and sadness = 0.74). Importantly, results also showed signiﬁcant main eﬀects of group [F(1,37) = 8.844, p = 0.005, η2 p = 0.193], with overall accuracy being higher in earthquake witnesses (mean = 0.90, SD = 0.26) than in controls (mean = 0.79, SD = 0.24), and of sex [F(1,37) = 4.369, p = 0.044, η2 p = 0.106], with females (mean = 0.88, SD = 0.26) being more accurate than males (mean = 0.81, SD = 0.25). No interaction was signiﬁcant (all p > 0.05). performed (p = 0.01 for the comparison between the two groups and 0.006 for the imaging/clinical correlations). Frontiers in Neuroscience \| www.frontiersin.org rs-fMRI Data The three orthogonal planes are centered on the peak values of the two clusters [MNI coordinates, respectively, (57, –33, 3) and (36, –66, –12)]. No signiﬁcant cluster emerged when probing the inverse (exposed > non-exposed) contrast. (B) Corresponding mean FC values are plotted against the scores obtained for the facial expression recognition task for exposed (red, ﬁlled marks) and non-exposed (blue, empty marks) subjects. The 95% conﬁdence intervals of the ﬁt are also shown in the corresponding colors. In these regions, the exposed subjects display a signiﬁcantly inverse correlation between the FC with the left dorsal visual seed and the score, whereas a direct correlation is present in the same regions in non-exposed subjects. FIGURE 1 \| (A) Regions showing a signiﬁcant interaction between the groups (non-exposed > exposed) and correlation of the functional connectivity to the dorsal visual seed of the VN with the behavioral scores for the facial expression recognition task. The two clusters are superimposed to the T1-weighted volume from one of the participants to the study normalized to the MNI space. Color-scale maximum is set to a T-value of 6. The three orthogonal planes are centered on the peak values of the two clusters [MNI coordinates, respectively, (57, –33, 3) and (36, –66, –12)]. No signiﬁcant cluster emerged when probing the inverse (exposed > non-exposed) contrast. (B) Corresponding mean FC values are plotted against the scores obtained for the facial expression recognition task for exposed (red, ﬁlled marks) and non-exposed (blue, empty marks) subjects. The 95% conﬁdence intervals of the ﬁt are also shown in the corresponding colors. In these regions, the exposed subjects display a signiﬁcantly inverse correlation between the FC with the left dorsal visual seed and the score, whereas a direct correlation is present in the same regions in non-exposed subjects. (Ganel et al., 2005; Said et al., 2011) when both implicit and explicit emotional processing are required (Gorno-Tempini et al., 2001; LeDoux, 2003; Litt et al., 2011; Brooks et al., 2012). Similarly, Mazza et al. (2012) investigated neural response to facial expressions implicitly presented during fMRI in a sample of L’Aquila earthquake witnesses aﬀected by PTSD. Results showed that subliminal presentation of emotional faces (happy and sad) was related to activation of the occipito-temporal cortex, amygdala, and insula. Interestingly, the neurofunctional model of facial processing developed by Haxby et al. rs-fMRI Data All participants were included in the analysis; when asked if they had fallen asleep even brieﬂy, they all conﬁrmed that they had remained awake the whole time. No signiﬁcant clusters of diﬀerent FC with any of the tested seeds emerged when comparing the two groups independently of behavioral performance. Diﬀerences were found in the between-group correlation of the score for the facial expression recognition task and the FC with the VN and DMN seeds (interaction analysis) (Table 3). For the left dorsal visual seed of the VN, diﬀerences emerged in the peripheral ventral occipital cortex on the right (Figure 1A, left) and in the middle temporal gyrus on the left (Figure 1A, right). For seeds of the DMN, diﬀerences were found in the precuneus for the medial prefrontal cortex (Figure 2A) and the left lower parietal (Figure 2B). These diﬀerences were due to a presence of an inverse correlation between FC and score in these regions in the exposed subjects, as opposed to the direct correlation detectable in the same regions in the non-exposed subjects (Figure 1B and right column of Figure 2). The results of the study show signiﬁcant between-group diﬀerences in the correlation of the score for the facial expression recognition task and the FC of the dorsal seed of the VN with the right occipito-temporal cortex and the left middle temporal cortex, and of the two parietal seeds of DMN, i.e., lower parietal and medial prefrontal cortex, with the precuneus bilaterally. These signiﬁcant between-group diﬀerences are consistent with growing data demonstrating chronic and stable changes in functional brain connectivity at rest in persons exposed to trauma (Lui et al., 2009; Long et al., 2014; Chen et al., 2015; Kleshchova et al., 2019). In particular, Kleshchova et al. (2019) suggested that September 2019 \| Volume 13 \| Article 1038 Frontiers in Neuroscience \| www.frontiersin.org 5 Pistoia et al. rs-fMRI in Earthquake Witnesses FIGURE 1 \| (A) Regions showing a signiﬁcant interaction between the groups (non-exposed > exposed) and correlation of the functional connectivity to the dorsal visual seed of the VN with the behavioral scores for the facial expression recognition task. The two clusters are superimposed to the T1-weighted volume from one of the participants to the study normalized to the MNI space. Color-scale maximum is set to a T-value of 6. Frontiers in Neuroscience \| www.frontiersin.org September 2019 \| Volume 13 \| Article 1038 rs-fMRI Data (2000) postulates that the core processing system not only involves the ventral occipito-temporal cortex but also the lateral (superior and middle) temporal cortex. Here, we actually found that the behavioral performance of earthquake witnesses was related to altered FC in the VN with the left middle temporal gyrus, consistent with Haxby et al.’s (2000) model, according to which this cortical region would be particularly involved in detecting the changeable aspects of faces, such as emotional expressions. resting neurofunctional changes in trauma-exposed persons are part of an exaggerated neural alerting response to threat that can be observed even in the absence of physical threat, likely due to a chronic trauma-related condition of hypervigilance. It is worth remembering here that our sample of persons exposed to earthquake was actually in a chronic condition since, as noted above, the main L’Aquila earthquake in 2009 was followed by continuous aftershocks in the later months and by additional earthquakes in 2016 and 2017. The diﬀerence we found between exposed and non-exposed participants in the relationship between facial expression recognition and the correlations of FC values in VN and DMN with occipito-temporal, lateral temporal, and medial parietal regions support the idea that the emotional expertise in earthquake witnesses involves neurofunctional changes in networks devoted to the processing of speciﬁc signals of potential threats, such as emotional faces. Indeed, although the occipito- temporal cortex represents a central node in the face identity recognition network (Haxby et al., 2000), a recent meta-analysis showed its involvement in the aﬀective representation of a face Facial expressions are actually among the most relevant signals conveying information on what is going on in other persons’ September 2019 \| Volume 13 \| Article 1038 6 rs-fMRI in Earthquake Witnesses Pistoia et al. FIGURE 2 \| Regions showing a signiﬁcant interaction between the group and the correlation of the functional connectivity to the medial prefrontal cortex (A) and the left lower parietal (B) seeds of the DMN with the scores for the facial expression recognition task. For both seeds, the FC with the precuneus (part of the DMN) showed an altered correlation with the scores in the exposed subjects. The clusters are superimposed to the T1-weighted volume from one of the participants to the study normalized to the MNI space. The color-scale maximum is set to a T-value of 6. rs-fMRI Data For each cluster, the three orthogonal planes are centered on the peak value MNI coordinates. No signiﬁcant cluster emerged when probing the inverse (exposed > non-exposed) contrast. On the right, the mean FC values of each cluster are plotted against the scores obtained for the facial expressions recognition task for exposed (red, ﬁlled marks) and non-exposed (blue, empty marks) subjects. The 95% conﬁdence intervals of the ﬁt are also shown in the corresponding colors. In the precuneus, the exposed subjects display a signiﬁcantly inverse correlation between the FC with both these DMN seeds and the score, whereas non-exposed subjects show a direct correlation. FIGURE 2 \| Regions showing a signiﬁcant interaction between the group and the correlation of the functional connectivity to the medial prefrontal cortex (A) and the left lower parietal (B) seeds of the DMN with the scores for the facial expression recognition task. For both seeds, the FC with the precuneus (part of the DMN) showed an altered correlation with the scores in the exposed subjects. The clusters are superimposed to the T1-weighted volume from one of the participants to the study normalized to the MNI space. The color-scale maximum is set to a T-value of 6. For each cluster, the three orthogonal planes are centered on the peak value MNI coordinates. No signiﬁcant cluster emerged when probing the inverse (exposed > non-exposed) contrast. On the right, the mean FC values of each cluster are plotted against the scores obtained for the facial expressions recognition task for exposed (red, ﬁlled marks) and non-exposed (blue, empty marks) subjects. The 95% conﬁdence intervals of the ﬁt are also shown in the corresponding colors. In the precuneus, the exposed subjects display a signiﬁcantly inverse correlation between the FC with both these DMN seeds and the score, whereas non-exposed subjects show a direct correlation. TABLE 3 \| Clusters of altered correlation of the connectivity with the scores for the facial expression recognition task (Interaction). For each cluster, corresponding p-values (corrected at cluster level for family-wise error) and size (in cubic centimeters of gray matter) are reported, along with the local maxima T-values and coordinates. TABLE 3 \| Clusters of altered correlation of the connectivity with the scores for the facial expression recognition task (Interaction). Frontiers in Neuroscience \| www.frontiersin.org ETHICS STATEMENT Future studies on earthquake witnesses are warranted to replicate the present results on a large sample; the size of the present group was small, although it was in line with the size of samples recruited in similar studies (e.g., Long et al., 2014; Kleshchova et al., 2019). Further, sensitivity of the study may have been limited by the relatively short scan duration which, however, was within the timeframe that has been shown to be required to stabilize the correlation strengths within and between the major networks (Van Dijk et al., 2010). While we preferred to keep acquisition short to reduce the risk of the subject falling asleep and/or moving, the increased S/N ratio achievable with longer acquisitions may be considered in future studies, to increase sensitivity. The research protocol was approved by the Internal Review Board of the University of L’Aquila (01/2017). The study was conducted in accordance with the ethical standards of the Helsinki Declaration and signed informed consent was obtained from all the participants. AUTHOR CONTRIBUTIONS All authors equally contributed to the planning, development, and drafting of the manuscript. rs-fMRI Data Therefore, we suggest that this default tendency to focus on the other person’s mental state could be enhanced in persons exposed to traumatic experiences, as in the case of earthquake witnesses, in order to search relevant social signals, allowing rapid identiﬁcation of possible threats in the environment (Zhang et al., 2014; Bell et al., 2017; Pistoia et al., 2018). This result ﬁts with ﬁndings from a seminal rs- fMRI study on witnesses of the Wenchun earthquake in China by Lui et al. (2009), who found a reduced temporal synchronization within the DMN in trauma victims, even immediately after trauma exposure. Moreover, a recent study investigating the correspondence between spontaneous neural activity in the DMN and the severity of PTSD symptoms showed that the at- rest activity of the left inferior parietal lobule was positively correlated with symptom severity, thus suggesting that its activity is involved in the cognitive biases observed in persons with PTSD (Disner et al 2018) Also, shorter TRs allowed by multiband acquisition (Feinberg and Setsompop, 2013), which was not available on our scanner, may in the future allow further boosting of the S/N ratio, overcoming the limitations derived from the relatively long sampling interval (3 s), which was used here to allow complete brain coverage while keeping a reasonable in-plane resolution with the available hardware. Notwithstanding these limitations, our ﬁndings suggest that emotional expertise in earthquake witnesses goes through a functional modiﬁcation of brain systems devoted to detection, identiﬁcation, and interpretation of emotional faces, including the occipito-temporal cortex and the medial parietal cortex. Since we observed a general increase in anxiety and anticipation of threats, as well as emotional uncertainty, such emotional expertise, although ﬁrst developing as a response of adaptive value, ends up being a maladaptive change to trauma, likely related to anxiety responses (see also Pistoia et al., 2018). It is possible to suggest that this emotional response is even more likely in young persons, in whom traumatic experiences tend to have a great impact on psychological functioning (Wang et al., 2015; Hong and Eﬀerth, 2016). The earthquake witnesses who participated in the present study were young teenagers in 2009. Thus, in future studies, it could be interesting to investigate “emotional expertise” after a natural disaster in persons who were exposed to the traumatic experience at a later stage of their life. DATA AVAILABILITY STATEMENT The datasets generated for this study are available on request to the corresponding author. rs-fMRI Data For each cluster, corresponding p-values (corrected at cluster level for family-wise error) and size (in cubic centimeters of gray matter) are reported, along with the local maxima T-values and coordinates. TABLE 3 \| Clusters of altered correlation of the connectivity with the scores for the facial expression recognition task (Interaction). For each cluster, corresponding p-values (corrected at cluster level for family-wise error) and size (in cubic centimeters of gray matter) are reported, along with the local maxima T-values and coordinates. Network Seed p-value (FWE) Size (cc) T X Y Z Anatomical labels Visual Left dorsal <10−5 1.4 6.33 −57 −33 3 Left middle temporal gyrus 4.29 −57 −24 0 Left middle temporal gyrus 3.82 −51 −18 −6 Left middle temporal gyrus <10−6 1.8 5.61 36 −66 −12 Right inferior occipital 5.16 24 −57 −6 Right lingual 4.39 30 −54 −12 Right fusiform DMN Left lower parietal <10−13 4.1 6.68 −3 −54 51 Left precuneus 6.14 12 −57 45 Right precuneus 5.72 −6 −48 45 Left precuneus Medial prefrontal cortex <10−11 3.7 5.54 12 −51 45 Right precuneus 5.32 9 −57 39 Right precuneus Anatomical labeling is according to Tzourio-Mazoyer et al. (2002). natomical labeling is according to Tzourio-Mazoyer et al. (2002). tasks, such as judgments on facial expressions, stories, or moving shapes (Frith and Frith, 2006, 2007). Results always show the activation of a set of regions including medial prefrontal and minds (Adolphs, 1999, 2003; LeDoux, 2003; Kanwisher and Yovel, 2006). Many imaging studies have examined the neural basis of understanding others’ minds by diﬀerent experimental September 2019 \| Volume 13 \| Article 1038 Frontiers in Neuroscience \| www.frontiersin.org Frontiers in Neuroscience \| www.frontiersin.org 7 Pistoia et al. rs-fMRI in Earthquake Witnesses parietal cortex, and posterior temporo-parietal areas around the temporo-parietal junction (Frith and Frith, 2006, 2007), a network of areas largely overlapping with the DMN (e.g., Mars et al., 2012). In particular, the left parietal and posterior midline nodes of the DMN are involved in processing emotional facial expressions, both in healthy individuals (Sreenivas et al., 2012) and in patients with diﬀerent psychopathological conditions including social phobia (Gentili et al., 2009) and schizophrenia (Salgado-Pineda et al., 2011). Schilbach et al. (2008) explored the relationship between the neural basis for social cognition and the DMN, and found that the core nodes of the DMN overlap with those involved in social cognition (Vogeley and Fink, 2003; Schilbach et al., 2006). rs-fMRI Data The authors suggested that the resting default state of the human brain is related to the predisposition of humans for social cognition as a default mental state. Consistently, here we demonstrated the involvement of key nodes of the DMN as the lateral parietal cortex, and the medial prefrontal and the parietal cortex. Therefore, we suggest that this default tendency to focus on the other person’s mental state could be enhanced in persons exposed to traumatic experiences, as in the case of earthquake witnesses, in order to search relevant social signals, allowing rapid identiﬁcation of possible threats in the environment (Zhang et al., 2014; Bell et al., 2017; Pistoia et al., 2018). This result ﬁts with ﬁndings from a seminal rs- fMRI study on witnesses of the Wenchun earthquake in China by Lui et al. (2009), who found a reduced temporal synchronization within the DMN in trauma victims, even immediately after trauma exposure. Moreover, a recent study investigating the correspondence between spontaneous neural activity in the DMN and the severity of PTSD symptoms showed that the at- rest activity of the left inferior parietal lobule was positively correlated with symptom severity, thus suggesting that its activity is involved in the cognitive biases observed in persons with PTSD (Disner et al., 2018). parietal cortex, and posterior temporo-parietal areas around the temporo-parietal junction (Frith and Frith, 2006, 2007), a network of areas largely overlapping with the DMN (e.g., Mars et al., 2012). In particular, the left parietal and posterior midline nodes of the DMN are involved in processing emotional facial expressions, both in healthy individuals (Sreenivas et al., 2012) and in patients with diﬀerent psychopathological conditions including social phobia (Gentili et al., 2009) and schizophrenia (Salgado-Pineda et al., 2011). Schilbach et al. (2008) explored the relationship between the neural basis for social cognition and the DMN, and found that the core nodes of the DMN overlap with those involved in social cognition (Vogeley and Fink, 2003; Schilbach et al., 2006). The authors suggested that the resting default state of the human brain is related to the predisposition of humans for social cognition as a default mental state. Consistently, here we demonstrated the involvement of key nodes of the DMN as the lateral parietal cortex, and the medial prefrontal and the parietal cortex. Bastien, C. H., Vallières, A., and Morin, C. M. (2001). Validation of the Insomnia Severity Index as an outcome measure for insomnia research. Sleep Med. 2, 297–307. doi: 10.1016/s1389-9457(00)00065-4 Bastien, C. H., Vallières, A., and Morin, C. M. (2001). Validation of the Insomnia Severity Index as an outcome measure for insomnia research. Sleep Med. 2, 297–307. doi: 10.1016/s1389-9457(00)00065-4 Beck, A. T. (1967). Depression: Clinical, Experimental, and Theoretical Aspects. New York, NY: Harper & Row. Bell, C. J., Colhoun, H. C., Frampton, C. M., Douglas, K. M., McIntosh, V. V. W., Carter, F. A., et al. (2017). Earthquake brain: altered recognition and misclassiﬁcation of facial expressions are related to trauma exposure but not REFERENCES doi: 10.3389/fnint.2012.00089 Castronovo, V., Galbiati, A., Marelli, S., Brombin, C., Cugnata, F., Giarolli, L., et al. (2016). Validation study of the Italian version of the Insomnia Severity Index (ISI). Neurol Sci. 37, 1517–1524. doi: 10.1007/s10072-016-2620-z Hong, C., and Eﬀerth, T. (2016). Systematic review on post-traumatic stress disorder among survivors of the wenchuan earthquake. Trauma Violence Abuse 17, 542–561. doi: 10.1177/1524838015585313 Chen, B., Xu, T., Zhou, C., Wang, L., Yang, N., Wang, Z., et al. (2015). Individual variability and test-retest reliability revealed by ten repeated resting-state brain scans over one month. PLoS One 10:e0144963. doi: 10.1371/journal.pone. 0144963 Kanwisher, N., and Yovel, G. (2006). The fusiform face area: a cortical region specialised for the perception of faces. Philos. Trans. R. Soc. Lond. B Biol. Sci. 361, 2109–2128. doi: 10.1098/rstb.2006.1934 Disner, S. G., Marquardt, C. A., Mueller, B. A., Burton, P. C., and Sponheim, S. R. (2018). Spontaneous neural activity diﬀerences in posttraumatic stress disorder: a quantitative resting-state meta-analysis and fMRI validation. Hum. Brain Mapp. 39, 837–850. doi: 10.1002/hbm.23886 Kleshchova, O., Rieder, J. K., Grinband, J., and Weierich, M. R. (2019). Resting amygdala connectivity and basal sympathetic tone as markers of chronic hypervigilance. Psychoneuroendocrinology 102, 68–78. doi: 10.1016/j.psyneuen. 2018.11.036 Dube, A., Moﬀatt, M., Davison, C., and Bartels, S. (2018). Health outcomes for children in haiti since the 2010 earthquake: a systematic review. Prehosp. Disaster Med. 33, 77–88. doi: 10.1017/S1049023X17007105 LeDoux, J. (2003). The emotional brain, fear, and the amygdala. Cell Mol. Neurobiol. 23, 727–738. Litt, A., Plassmann, H., Shiv, B., and Rangel, A. (2011). Dissociating valuation and saliency signals during decision-making. Cereb. Cortex 21, 95–102. doi: 10.1093/cercor/bhq065 Eisma, M. C., Lenferink, L. I. M., Chow, A. Y. M., Chan, C. L. W., and Li, J. (2019). Complicated grief and post-traumatic stress symptom proﬁles in bereaved earthquake survivors: a latent class analysis. Eur. J. Psychotraumatol. 10:1558707. doi: 10.1080/20008198.2018.1558707 Long, J., Huang, X., Liao, Y., Hu, X., Hu, J., Lui, S., et al. (2014). Prediction of post- earthquake depressive and anxiety symptoms: a longitudinal resting-state fMRI study. Sci. Rep. 4:6423. doi: 10.1038/srep06423 Ekman, P. (1993). Facial expression and emotion. Am. Psychol. 48, 384–392. Ekman, P., and Friesen, W. V. (1976). Pictures of Facial Aﬀect. Palo Alto, CA: Consulting Psychologists. Lui, S., Huang, X., Chen, L., Tang, H., Zhang, T., Li, X., et al. (2009). High-ﬁeld MRI reveals an acute impact on brain function in survivors of the magnitude 8.0 earthquake in China. Proc. Natl. Acad. REFERENCES Sci. U.S.A. 106, 15412–15417. doi: 10.1073/pnas.0812751106 Eysenck, S. B. G., Eysenck, H. J., and Barrett, P. (1985). A revised version of the psychoticism scale. Pers. Individ. Dif. 6, 21–29. doi: 10.1016/0191-8869(85) 90026-1 Fani, N., Jovanovic, T., Ely, T. D., Bradley, B., Gutman, D., Tone, E. B., et al. (2012). Neural correlates of attention bias to threat in post-traumatic stress disorder. Biol. Psychol. 90, 134–142. doi: 10.1016/j.biopsycho.2012.03.001 Mars, R. B., Neubert, F. X., Noonan, M. P., Sallet, J., Toni, I., and Rushworth, M. F. (2012). On the relationship between the “default mode network” and the “social brain”. Front. Hum. Neurosci. 6:189. doi: 10.3389/fnhum.2012.00189 Mather, M., and Carstensen, L. L. (2005). Aging and motivated cognition: the positivity eﬀect in attention and memory. Trends Cogn. Sci. 9, 496–502. doi: 10.1016/j.tics.2005.08.005 Farooqui, M., Quadri, S. A., Suriya, S. S., Khan, M. A., Ovais, M., Sohail, Z., et al. (2017). Posttraumatic stress disorder: a serious post-earthquake complication. Trends Psychiatry Psychother. 39, 135–143. doi: 10.1590/2237-6089-2016-0029 Mattavelli, G., Cattaneo, Z., and Papagno, C. (2011). Transcranial magnetic stimulation of medial prefrontal cortex modulates face expressions processing in a priming task. Neuropsychologia 49, 992–998. doi: 10.1016/j.neuropsychologia.2011.01.038 Feinberg, D. A., and Setsompop, S. K. (2013). Ultra-fast MRI of the human brain with simultaneous multi-slice imaging. J. Magn. Reson. 229, 90–100. doi: 10. 1016/j.jmr.2013.02.002 Freeston, M., Rhéaume, J., Letarte, H., Dugas, M. J., and Ladouceur, R. (1994). Why do people worry? Pers. Individ. Dif. 17, 791–802. doi: 10.1016/0191-8869(94) 90048-5 j y g Mattavelli, G., Rosanova, M., Casali, A. G., Papagno, C., and Romero Lauro, L. J. (2016). Timing of emotion representation in right and left occipital region: evidence from combined TMS-EEG. Brain Cogn. 106, 13–22. doi: 10.1016/j. bandc.2016.04.009 Friston, K. J., Williams, S., Howard, R., Frackowiak, R. S. J., and Turner, R. (1996). Movement-related eﬀects in fMRI timeseries. Magn. Reson. Med 35, 346–355. doi: 10.1002/mrm.1910350312 Mazza, M., Catalucci, A., Mariano, M., Pino, M. C., Tripaldi, S., Roncone, R., et al. (2012). Neural correlates of automatic perceptual sensitivity to facial aﬀect in posttraumatic stress disorder subjects who survived L’Aquila eartquake of April 6, 2009. Brain Imaging Behav. 6, 374–386. doi: 10.1007/s11682-012- 9151-x Frith, C. D., and Frith, U. (2006). The neural basis of mentalizing. Neuron 50, 531–534. doi: 10.1016/j.neuron.2006.05.001 Frith, C. D., and Frith, U. (2007). Social cognition in humans. Curr. Biol. 17, R724–R732. O’Leary, V. E., and Ickovics, I. R. (1995). REFERENCES Bastien, C. H., Vallières, A., and Morin, C. M. (2001). Validation of the Insomnia Severity Index as an outcome measure for insomnia research. Sleep Med. 2, 297–307. doi: 10.1016/s1389-9457(00)00065-4 Adolphs, R. (1999). Social cognition and the human brain. Trends Cogn. Sci. 3, 469–479. Beck, A. T. (1967). Depression: Clinical, Experimental, and Theoretical Aspects. New York, NY: Harper & Row. Adolphs, R. (2003). Cognitive neuroscience of human social behaviour. Nat. Rev. Neurosci. 4, 165–178. doi: 10.1038/nrn1056 Bell, C. J., Colhoun, H. C., Frampton, C. M., Douglas, K. M., McIntosh, V. V. W., Carter, F. A., et al. (2017). Earthquake brain: altered recognition and misclassiﬁcation of facial expressions are related to trauma exposure but not Ashburner, J., and Friston, K. J. (2005). Uniﬁed segmentation. Neuroimage 26, 839–851. doi: 10.1016/j.neuroimage.2005.02.018 September 2019 \| Volume 13 \| Article 1038 8 Frontiers in Neuroscience \| www.frontiersin.org Pistoia et al. rs-fMRI in Earthquake Witnesses posttraumatic stress disorder. Front. Psychiatry 8:278. doi: 10.3389/fpsyt.2017. 00278 posttraumatic stress disorder. Front. Psychiatry 8:278. doi: 10.3389/fpsyt.2017. 00278 Greco, V., and Roger, D. (2001). Coping with uncertainty: the construction and validation of a new measure. Pers. Individ. Dif. 31, 519–534. doi: 10.1016/s0191- 8869(00)00156-2 Brooks, S. J., Savov, V., Allzén, E., Benedict, C., Fredriksson, R., and Schiöth, H. B. (2012). Exposure to subliminal arousing stimuli induces robust activation in the amygdala, hippocampus, anterior cingulate, insular cortex and primary visual cortex: a systematic meta-analysis of fMRI studies. Neuroimage 59, 2962–2973. doi: 10.1016/j.neuroimage.2011.09.077 Harmer, C. J., Thilo, K. V., Rothwell, J. C., and Goodwin, G. M. (2001). Transcranial magnetic stimulation of medial prefrontal cortex impairs processing of angry Harmer, C. J., Thilo, K. V., Rothwell, J. C., and Goodwin, G. M. (2001). Transcranial magnetic stimulation of medial prefrontal cortex impairs processing of angry facial expression. Nat. Neurosci. 4, 17–18. doi: 10.1038/82854 facial expression. Nat. Neurosci. 4, 17–18. doi: 10.1038/82854 Haxby, J. V., Hoﬀman, E. A., and Gobbini, M. I. (2000). The distributed human neural system for face perception. Trends Cogn. Sci. 4, 223–233. doi: 10.1016/ s1364-6613(00)01482-0 j g Calhoun, L. G., and Tedeschi, R. G. (2006). “The foundations of posttraumatic growth: An expanded framework,” in Handbook of Posttraumatic Growth, eds L. G. Calhoun, and R. G. Tedeschi, (Mahwah, NJ: Erlbaum), 1–23. Hayes, J. P., Vanelzakker, M. B., and Shin, L. M. (2012). Emotion and cognition interactions in PTSD: a review of neurocognitive and neuroimaging studies. Front. Integr. Neurosci. 6:89. REFERENCES Resilience and thriving in response to challenge: an opportunity for a paradigm shift in women’s health. Womens Health 1, 121–142. Ganel, T., Valyear, K. F., Goshen-Gottstein, Y., and Goodale, M. A. (2005). The involvement of the “fusiform face area” in processing facial expression. Neuropsychologia 43, 1645–1654. doi: 10.1016/j.neuropsychologia.2005. 01.012 Pedrabissi, L., and Santiniello, M. (1989). Manuale Dell’Adattamento Italiano Dello STAI Forma Y. Firenze: Organizzazioni Speciali. Geng, F., Liang, Y., Shi, X., and Fan, F. (2018). A prospective study of psychiatric symptoms among adolescents after the wenchuan earthquake. J. Trauma Stress 31, 499–508. doi: 10.1002/jts.22307 Petrocchi, N., Tenore, K., Couyoumdjian, A., and Gragnani, A. (2015). The anxiety sensitivity index-3: factor structure and psychometric properties in Italian clinical and non-clinical samples. Boll. Psicol. Appl. 269, 53–64. Gentili, C., Ricciardi, E., Gobbini, M. I., Santarelli, M. F., Haxby, J. V., Pietrini, P., et al. (2009). Beyond amygdala: default mode network activity diﬀers between patients with social phobia and healthy controls. Brain Res. Bull. 79, 409–413. doi: 10.1016/j.brainresbull.2009.02.002 Phillips, M. L., Bullmore, E. T., Howard, R., Woodruﬀ, P. W. R., Wright, I. W., Williams, S. C., et al. (1998). Investigation of facial recognition memory and happy and sad facial expression perception: an fMRI study. Psychiatry Res. 83, 127–138. doi: 10.1016/s0925-4927(98)00036-5 Gorno-Tempini, M. L., Pradelli, S., Seraﬁni, M., Pagnoni, G., Baraldi, P., Porro, C., et al. (2001). Explicit and incidental facial expression processing: an fMRI study. Neuroimage 14, 465–473. doi: 10.1006/nimg.2001.0811 Picconi, L., Jackson, C. J., Balsamo, M., Tommasi, M., and Saggino, A. (2018). Factor structure and measurement invariance across groups of the Italian September 2019 \| Volume 13 \| Article 1038 Frontiers in Neuroscience \| www.frontiersin.org 9 Pistoia et al. rs-fMRI in Earthquake Witnesses validation of the anxiety sensitivity index-3. Psychol. Assess. 19, 176–188. doi: 10.1037/1040-3590.19.2.176 Eysenck Personality Questionnaire - Short form (EPP-S). Pers. Individ. Dif. 123, 76–80. doi: 10.1016/j.paid.2017.11.013 validation of the anxiety sensitivity index-3. Psychol. Assess. 19, 176–188. doi: 10.1037/1040-3590.19.2.176 Pistoia, F., Conson, M., Carolei, A., Dema, M. G., Splendiani, A., Curcio, G., et al. (2018). Post-earthquake distress and development of emotional expertise in young adults. Front. Behav. Neurosci. 12:91. doi: 10.3389/fnbeh.2018.00091 Todd, R. M., MacDonald, M. J., Sedge, P., Robertson, A., Jetly, R., Taylor, M. J., et al. (2015). Soldiers with Posttraumatic Stress Disorder see a world full of threat: magnetoencephalography reveals enhanced tuning to combat-related cues. Biol. Psychiatry 78, 821–829. doi: 10.1016/j.biopsych.2015.05.011 Psychiatry 78, 821–829. REFERENCES doi: 10.1016/j.biopsych.2015.05.011 Pistoia, F., Conson, M., Trojano, L., Grossi, D., Ponari, M., Colonnese, C., et al. (2010). Impaired conscious recognition of negative facial expressions in patients with locked-in syndrome. J. Neurosci. 30, 7838–7844. doi: 10.1523/ JNEUROSCI.6300-09.2010 Tzourio-Mazoyer, N., Landeau, B., Papathanassiou, D., Crivello, F., Etard, O., Delcroix, N., et al. (2002). Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain. Neuroimage 15, 273–289. doi: 10.1006/nimg.2001.0978 Power, J. D., Barnes, K. A., Snyder, A. Z., Schlaggar, B. L., and Petersen, S. E. (2012). Spurious but systematic correlations in functional connectivity MRI networks arise from subject motion. Neuroimage 59, 2142–2154. doi: 10.1016/j. neuroimage.2011.10.018 Van Dijk, K. R., Hedden, T., Venkataraman, A., Evans, K. C., Lazar, S. W., and Buckner, R. L. (2010). Intrinsic functional connectivity as a tool for human connectomics: theory, properties, and optimization. J. Neurophysiol. 103, 297– 321. doi: 10.1152/jn.00783.2009 Raﬁey, H., Alipour, F., LeBeau, R., and Salimi, Y. (2019). Prevalence and determinants of PTSD 3 years after an earthquake in Iran. Commun. Ment. Health J. 55, 542–547. doi: 10.1007/s10597-019-00384-x Vogeley, K., and Fink, G. R. (2003). Neural correlates of the ﬁrst-person- perspective. Trends Cogn. Sci. 7, 38–42. doi: 10.1016/s1364-6613(02)00 003-7 Said, C. P., Haxby, J. V., and Todorov, A. (2011). Brain systems for assessing the aﬀective value of faces. Philos. Trans. R. Soc. Lond. B Biol. Sci. 366, 1660–1670. doi: 10.1098/rstb.2010.0351 Wang, L., Zhang, L., Armour, C., Cao, C., Qing, Y., Zhang, J., et al. (2015). Assessing the underlying dimensionality of DSM-5 PTSD symptoms in Chinese adolescents surviving the 2008 Wenchuan earthquake. J. Anxiety Disord. 31, 90–97. doi: 10.1016/j.janxdis.2015.02.006 Salgado-Pineda, P., Fakra, E., Delaveau, P., McKenna, P. J., Pomarol-Clotet, E., and Blin, O. (2011). Correlated structural and functional brain abnormalities in the default mode network in schizophrenia patients. Schizophrenia Res. 125, 101–109. doi: 10.1016/j.schres.2010.10.027 Whitﬁeld-Gabrieli, S., and Nieto-Castanon, A. (2012). Conn: a functional connectivity toolbox for correlated and anticorrelated brain networks. Brain Connect. 2, 125–141. doi: 10.1089/brain.2012.0073 Schilbach, L., Eickhoﬀ, S. B., Rotarska-Jagiela, A., Fink, G. R., and Vogeley, K. (2008). Minds at rest? Social cognition as the default mode of cognizing and its putative relationship to the default system of the brain. Consciousness Cogn. 17, 457–467. doi: 10.1016/j.concog.2008.03.013 Zhang, Y., Kong, F., Han, L., Najam, U., Hasan, A., and Chen, H. (2014). Attention bias in earthquake-exposed survivors: an event-related potential study. Int. J. Psychophysiol. 94, 358–364. doi: 10.1016/j.ijpsycho.2014. Frontiers in Neuroscience \| www.frontiersin.org September 2019 \| Volume 13 \| Article 1038 REFERENCES 09.004 Schilbach, L., Wohlschlaeger, A. M., Kraemer, N. C., Newen, A., Shah, N. J., Fink, G. R., et al. (2006). Being with virtual others: neural correlates of social interaction. Neuropsychologia 44, 718–730. doi: 10.1016/j.neuropsychologia. 2005.07.017 Zhao, K., Zhao, J., Zhang, M., Cui, Q., and Fu, X. (2017). Neural responses to rapid facial expressions of fear and surprise. Front. Psychol. 8:761. doi: 10.3389/fpsyg. 2017.00761 Spielberger, C. D., Gorsuch, R. L., Lushene, R., Vagg, P. R., and Jacobs, G. A. (1983). Manual for the State-Trait Anxiety Inventory. Palo Alto, CA: Consulting Psychologists Press. Conﬂict of Interest: The authors declare that the research was conducted in the absence of any commercial or ﬁnancial relationships that could be construed as a potential conﬂict of interest. Conﬂict of Interest: The authors declare that the research was conducted in the absence of any commercial or ﬁnancial relationships that could be construed as a potential conﬂict of interest. Sreenivas, S., Boehm, S. G., and Linden, D. E. (2012). Emotional faces and the default mode network. Neurosci. Lett. 506, 229–234. doi: 10.1016/j.neulet.2011. 11.012 Copyright © 2019 Pistoia, Conson, Quarantelli, Panebianco, Carolei, Curcio, Sacco, Saporito, Di Cesare, Barile, Masciocchi and Splendiani. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. Tang, W., Lu, Y., Yang, Y., and Xu, J. J. (2018). An epidemiologic study of self- reported sleep problems in a large sample of adolescent earthquake survivors: the eﬀects of age, gender, exposure, and psychopathology. Psychosom. Res. 113, 22–29. doi: 10.1016/j.jpsychores.2018.07.006 Taylor, S., Zvolensky, M. J., Cox, B. J., Deacon, B., Heimberg, R. G., Ledley, D. R., et al. (2007). Robust dimensions of anxiety sensitivity: development and initial September 2019 \| Volume 13 \| Article 1038 Frontiers in Neuroscience \| www.frontiersin.org Frontiers in Neuroscience \| www.frontiersin.org 10
https://openalex.org/W3089808737	https://www.researchsquare.com/article/rs-17336/latest.pdf	English	null	Mercury in the terrestrial environment: a review	Environmental sciences Europe	2,020	cc-by	20,105	Mercury in the terrestrial environment: a review Barbara Gworek Instytut Ochrony Srodowiska Wojciech Dmuchowski Instytut Ochrony Srodowiska Aneta Baczewska-Dąbrowska (  a.baczewska-dabrowska@obpan.pl ) Polska Akademia Nauk Ogrod Botaniczny Centrum Zachowania Roznorodnosci Biologicznej https://orcid.org/0000-0002-6606-6112 Instytut Ochrony Srodowiska Aneta Baczewska-Dąbrowska (  a.baczewska-dabrowska@obpan.pl ) Polska Akademia Nauk Ogrod Botaniczny Centrum Zachowania Roznorodnosci Biologicznej https://orcid.org/0000-0002-6606-6112 Instytut Ochrony Srodowiska Aneta Baczewska-Dąbrowska (  a.baczewska-dabrowska@obpan.pl ) Polska Akademia Nauk Ogrod Botaniczny Centrum Zachowania Roznorodnosci Biologicznej https://orcid.or k Ogrod Botaniczny Centrum Zachowania Roznorodnosci Biologicznej https://orcid.org/0000-0002-6606-6112 Mercury in the terrestrial environment: a review Barbara Gworek Instytut Ochrony Srodowiska Wojciech Dmuchowski Instytut Ochrony Srodowiska Aneta Baczewska-Dąbrowska (  a.baczewska-dabrowska@obpan.pl ) Polska Akademia Nauk Ogrod Botaniczny Centrum Zachowania Roznorodnosci Biologicznej https://orcid.org/0000-0002-6606-6112 Review Keywords: mercury, deposition and emissions from land, soil pollution, content in plants Posted Date: September 9th, 2020 DOI: https://doi.org/10.21203/rs.3.rs-17336/v3 License:   This work is licensed under a Creative Commons Attribution 4.0 International License. Read Full License Version of Record: A version of this preprint was published on October 2nd, 2020. See the published version at https://doi.org/10.1186/s12302 Background Mercury is recognized as a toxic, persistent, and mobile contaminant; it does not degrade in the environment and becomes mobile because of the volatility of the element and several of its compounds. Moreover, mercury has the ability to be transported within air masses over very long distances [1]. Over the last few decades, considerable scientific knowledge has been developed on the sources and emissions of mercury, its pathways and cycling through the environment, human exposure, and impacts to the environment and human health [2]. Hg is the only element on the periodic table to have its own environmental convention, i.e., the Minamata Convention on Mercury, thus highlighting the importance of the Hg pollution issue [3]. An improved understanding of the global mercury (Hg) cycle is important for our capacity to predict how regulatory efforts to reduce current emissions to air, water and land will affect Hg concentrations in environmental compartments, biota and humans. Hg is released into the environment through human activities and via natural sources and processes, such as volcanoes and rock weathering. Following its release, Hg is transported and recycled between the major environmental compartments, i.e., air, soil and water, until it is eventually removed from the system through burial in coastal and deep ocean sediments, lake sediments, and subsurface soils [2,4]. Hg is considered to be a peculiar chemical element because it displays particularly strong chemical and biological activity as well as variability in form (liquid and gaseous). Hg compounds with very different chemical and physical properties are included in various cycles of its natural circulation [5-6]. Hg is a globally distributed pollutant due to characteristics such as low melting and boiling points, conversions between chemical forms and participation in biological cycles. As a result of anthropogenic emissions, the global atmospheric Hg deposition rate is approximately three times higher than that in preindustrial times and has increased by a factor of 2-10 in and around the most industrialized regions [7]. Hg-contaminated land environments pose a risk to global public health, with Hg being listed as one of the ‘ten leading chemicals of concern’ [8]. In 2013, the United Nations (UN) introduced the ‘Minamata Convention on Mercury’ [9], which aims for a more global effort for managing the risk presented by Hg to human health and the environment. The Minamata Convention has as of today 123 parties and the convention entered into force 16 August 2017. Review Keywords: mercury, deposition and emissions from land, soil pollution, content in plants se:   This work is licensed under a Creative Commons Attribution 4.0 International License. Read Full License Version of Record: A version of this preprint was published on October 2nd, 2020. See the published version at https://doi.org/10.1186/s12302-020-00401-x. ersion of Record: A version of this preprint was published on October 2nd, 2020. See the published version at https Page 1/24 Background This concerted action, if successful, will have great implications for public health for decades to come; however, there are many hurdles on the way to achieving this goal [10-11]. Our understanding of the critical processes driving global Hg cycling, particularly those that affect the large-scale exchange of Hg among major environmental compartments, has advanced substantially over the past decade. This progress has been driven by major advances in three interconnected areas: new data, new models, and new analytical tools and techniques [12]. The main task for improving our knowledge of mercury sources, fates, impacts, and emission control options was defined at the 14th ICMGP Conference in 2019 [13] as “Bridging knowledge on global mercury with environmental responsibility, human welfare and policy response”. The main four questions that require urgent answers within this subject are defined as follows: 1. How is mercury biogeochemical cycling changing on global, regional, and local scales in response to perturbations caused by major anthropogenic drivers of environmental change? 1. How is mercury biogeochemical cycling changing on global, regional, and local scales in response to perturbations caused by major anthropogenic drivers of environmental change? 2. What is the relative risk of mercury exposure to human health and wildlife in the context of human welfare? 2. What is the relative risk of mercury exposure to human health and wildlife in the context of human welfare? 3. How can technological development contribute to the reduction in mercury exposure and improvement of e g p y p p p y 4. How can scientific knowledge contribute to the implementation and effectiveness evaluation of the Minamata Convention and other regulatory agreements, and what is the importance of integrating and implementing emerging and future mercury research into the policy making? 4. How can scientific knowledge contribute to the implementation and effectiveness evaluation of the Minam agreements, and what is the importance of integrating and implementing emerging and future mercury researc Abstract Environmental contamination by mercury is and will continue to be a serious risk for human health. Pollution of the terrestrial environment is particularly important as it is a place of human life and food production. Publication presents a review of the literature on issues related to Hg pollution of the terrestrial environment: soil and plants and their transformations. Different forms of atmospheric Hg may be deposited on surfaces by way of wet and dry processes. These forms may be sequestered within terrestrial compartments or emitted back into the atmosphere, and the relative importance of these processes is dependent on the form of Hg, the surface chemistry, and the environmental conditions. On the land surface, Hg deposition mainly occurs in the oxidized form (Hg 2+ ), and its transformations are associated primarily with the oxidation-reduction potential of the environment and the biological and chemical processes of methylation. The deposition of Hg on ground with low vegetation as 3–5 times lower than that in forests. The estimation of Hg emissions from soil and plants, which occur mainly in the Hg 0 form, is very difficult. Generally, the largest amounts of Hg are emitted from tropical regions, and the lowest levels are from the polar regions. Methods While developing this paper, the method recommended by Liberati et al. [14] was used to some extent. However, the strict use of this method was not possible due to the specifics of our publication. The problems of pollution and Hg transformation in the environment caused by significant threats to human health represent the subject of research by numerous research teams around the world. However, many publications do not meet the criterion of quality of research results. Therefore, the following criteria were used to eliminate publications: 1. Accurate descriptions of the research methods were lacking; Page 2/24 Page 2/24 2. Methods that did not guarantee the quality of the results were used; 3. The latest publications were selected, but historical works were not omitted; and 4. Publications from various countries and regions were cited. Re-emissions from historical deposits from natural and anthropogenic sources. Environmental archives offer an opportunity to reconstruct temporal trends in atmospheric Hg deposition at various timescales. Lake sediment, peat, ice cores, tree rings, and Hg stable isotope measurements are offering new insights into historical Hg cycling. Preindustrial Hg deposition has been studied over decadal to millennial timescales extending as far back as the Late Pleistocene. Exploitation of mercury deposits (mainly cinnabar) first began during the Mid- to Late Holocene in South America, Europe, and Asia but increased dramatically during the Colonial era (1532-1900) for silver production [19-20]. Artisanal gold mining is now thought responsible for over half the global stream flux of Hg, followed by the burning of coal [2,21]. Fluxes of Hg to the air occur via the volatilization of Hg0 as well as through wind entrainment of Hg bound to dust particles (often referred to as fugitive dust emissions). The relative magnitude of these two types of Hg emission sources varies depending on site-specific conditions. However, due to the long atmospheric lifetime of Hg0, these emissions mostly contribute to the global pool of Hg, whereas the fugitive dust emissions impacts are more local [22]. The update global mercury budget shows the impact of human activities on the mercury cycle and the resulting increase in mercury accumulated in soil and oceans. The updated global mercury budget in 2018 [2] is shown in the Table 1. The annual deposits were 480 tons higher than the emissions, indicating considerable enrichment in the environment. In general [23-24], Hg0 emissions from undisturbed nongeologically thermally enriched areas are <1 ng m−2 h−1, whereas fluxes from contaminated sites can be several orders of magnitude higher at >5000 ng m−2 h−1. For contaminated sites that cover large spatial areas (such as mining operations), the annual emissions from the entire surface area have been shown to range from 19 to 105 kg yr−1 from active industrial gold mines and 51 kg yr−1 from a large abandoned Hg mining area. Table 1. The update global mercury budget [2] (Mg/yr). Emission: i) Natural (geogenic) 500; ii) Re emission/re- mobilization (natural and legasy): - biomass burning 600 - from soil and vegetation 1000 iii) Anthropogenic 2500 (2000-3000) iv) Ocean, net evasion (gaseous elemental Hg) 3400 (2900-4000) total 7000 Deposition: i) to land / freshwater 3600 ii) to ocean 3880 total 7480 Hg emission The advances achieved over the last decade for the assessment of Hg emissions from major man-made and natural sources have contributed to improvements in the assessments of the impacts of atmospheric deposition of Hg on the terrestrial environment [15]. The assessment of Hg emissions poses serious methodological problems. In estimating these impacts, state institutions mainly focus on inventories of their sources, while international organizations apply different models and use emission factors and statistical data of industrial production and consumption of Hg-containing materials. It is particularly difficult to distinguish natural and anthropogenic emissions from re-emissions from the land and oceans [16-17]. Hg emission sources include both natural processes unfolding in the biosphere and anthropogenic sources. In 2008, the following classification was adopted in a UNEP report [18], which distinguished three emission sources: – Current emissions from natural sources; – Current emissions from natural sources; – Current emissions from anthropogenic sources; and Re-emissions from historical deposits from natural and anthropogenic sources. Re-emissions from historical deposits from natural and anthropogenic sources. Table 1. The update global mercury budget [2] (Mg/yr). Various models are applied to estimate Hg emission levels into the atmosphere, However, the emission levels determined by using these models differ substantially. Travnikov et al. [25] compared the global emission levels from natural and anthropogenic sources determined by using four models [26-29]. The models differed significantly in their estimations of global total emissions, with values ranging from 4,000 Mg yr-1 to 9,230 Mg yr-1, of which natural emissions and re-emission ranged from 45% to 66% of the total emissions. A later summary of the four models [30] estimated that global anthropogenic emissions were 1,870 Mg yr-1, although the global natural emissions and re-emissions were already significantly different from each other at 3,995 Mg yr-1 to 8,600 Mg yr-1 [31]. The proportions between the amounts of natural emissions and anthropogenic emissions are not precisely determined. The ratios vary by authors and encompass a relatively wide range from 0.8-1.8 [32-34]. The relatively broad emission ranges presented in various studies are caused by the following factors: volatility Page 3/24 of Hg compounds, distribution of sources, low levels of Hg concentrations in the air, concentrations that are much lower than levels of other basic pollutants and difficulty of determination [35-36]. of Hg compounds, distribution of sources, low levels of Hg concentrations in the air, concentrations that are much lower than levels of other basic pollutants and difficulty of determination [35-36]. Annual mercury emissions in selected regions of the world in 2000-2015 are shown in Table 2. Despite the many applicable global, regional and national programs and conventions aimed at reducing Hg emissions, the global THg emissions in the years 2000-2015 increased by 1.8% [11]. Emission reductions only took place in North America and Europe without the former USSR, and the largest increase was in Central America (5.4%) and South Asia (4.0%). Table 2. Annual emissions of mercury by selected world region [11] (Mg/yr). 2000 2015 2000- 2015 growth (%/yr) 2010- 2015 growth (%/yr) USA 127.7 42.7 -66,6 -10.2 OECD Europe 106.2 38.0 -64,2 -5.8 Canada 12.7 8.3 -34,6 -3.2 Eastern Europe 49.4 33.6 -32,0 -1.3 Central America 33.5 34,3 2,4 5.4 South Asia 120.8 191,6 58,6 4.6 Eastern Africa 19.1 72.9 281,7 4.0 East Asia 532.7 1012,3 90,0 2.6 South America 239.0 275.8 15,4 0.7 Southeast Asia 224,6 187,5 -16,5 2,2 Global Total 1964.1 2389.8 21,7 1.8 able 2. Annual emissions of mercury by selected world region [11] (Mg/yr). Table 1. The update global mercury budget [2] (Mg/yr). Emission values [11] for the aggregated source categories and their composition are shown in Table 3. Three categories had the highest absolute value in 2015: gold ASMG (775.1 Mg/yr), coal combustion (558.3 Mg/yr) and cement production (206.3 Mg/yr). From 2010-2015, the largest increase in emissions was demonstrated by cement production (6.3%), gold production at a large scale (5.2%) and industrial metal extraction (4.6%). The reductions [37] were in the following fields: dental (-5.6%) and electrical equipment (-5.2%). This finding is consistent with the continued expansion of the global economy but is less than the growth in world GDP over the same period (5.7%/yr), suggesting continuous improvement in pollutant emissions per unit of production. Table 3. Annual emissions of mercury by selected source category [11] (Mg/yr). Source category 2000 2015 2000- 2015 growth (%/yr) 2010- 2015 growth (%/yr) Gold ASMG 583,7 775.1 32,8 1.3 Coal combustion 359.9 558.3 55,1 0.7 Cement production 74.3 206.3 177,7 6.3 Waste burning 201.9 165,6 -18,0 2.1 Municipal waste 202.6 140.6 -30,6 2.5 Gold, large-scale 82.7 112.1 35,6 5.2 Zinc smelting 81.7 103.6 26,8 1.4 Copper smelting 53.7 70.0 30,4 3.2 Dental 49.4 16.1 -67,4 -5.6 Electrical equipment 118.7 62.0 -94,8 -5.2 Global Total 1964.1 2389.8 21,7 1.8 Annual emissions of mercury by selected source category [11] (Mg/yr). Exchange of mercury between atmospheric and terrestrial ecosystems [60] estimated that evasion of Hg linked to the decomposition of soil organic carbon pools and subsequent liberation of HgII sorbed to soil organic matter is over 700 t/y, thus reflecting the large pool of Hg stored in terrestrial ecosystems globally (over 240 kgt) In total this study estimated that 56% of Hg deposited to terrestrial soil organic carbon pools and subsequent liberation of HgII sorbed to soil organic matter is over 700 t/y, thus reflecting the large pool of Hg stored in terrestrial ecosystems globally (over 240 kgt). In total, this study estimated that 56% of Hg deposited to terrestrial ecosystems is reemitted. Similarly, Graydon et al. [61] found that 45–70% of isotopically labeled HgII wet-deposited to a forested watershed had been reemitted to the atmosphere after a year. Litter deposition is the predominant source of Hg in soil. Forest litter horizons show significant increases in mass-dependent fractionation (MDF) during decomposition concurrent with augmented total Hg mass, and this relationship is most significant at high- elevation sites [62-63]. Measurements of mercury in litterfall and throughfall have been increasingly used to provide knowledge of mercury deposition over forest canopies. The majority of mercury in litterfall is considered to be from the stomatal uptake of Hg0 and can be used as a rough and conservative estimation of atmospheric mercury dry deposition (the portion that is retained in leaves). Mercury in throughfall also includes a portion of previously dry-deposited mercury (the portion that is washed off from the canopy). Concurrent measurements of litterfall, throughfall, and open-space wet deposition measurements can be used to estimate dry deposition on seasonal or longer time scales, whereby dry deposition is approximated as litterfall plus throughfall minus open-space wet deposition [64-65]. On the land surface, Hg deposition is mainly in the oxidized form (Hg2+), and its transformations are associated primarily with the oxidation-reduction potential of the environment and with the biological and chemical processes of methylation. For soils in which oxidizing conditions predominate, the Hg2+ and Hg2 2+ forms dominate, and in soils with reducing conditions, Hg and sulfur compounds are mainly present. Methyl-Hg compounds are most commonly found in soils with transient conditions [5]. Mercury emissions from natural sources The estimate of Hg emissions from natural sources includes contributions from primary natural sources and the re-emission processes of historically deposited Hg over land and sea surfaces. Hg emitted from volcanoes, geothermal sources and topsoil enriched in Hg pertains to primary natural sources, whereas the re-emission of previously deposited Hg on vegetation, land or water surfaces is primarily related to land use changes, biomass burning, meteorological conditions and gaseous Hg exchange mechanisms at the air- water/topsoil/snow-ice pack interfaces [15, 38]. A characteristic feature of natural Hg emissions compared to anthropogenic sources is their distributed nature and wide range of distribution. Hg from natural sources is introduced on a global scale into the atmosphere, and Hg from anthropogenic sources is deposited mainly locally and regionally; therefore, determining emission levels and applying effective control methods are difficult. Natural Hg emission processes also include re-emission of Hg previously deposited from the atmosphere in the process of wet and dry deposition from both natural and anthropogenic sources, which increases the difficulty of estimating Hg emissions from natural sources [34-35, 39]. Annual global Hg emissions from natural sources on land are estimated by various authors and cover a wide value range, e.g., total emissions of 1,600-2,500 Mg yr-1, including re-emissions of 790-2,000 Mg yr-1. According to most authors, re-emissions were higher than Page 4/24 Page 4/24 Page 4/24 primary emissions [40-42]. Volcanoes Volcanoes and geothermal activities are important sources of Hg pollution in terrestrial environments. Hg is emitted from volcanoes primarily as gaseous Hg0, and the Hg/SO2 ratio is generally adopted to estimate Hg emissions. The annual average of Hg released to the atmosphere without episodic strong eruptions for volcanoes and geothermal activities is ~75-112 Mg yr-1 of Hg, accounting for approximately < 2% of the contribution from natural sources [15, 43-45]. In volcanic plumes, Hg is present both in the gas phase as elemental Hg0 and reactive HgII and in the particle phase as Hgp forms. The proportions of these species are highly variable. HgII and Hgp typically amount to < 5% of THg, with Hg° as the most abundant form [46-48]. Hg levels in volcanic ash nanoparticles (36±4 mg kg-1) are dramatically higher than their bulk concentrations (0.08 mg kg-1) [49]. Many areas of geothermal activity have long been associated with elevated levels of Hg in the soil and air in places such as Hawaii, Iceland, western parts of the United States and New Zealand [43, 50]. The average annual global Hg emissions estimate from biomass burning (emissions from wildfires: forests, savannas and grasslands) for 1997-2006 was 675 (±240) Mg yr -1, which accounts for 8% of all current anthropogenic and natural emissions. The largest Hg emissions are from tropical and boreal Asia, followed by Africa and South America [51-52]. The important factor for Hg concentrations in forest soils is the time since stand-replacing fires have occurred, and high soil burn severity has the potential to reduce the concentrations of Hg in burned soils for tens to hundreds of years [53-54]. In a specific emission source in Nisyros Island (Greece), Hg concentrations in fumarolic gases in Nisyros Island (Greece) ranged from 10,500 to 46,300 ng/m3, while Hg concentrations in the air ranged from high background values in the Lakki Plain caldera (10-36 ng/m3) up to 7100 ng/m3 in the fumarolic areas [55]. Exchange of mercury between atmospheric and terrestrial ecosystems Exchange of mercury between atmospheric and terrestrial ecosystems Air-, soil- or vegetation-covered exchange fluxes are an important part of global and regional biogeochemical cycles [56]. Much of the HgII deposited in precipitation or taken up by plants is reduced to Hg0 and may be released back to the atmosphere. Recent vegetation and soil Hg studies suggested that vegetation Hg0 uptake dominates (50-80%) Hg net deposition at terrestrial sites [57-59]. g g gg g g p ( ) g p The different forms of atmospheric Hg may be deposited on surfaces by way of wet and dry processes. These forms may be sequestered within terrestrial compartments or emitted back to the atmosphere, with the relative importance of these processes being dependent on the form of Hg, surface chemistry, and environmental conditions. Many models assume that the net GEM (gaseous elemental mercury) exchange with soil surfaces is zero; however, as discussed below, some components are assimilated into foliage over the growing season and accumulate in soils [59]. Smith-Downey et al. [60] estimated that evasion of Hg linked to the decomposition of soil organic carbon pools and subsequent liberation of HgII sorbed to soil organic matter is over 700 t/y, thus reflecting the large pool of Hg stored in terrestrial ecosystems globally (over 240 kgt). In total, this study estimated that 56% of Hg deposited to terrestrial ecosystems is reemitted. Similarly, Graydon et al. [61] found that 45–70% of isotopically labeled HgII wet-deposited to a forested watershed had been reemitted to the atmosphere after a year. The different forms of atmospheric Hg may be deposited on surfaces by way of wet and dry processes. These forms may be sequestered within terrestrial compartments or emitted back to the atmosphere, with the relative importance of these processes being dependent on the form of Hg, surface chemistry, and environmental conditions. Many models assume that the net GEM (gaseous elemental mercury) exchange with soil surfaces is zero; however, as discussed below, some components are assimilated into foliage over the growing season and accumulate in soils [59]. Smith-Downey et al. Exchange of mercury between atmospheric and terrestrial ecosystems Table 5 shows a summary of total Hg (THg) ﬂuxes from terrestrial regions [82], and Table cases, the range of ﬂuxes) for various ecosystems measured by a number of investigators. al Hg (THg) ﬂuxes from terrestrial regions [82], and Table 6 shows the average ﬂuxes (or, in som ecosystems measured by a number of investigators. Table 5. Summary of mercury ﬂuxes from terrestrial regions [82] Region Evasion (average) (Mg yr-1) Ratio1 (%) Forest 342 7,5 Tundra/Grassland/Savannah/Prairie/Chaparral 448 9,9 Desert/Metalliferous/Non-vegetated Zones 546 12,0 Agricultural areas 128 2,8 Evasion after mercury Depletion Events2 200 4,4 Total 1664 - Volcanoes and geothermal areas 90 - Biomass burning 675 - Table 5. Summary of mercury ﬂuxes from terrestrial regions [82] 1calculated over the total evasion from natural sources which sum 4532 Mg yr-1 1calculated over the total evasion from natural sources which sum 4532 Mg yr-1 2 2Friedli [42] and Mason [68] distinguish as a natural source of mercury emission to the atmospheric air of regions where there are episodes of sudden decreases in mercury concentrations in the air by deposition to the ground and then reemission. Exchange of mercury between atmospheric and terrestrial ecosystems For GEM (Hg0), the residence time is estimated at 6 to 18 months, while GOM (gaseous mercury in oxidized form) and TPM (total particulate mercury) are quickly removed from the air through wet and dry deposition, and their residence times are estimated to be hours or days at most [66-67]. Given the long time for removal from the air, GEM can be transported over large distances [16]. The particular Hg species are characterized by different dry deposition rates, which also determine their residence times. The dry residence times of the different Hg species form the following series [68]: GEM 0.19 cm s-1 < TPM 2:1 cm s-1 < GOM 7:6 cm s-1 GEM 0.19 cm s-1 < TPM 2:1 cm s-1 < GOM 7:6 cm s-1 GEM 0.19 cm s-1 < TPM 2:1 cm s-1 < GOM 7:6 cm s-1 Page 5/24 Page 5/24 Page 5/24 According to Marsik et al. [69], the dry deposition rates of GOM and GEM are much higher during daytime than nighttime. As with Lindberg et al. [68], these authors explain this fact by the closure of plant stomata at night. The deposition rates also depend on the type of surface [70]. Caffrey et al. [71] determined that the deposition rates of particulate air pollutants on the ground with low vegetation were 3-5 times lower than those in forests. Deposition is also affected by the weather conditions, air humidity, insolation and atmospheric precipitation. Research by Converse et al. [72] in an uncontaminated high-elevation wetland meadow in Shenandoah National Park, Virginia (USA) showed the highest Hg deposition occurred in spring (4.8 ng m−2 h−1), with a decrease occurring in summer (2.5 ng m−2 h−1) to near zero flux in fall (0.3 ng m−2 h−1), followed by an increase in winter emissions (4.1 ng m−2 h−1). These studies also suggest that stomatal processes are not the dominant mechanism for ecosystem-level GEM exchange. Table 4 shows a summary of biome-level Hg depositions and soil Hg turnover times [60]. The concentration of Hg in soils is therefore a function of the deposition rate and carbon turnover time. High soil concentrations in desert ecosystems are driven by a combination of higher deposition and extremely slow Hg turnover. Tropical and temperate lifetimes are similar despite the faster carbon turnover in tropical systems due to the relative balance between Hg provided by wet deposition and leaf uptake. Table 4. Exchange of mercury between atmospheric and terrestrial ecosystems Summary of Biome Level Hg Deposition and Soil Hg Turnover Times [60]. Table 4. Summary of Biome Level Hg Deposition and Soil Hg Turnover Times [60]. Biome Mean Hg deposition (g m-2) Mean soil Hg turnover time* (years) Preindustrial Present day Preindustrial Present day Tropical forest 0.9 3.7 234 126 Temperate forest 0.8 2.9 250 151 Boreal forest 0.5 1.5 998 560 Grassland 1.0 3.5 522 269 Tundra 0.3 0.8 1108 702 Desert 0.5 1.4 2387 1748 with respect to respiration with respect to respiration. Emissions from soils have the form of GEM and depend on many factors [73-77]: Emissions from soils have the form of GEM and depend on many factors [73-77]: – The properties of soils, e.g., Hg content, the contents of organic compounds, and satu – The concentrations of oxidants, mainly ozone, in the air; and – The concentrations of oxidants, mainly ozone, in the air; and – The weather conditions, e.g., solar radiation, temperature, humidity and winds. – The weather conditions, e.g., solar radiation, temperature, humidity and winds. Soil Hg fluxes are significantly lower in dark conditions than light conditions for al – The weather conditions, e.g., solar radiation, temperature, humidity and winds. Soil Hg fluxes are significantly lower in dark conditions than light conditions for all sites except grassland [64]. , g , , p , y oil Hg fluxes are significantly lower in dark conditions than light conditions for all sites except gras Soil Hg fluxes are significantly lower in dark conditions than light conditions for all sites except grassland [64]. It is most difficult to estimate Hg emissions from plants, and these emissions mainly occur in th 78] suggested the following hierarchy of environmental parameters that inﬂuence Hg ﬂux: It is most difficult to estimate Hg emissions from plants, and these emissions mainly occur in the form of Hg0 [79-81]. Ericksen et al. [78] suggested the following hierarchy of environmental parameters that inﬂuence Hg ﬂux: soil moisture > light > air concentration > relative humidity > temperature. Table 5 shows a summary of total Hg (THg) ﬂuxes from terrestrial regions [82], and Table 6 shows the average ﬂuxes (or, in some cases, the range of ﬂuxes) for various ecosystems measured by a number of investigators. Exchange of mercury between atmospheric and terrestrial ecosystems These phenomena occur mainly in the arctic regions and Antarctica, and the emission from this source is estimated at 200 Mg ye-1 When analyzing the data in Table 6, the following generalizations can be made [82-83]: When analyzing the data in Table 6, the following generalizations can be made [82-83]: When analyzing the data in Table 6, the following generalizations can be made [82-83]: i) The largest amounts of Hg are emitted from tropical regions (45%), followed by the temperate zones (41%), with the lowest emissions When analyzing the data in Table 6, the following generalizations can be made [82-83]: i) The largest amounts of Hg are emitted from tropical regions (45%), followed by the temperate zones (41%), with the lowest emissions from the polar regions (8%), and emissions from volcanoes and geothermal areas account for 5%; i) The largest amounts of Hg are emitted from tropical regions (45%), followed by the temperate zones (41%), with the lowest emissions from the polar regions (8%), and emissions from volcanoes and geothermal areas account for 5%; ii) Areas with vegetation can be ranked according to the size of their emissions as follows: forests > other areas (tundra, savannas, and chaparral) > agricultural areas > grassland ecosystems; and ii) Areas with vegetation can be ranked according to the size of their emissions as follows: forests > other areas (tundra, savannas, and chaparral) > agricultural areas > grassland ecosystems; and (iii) Land areas devoid of vegetation emit more Hg than do areas with plants. Deforestation can increase GEM emissions due to higher solar radiation and increased temperature at the soil surface [84-85]. Table 6. Average ﬂuxes, or in some cases the range of ﬂuxes, for various ecosystems measured by a number of investigators. Species/ecosystem Flux1 nmol m-2 mth-1 References. Ground Level Forest Floor Sweden 1.4-1.7 Hanson et al. [83] Lindberg et al. [68] Model Estimates Hard wood forest Max 4 Bash et al. [87] Model Estimates Forest Soil 4 Bash et al. [87] Agricultural crops Max 11 Bash et al. [87] Temperate forest 5.0 Rea et al. [88] Deforested site 50 Magarelli and Fostier [89] Desert Soils 3.6-9.8 Magarelli and Fostier [89] High Hg regions Max 1500 Gustin and Lindberg [90] Maple 20 Hanson et al. [83] Oak 16.4 Hanson et al. [83] Spruce 6.1 Hanson et al. [83] Prairie grass 12.5 Obrist et. al. [53] Typha sp. 60 Gustin et al. Exchange of mercury between atmospheric and terrestrial ecosystems Areas of land devoid of vegetation emit more Hg than those with plants. The annual averaged fluxes in the subtropical forest zones in China from soil in the forests were eas of land devoid of vegetation emit more Hg than those with plants. The annual averaged fluxes in the subtropical forest zones a from soil in the forests were g m-2h−1, and for open-air sites, they were 20.7 ng m-2 h−1 [96]. Soil Hg fluxes were significantly lower in dark conditions than in nditions. In grassland sites, the mean soil Hg flux was 0.6±0.9 ng.m−2 h−1 in darkness, 1.0±0.7 ng.m−2 h−1 in light, and 14.2 ng m-2h−1, and for open-air sites, they were 20.7 ng m-2 h−1 [96]. Soil Hg fluxes were significantly lower in dark conditions than in light conditions. In grassland sites, the mean soil Hg flux was 0.6±0.9 ng.m−2 h−1 in darkness, 1.0±0.7 ng.m−2 h−1 in light, and 0.9±0.7 ng.m−2 h−1 overall [64]. Cropland is an important component of terrestrial ecosystems. It is estimated that 33% of natural-source atmospheric Hg comes from the emissions at cropland surfaces [97]. The emission of Hg from cropland soil greatly affects the global Hg cycle. Combinations of different crop cultivars and planting densities will result in different light transmittance under canopies, which directly affects the solar and heat radiation flux received by the soil surface below crops. In turn, this might lead to differences in the soil–air total gaseous mercury (TGM) exchange under different cropping patterns. The light transmittance under the canopy was the key control on soil–air TGM exchange fluxes. High light transmittance can enhance soil TGM emission rates and increase the magnitude of diurnal variations in soil–air TGM exchange fluxes. The estimated annual average soil–air TGM exchange flux was 5.46±21.69 ng m−2 h−1 in corn–wheat rotation cropland with 30 cm row spacing [98-99]. soil–air TGM exchange fluxes. The estimated annual average soil–air TGM exchange flux was 5.46±21.69 ng m−2 h−1 in corn–wheat rotation cropland with 30 cm row spacing [98-99]. The bidirectional exchange of Hg between the atmosphere and terrestrial surfaces is better understood because of advancements in research that are primarily associated with the interpretation from Hg isotopes, and the latest estimates place land surface Hg re- emission at values lower than previously thought [100]. Exchange of mercury between atmospheric and terrestrial ecosystems [91] Average global soil 1.5 Selin et al. [92] e cases the range of ﬂuxes, for various ecosystems measured by a number of investigators. Values have all been converted to a common ﬂux unit of nmol m-2 month-1. Results from the older literature are combined The overall background soil Hg flux in the United States is estimated to be 0.9±0.2 ng/m2/h [79], and in areas with significant Hg pollution, soil emissions are much larger. In the canton of Valais, Switzerland, elemental Hg (Hg0) is undetectable in soil, although substantial Hg0 emissions were found to occur (20-1,392 ng.m−2 h−1) [94]. as are of particular concern with respect to the global Hg cycle due to the following [95]: Urban areas are of particular concern with respect to the global Hg cycle due to the followin Urban areas are of particular concern with respect to the global Hg cycle due to the following [95]: Frequently high terrestrial Hg concentrations and the physically and chemically diverse nature o Frequently high terrestrial Hg concentrations and the physically and chemically diverse nature of urban surface covers; Frequently high terrestrial Hg concentrations and the physically and chemically diverse nature of urban surface covers; Highly variable time series concentrations of ambient atmospheric Hg as a result of regional and local emissions; and Frequently high terrestrial Hg concentrations and the physically and chemically diverse nature of urban surface covers; Highly variable time series concentrations of ambient atmospheric Hg as a result of regional and local emissions; and Urban meteorology (i.e., heat island effect). Frequently high terrestrial Hg concentrations and the physically and chemically diverse nature of urban surface covers; Highly variable time series concentrations of ambient atmospheric Hg as a result of regional and local emissions; and Urban meteorology (i.e., heat island effect). Highly variable time series concentrations of ambient atmospheric Hg as a result of regional and g y p g g Urban meteorology (i.e., heat island effect). oosa, Alabama (USA), Hg fluxes on bare undisturbed soil surfaces were as follows (median) [95]: In the city of Tuscaloosa, Alabama (USA), Hg fluxes on bare undisturbed soil surfaces were as follows (median) [95]: In the city of Tuscaloosa, Alabama (USA), Hg fluxes on bare undisturbed soil surfaces were Residential site - 4.45 ng.m−2 h−1; Industrial site - 1.40 ng.m−2 h−1; Commercial site - 2.14 ng.m−2 h−1; and Mixed land use site – 0.87 ng.m−2 h−1. Methylmercury might lead to increases in biological activity as well as higher Hg2+ methylation rates [111]. The direct conversion of insoluble HgS species to MeHg in anaerobic soils is generally believed to be low, although this condition can change when environmental conditions favor HgS complexation [112]. The redox potential also seems to be a key factor, as suboxic and mildly reducing conditions seem to promote high Hg2+ methylation rates, while anoxic and strongly reducing conditions might lead to elevated sulfide concentrations that eventually prevent Hg2+ from being available for methylation of some methylating bacteria, including SRB (sulphate reducing bacteria, e.g., Desulfobacter sp.), and some that control the availability of Hg2+ for methylation (e.g., Deltaproteobacteria or Clostridia) [113-114]. S plays a major role in influencing Hg2+ methylation by directly affecting the activity of some methylating bacteria, such as SRB, and controlling the availability of Hg2+ for methylation [115-116]. The paddy soils in Hg mining areas have a high methylation ability and may eventually result in heavily biological effects on the local residents through the food chains, such as rice containing high methyl-Hg [117]. In the Wuchuan Hg mining areas (Guizhou, China), soil samples present THg values ranging from 0.33 to 320 mg kg−1 and methyl-Hg values ranging from 0.69 to 20 ng g−1. The rice grain samples contain elevated methyl-Hg concentrations ranging from 4.2 to 18 ng g−1, while corn grain contained only 0.5-2.0 ng g−1 [118]. Research carried out in areas with coal-fired power plants in Hunan (China) [119] shows that in the soil samples, THg varied from 0.068 to 0.220 mg kg−1 (mean value of 0.130±0,040 mg kg−1), and methyl-Hg ranged from 0.30 to 3.5 μg kg−1 (mean 1.6±1.0 μg kg−1). In rice samples, the Hg(II) concentrations varied from 0.002 to 0.022 mg kg−1 (mean 0.057 mg kg−1), and methyl-Hg concentrations varied from 1.7 to 3.8 ng gg−1 (mean of 2.4±0.72 ng g−1). Meng et al. [107] showed that rice had high affinity for methyl-Hg and that the concentrations in rice seeds may be 2 to 3 orders of magnitude higher in Hg mining sites than in other local edible crop plants. Freshly deposited Hg is more likely to methylate and be incorporated in rice than stored Hg [100]. Forest fires cause a significant reduction in mercury content in soil. Burned soils in northwestern Ontario (Canada) had 82% less methyl-Hg than fresh soils [120]. Methylmercury Current climate change has had a significant impact on Hg transformation processes, especially in the Arctic. The very large mass of mainly natural Hg found in northern permafrost deposits, which is projected to be released with further climate warming, may profoundly affect biotic Hg levels around the Northern Hemisphere, especially because large amounts of organic carbon, which may stimulate Hg methylation rates, will be simultaneously released [121]. Methylmercury High doses of organic compounds of Hg, particularly methyl-Hg, can be fatal to humans and wildlife, and even relatively low doses can seriously affect the nervous system of organisms. Hg has also been linked to harmful effects on the cardiovascular, immune and Page 7/24 Page 7/24 reproductive systems. Methyl-Hg passes through both the placenta and blood-brain barrier; therefore, the exposure of women of child- bearing age and of children to methylmercury is of great concern [1]. Methyl-Hg can be both biotically and abiotically produced in the environment. Methylation of Hg tends to occur in environments with low oxygen levels, low pH, Hg bioavailability, temperature, redox potential and high levels of dissolved organic compounds and in environments favored by sulfate-reducing bacteria, which are largely responsible for methylation. These conditions are found primarily in deep sea environments, coastal marine sediments, and some freshwater lakes as well as soils. These conditions are also characteristic for paddy soil [101-102]. Organic Hg is much more toxic to living organisms than inorganic Hg [4, 103-104]. The content of methyl-Hg in soils and plants is significantly lower than that of THg; however, due to its much higher toxicity, methyl-Hg is particularly dangerous for living organisms. In soils in the coniferous boreal forests of Sweden, the background level of pollution from methyl-Hg accounted for 0.35-0.59% of THg [105]. In the Idrija Hg mining area of Slovenia, a heavily polluted region, methyl-Hg accounted for 0.003% of THg, and its background presence in controls was 0.17% [106]. The content of methyl-Hg relative to THG was 1.9% in the roots of rice under background conditions and 0.55% in the leaves; in areas of Hg mining sites, the concentrations were 0.07% in the roots and 0.01% in the leaves; and in areas with artisanal Hg mining sites, the concentrations were 0.63% in the roots and 0.02% in the leaves [107]. Methyl-Hg in heavily contaminated soil in the Rhône Valley (Switzerland) accounted for < 0.8% of THg [108]. A number of factors that control microbial activity and/or the geochemical speciation of inorganic Hg2+ govern MeHg formation in the environment [109]. Microorganisms that live in soil can transform inorganic Hg(II) species into Hg0 by using the enzyme Hg reductase, which is found in various bacteria, such as Pseudomonas sp., Staphylococcus aureus, Thiobacillus and many others [110]. Increases in temperature might lead to increases in biological activity as well as higher Hg2+ methylation rates [111]. Mercury in soil Table 7 presents examples of the concentrations of Hg in the soils in the vicinity of industrial emission sources according to different authors, The highest Hg contents were found in soils near Hg mines: Almaden in Spain, with 2,000 years of mining and ore processing (<8,889 mg kg−1) [142]; Idrija in Slovenia, with 500 years of mining activity (<2,759 mg kg−1) [106, 139]; and in Alaska (5,326 mg kg-1) [136]. Chlor-alkali plants are also an important source of environmental Hg pollution. Bernaus et al. [155] estimated that Hg levels in the soil around a chlor-alkali plant in the Netherlands were as high as 1,150 mg kg-1. According to Richardson and Moore [170], in the urban environment, the diversity of Hg content in soils is relatively high because of the diversity of land functions in towns. Urban soils were found to accumulate higher concentrations and pools of Hg than their rural montane counterparts across New York and southern New England, which highlights the importance of soils in urban systems for sequestering Hg and preventing its movement towards riparian and aquatic ecosystems, where it can bioaccumulate. Moreover, soil Hg concentrations were poorly correlated with pH, loss-on-ignition, and clay content, Instead, proximity to local industrial and agricultural sources proved a significant influence on Hg accumulation. The lowest median results were determined for soils in Changchun, China, at 0.018 mg kg-1, with a range of 0.012-0.036 mg kg-1 [171] and in Oslo, Norway (0.06 mg kg-1, with a range of 0.01-2.3 mg kg-1) [172], while the highest concentrations were in Palermo, Italy (median value of 1.85 mg kg-1, with a range of 0.004-2.61 mg kg-1) [173], and Glasgow, Scotland (1.2 mg kg-1, with a range of 0.312-5.2 mg kg-1) [174]. Of note is the reduction of 270% (median from 0.68 mg kg-1 to 0.37) in Hg pollution from 1987-2009 for soils from Beijing, China [175]. j g, [ ] In agricultural soils, pollution by Hg was relatively low, as indicated by the low median and average values at usually below 0.1 mg kg-1 (e.g., Scandinavia – Ottesen et al. [176]; Poland – Loska et al. [177]; Iran – Ahmadi et al. [178]). However, the ranges of the results were relatively wide, and the maximum values often exceeded 1 mg kg-1, which may indicate a threat to food production due to the need to protect human health in some areas with higher Hg soil levels. Mercury in soil In Europe, pastures were slightly more polluted with Hg than plowed fields [177]. Soils in forest environments contained low levels of Hg. Average and median values did not exceed 1 mg kg-1, although compared with agricultural soils, the maximum values were also lower than 1 mg kg-1. Mineral forest soils contained less Hg than organic ones (USA – Woodruff and Cannon, [54] Czech Republic – Navrátil et al. [179]; Sweden – Åkerblom et al. [180]. According to Obrist et al. [129], soil Hg concentrations significantly differed among land covers following the order: forested upland > planted/cultivated > herbaceous upland/shrubland > barren soils forested upland > planted/cultivated > herbaceous upland/shrubland > barren soils. Concentrations in forests were an average of 2.5 times higher than those in barren locations. Concentrations in forests were an average of 2.5 times higher than those in barren locations. The global distributions of soil Hg storage and emissions for both preindustrial and present-day simulations in different biomes are shown in Table 4 [60]. The relatively low soil Hg concentrations in boreal and arctic ecosystems are driven by extremely low deposition. The high soil concentrations in desert ecosystems are driven by a combination of higher deposition and extremely slow Hg turnover. The concentration of Hg in soils is therefore a function of the deposition rate and carbon turnover time. Physical and chemical properties of the soil affect the Hg cycle in the environment. The soil aggregate size fractions have significant effects on the Hg content in soil. The concentrations of Hg and other heavy metals in soils and sediments generally tend to increase with decreasing grain size, which is due to the propensity of metals to bind with finer particles [181]. Generally, higher values of Hg in soil are found in the fraction at <63 µm [182]. In the Amazonian areas without anthropogenic sources, the fine fraction (< 53 μm) of podzolized soils had higher Hg contents than clayey soils [183]. In a temperate forest podzol, Hg mean values increased as the aggregate sizes decrease, as follows: ) > fine silt (130 ng g-1) > coarse silt (80 ng g-1) > fine sand (32 ng g-1) > coarse sand (14 ng g-1). clay (170 ng g-1) > fine silt (130 ng g-1) > coarse silt (80 ng g-1) > fine sand (32 ng g-1) > coarse sand (14 ng g-1). Mercury in soil Kabata Pendias and Pendias [5] defined a narrower range of 0.05-0.3 mg kg-1, although some volcanic and organic soils, especially in Canada, may contain higher values, and in the vicinity of industrial emission sources, the values can be extremely high. Obrist et al. [129] showed that a dataset with more All results for Hg soil content, which are presented in the next part of this publication, pertain to the topsoil layer. Generally, the average background concentration of Hg in soil ranges from 0.03 to 0.1 mg kg−1, with an average value of 0.06 mg kg−1, whereas Hg- contaminated sites often have soil concentrations that are 2- to 4-orders of magnitude higher [127-128]. Kabata Pendias and Pendias [5] defined a narrower range of 0.05-0.3 mg kg-1, although some volcanic and organic soils, especially in Canada, may contain higher values, and in the vicinity of industrial emission sources, the values can be extremely high. Obrist et al. [129] showed that a dataset with more than 1,900 randomly selected sampling points across the western USA indicated median Hg concentrations of 0.019 mg kg−1 and an average value of 24 mg kg−1, with only 1% of soil samples exceeding background values (e.g., >0.10 mg kg−1). ue of 24 mg kg−1, with only 1% of soil samples exceeding background values (e.g., >0.10 mg kg−1) The LUCAS Topsoil Survey of the European Union organization collected over 23,000 topsoil samples (upper 20 cm) from land in all European Union countries (28) except for Croatia [130]. The average for European topsoil Hg concentrations was 0.04 mg kg−1, with a range of 0-159 mg kg−1. Studies have identified highly polluted, isolated sites, and the larger historical and recent industrial and Hg mining areas show elevated concentrations of Hg. Historically, mining for gold and Hg led to high Hg concentrations in these mining areas, which may explain the high Hg concentrations in some samples from Central Italy, Northwest England and Eastern Slovakia. Moreover, the natural/background Hg level was 0.08 mg/kg in Brazil [131], 0.05 mg/kg in India [132], 0.23 mg/kg in New Zealand [133], 0.11 mg/kg in the Norwegian Arctic [134] and 0.4 mg/kg in Paris [136]. Most soil Hg was found as soil matrix-bound divalent Hg (HgII), whereas elemental Hg (Hg0) was undetectable in soils [94,108,]. Mercury in soil Mercury has a relatively long half-life in surface soils because of its recycling between the surface environment and atmosphere. Permanent removal of anthropogenic Hg from the biologically active part of the environment will only occur once it is buried in mineral soils [4]. Soil plays an important role in biogeochemical Hg circulation because it accumulates this element and is a source for other environmental components. Hg occurs naturally in soils from geologic sources [12] or as the result of natural events such as forest fires and volcanic eruptions [49]. The total amount worldwide of Hg accumulated in the soils of terrestrial environments is estimated at 200- 300 Gg [122-124]. Smith-Downey et al. [60] suggested that organically bound Hg in preindustrial soils is 200 Gg and that a 20% increase in organically bound soil Hg (to 240 Gg) has occurred from preindustrial steady state conditions to the present day. In the 2013 Technical Background Report for the Global Mercury Assessment [4], based on a global model and budget developed by Mason et al. [124], human activities were estimated to cumulatively increase atmospheric Hg concentrations by 300-500% over the past century. Because of the naturally high Hg amount present in soil, the average Hg increase was only 20% in surface organic soil and negligible in mineral soils. The revolatilization of "legacy Hg" (i.e., Hg from historical sources of pollution) from soil and ocean and its long residence time in those compartments contribute to maintaining atmospheric Hg concentrations and deposition rates at higher levels than those supported by current primary emissions [125]. Recent estimates of the anthropogenic and natural Hg contents in global soils Page 8/24 Page 8/24 (organic layers) (data in kilotons) included 182 natural and 89 anthropogenic sources based on Amos et al. [126] and 130 natural and 20 anthropogenic sources according to the AMAP/UN Environment [4]. (organic layers) (data in kilotons) included 182 natural and 89 anthropogenic sources based on Amos et al. [126] and 130 natural and 20 anthropogenic sources according to the AMAP/UN Environment [4]. All results for Hg soil content, which are presented in the next part of this publication, pertain to the topsoil layer. Generally, the average background concentration of Hg in soil ranges from 0.03 to 0.1 mg kg−1, with an average value of 0.06 mg kg−1, whereas Hg- contaminated sites often have soil concentrations that are 2- to 4-orders of magnitude higher [127-128]. plants Vegetation affects environmental factors at the ground surface by reducing solar radiation, temperature, and wind velocity and serves as a surface for Hg uptake [85]. Many studies have recognized the essential role of terrestrial plants in the biogeochemical cycling of Hg (e.g., Gustin et al. [190]; Fantozzi et al. [191]; Mazur et al. [192]). Approximately 80% of total Hg accumulated in the aboveground biomass is found in the leaves, and approximately 1% of that Hg is methylated. The concentrations of Hg in aspen tissue grown in high-Hg soil increases in the following order [93]: Approximately 80% of total Hg accumulated in the aboveground biomass is found in the leaves, and appro oncentrations of Hg in aspen tissue grown in high-Hg soil increases in the following order [93]: stems < branches < petioles < roots < leaves stems < branches < petioles < roots < leaves Research conducted by Leonard et al. [193] in Nevada (USA) in an area with high levels of Hg contamin latifolium, 70% of the Hg taken up by the roots during the growing season was emitted to the atmosphere. Research conducted by Leonard et al. [193] in Nevada (USA) in an area with high levels of Hg contaminat latifolium, 70% of the Hg taken up by the roots during the growing season was emitted to the atmosphere. Nevada (USA) in an area with high levels of Hg contamination revealed that for the plant species Lepidium during the growing season was emitted to the atmosphere. The main source of Hg in leaves comes from air pollution with Hg0 and not from soil contamination [190, 194-195]. The studies by Fleck et al. [196] of Pinus resinosa have shown that neither woody tissue Hg nor any amount of Hg in the soil or forest floor were closely related to foliar levels, while for some relationships, the opposite was true. The authors interpret these data as indicating that Hg in plant tissues is derived directly from the atmosphere and not from the soil. It is estimated that in highly contaminated soils, generally less than 2% of the Hg present is available for plants [143]. Total leaf concentrations of Hg varied among species and were most closely correlated with the number of stomates per sample, thus supporting the hypothesis that stomatal uptake of atmospheric Hg (most likely Hg0) is a potential uptake pathway [197]. Research by Arnold et al. plants [198] also indicated the importance of the nonstomatal pathway for the uptake of total gaseous Hg (TGM). 2% of the Hg present is available for plants [143]. Total leaf concentrations of Hg varied among species and were most closely correlated with the number of stomates per sample, thus supporting the hypothesis that stomatal uptake of atmospheric Hg (most likely Hg0) is a potential uptake pathway [197]. Research by Arnold et al. [198] also indicated the importance of the nonstomatal pathway for the uptake of total gaseous Hg (TGM). Plants growing beyond the influence of high Hg emissions contained less than 100 ng g-1 THg. Plants growing in the vicinity of factories are large emitters of Hg, such as those around Hg mining sites (e.g., Moreno-Imenez et al. [199]; Qian et al. [200]) and chlor- alkali mining sites [201]. Au mining sites [202-203] may also contain extremely high Hg contents. Mushrooms have been identified as organisms that accumulate more Hg than other plants [204]. A synthesis of published vegetation Hg data from the western United States showed that aboveground biomass concentrations followed the order [129]: leaves (26 μg kg−1) ~ branches (26 μg kg−1) > bark (16 μg kg−1) > bole wood (1 μg kg−1) leaves (26 μg kg−1) ~ branches (26 μg kg−1) > bark (16 μg kg−1) > bole wood (1 μg kg−1) Hg concentrations in leaves were monitored from the emergence to senescence and showed a strong positive correlation with leaf age [195, 197, 205]. Hg concentrations in leaves were monitored from the emergence to senescence and showed a strong positive correlation with leaf age [195, 197, 205]. Mercury in soil Total Hg enrichment in clay-sized aggregates were 2 to 11 times higher than the values shown by the bulk soil (< 2 mm) [184]. In a heavily polluted area near the Wanshan Hg mine (China), the fine soil aggregate size fractions < 231 μm showed higher total Hg Page 9/24 concentrations and higher soil organic matter content than did the larger aggregate size fractions (231 to 2,000 μm) [185]. concentrations and higher soil organic matter content than did the larger aggregate size fract Humic acid influences Hg transport and transformation in soil-plant systems, especially for soils having low clay content. Humic acid reduces the amount of available Hg in soil and prevents Hg from being transported into plants or leached from the soil. Leaching can result in Hg leaking into natural water systems under normal environmental conditions. In practice, humic acid can be used to control Hg transportation into food chains from soil heavily polluted by Hg [186]. The chemical and mineralogical properties of soil affect oxidation and retention of atmospheric Hg. Abiotic Hg oxidation occurs because organic matter has -SH groups, which have a high affinity for Hg ions, and Hg oxidation is favored in the presence of compounds with high affinities for the Hg ion [131]. A microbial contribution to Hg oxidation was first proposed by Smith et al. [187], who demonstrated that typical soil bacteria (Bacillus and Streptomyces) can oxidize elemental Hg to Hg2+ through enzymatic paths. Recent studies have shown that Hg can also be oxidized by anaerobic bacteria [188-189]. The soil microbial community is very sensitive to Hg concentrations, and this sensitivity is influenced not only by soil properties but also by the plant species growing in the soil. A level of 0.36.mg.kg-1 of Hg in soils is proposed to be a critical concentration above which plant and soil organisms will be affected [131]. Toxic effects on plants Hg does not have any beneficial effects on organisms and is thus regarded as the “main threat” since it is very harmful to both plants and animals; pollutes the air, water and soil; and is toxic [206]. Mercury has toxic effects on plants, even at low concentrations, and leads to growth retardation [207] and many other adverse effects [208]. Hg in plants is strongly bound to sulfhydryl/thiol groups of proteins and forms SHgS. Hg toxicity in plants occurs via its binding to SH groups of proteins, displacement of essential elements and disruption of the protein structure [209]. This biochemical property probably determines the toxic effects on plants [5, 210-211]. Studies of the toxic effects of Hg on soil organisms and native plants in fields are limited. The effects of Hg are usually examined in sterile and much-simplified laboratory conditions, which may differ from field conditions to varying degrees [212]. The field study of Moreno-Jiménez et al. [199] was conducted in the mining district of Almadén (Spain), which is a cinnabar (HgS) enriched zone, from which one-third of the total Hg produced worldwide is extracted. Mining activity began more than 2,000 years ago, and no other region in the world has been influenced by Hg for such a long period. The region is considered to be one of the regions most polluted by Hg in the world. Hg concentrations in the field plants Rumex induratus and Marrubium vulgare grown in these soils can be considered phytotoxic, although no symptoms of Hg toxicity have been observed in any of the studied plant species. In most contaminated soils and mine tailings, Hg is not readily available for plant uptake [213]. The absorption of organic and inorganic Hg from soil by plants is low, and there is a barrier to Hg translocation from plant roots to tops. Thus, large increases in soil Hg levels produce only modest increases in plant Hg levels by direct uptake from soil [212]. In Page 10/24 Page 10/24 terrestrial vegetation, Hg in the aboveground biomass originates primarily from the atmosphere, whereas Hg in the roots comes from the soil [67,214]. The research conducted by Lomonte et al. [215] suggested the existence of Hg stress-activated defense mechanisms in plants and hypothesized that these mechanisms were likely the reason for the increased production of sulfur compounds in the tested plant species, which stimulated their growth. Toxic effects on plants Hg has very limited solubility in soil, low availability for plant uptake and no known biological function, which may explain why Hg-hyperaccumulating plants have not yet been identified, meaning that a method for Hg phytoremediation in soils contaminated with Hg has not yet been developed [215]. However, studies suggesting the use of transgenic plants for phytoremediation have been published recently [216-217]. The significant toxic effect of Hg on plants is the generation of reactive oxygen species (ROS) [218], e.g., superoxide anion radicals, H2O2, and hydroxyl radicals (OH.) [219-220]. Detoxification mechanisms to combat Hg-induced oxidative stress include enzymatic antioxidants and some nonenzymatic antioxidants, such as the following: glutathione [221], phytochelatin [222], salicylic acids [223], ascorbic acid [224] selenium [225] proline [226] and tocopherols [227] This process is correlated with the disruption of biomembrane antioxidants and some nonenzymatic antioxidants, such as the following: glutathione [221], phytochelatin [222], salicylic acids [223], ascorbic acid [224], selenium, [225], proline [226] and tocopherols [227]. This process is correlated with the disruption of biomembrane lipids and cellular metabolism, resulting in plant injury [228]. Increasing levels of mercury species in the soil exert a wide range of adverse effects on the growth and metabolism of plants [206, 229-230], such as reduced photosynthesis, transpiration, water uptake, chlorophyll synthesis [228, 231-232] and increased lipid peroxidation (Cho and Park [219]). A high Hg content in plants affects the activity of most enzymes. The total activity of stress indicators such as superoxide dismutase (SOD), peroxidase (POD) and ascorbate peroxides (APX) increased after Hg treatment, but the vast majority of enzymes were inhibited at higher concentrations (e.g., Manikandan et al. [233]; Mahbub et al. [234]; Zhou et al. [235]). Conclusions And Commentary This ratio, which is dependent on the authors, has been estimated over a relatively wide range of 0.8-1.8. The proportions between the amounts of natural emissions and anthropogenic emissions have not been precisely determined. This ratio, which is dependent on the authors, has been estimated over a relatively wide range of 0.8-1.8. The proportions between the amounts of natural emissions and anthropogenic emissions have not been precis dependent on the authors, has been estimated over a relatively wide range of 0.8-1.8. Annual global Hg emissions from natural sources on land are estimated by various authors over a wide range, with total emissions of 1,600-2,500 Mg yr-1, including re-emissions of 790-2,000 Mg yr-1. The low share of Hg taken up by plants from the soil is beneficial from perspective of protecting food against contamination, although it also limits the possibility of using plants in the phytoremediation of contaminated soils. Annual global Hg emissions from natural sources on land are estimated by various authors over a wide range, with total emissions of 1,600-2,500 Mg yr-1, including re-emissions of 790-2,000 Mg yr-1. The low share of Hg taken up by plants from the soil is beneficial from perspective of protecting food against contamination, although it also limits the possibility of using plants in the phytoremediation of contaminated soils. mainly in the oxidized form (Hg2+), and its transformations are associated primarily with the oxidation-reduction ological and chemical processes of methylation. On the land surface, Hg deposition is mainly in the oxidized form (Hg2+), and its transformations are associated primarily with the oxidation-reduction potential of the environment and the biological and chemical processes of methylation. On the land surface, Hg deposition is mainly in the oxidized form (Hg2+), and its transformations are associated primarily with the oxidation-reduction potential of the environment and the biological and chemical processes of methylation. The main source of Hg in plant leaves comes from air pollution with Hg0 and not from soil contamination. It is from plants, which mainly occur in the form of Hg0. The main source of Hg in plant leaves comes from air pollution with Hg0 and not from soil contamination. It is very difficult to estimate Hg emissions from plants, which mainly occur in the form of Hg0. The main source of Hg in plant leaves comes from air pollution with Hg0 and not from soil contamination. Conclusions And Commentary mospheric contamination by mercury continues to be one of the most important environmental problems in the m Atmospheric contamination by mercury continues to be one of the most important environmental problems in t conclusions can be drawn from this review of the literature and are accompanied by the authors' critical comme Atmospheric contamination by mercury continues to be one of the most important environmental problems in the modern world. The following general conclusions can be drawn from this review of the literature and are accompanied by the authors' critical commentary: ns can be drawn from this review of the literature and are accompanied by the authors' critical commentary: Models differ significantly in their estimations of global total Hg emissions—from 4,000 Mg yr-1 to 9,230 Mg yr-1—of which natural emissions and re- emissions ranged from 45% to 66%. Models differ significantly in their estimations of global total Hg emissions—from 4,000 Mg yr-1 to 9,230 Mg yr-1—of which natural emissions and re- emissions ranged from 45% to 66%. Many factors contribute to such large differences in the assessments of the level of global emissions: (i) methodological difficulties exist in assessing re- emission from heavily polluted areas under the influence of contemporary and historical emissions and areas with background pollution; (ii) the transformation of various forms of Hg depends on many difficult to evaluate processes, which makes estimating emissions difficult; and (iii) unusual phenomena associated with the transformation of various Hg forms (e.g., mercury depletion events (MDEs), which consist of episodes of sudden drops in total gaseous mercury concentrations in the air in the Antarctic and Arctic, can occur. Despite the many applicable global, regional and national programs and conventions aimed at reducing Hg emissions, global total Hg emissions in the years 2000-2015 increased by 1.8%. In many countries, including those with a high national income, there is a lack of understanding by society at large and politicians about the need to reduce emissions. Despite the many applicable global, regional and national programs and conventions aimed at reducing Hg emissions, global total Hg emissions in the years 2000-2015 increased by 1.8%. In many countries, including those with a high national income, there is a lack of understanding by society at large and politicians about the need to reduce emissions. The proportions between the amounts of natural emissions and anthropogenic emissions have not been precisely determined. Conclusions And Commentary The possibility of using contaminated soil in phytoremediation by transgenic plants is promising and a future research direction. A large number of scientific publications have been devoted to the problem of Hg environmental pollution. However, these studies face many difficulties: (i) analytical difficulties exist, which are caused by very low Hg contents in all elements of the environment; (ii) the need to determine a specific form of Hg prohibits providing results in total Hg; and (iii) the form in which Hg occurs depends on many environmental factors that must be accurately recognized and described. Unfortunately, many scientific publications do not meet these requirements. A large number of scientific publications have been devoted to the problem of Hg environmental pollution. However, these studies face many difficulties: (i) analytical difficulties exist, which are caused by very low Hg contents in all elements of the environment; (ii) the need to determine a specific form of Hg prohibits providing results in total Hg; and (iii) the form in which Hg occurs depends on many environmental factors that must be accurately recognized and described. Unfortunately, many scientific publications do not meet these requirements. Abbreviations Conclusions And Commentary It is very difficult to estimate Hg emissions from plants, which mainly occur in the form of Hg0. Methyl-Hg can be produced both biotically and abiotically in the environment. Methylation of Hg tends to occur in environments with low oxygen levels, low pH, Hg bioavailability, temperature, redox potential and high dissolved organic compound levels, and environments favored by SRB are largely responsible for this methylation. Methyl-Hg can be produced both biotically and abiotically in the environment. Methylation of Hg tends to occur in environments with low oxygen levels, low pH, Hg bioavailability, temperature, redox potential and high dissolved organic compound levels, and environments favored by SRB are largely responsible for this methylation. Rice growing conditions mean that the Hg methylation process is extremely intensive. Consequently, rice may contain significantly more Hg than other crops, which is particularly dangerous because rice in many regions of the world is the basis for feeding the population. Rice growing conditions mean that the Hg methylation process is extremely intensive. Consequently, rice may contain significantly more Hg than other crops, which is particularly dangerous because rice in many regions of the world is the basis for feeding the population. Rice growing conditions mean that the Hg methylation process is extremely intensive. Consequently, rice m crops, which is particularly dangerous because rice in many regions of the world is the basis for feeding the Hg has very limited solubility in soil and low availability for plant uptake, and it does not have any known biological function. These factors may explain why Hg-hyperaccumulating plants have not yet been identified, meaning that an effective phytoremediation methods for soil contaminated with Hg has not yet been developed, which may explain why Hg hyperaccumulator plants have not found practical use in phytoremediation of contaminated soils. Such work may be more applicable in wetland environments. The possibility of using contaminated soil in phytoremediation by transgenic plants is promising and a future research direction. Hg has very limited solubility in soil and low availability for plant uptake, and it does not have any known biological function. These factors may explain why Hg-hyperaccumulating plants have not yet been identified, meaning that an effective phytoremediation methods for soil contaminated with Hg has not yet been developed, which may explain why Hg hyperaccumulator plants have not found practical use in phytoremediation of contaminated soils. Such work may be more applicable in wetland environments. Consent for publication Not applicable. Not applicable; presented information is based on previously published data only. Not applicable; presented information is based on previously published data only. Funding The work was financed from the Own Research Fund Institute of Environment Protection - National Research Institute. Acknowledgements Not applicable. Competing interests The authors declare that they have no competing interests. Authors’ contributions WD has been responsible for the concept of the manuscript and drafted the manuscript. BG and AH B-D helped to further elaborate the manuscript. All authors improved the final manuscript. Author details 1 Institute of Environmental Protection – National Research Institute, Warsaw, Poland Abbreviations Page 11/24 Page 11/24 Hg – mercury Hg mercury UN - United Nations UNEP - United Nations Environment Programme GEM - gaseous elemental mercury Hg0 GOM - gaseous mercury in oxidized form TPM -total particulate mercury TGM - total gaseous Hg ROS - reactive oxygen species OH - hydroxyl radicals SOD - superoxide dismutase POD - peroxidase APX - ascorbate peroxides 7. Hylander LD, Meili M (2003) 500 years of mercury production: global annual inventory by region until 2000 and associated emissions. Sci Total Environ 304:13-27. https://doi.org/10.1016/S0048-9697(02)00553-3 7. Hylander LD, Meili M (2003) 500 years of mercury production: global annual inventory by region until 2000 and associated emissions. Sci Total Environ 304:13-27. https://doi.org/10.1016/S0048-9697(02)00553-3 8. WHO (2017) Ten chemicals of major health concern. Retrieved from www.who.int/ipcs/assessment/public_health/chemicals_phc/en/index.html 9. Minamata Convention on Mercury. (2020) Available via http://www.mercuryconvention.org/Portals/ 11/documents/conventionText/Minamata Minamata Convention on Mercury. (2020) Available via http://www.mercuryconvention.org/Portals/ 11/documen 10. Selin H, Keane SE, Wang S, Selin NE, Davis K, Bally D (2018) Linking science and policy to support the implementation of the Minamata Convention on Mercury. Ambio 47:198-215. https://doi.org/10.1007/s13280-017-1003-x 11. Streets DG, Horowitz HM, Lu Z, Levin L, Thackray CP, Sunderland EM (2019) Global and regional trends in mercury emissions and concentrations. 2010– 2015. Atmos Environ 201:417-427. https://doi.org/10.1016/j.atmosenv.2018.12.031 12. Obrist D, Kirk JL, Zhang L, Sunderland EM, Jiskra M, Selin NE (2018) A review of global environmental mercury processes in response to human and natural perturbations: Changes of emissions, climate, and land use. Ambio 47:116-140. https://doi.org/10.1007/s13280-017-1004-9 13. ICMGP 2019)14th International Conference on Mercury as a Global Pollutan, 8-13September 2019 Cracow, Poland, Available via http:// www.mercury2019krakow.com, 14. Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gøtzsche PC, Ioannidis JP, et al. (2009) The PRISMA statement for reporting systematic reviews and meta- analyses of studies that evaluate health care interventions: explanation and elaboration. J Clin Epidemiol 62:1-34. https://doi.org/10.1016/j.jclinepi.2009.06.006 15. Pirrone N, Cinnirella S, Feng X, Finkelman RB, Friedli HR, Leaner J, et al. (2010) Global mercury emissions to the atmosphere from anthropogenic and natural sources. Atmos Chem and Phys 10:5951-5964. https://doi.org/ 10.5194/acp-10-5951-2010 16. Pacyna EG, Pacyna JM, Sundsetha K, Munthec J, Kindbomc K, Wilsond S, et al. (2010) Global emission of mercury to the atmosphere from anthropogenic sources in 2005 and projections to 2020. Atmos Environ 44:2487–2499. https://doi.org/10.1016/j.atmosenv.2009.06.009 17. Gworek B, Dmuchowski W, Baczewska AH, Brągoszewska P, Bemowska-Kałabun O, Wrzosek-Jakubowska J (2017) Air Contamination by Mercury, Emissions and Transformations - a Review. Water Air Soil Poll 228:123. https://doi.org/10.1007/s11270-017-3311-y 8. UNEP Chemical Branch (2008) The global atmospheric mercury assessment: sources, emissions and transport. 18. UNEP Chemical Branch (2008) The global atmospheric mercury assessment: sources, emissions and transport. Geneva: UNEP-Chemicals. 19. Cooke CA, Martínez-Cortizas A, Bindler R, Gustin MS (2020) Environmental archives of atmospheric Hg deposition–A review. Sci Total Environ 709:134800. https://doi.org/10.1016/j.scitotenv.2019.134800 20. References Page 12/24 Page 12/24 1. Pacyna JM (2020) Recent advances in mercury research. Sci Total Environ 738:139955-139955. https://doi: 10.1016/j.scitotenv.2020.139955 2 UN Environment (2019) Global Mercury Assessment 2018 UN Environment Programme Chemical and Health Branch Geneva Switzerland 1. Pacyna JM (2020) Recent advances in mercury research. Sci Total Environ 738:139955-139955. https://doi: 10.1016/j.scitotenv.2020.139955 1. Pacyna JM (2020) Recent advances in mercury research. Sci Total Environ 738:139955-139955. https://doi: 10.1016/j.scitotenv.2020.139955 2. UN Environment (2019) Global Mercury Assessment 2018. UN Environment Programme, Chemical and Health Branch Geneva, Switzerland. 2. UN Environment (2019) Global Mercury Assessment 2018. UN Environment Programme, Chemical and Health 2. UN Environment (2019) Global Mercury Assessment 2018. UN Environment Programme, Chemical and Health Branch Geneva, Switzerland. 3. Bank MS (2020). The mercury science-policy interface: History, evolution and progress of the Minamata Convention. Sci Total Environ 722:137832. https://doi.org/10.1016/j.scitotenv.2020.137832 4. AMAP/UN Environment (2019) Technical Background Report for the Global Mercury Assessment 2018. Arctic Monitoring and Assessment Programme, Oslo, Norway/UN Environment Programme, Chemicals and Health Branch, Geneva, Switzerland. 5. Kabata-Pendias A, Pendias H (2001) Trace elements in soils and plants (3rd ed.). CRC Press, Boca Raton. https://doi.org/10.1201/9781420039900 5. Kabata-Pendias A, Pendias H (2001) Trace elements in soils and plants (3rd ed.). CRC Press, Boca Raton. http 5. Kabata-Pendias A, Pendias H (2001) Trace elements in soils and plants (3rd ed.). CRC Press, Boca Raton. https://doi.org/10.1201/9781420039900 6 Newman MC (2014) Fundamentals of ecotoxicology: the science of pollution 4th Edition CRC Press Boca Raton https://doi org/10 1201/b17658 6. Newman MC (2014) Fundamentals of ecotoxicology: the science of pollution. 4th Edition, CRC Press, Boca Raton. https://doi.org/10.1201/b17658 Newman MC (2014) Fundamentals of ecotoxicology: the science of pollution. 4th Edition, CRC Press, Boca Raton 7. Hylander LD, Meili M (2003) 500 years of mercury production: global annual inventory by region until 2000 and associated emissions. Sci Total Environ 304:13-27. https://doi.org/10.1016/S0048-9697(02)00553-3 Miszczak E, Stefaniak S, Michczyński A, Steinnes E, Twardowska I (2020) A novel approach to peatlands as archives of total cumulative spatial pollution loads from atmospheric deposition of airborne elements complementary to EMEP data: priority pollutants (Pb, Cd, Hg). Sci Total Environ 705:135776. https://doi.org/10.1016/j.scitotenv.2019.135776 21. Zhu C, Tian H, Hao J (2020) Global anthropogenic atmospheric emission inventory of twelve typical hazardous trace elements, 1995–2012.Atmos Environ 220:117061. https://doi.org/10.1016/j.atmosenv.2019.117061 22. Eckley C, Blanchard P, McLennan D, Mintz R, Sekela M (2015) Soil–air mercury flux near a large industrial emission source before and after closure (Flin Flon, Manitoba, Canada). . Environ Sci Techno, 49:9750-9757. https://doi.org/10.1021/acs.est.5b01995 23. Eckley CS, Gilmour CC, Janssen S, Luxton TP, Randall PM, Whalin L, et al. (2020) The assessment and remediation of mercury contaminated sites: A review of current approaches. Sci Total Environ 707:136031. https://doi.org/10.1016/j.scitotenv.2019.136031 Page 13/24 Page 13/24 Page 13/24 24. Agnan Y, Le Dantec T, Moore CW, Edwards GC, Obrist D (2016) New constraints on terrestrial surface–atmosphere fluxes of gaseous elemental mercury using a global database. Environ Sci Technol 50:507-524. https://doi.org/10.1021/acs.est.5b04013 25. Travnikov O, Dastoor A, Bullock R, Christensen JH (2008) Modeling atmospheric transport and deposition. in: AMAP/UNEP, Technical Background Report to the Global Atmospheric Mercury Assessment. Arctic Monitoring and Assessment Programme, UNEP Chemical Branch, pp. 79-107. 26. Dastoor AP, Larocque Y (2004) Global circulation of atmospheric mercury: a modelling study. Atmos Environ 38:147-161. https://doi.org/10.1016/j.atmosenv.2003.08.037 27. Henze DK, Hakami A, Seinfeld JH (2007) Development of the adjoint of GEOS-Chem. Atmos Chem Phys 79:2413-2433. https://resolver.caltech.edu/CaltechAUTHORS:HENacp07 28. Seigneur C, Vijayaraghavan K, Lohman K, Levin L (2009) The AER/EPRI global chemical transport model for mercury (CTM-HG). in: Mercury Fate and Transports in the Global Atmosphere, Pirrone N. and Mason R. (Eds.), Springer, New York, pp. 589-602. 29. Travnikov O, Ilyin I (2009) The EMEP/MSC-E mercury modeling system. in: Pirrone N, Mason R (Eds.), Mercury Fate and Transports in the Global Atmosphere, Springer, New York, pp. 571-587. https://doi.org/10.1007/978-0-387-93958-2_20 30. Travnikov O, Angot H, Artaxo P, Bencardino M, Bieser J, D'Amore F, et al. (2017). Multi-model study of mercury dispersion in the atmosphere: atmospheric processes and model evaluation. Atmos Chem Phys 5271-5295. https://doi.org/10.5194/acp-17-5271-201 30. Travnikov O, Angot H, Artaxo P, Bencardino M, Bieser J, D'Amore F, et al. (2017). Multi-model study of mercury dispersion in the atmosphere: atmospheric processes and model evaluation. Atmos Chem Phys 5271-5295. https://doi.org/10.5194/acp-17-5271-201 31. De Simone F, Gencarelli CN, Hedgecock IM, Pirrone N (2014) Global atmospheric cycle of mercury: a model study on the impact of oxidation mechanisms. Environ Sci Pollut R 21:4110-4123. https://doi.org/10.1007/s11356-013-2451-x. 32. Nriagu JO, Pacyna JM (1998) Quantitative assessment of worldwide contamination of air, water and soils by trace metals. Nature 333:134–139. https://doi org/10 1038/333134a0 32. Nriagu JO, Pacyna JM (1998) Quantitative assessment of worldwide contamination of air, water and soils by trace metals. Nature 333:134–139. https://doi.org/10.1038/333134a0 33. Shetty SK, Lin CJ, Streets DG, Jang C (2008) Model estimate of mercury emission from natural sources in East Asia. Atmos Environ 42:8674–8685. https://doi.org/10.1016/j.atmosenv.2008.08.026 34. Liu G, Cai Y, O’Driscoll N, Feng X, Jiang G. (2012) Overview of Mercury in the Environment. in: Liu G, Cai Y, O’Driscoll N (Eds.), Environmental Chemistry and Toxicology of Mercury. J. Wiley & Sons, Inc., Hoboken, USA, pp. 1-12. 35. Gustin MS, Lindberg SE, Austin K, Coolbaugh M, Vette A, Zhang H (2000) Assessing the contribution of natural sources to regional atmospheric mercury budgets. Sci Total Environ 259:61–71. https://doi.org/10.1016/S0048-9697(00)00556-8 36. Gustin MS, Lindberg SE, Weisberg PJ (2008) An update on the natural sources and sinks of atmospheric mercury. Appl Geoch23:482–493. https://doi.org/10.1016/j.apgeochem.2007.12.010 37. Streets DG, Horowitz HM, Jacob DJ, Lu Z, Levin L, Ter Schure AF, et al. (2017) Total mercury released to the environment by human activities.Environ Sci Technol 51:5969-5977. https://doi.org/10.1021/acs.est.7b00451 38. Mason RP (2009) Mercury Emissions from Natural Processes and their Importance in the Global Mercury Cycle. in: Pirrone N, Mason R (Eds.), Mercury Fate and Transports in the Global Atmosphere, Springer, New York, pp. 173-191. https://doi.org/10.1007/978-0-387-93958-2_7 39. Wang Y, Greger M (2004) Clonal differences in mercury tolerance, accumulation, and distribution in willow. J Environ Qual 33:1779-1785. https://doi.org/10.2134/jeq2004.1779 40. Bergan T, Gallardo L, Rohde H (1999) Mercury in the global troposphere - a three-dimensional model study. Atmos Environ 33:1575-1585. https://doi.org/10.1016/S1352-2310(98)00370-7 41. Mason RP, Sheu GR (2002) Role of the ocean in the global mercury cycle. Global Biogeoch Cy 16:1093. https://doi.org/10.1029/2001GB001440. 42. Lamborg CH, Fitzgerald WF, O’Donnell J, Torgersen T (2002) A non-steady-state compartmental model of global-scale mercury biogeochemistry with interhemispheric atmospheric gradients. Geochim Cosmochim Ac66:1105-1118. https://doi.org/10.1016/S0016-7037(01)00841-9 43. Nriagu J, Becker C (2003) Volcanic emissions of mercury to the atmosphere: global and regional inventories Sci Total Environ 304:3-12. https://doi.org/10.1016/S0048-9697(02)00552-1 44. Pyle DM Macher RA (2003) The importance of volcanic emissions for the global atmospheric mercury cycle. Atmos Environ 37:5115-5124. 30. Travnikov O, Angot H, Artaxo P, Bencardino M, Bieser J, D'Amore F, et al. (2017). Multi-model study of mercury dispersion in the atmosphere: atmospheric processes and model evaluation. Atmos Chem Phys 5271-5295. https://doi.org/10.5194/acp-17-5271-201 https://doi.org/10.1016/j.atmosenv.2003.07.011 Page 14/24 Page 14/24 Page 14/24 45. Bagnato E, Aiuppa A, Parello F, Allard P, Shinohara H, Liuzzo M, et al. (2011) New clues on the contribution of Earth’s volcanism to the global mercury cycle. B Volcanol 73:497-510. https://doi.1007/s00445-010-0419- 46. Bagnato E, Aiuppa A, Parello F, Calabrese S, D’Alessandro W, Mather TA, et al. (2007) Degassing of gaseous (elemental and reactive) and particulate mercury from Mount Etna volcano (Southern Italy). Atmos Environ 41:377-7388. https://doi.org/10.1016/j.atmosenv.2007.05.060 47. Witt MLI, Mather TA, Pyle DM, Aiuppa A, Bagnato E, Tsanev VI (2008) Mercury and halogen emissions from Masaya and Telica volcanoes, Nicaragua. J Geophys Res Solid Earth 113(B6). https://doi.org/10.1029/2007JB005401 48. Martin RS, Witt MLI Pyle DM, Mather TA, Watt SFL, Bagnato E. et al. (2011) Rapid oxidation of mercury (Hg) at volcanic vents: Insights from high temperature thermodynamic models of Mt Etna's emissions. Chem Geol 283:279-286. https://doi.org/10.1016/j.chemgeo.2011.01.027 49. Ermolin MS, Fedotov PS, Malik NA, Karandashev VK (2018) Nanoparticles of volcanic ash as a carrier for toxic elements on the global scale. Chemosphere 200:16-22. https://doi.org/10.1016/j.chemosphere.2018.02.089 50. Coolbaugh M, Gustin M, Rytuba J (2002) Annual emissions of mercury to the atmosphere from natural sources in Nevada and California. Environ Geol 42:338-349. https://doi 10.1007/s00254-002-0557-4. 51. Cinnirella S, Pirrone N (2006) Spatial and temporal distributions of mercury emissions from forest fires in Mediterranean region and Russian federation. Atmos Environ 40:7346-7361. https://doi.org/10.1016/j.atmosenv.2006.06.051 52. Friedli HR, Arellano AF, Cinnirella S, Pirrone N (2009) Mercury emissions from global biomass burning: spatialand temporal distribution. in: Mason R., Pirrone N. (Eds), Mercury Fate and Transport in the Global Atmosphere. Springer, Boston, MA. https://doi.org/10.1007/978-0-387-93958-2_8 53. Obrist D, Gustin MS, Arnone III JA, Johnson DW, Schorran DE, Verburg PS (2005) Measurements of gaseous elemental mercury fluxes over intact tallgrass prairie monoliths during one full year. Atmos. Environ., 39, 957-965. https://doi.org/10.1016/j.atmosenv.2004.09.081 54. Woodruff LG, Cannon WF (2010) Immediate and long-term fire effects on total mercury in forests soils of northeastern Minnesota. Environ Sci Technol 44:5371-5376. https://doi.org/10.1021/es100544d 54. Woodruff LG, Cannon WF (2010) Immediate and long-term fire effects on total mercury in forests soils of northeastern Minnesota. Environ Sci Technol 44:5371-5376. https://doi.org/10.1021/es100544d 55. Gagliano AL, Calabrese S, Daskalopoulou K, Cabassi J, Capecchiacci F, Tassi F, et al. (2019) Degassing and Cycling of Mercury at Nisyros Volcano (Greece). Geofluids, 2019, ID4783513. https://doi.org/10.1155/2019/4783514 56. Pierce AM, Moore CW, Wohlfahrt G, Hörtnagl L, Kljun N, Obrist D (2015) Eddy covariance flux measurements of gaseous elemental mercury using cavity ring-down spectroscopy. Environ Sci Technol 49:1559-1568. https://doi.org/10.1021/es505080z 56. Pierce AM, Moore CW, Wohlfahrt G, Hörtnagl L, Kljun N, Obrist D (2015) Eddy covariance flux measurements of gaseous elemental mercury using cavity ring-down spectroscopy. Environ Sci Technol 49:1559-1568. https://doi.org/10.1021/es505080z 57. Obrist D, Johnson DW, Lindberg SE (2009) Mercury concentrations and pools in four Sierra Nevada forest sites, and relationships to organic carbon and nitrogen. Biogeosciences 6:765-777. https://doi.org/10.1016/j.atmosenv.2004.09.081 57. Obrist D, Johnson DW, Lindberg SE (2009) Mercury concentrations and pools in four Sierra Nevada forest sites, and relationships to organic carbon and nitrogen. Biogeosciences 6:765-777. https://doi.org/10.1016/j.atmosenv.2004.09.081 58. Jiskra M, Sonke JE, Obrist D, Bieser J, Ebinghaus R, Myhre CL et al. (2018) A vegetation control on seasonal variations in global atmospheric mercury concentrations. Nat Geosci 11:244-250. https://doi.org/10.1038/s41561-018-0078-8 59. Gustin MS (2012) Exchange of mercury between the atmosphere and terrestrial ecosystems. in: Liu, G., Cai, Y., O'Driscoll, N. (Eds.), Environmental chemistry and toxicology of mercury. John Wiley & Sons, pp. 423-451 60. Smith-Downey NV, Sunderland EM, Jacob DJ (2010) Anthropogenic impacts on global storage and emissions of mercury from terrestrial soils: Insights from a new global model. J Geophys Res: Biogeosci 115(G3). https://doi.org/10.1029/2009JG001124 61. Graydon JA, St. Louis VL, Lindberg SE, Sandilands KA, Rudd JW, Kelly CA et al. (2012) The role of terrestrial vegetation in atmospheric Hg deposition: Pools and fluxes of spike and ambient Hg from the METAALICUS experiment. Global Biogeochem Cy 26:GB1022. https://doi.org/10.1029/2011GB004031 62. Zheng W, Obrist D, Weis D, Bergquist BA (2016) Mercury isotope compositions across North American forests. Global Biogeoch Cy 30:1475-1492. https://doi.org/10.1002/2015GB005323 63. Sommar J, Osterwalder S, Zhu W (2020) Recent advances in understanding and measurement of Hg in the environment: Surface-atmosphere exchange of gaseous elemental mercury (Hg0). Sci Total Environ 137648. https://doi.org/10.1016/j.scitotenv.2020.137648 Page 15/24 64. Wright LP, Zhang L, Marsik FJ (2016) Overview of mercury dry deposition, litterfall, and throughfall studies. Atmos Chem Phys 16:13399. https://doi.10.5194/acp-16-13399-2016 Page 15/24 Page 15/24 65. Lyman SN, Cheng I, Gratz LE, Weiss-Penzias P, Zhang L (2020) An updated review of atmospheric mercury. Sci Total Environ 707:135575. https://doi.org/10.1016/j.scitotenv.2019.135575 66. Skov H, Bullock R, Christensen JH, Sørensen LL (2008). Atmospheric pathways. in: AMAP/UNEP, Technical Background Report to the Global Atmospheric Mercury Assessment. Arctic Monitoring and Assessment Programme, UNEP Chemical Branch, pp. 64-72. 67. Selin NE, Jacob DJ, Park RJ., Yantosca RM, Strode S, Jaeglé L, et al. (2007) Chemical cycling and deposition of atmospheric mercury: Global constraints from observations, J Geophys Res 112:D02308, 14 PP. https://doi.org/10.1029/2006JD007450 68. Lindberg SE, Dong W, Meyers T. (2002) Transpiration of gaseous elemental mercury through vegetation in a subtropical wetland in Florida. Atmos Environ, 36:5207-5219. https://doi.org/10.1016/S1352-2310(02)00586-1 69. Marsik FJ, Keeler GJ, Landis MS (2007) The dry-deposition of speciated mercury to the Florida Everglades: Measurements and modeling. Atmos Environ 41:136-149. https://doi.org/10.1016/j.atmosenv.2006.07.032 70. 57. Obrist D, Johnson DW, Lindberg SE (2009) Mercury concentrations and pools in four Sierra Nevada forest sites, and relationships to organic carbon and nitrogen. Biogeosciences 6:765-777. https://doi.org/10.1016/j.atmosenv.2004.09.081 Hanson PJ, Lindberg SE, Tabberer TA, Owens JA, Kim KH (1995) Foliar exchange of mercury vapor: evidence for a compensation point. Water Air Soil Poll 80:373-382. https://doi.org/10.1007/BF01189687 71. Caffrey PF, Ondov JM, Zufall MJ, Davidson CI (1998) Determination of size-dependent dry particle deposition velocities with multiple intrinsic elemental tracers. Environ Sci Technol32:1615-1622. https://doi.org/10.1021/es970644f 72. Converse AD, Riscassi AL, Scanlon TM (2010) Seasonal variability in gaseous mercury fluxes measured in a high-elevation meadow. Atmos Environ 44:2176-2185. https://doi.org/10.1016/j.atmosenv.2010.03.024 73. Carpi A, Fostier AH, Orta OR, dos Santos JC, Gittings M (2014.) Gaseous mercury Carpemissions from soil following forest loss and land use changes: Field experiments in the United States and Brazil. Atmos Environ 96:423-429. https://doi.org/10.1016/j.atmosenv.2014.08.004 74. Zhang H, Lindberg SE, Marsik FJ, Keeler GJ (2001) Mercury air/surface exchange kinetics of background soils of the Tahquamenon River watershed in the Michigan Upper Peninsula. Water Air Soil Poll 126:151-169. 75. Ferrari CP, Dommerguea A, Veysseyrea A, Planchona F, Boutrona CF (2002) Mercury speciation in the French seasonal snow cover. Sci Total Environ 287, 61-69. https://doi.org/10.1016/S0048-9697(01)00999-8 76. Gustin MS, Biester H, Kim CS (2002) Investigation of the light-enhanced emission of mercury from naturally enriched substrates. Atmos Environ 36:3241- 3254. https://doi.org/10.1016/S1352-2310(02)00329-1 77. Poissant L, Pilote M, Xu X, Zhang H, Beauvais C (2004) Atmospheric mercury speciation and deposition in the Bay St. Francois wetlands. J Geophys Res 109:1-14. https://doi.org/10.1029/2003JD004364 78. Ericksen JA, Gustin MS, Xin M, Weisberg PJ, Fernandez GCJ (2006) Air–soil exchange of mercury from background soils in the United States. Sci Total Environ 366:851-863. https://doi.org/10.1016/j.scitotenv.2005.08.019 79. Gustin MS (2003) Are mercury emissions from geologic sources significant? A status report. Sci. Total Environ., 304, 153-167. https://doi.org/10.1016/S0048-9697(02)00565-X 80. Gustin MS, Lindberg SE (2005) Terrestial Hg Fluxes: Is the Next Exchange Up, Down, or Neither?. in: Pirrone, N., Mahaffey, K.R. (Eds.), Dynamics of Mercury Pollution on Regional and Global Scales, Springer, Boston, MA, pp. 241-259. https://doi.org/10.1007/0-387-24494-8_11 81. Stamenkovic J, Gustin MS (2007) Evaluation of use of EcoCELL technology for quantifying total gaseous mercury fluxes over background substrates. Atmos Environ 41:3702-3712. https://doi.org/10.1016/j.atmosenv.2006.12.037 82. Mason RP (2009) Mercury Emissions from Natural Processes and their Importance in the Global Mercury Cycle. in: Pirrone N, Mason R (Eds.), Mercury Fate and Transports in the Global Atmosphere, Springer, New York, pp.173-191. https://doi.org/10.1007/978-0-387-93958-2_7 83. Mason RP, Pirrone N, Hedgecock I, Suzuki N, Levin L (2010) Conceptual overview. in:Pirrone N., Keating T. 57. Obrist D, Johnson DW, Lindberg SE (2009) Mercury concentrations and pools in four Sierra Nevada forest sites, and relationships to organic carbon and nitrogen. Biogeosciences 6:765-777. https://doi.org/10.1016/j.atmosenv.2004.09.081 (Eds.), Hemispheric transport of air pollution -part B., United Nations Publication, New York, Geneva, pp. 1-19. https://doi.org/10.18356/38ccc958-en 84. Carpi A, Lindberg SE (1997) Sunlight-Mediated Emission of Elemental Mercury from Soil Amended with Municipal Sewage Sludge. Environ Sci Technol31:2085-2091. https://doi.org/10.1021/Es960910+ 85. Zhu W, Lin CJ, Wang X, Sommar J, Fu X, Feng X (2016) Global observations and modeling of atmosphere-surface exchange of elemental mercury: a critical review. Atmos Chem Phys 16:4451-4480. https://doi.org/10.5194/acp-16-4451-2016 86. Hanson PJ, Lindberg SE, Tabberer TA, Owens JA, Kim KH (1995) Foliar exchange of mercury vapor: evidence for a compensation point. Water Air Soil Poll 80:373-382. https://doi.org/10.1007/BF01189687 Page 16/24 87. Bash JO, Miller DR, Meyer TH, Bresnahan PA (2004) Northeast United States and Southeast Canada natural mercury emissions estimated with a surface emission model. Atmos Environ 38:5683-5692. https://doi.org/10.1016/j.atmosenv.2004.05.058 88. Rea AW, Lindberg SE, Scherbatskoy T, Keeler GJ (2002). Mercury accumulation in foliage over time in two northern mixed-hardwood forests. Water Air Soil Pollut 133:49-67. https://doi.org/10.1023/A:1012919731598 89. Magarelli G, Fostier AH (2005) Influence of deforestation on the mercury air/soil exchange in the Negro River Basin, Amazon. Atmos Environ 39:7518-7528. https://doi.org/10.1016/j.atmosenv.2005.07.067 90. Selin MS, Lindberg SE (2005) Terrestial Hg Fluxes: Is the Next Exchange Up, Down, or Neither?. in: Pirrone, N., Mahaffey, K.R. (Eds.), Dynamics of Mercury Pollution on Regional and Global Scales, Springer, Boston, MA, pp. 241-259. https://doi.org/10.1007/0-387-24494-8_11 91. Gustin MS, Engle M, Ericksen J, Lyman S, Stamenkovic J, Xin M (2006) Mercury exchange between the atmosphere and low mercury containing substrates. Appl Geoch 21:1913-1923. https://doi.org/10.1016/j.apgeochem.2006.08.007 92. Selin NE, Jacob DJ, Yantosca RM, Strode S, Jaeglé L, Sunderland EM (2008) Global 3D land, ocean, atmosphere model for mercury: Present day versus preindustrial cycles and anthropogenic enrichment factors for deposition. Global Biogeoch Cy 22:1-13. https://doi.org/10.1029/2007GB003040 93. Ericksen JA, Gustin MS, Schorran DE, Johnson DW, Lindberg SE, Coleman JS (2003) Accumulation of atmospheric mercury in forest foliage. Atmos Environ 37:1613-1622. https://doi.org/10.1016/S1352-2310(03)00008-6 94. Osterwalder S, Huang JH, Shetaya WH, Agnan Y, Frossard A, Frey, B et al. (2019) Mercury emission from industrially contaminated soils in relation to chemical, microbial, and meteorological factors. Environ. Pollut., 250, 944-952. https://doi.org/10.1016/j.envpol.2019.03.093 95. Gabriel MC, Williamson DG, Brooks S, Zhang H, Lindberg S (2005) Spatial variability of mercury emissions from soils in a southeastern US urban environment. Environ Geol 48:955-964. https://doi.org/10.1007/s00254-005-0043-x 96. 57. Obrist D, Johnson DW, Lindberg SE (2009) Mercury concentrations and pools in four Sierra Nevada forest sites, and relationships to organic carbon and nitrogen. Biogeosciences 6:765-777. https://doi.org/10.1016/j.atmosenv.2004.09.081 Ma M, Wang D, Sun R, Shen Y, Huang L (2013) Gaseous mercury emissions from subtropical forested and open field soils in a national nature reserve, southwest China Atmos Environ 64:116-123. https://doi.org/10.1016/j.atmosenv.2012.09.038 97. Wang X, Lin CJ, Yuan W, Sommar J, Zhu W, Feng X (2016) Emission-dominated gas exchange of elemental mercury vapor over natural surfaces in China. Atmos Chem Phys 16:11125–11143. doi:10.5194/acp-16-11125-2016 98. Sommar J, Zhu W, Shang L, Lin CJ, Feng XB (2015) Seasonal variations in metallic mercury (Hg 0) vapor exchange over biannual wheat-corn rotation cropland in the North China Plain. Biogeosciences Discuss 12:16105–16158. doi:10.5194/bgd-12-16105-2015 99. Gao Y, Wang Z, Zhang X, Wang C (2020) Observation and estimation of mercury exchange fluxes from soil under different crop cultivars and planting densities in North China Plain. Environ Pollut 259:113833. https://doi.org/10.1016/j.envpol.2019.113833 100. Bishop K, Shanley JB, Riscassi A, de Wit HA, Eklöf K, Meng, B, et al. (2020) Recent advances in understanding and measurement of mercury in the environment: Terrestrial Hg cycling. Sci Total Environ 721:137647. https://doi.org/10.1016/j.scitotenv.2020.137647 101. Qiu G, Feng X, Wang S, Shang L (2006) Environmental contamination of mercury from Hg-mining areas in Wuchuan, northeastern Guizhou, China. Environ Pollut 142:549-558. https://doi.org/10.1016/j.envpol.2005.10.015 102. Li Q, Tang L, Qiu G, Liu C (2020) Total mercury and methylmercury in the soil and vegetation of a riparian zone along a mercury-impacted reservoir. Sci Total Environ 738:139794. https://doi.org/10.1016/j.scitotenv.2020.139794 103. Goulet RR, Holmes J, Tessier A, Wang F, Siciliano SD, Page B et al. (2007) Mercury methylation in sediments of a riverine marsh: the role of redox conditions sulfur chemistry and microbial communities. Geochim Cosmochim Ac.71:3396-3406. 104. Windham-Myers L, Marvin-DiPasquale M, Kakouros E, Agee JL, Kieu LH, Stricker CA, et al. (2014) Mercury cycling in agricultural and managed wetlands of California, USA: Seasonal influences of vegetation on mercury methylation, storage, and transport. Sci Total Environ 484:308-318. https://doi.org/10.1016/j.scitotenv.2013.05.027 105. Kronberg RM, Jiskra M, Wiederhold JG, Bjorn E, Skyllberg U (2016) Methyl mercury formation in hillslope soils of boreal forests: The role of forest harvest and anaerobic microbes. Environ Sci Technol 50:9177-9186. https://doi.org/10.1021/acs.est.6b00762 106. Gnamuš A, Byrne AR, Horvat M (2000) Mercury in the soil-plant-deer-predator food chain of a temperate forest in Slovenia. Environ Sci Technol 34:3337- 3345. https://doi.org/10.1021/es991419w 106. Gnamuš A, Byrne AR, Horvat M (2000) Mercury in the soil-plant-deer-predator food chain of a temperate forest in Slovenia. Environ Sci Technol 34:3337- 3345. https://doi.org/10.1021/es991419w 107. 109. Lima FRD, Martins GC, Silva AO, Vasques ICF, Engelhardt MM, Cândido GS, et al. (2019) Critical mercury concentration in tropical soils: Impact on plants and soil biological attributes. Sci Total Environ 666:472-479. https://doi.org/10.1016/j.scitotenv.2019.02.216 109. Lima FRD, Martins GC, Silva AO, Vasques ICF, Engelhardt MM, Cândido GS, et al. (2019) Critical mercury concentration in tropical soils: Impact on plants and soil biological attributes. Sci Total Environ 666:472-479. https://doi.org/10.1016/j.scitotenv.2019.02.216 110. Gonzalez-Raymat H, Liu G, Liriano C, Li Y, Yin Y, Shi J et al. (2017) Elemental mercury: Its unique properties affect its behavior and fate in the environment. Environ Pollut 229 :69-86. https://doi.org/10.1016/j.envpol.2017.04.101 111. Xu J, Buck M, Eklöf K, Ahmed OO, Schaefer JK, Bishop K, et al. (2019) Mercury methylating microbial communities of boreal forest soils. Sci Rep 9:518. https://doi.org/10.1038/s41598-018-37383-z 112. Qianqian LING, Yingying GUO, Yong LIANG, Yongguang YIN, Yong CAI (2020) Microbial uptake of HgS nanoparticles and its effect on mercury methylation. Environ Chem 2:292-300. https://doi:10.7524/j.issn.0254-6108.2019112601 113. Bigham GN, Murray KJ, Masue-Slowey Y, Henry EA (2017) Biogeochemical controls on methylmercury in soils and sediments: Implications for site management. Integr Environ Assess Manag 13:249-263. https://doi.org/10.1002/ieam.1822 114. Shahid M, Khalid S, Bibi I, Bundschuh J, Niazi NK, Dumat C (2020) A critical review of mercury speciation, bioavailability, toxicity and detoxification in soil-plant environment: Ecotoxicology and health risk assessment. Sci Total Environ 711:134749. https://doi.org/10.1016/j.scitotenv.2019.134749 115. Zhao L, Qiu G, Anderson CW, Meng B, Wang D, Shang L, et al. (2016) Mercury methylation in rice paddies and its possible controlling factors in the Hg mining area, Guizhou province, Southwest China. Environ Pollut 215:1-9. https://doi.org/10.1016/j.envpol.2016.05.001 116. Eklöf K, Bishop K, Bertilsson S, Björn E, Buck M, Skyllberg U, et al. (2018) Formation of mercury methylation hotspots as a consequence of forestry operations. Sci Total Environ 613:1069-1078. https://doi.org/10.1016/j.scitotenv.2017.09.151 117. Qiu G, Feng X, Wang S, Shang L (2005) Mercury and methylmercury in riparian soil, sediments, mine-waste calcines, and moss from abandoned Hg mines in east Guizhou province, southwestern China. Appl Geoch 20:627-638. https://doi.org/10.1016/j.apgeochem.2004.09.006 118. Qiu G, Feng X, Wang S, Shang L (2006) Environmental contamination of mercury from Hg-mining areas in Wuchuan, northeastern Guizhou, China. Environ Pollut 142:549-558. https://doi.org/10.1016/j.envpol.2005.10.015 119. Xu X, Meng B, Zhang C, Feng X, Gu C, Guo J, et al. (2017) The local impact of a coal-fired power plant on inorganic mercury and methyl-mercury distribution in rice (Oryza sativa L.). Environ Pollut 223:11-18. https://doi.org/10.1016/j.envpol.2016.11.042 120. Mailman M, Bodaly RA (2005) Total mercury, methyl mercury, and carbon in fresh and burned plants and soil in Northwestern Ontario. Environ Pollut 138:161-166. https://doi.org/10.1016/j.envpol.2005.02.005 121. Schuster PF, Schaefer KM, Aiken GR Antweiler RC, Dewild JF., Gryziec JD, et al. 108. Gilli R, Karlen C, Weber M, Rüegg J, Barmettler K, Biester H, et al. (2018) Speciation and mobility of mercury in soils contaminated by legacy emissions from a chemical factory in the Rhône valley in canton of Valais, Switzerland. Soil Systems, 2:44. https://doi.org/10.3390/soilsystems2030044 108. Gilli R, Karlen C, Weber M, Rüegg J, Barmettler K, Biester H, et al. (2018) Speciation and mobility of mercury in soils contaminated by legacy emissions from a chemical factory in the Rhône valley in canton of Valais, Switzerland. Soil Systems, 2:44. https://doi.org/10.3390/soilsystems2030044 57. Obrist D, Johnson DW, Lindberg SE (2009) Mercury concentrations and pools in four Sierra Nevada forest sites, and relationships to organic carbon and nitrogen. Biogeosciences 6:765-777. https://doi.org/10.1016/j.atmosenv.2004.09.081 Meng B, Feng X, Qiu G, Cai Y, Wang D, Li P, et al. (2010) Distribution patterns of inorganic mercury and methylmercury in tissues of rice (Oryza sativa L.) plants and possible bioaccumulation pathways. J Agr Food Chem 58:4951-4958. https://doi.org/10.1021/jf904557x 107. Meng B, Feng X, Qiu G, Cai Y, Wang D, Li P, et al. (2010) Distribution patterns of inorganic mercury and methylmercury in tissues of rice (Oryza sativa L.) plants and possible bioaccumulation pathways. J Agr Food Chem 58:4951-4958. https://doi.org/10.1021/jf904557x Page 17/24 Page 17/24 109. Lima FRD, Martins GC, Silva AO, Vasques ICF, Engelhardt MM, Cândido GS, et al. (2019) Critical mercury concentration in tropical soils: Impact on plants and soil biological attributes. Sci Total Environ 666:472-479. https://doi.org/10.1016/j.scitotenv.2019.02.216 (2018) Permafrost stores a globally significant amount of mercury. Geophys Res Lett 45:1463-1471. https://doi.org/10.1002/2017GL075571 122. Hararuk O, Obrist D, Luo Y (2013) Modelling the sensitivity of soil mercury storage to climate-induced changes in soil carbon pools. Biogeosciences 10:2393-2407. https://doi.org/10.5194/bg-10-2393-2013 123. Amos HM, Sonke JE, Obrist D, Robins N, Hagan N, Horowitz HM, Mason RP∇, et al. (2015) Observational and modeling constraints on global anthropogenic enrichment of mercury. Environ Sci Technol 49:4036-4047. https://doi.org/10.1021/es5058665 124. Mason RP, Choi AL, Fitzgerald WF, Hammerschmidt CR, Lamborg CH, Soerensen AL, et al. (2012) Mercury biogeochemical cycling in the ocean and policy implications. Environ Res 119:101-117. https://doi.org/10.1016/j.envres.2012.03.013 125. Fitzgerald WF, Lamborg CH (2014) Geochemistry of mercury in the environment. In: Treatise on Geochemis 126. Amos HM, Jacob DJ, Streets DG, Sunderland EM (2013) Legacy impacts of all‐time anthropogenic emissions on the global mercury cycle. Global Biogeoch Cy 27: 410-421. https://doi.org/10.1002/gbc.20040 127. Wang J, Feng X, Anderson CW, Xing Y, Shang L (2012) Remediation of mercury contaminated sites–a review. J Hazard Mater 221:1-18. https://doi.org/10.1016/j.jhazmat.2012.04.035 128. Li S, Jia Z (2018) Heavy metals in soils from a representative rapidly developing megacity (SW China): Levels, source identification and apportionment. Catena 163:414-423. https://doi.org/10.1016/j.catena.2017.12.035 Page 18/24 129. Obrist D., Pearson C, Webster J, Kane T, Lin CJ, Aiken GR, et al. (2016) A synthesis of terrestrial mercury in the western United States: Spatial distribution defined by land cover and plant productivity. Sci Total Environ 568:522-535. https://doi.org/10.1016/j.scitotenv.2015.11.104 130. Tóth G, Hermann T, Szatmári G, Pásztor L (2016) Maps of heavy metals in the soils of the European Union and proposed priority areas for detailed assessment. Sci Total Environ 565:1054-1062. https://doi.org/10.1016/j.scitotenv.2016.05.115 131. Lima LC, Egreja Filho FB, Linhares LA, Windmoller CC, Yoshida MI (2015) Accumulation and oxidation of elemental mercury in tropical soils. Chemosphere 134:181-191. https://doi.org/10.1016/j.chemosphere.2015.04.020 132. Sharma R, Ramteke S, Patel KS, Kumar S, Sarangi B, Agrawal SG, et al. (2015) Contamination of Lead and Mercury in Coal Basin of India. J Environ Prot 6:1430-1441. https//doi:10.4236/jep.2015.612124 133.Taylor M (2014) Soil quality monitoring in the Waikato Region 2012. WRC Technical Report 2013/49, Waikato Regional Council, Hamilton, New Zealand. http://www. waikatoregion. govt. nz/tr201349/. Accessed 3 Sept, 2014 134.Halbach K, Mikkelsen Ø, Berg T, Steinnes E (2017) The presence of mercury and other trace metals in surface soils in the Norwegian Arctic. Chemosphere 188:567-574. https://doi.org/10.1016/j.chemosphere.2017.09.012 135. Gasperi J, Ayrault S, Moreau-Guigon E, Alliot F, Labadie P, Budzinski H, et al. 109. Lima FRD, Martins GC, Silva AO, Vasques ICF, Engelhardt MM, Cândido GS, et al. (2019) Critical mercury concentration in tropical soils: Impact on plants and soil biological attributes. Sci Total Environ 666:472-479. https://doi.org/10.1016/j.scitotenv.2019.02.216 (2016) Contamination of soils by metals and organic micropollutants: case study of the Parisian conurbation. Environ Sci Pollut Res 25:23559–23573. https://doi.org/10.1007/s11356-016-8005-2 136. Bailey EA, Gray JE, Theodorakos PM (2002) Mercury in vegetation and soils at abandoned mercury mines in southwestern Alaska. USA. Geochem Explor Env A 2:275-285. https://doi.org/10.1144/1467-787302-032 137. Gray JE, Theodorakos PM, Fey DL, Krabbenhoft DP (2015) Mercury concentrations and distribution in soil, water, mine waste leachates, and air in and around mercury mines in the Big Bend region, Texas, Environ Geochem Hlth. 37:35-48. https://doi.org/10.1007/s10653-014-9628-1 138. Wu G, Kang H, Zhang X, Shao H, Chu L, Ruan C (2010) A critical review on the bio-removal of hazardous heavy metals from contaminated soils: issues, progress, eco-environmental concerns and opportunities. J Hazard Mater 174:1-8. https://doi.org/10.1016/j.jhazmat.2009.09.113 139. Gosar M, Šajn R, Biester H (2006) Binding of mercury in soils and attic dust in the Idrija mercury mine area (Slovenia). Sci Total Environ 369:150-162. https://doi.org/10.1016/j.scitotenv.2006.05.006 140. Teršič T, Gosar M, Šajn R (2009) Impact of mining activities on soils and sediments at the historical mining area in Podljubelj, NW Slovenia. J Geochem Explor 100:1-10. https://doi.org/10.1016/j.gexplo.2008.02.005 141. Banásová V (1999) Vegetation on contaminated sites near an Hg mine and smelter. in: R. Ebinghaus R, Turner RR, de Lacedra LD, Vasiljev O., Salomons W (Eds.), Mercury Contaminated Sites. Springer, Berlin, Heidelberg, pp. 321-335. https://doi.org/10.1007/978-3-662-03754-6 142. Higueras P, Oyarzun R, Biester H, Lillo J, Lorenzo S (2003) A first insight into mercury distribution and speciation in soils from the Almadén mining district, Spain. J Geochem Explor 80(1):95-104. https://doi.org/10.1016/S0375-6742(03)00185-7 143. Dago A, González I, Ariño C, Martínez-Coronado A, Higueras P, Díaz-Cruz JM, et al. (2014) Evaluation of mercury stress in plants from the Almadén mining district by analysis of phytochelatins and their Hg complexes. Environ Sci Technol 8:6256-6263. https://doi.org/10.1021/es405619y 144. Boente C, Albuquerque MTD, Gerassis S, Rodríguez-Valdés E, Gallego JR (2019) A coupled multivariate statistics, geostatistical and machine-learning approach to address soil pollution in a prototypical Hg-mining site in a natural reserve. Chemosphere 218:767-777. https://doi.org/10.1016/j.chemosphere.2018.11.172 145. Wahsha M, Maleci L, Bini C (2019) The impact of former mining activity on soils and plants in the vicinity of an old mercury mine (Vallalta, Belluno, NE Italy). Geochem-Explor Envir Anal 19:171-175. https://doi.org/10.1144/geochem2018-040 146. Gemici Ü, Tarcan G, Somay AM, Akar T (2009) Factors controlling the element distribution in farming soils and water around the abandoned Halıköy mercury mine (Beydağ, Turkey). Appl Geoch 24:1908-1917. https://doi.org/10.1016/j.apgeochem.2009.07.004 147. 109. Lima FRD, Martins GC, Silva AO, Vasques ICF, Engelhardt MM, Cândido GS, et al. (2019) Critical mercury concentration in tropical soils: Impact on plants and soil biological attributes. Sci Total Environ 666:472-479. https://doi.org/10.1016/j.scitotenv.2019.02.216 Xiao R, Wang S, Li R, Wang JJ, Zhang Z (2017) Soil heavy metal contamination and health risks associated with artisanal gold mining in Tongguan, Shaanxi, China. Ecotox Environ Safe 141:17-24. https://doi.org/10.1016/j.ecoenv.2017.03.002 148. Yin D, He T, Yin R, Zeng L (2018) Effects of soil properties on production and bioaccumulation of methylmercury in rice paddies at a mercury mining area, China. J Environ Sci 68:194-205. https://doi.org/10.1016/j.jes.2018.04.028 148. Yin D, He T, Yin R, Zeng L (2018) Effects of soil properties on production and bioaccumulation of methylmercury in rice paddies at a mercury mining area, China. J Environ Sci 68:194-205. https://doi.org/10.1016/j.jes.2018.04.028 149. Chiarantini L, Rimondi V, Benvenuti M, Beutel MW, Costagliola P, Gonnelli C, et al. (2016) Black pine (Pinus nigra) barks as biomonitors of airborne mercury pollution. Sci Total Environ 569:105-113. https://doi.org/10.1016/j.scitotenv.2016.06.029 149. Chiarantini L, Rimondi V, Benvenuti M, Beutel MW, Costagliola P, Gonnelli C, et al. (2016) Black pine (Pinus nigra) barks as biomonitors of airborne mercury pollution. Sci Total Environ 569:105-113. https://doi.org/10.1016/j.scitotenv.2016.06.029 150. Hissler C, Probst JL (2006) Impact of mercury atmospheric deposition on soils and streams in a mountainous catchment (Vosges, France) polluted by chlor-alkali industrial activity: the important trapping role of the organic matter. Sci Total Environ 361:163-178. https://doi.org/10.1016/j.scitotenv.2005.05.023 Page 19/24 Page 19/24 Page 19/24 Page 19/24 151. Grangeon S, Guédron S, Asta J, Sarret G, Charlet L (2012) Lichen and soil as indicators of an atmospheric mercury contamination in the vicinity of a chlor-alkali plant (Grenoble, France). Ecol Indic 13:178-183. https://doi.org/10.1016/j.ecolind.2011.05.024 152. Ullrich SM, Ilyushchenko MA, Kamberov IM, Tanton TW (2007) Mercury contamination in the vicinity of a derelict chlor-alkali plant. Part I: Sediment and water contamination of Lake Balkyldak and the River Irtysh. Sci Total Environ 381:1-16. https://doi.org/10.1016/j.scitotenv.2007.02.033 153. Esbrí JM, López-Berdonces MA, Fernández-Calderón S, Higueras P, Díez S (2015) Atmospheric mercury pollution around a chlor-alkali plant in Flix (NE Spain): an integrated analysis. Environ. Sci. Pollut. R., 22, 4842-4850. https://doi.org/10.1007/s11356-014-3305-x 154. Biester H, Müller G, Schöler HF (2002) Binding and mobility of mercury in soils contaminated by emissions from chlor-alkali plants. Sci Total Environ 284:191-203. https://doi.org/10.1016/S0048-9697(01)00885-3 155. Bernaus A, Gaona X, van Ree D, Valiente M (2006) Determination of mercury in polluted soils surrounding a chlor-alkali plant: direct speciation by X-ray absorption spectroscopy techniques and preliminary geochemical characterisation of the area. Anal Chim Acta 565:73-80. https://doi.org/10.1016/j.aca.2006.02.020 156. Reis AT, Rodrigues SM, Araújo C, Coelho JP, Pereira E, Duarte AC (2009) Mercury contamination in the vicinity of a chlor-alkali plant and potential risks to local population. Sci Total Environ 407:2689-2700. https://doi.org/10.1016/j.scitotenv.2008.10.065 157. Zheng N, Liu J, Wang Q, Liang Z (2011) Mercury contamination due to zinc smelting and chlor-alkali production in NE China. Appl Geoch 26(2):188-193. https://doi.org/10.1016/j.apgeochem.2010.11.018 158. Song Z, Li P, Ding L, Li Z, Zhu W, He T, Feng X (2018) Environmental mercury pollution by an abandoned chlor-alkali plant in Southwest China. J Geochem Explor 194:81-87. https://doi.org/10.1016/j.gexplo.2018.07.017 159. Zhu W, Li Z, Li P, Yu B, Lin CJ, Sommar J, et al. (2018) Re-emission of legacy mercury from soil adjacent to closed point sources of Hg emission. Environ Pollut 242:718-727. https://doi.org/10.1016/j.envpol.2018.07.002 160. Reis AT, Rodrigues SM, Davidson CM, Pereira E, Duarte AC (2010) Extractability and mobility of mercury from agricultural soils surrounding industrial and mining contaminated areas. Chemosphere 81:1369-1377. https://doi.org/10.1016/j.chemosphere.2010.09.030 161. Liang J, Feng C, Zeng G, Gao X, Zhong M, Li X, et al. (2017) Spatial distribution and source identification of heavy metals in surface soils in a typical coal mine city, Lianyuan, China. Environ Pollut 225:681-690. https://doi.org/10.1016/j.envpol.2017.03.057 162. Wu Q, Wang S, Wang L, Liu F, Lin CJ, Zhang L, et al. (2014) Spatial distribution and accumulation of Hg in soil surrounding a Zn/Pb smelter. Page 19/24 Sci Total Environ 496:668-677. https://doi.org/10.1016/j.scitotenv.2014.02.067 163. Wang D, Shi X, Wei S (2003) Accumulation and transformation of atmospheric mercury in soil. Sci Total Environ 304:209-214. https://doi.org/10.1016/S0048-9697(02)00569-7 164. Navrátil T, Burns DA, Nováková T, Kaňa, J, Rohovec J, Roll M et al. (2018) Stability of mercury concentration measurements in archived soil and peat samples. Chemosphere 208:707-711. https://doi.org/10.1016/j.chemosphere.2018.06.033 165. Boszke L, Kowalski A, Siepak J (2004.) Grain size partitioning of mercury in sediments of the middle Odra River (Germany/Poland). Water Air Soil Poll 159:125-138. https://doi.org/10.1023/B:WATE.0000049171.22781.bd 166. Khan MN, Wasim AA, Sarwar A, Rasheed MF (2011) Assessment of heavy metal toxicants in the roadside soil along the N-5, National Highway, Pakistan. Environ Monit Assess 182:587-595. https://doi.org/10.1007/s10661-011-1899-8 167. Beckers F, Awad YM, Beiyuan J, Abrigata J, Mothes S, Tsang D.C, et al., (2019) Impact of biochar on mobilization, methylation, and ethylation of mercury under dynamic redox conditions in a contaminated floodplain soil.Environ Inter 127:276-290. https://doi.org/10.1016/j.envint.2019.03.040 168. Bonanno G, Giudice RL, Pavone P (2012) Trace element biomonitoring using mosses in urban areas affected by mud volcanoes around Mt. Etna. The case of the Salinelle, Italy. Monit Assess 184:5181-5188. https://doi.10.1007/s10661-011-2332-z 169. Gosar M, Šajn R, Biester H (2006) Binding of mercury in soils and attic dust in the Idrija mercury mine area (Slovenia). Sci Total Environ 369:150-162. https://doi.org/10.1016/j.scitotenv.2006.05.006 170. Richardson JB, Moore L (2020) A tale of three cities: Mercury in urban deciduous foliage and soils across land-uses in Poughkeepsie NY, Hartford CT, and Springfield MA USA. Sci Total Environ 715:136869. https://doi.org/10.1016/j.scitotenv.2020.136869 171. Cheng H, Li M, Zhao C, Li K, Peng M, Qin A, Cheng X (2014) Overview of trace metals in the urban soil of 31 metropolises in China. J Geochem Explor 139:31-52. https://doi.org/10.1016/j.gexplo.2013.08.012 172. Tijhuis L, Brattli B, Sæther OM (2002) A geochemical survey of topsoil in the city of Oslo, Norway. Environ Geochem Hlth 24:67-94. https://doi.org/10.1023/A:1013979700212. 172. Tijhuis L, Brattli B, Sæther OM (2002) A geochemical survey of topsoil in the city of Oslo, Norway. Environ Geochem Hlth 24:67-94. https://doi.org/10.1023/A:1013979700212. Page 20/24 Page 20/24 173. Manta DS, Angelone M, Bellanca A, Neri R, Sprovieri M (2002) Heavy metals in urban soils: a case study from the city of Palermo (Sicily), Italy. Sci Total Environ 300:229-243. https://doi.org/10.1016/S0048-9697(02)00273-5 174. Rodrigues S, Pereira ME, Duarte AC, Ajmone-Marsan F, Davidson CM, Grčman H, et al. (2006) Mercury in urban soils: a comparison of local spatial variability in six European cities. Sci Total Environ 368:926-936. https://doi.org/10.1016/j.scitotenv.2006.04.008 175. Page 19/24 Cheng H, Zhao C, Liu F, Yang K, Liu Y, Li M, et al. (2013) Mercury drop trend in urban soils in Beijing, China, since 1987. J Geochem Explor 124:195-202. https://doi.org/10.1016/j.gexplo.2012.09.007 176. Ottesen RT, Birke M, Finne TE, Gosar M, Locutura J, Reimann C, et al. (2013) Mercury in European agricultural and grazing land soils. Appl Geoch 33:1-12. https://doi.org/10.1016/j.apgeochem.2012.12.013 177. Loska K, Wiechuła D, Korus I (2004) Metal contamination of farming soils affected by industry. Environ Int 30:159-165. https://doi.org/10.1016/S0160- 4120(03)00157-0 178. Ahmadi M, Akhbarizadeh R, Haghighifard NJ, Barzegar G, Jorfi S (2019) Geochemical determination and pollution assessment of heavy metals in agricultural soils of south western of Iran J Environ Health Sci. Eng, 1-13. https://doi.org/10.1007/s40201-019-00379-6 179. Navrátil T, Shanley J, Rohovec J., Hojdová M, Penížek V, Buchtová J (2014) Distribution and pools of mercury in Czech forest soils. Water Air Soil Poll 225:1829. https://doi.org/10.1007/s11270-013-1829-1 180. Åkerblom S, Meili M, Bringmark L, Johansson K, Kleja DB, Bergkvist B (2008) Partitioning of Hg between solid and dissolved organic matter in the humus layer of boreal forests. Water Air Soil Poll, 189: 239-252. https//doi.10.1007/s11270-007-9571-1 181. Boszke L, Kowalski A, Siepak J (2004.) Grain size partitioning of mercury in sediments of the middle Odra River (Germany/Poland). Water Air Soil Poll 159:125-138. https://doi.org/10.1023/B:WATE.0000049171.22781.bd 182. Fiorentino JC, Enzweiler J, Angélica RS (2011) Geochemistry of mercury along a soil profile compared to other elements and to the parental rock: evidence of external input. Water Air Soil Poll 221:63-75. https://doi.org/10.1007/s11270-011-0769-x 183. do Valle CM, Santana GP, Augusti R, Egreja Filho FB, Windmöller CC (2005) Speciation and quantification of mercury in Oxisol, Ultisol, and Spodosol from Amazon (Manaus, Brazil). Chemosphere 58:779-792. https://doi.org/10.1016/j.chemosphere.2004.09.005 184. Armesto AG, Bibián-Núñez L, Campillo-Cora C, Pontevedra-Pombal X, Arias-Estévez M, Nóvoa-Muñoz JC. (2018) Total mercury distribution among soil aggregate size fractions in a temperate forest podzol. Span J Soil Sci 8:190-202. https://doi.org/10.3232/SJSS.2018.V8.N1.05 185. Yin R, Gu C, Feng X, Hurley JP, Krabbenhoft DP, Lepak RF, et al. (2016) Distribution and geochemical speciation of soil mercury in Wanshan Hg mine: Effects of cultivation. Geoderma 272:32-38. https://doi.org/10.1016/j.jes.2018.04.028 186. Wang DY, Qing CL, Guo TY, Guo YJ (1997) Effects of humic acid on transport and transformation of mercury in soil-plant systems. Water Air Soil Poll 95:35-43. https://doi.org/10.1007/BF02406154 187. Smith T, Pitts K, McGarvey JA, Summers AO (1998) Bacterial oxidation of mercury metal vapor, Hg(0). Appl Environ Microbiol 64:1328-1332. 187. Page 19/24 Smith T, Pitts K, McGarvey JA, Summers AO (1998) Bacterial oxidation of mercury metal vapor, Hg(0). App vey JA, Summers AO (1998) Bacterial oxidation of mercury metal vapor, Hg(0). Appl Environ Microbiol 64:1328-13 188. Colombo MJ, Ha J, Reinfelder JR, Barkay T, Yee N (2013) Anaerobic oxidation of Hg (0) and methylmercury formation by Desulfovibrio desulfuricans ND132. Geochim Cosmochim. Ac 112:166-177. https://doi.org/10.1016/j.gca.2013.03.001 189. Colombo MJ, Ha J, Reinfelder JR, Barkay T, Yee N (2014) Oxidation of Hg(0) to Hg(II) by diverse anaerobic bacteria. Chem Geol 363:334-340. https://doi.org/10.1016/j.chemgeo.2013.11.020 190. Gustin MS, Ericksen JA, Schorran DE, Johnson DW, Lindberg SE, Coleman JS (2004) Application of controlled mesocosms for understanding mercury air- soil - plant exchange. Environ Sci Technol 38:6044-6050. https://doi.org/10.1021/es0487933 191. Fantozzi L, Ferrara R, Dini F, Tamburello L, Pirrone N, Sprovieri F (2013) Study on the reduction of atmospheric mercury emissions from mine waste enriched soils through native grass cover in the Mt. Amiata region of Italy. Environ Res 125:69-74. https://doi.org/10.1016/j.envres.2013.02.004 192. Mazur M, Mitchell CPJ, Eckley CS, Eggert SL, Kolka RK, Sebestyen SD, et al. (2014) Gaseous mercury fluxes from forest soils in response to forest harvesting intensity: A field manipulation experiment. Sci Total Environ 496:678-687. https://doi.org/10.1016/j.scitotenv.2014.06.058 193. Leonard TL, Taylor Jr, GE, Gustin MS, Fernandez GC (1998) Mercury and plants in contaminated soils: 1. Uptake, partitioning, and emission to the atmosphere. Environ Toxicol Chem 17:2063-2071. https://doi.org/10.1002/etc.5620171024 194. Frescholtz TF, Gustin MS, Schorran DE, Fernandez GC (2003) Assessing the source of mercury in foliar tissue of quaking aspen. Environ Toxicol Chem22:2114-2119. https://doi.org/10.1002/etc.5620220922 195. Assad M, Parelle J, Cazaux D, Gimbert F, Chalot M, Tatin-Froux F (2016) Mercury uptake into poplar leaves. Chemosphere 146:1-7. https://doi.org/10.1016/j.chemosphere.2015.11.103 Page 21/24 196. Fleck JA, Grigal DF, Nater EA (1999) Mercury uptake by trees: An observational experiment. Water Air Soil Poll., 115, 513-523. https://doi.org/10.1023/A:1005194608598 196. Fleck JA, Grigal DF, Nater EA (1999) Mercury uptake by trees: An observational experiment. Water Air Soil Poll., 115, 513-523. https://doi.org/10.1023/A:1005194608598 197. Laacouri A, Nater EA, Kolka RK (2013) Distribution and uptake dynamics of mercury in leaves of common deciduous tree species in Minnesota, USA. Environ Sci Technol 47:10462-10470. https://doi.org/10.1021/es401357z 198. Arnold J, Gustin MS, Weisberg PJ (2018) Evidence for nonstomatal uptake of Hg by aspen and translocation of Hg from foliage to tree rings in Austrian pine. Environ Sci Technol 52:1174-1182. https://doi.org/10.1021/acs.est.7b04468 199. Page 19/24 Bot Rev 66:379-422. https://doi.org/10.1007/BF02868923 213. Moreno FN, Anderson CW, Stewart RB, Robinson BH (2004) Phytoremediation of mercury-contaminated mine tailings by induced plant-mercury accumulation. Environ Pra 6:65-175. https://doi.org/10.1017/S1466046604000274 214. Obrist D (2007) Atmospheric mercury pollution due to losses of terrestrial carbon pools? Biogeochemistry 85:119-123. https://doi.org/10.1007/s10533- 007-9108-0 215. Lomonte C, Doronila AI, Gregory D, Baker AJ, Kolev SD (2010) Phytotoxicity of biosolids and screening of selected plant species with potential for mercury phytoextraction. J Hazard Mater173:494-501. https://doi.org/10.1016/j.jhazmat.2009.08.112 216. Fasani E, Manara A, Martini F, Furini A, DalCorso G (2018) The potential of genetic engineering of plants for the remediation of soils contaminated with heavy metals. Plant Cell Environ 41:1201-1232. https://doi.org/10.1111/pce.12963 Page 19/24 Moreno-Jiménez E, Gamarra R, Carpena-Ruiz RO, Millán R, Peñalosa JM, Esteban E (2006) Mercury bioaccumulation and phytotoxicity in two wild plant species of Almadén area. Chemosphere 63:1969-1973. https://doi.org/10.1016/j.chemosphere.2005.09.043 200. Qian X, Wu Y, Zhou H, Xu X, Xu Z, Shang L, et al. (2018) Total mercury and methylmercury accumulation in wild plants grown at wastelands composed of mine tailings: Insights into potential candidates for phytoremediation. Environ Pollut 239:757-767. https://doi.org/10.1016/j.envpol.2018.04.105 201. De Temmerman L, Waegeneers N, Claeys N, Roekens E (2009) Comparison of concentrations of mercury in ambient air to its accumulation by leafy vegetables: An important step in terrestrial food chain analysis. Environ Poll 157:1337-1341. https://doi.org/10.1016/j.envpol.2008.11.035 202. Egler SG., Rodrigues-Filho S, Villas-Bôas RC, Beinhoff C (2006) Evaluation of mercury pollution in cultivated and wild plants from two small communities of the Tapajós gold mining reserve, Pará State, Brazil. Sci Total Environ 368(1):424-433. 203. Svoboda L, Havlíčková B, Kalač P (2006) Contents of cadmium, mercury and lead in edible mushrooms growing in a historical silver-mining area. Food Chem96:580-585. https://doi.org/10.1016/j.foodchem.2005.03.012 204. Falandysz J, Mędyk M, Treu R (2018) Bio-concentration potential and associations of heavy metals in Amanita muscaria (L.) Lam. from northern regions of Poland. Environ Sci Pollut R., 25(25):25190-25206. https://doi.org/10.1007/s11356-018-2603-0 205. Millhollen AG, Gustin MS, Obrist D (2006) Foliar mercury accumulation and exchange for three tree species. Environ Sci Technol 40:6001-6006. https://doi.org/10.1021/es0609194 206. Asati A, Pichhode M, Nikhil K (2016) Effect of heavy metals on plants: an overview. Int J Appl Innov Eng Mgmt 5:56-66. 207. Ahammad SJ, Sumithra S, Senthilkumar P (2018) Mercury uptake and translocation by indigenous plants. Rasayan J Chem. 11: 1-12. http//dx.doi.org/10.7324/RJC.2018.1111726 208. Shahid M, Khalid S, Bibi I, Bundschuh J, Niazi NK, Dumat C (2020) A critical review of mercury speciation, bioavailability, toxicity and detoxification in soil-plant environment: ecotoxicology and health risk assessment. Sci Total Environ 711:134749. https://doi.org/10.1016/j.scitotenv.2019.134749 209. Safari F, Akramian M, Salehi-Arjmand H, Khadivi A (2019) Physiological and molecular mechanisms underlying salicylic acid-mitigated mercury toxicity in lemon balm (Melissa officinalis L.). Ecotoxicol Environ Saf 183:109542. https://doi.org/10.1016/j.ecoenv.2019.109542 210. Zhou ZS, Wang SJ, Yang ZM (2008) Biological detection and analysis of mercury toxicity to alfalfa (Medicago sativa) plants. Chemosphere 70:1500- 1509. https://doi.org/10.1016/j.chemosphere.2007.08.028 211. Azevedo R, Rodriguez E (2012) Phytotoxicity of mercury in plants: a review. J Bot ID 848614. http://dx.doi.org/10.1155/2012/848614 2. Patra M, Sharma A (2000) Mercury toxicity in plants. Bot Rev 66:379-422. https://doi.org/10.1007/BF0286892 212. Patra M, Sharma A (2000) Mercury toxicity in plants. Metals and Metalloids. Academic Press, pp 43-62. doi.org.1016/C2017-0- 01241-7 Nagajyoti PC, Lee KD, Sreekanth TVM (2010) Heavy metals, occurrence and toxicity for plants: a review. Environ Chem Lett 8:199-216. https://doi.org/10.1007/s10311-010-0297-8 231. Marrugo-Negrete J, Durango-Hernández J, Pinedo-Hernández J, Enamorado-Montes G, Díez S (2016) Mercury uptake and effects on growth in Jatropha curcas. J Environ Sci 48:120-125. https://doi.org/10.1016/j.jes.2015.10.036 232. Teixeira DC, Lacerda LD, Silva-Filho EV (2018) Foliar mercury content from tropical trees and its correlation with physiological parameters in situ. Environ Pollut 242:1050-1057. https://doi.org/10.1016/j.envpol.2018.07.120 233. Manikandan R, Sahi SV, Venkatachalam P (2015) Impact assessment of mercury accumulation and biochemical and molecular response of Mentha arve 234. Mahbub KR, Krishnan K, Naidu R, Andrews S, Megharaj M (2017) Mercury toxicity to terrestrial biota. Ecol Indic 74:451-462. https://doi.org/10.1016/j.atmosenv.2005.07.067. 235. Zhou ZS, Huang SQ, Guo K, Mehta SK, Zhang PC, Zhi MY et al. (2007) Metabolic adaptations to mercury-induced oxidative stress in roots of Medicago sativa L. J Inorg Biochem 101:1-9. https://doi.org/10.1016/j.jinorgbio.2006.05.011 235. Zhou ZS, Huang SQ, Guo K, Mehta SK, Zhang PC, Zhi MY et al. (2007) Metabolic adaptations to mercury-induced oxidative stress in roots of Medicago sativa L. J Inorg Biochem 101:1-9. https://doi.org/10.1016/j.jinorgbio.2006.05.011 235. Zhou ZS, Huang SQ, Guo K, Mehta SK, Zhang PC, Zhi MY et al. (2007) Metabolic adaptations to mercury-induced oxidative stress in roots of Medicago sativa L. J Inorg Biochem 101:1-9. https://doi.org/10.1016/j.jinorgbio.2006.05.011 217. Ahmed I, Sebastain A, Prasad MNV, Kirti PB (2019) Emerging Trends in Transgenic Technology for Phytoremediation of Toxic Metals and Metalloids. In: Prasad MNV (ed) Transgenic Plant Technology for Remediation of Toxic Page 22/24 Page 22/24 Metals and Metalloids. Academic Press, pp 43-62. doi.org.1016/C2017-0- 01241-7 218. Kim Y-O, Bae H-J, Cho E, Kang H (2017) Exogenous Glutathione Enhances Mercury Tolerance by Inhibiting Mercury Entry into Plant Cells. Front. Plant Sci 8:683. https://doi.org/10.3389/fpls.2017.00683 219. Cho UH, Park JO (2000) Mercury-induced oxidative stress in tomato seedlings. Plant Sci 156:1-9. https://do 219. Cho UH, Park JO (2000) Mercury-induced oxidative stress in tomato seedlings. Plant Sci 156:1-9. https://doi.org/10.1016/S0168-9452(00)00227-2 220. Israr M, Sahi SV (2006) Antioxidative responses to mercury in the cell cultures of Sesbania drummondii. Plant Physiol Bioch 44:590-595. https://doi.org/10.1016/j.plaphy.2006.09.021 221. Zhang T, Lu Q, Su C, Yang Y, Hu D, Xu Q (2017) Mercury induced oxidative stress, DNA damage, and activation of antioxidative system and Hsp70 induction in duckweed (Lemna minor). Ecotoxicol Environ Saf 143:46-56. https://doi.org/10.1016/j.ecoenv.2017.04.058 222. Gómez-Jacinto V, García-Barrera T, Gómez-Ariza JL, Garbayo-Nores I, Vílchez-Lobato C (2015) Elucidation of the defence mechanism in microalgae Chlorella sorokiniana under mercury exposure. Identification of Hg–phytochelatins. Chem-Biol Interact 238:82-90. https://doi.org/10.1016/j.cbi.2015.06.013 223. Wani AB, Chadar H, Wani AH, Singh S, Upadhyay N (2017) Salicylic acid to decrease plant stress. EnvironChem Lett 15:101–123. https://doi.org/10.1007/s10311-016-0584-0 224. Kováčik J, Rotková G, Bujdoš M, Babula P, Peterková V, Matúš P (2017) Ascorbic acid protects Coccomyxa subellipsoidea against metal toxicity through modulation of ROS/NO balance and metal uptake. J Hazar Mat 339:200-207. https://doi.org/10.1016/j.jhazmat.2017.06.035 225. Zhou X, Gao A, Zhang C, Xu W, Shi X (2017) Exogenous selenium alleviates mercury toxicity by preventing oxidative stress in rice (Oryza sativa) seedlings. Int J Agric Biol 19:1593-1600. 226. Seneviratne M, Rajakaruna N, Rizwan M, Madawala HMSP, Yong Sik Ok YS, Vithanage M (2019) Heavy metal-induced oxidative stress on seed germination and seedling development: a critical review. Environ Geochem Health 41:1813–1831. https://doi.org/10.1007/s10653-017-0005-8 227. Hasanuzzaman M, Hossain MA, da Silva JAT, Fujita M (2012) Plant Response and Tolerance to Abiotic Oxidative Stress: Antioxidant Defense Is a Key Factor. In: Venkateswarlu B, Shanker A, Shanker C, Maheswari M (eds) Crop Stress and its Management: Perspectives and Strategies. Springer, Dordrecht. pp. 261-315. https://doi.org/10.1007/978-94-007-2220-0_ 228. Cargnelutti D, Tabaldi LA, Spanevello RM, de Oliveira Jucoski G, Battisti V, Redin M, et al. (2006) Mercury toxicity induces oxidative stress in growing cucumber seedlings. Chemosphere 65:999-1006. https://doi.org/10.1016/j.chemosphere.2006.03.037 229. Patra M, Bhowmik N, Bandopadhyay B, Sharma A (2004) Comparison of mercury, lead and arsenic with respect to genotoxic effects on plant systems and the development of genetic tolerance. Environ Exp Bot 52:199-223. https://doi.org/10.1016/j.envexpbot.2004.02.009 230. Table 7 Table 7. Some examples of surface soil layer contamination with mercury in regions of important sources of emissions by various authors. Page 23/24 Page 23/24 Page 23/24 Country Location source of pollution period Total Hg mg kg-1 Mean/ Median author USA Alaska Hg-mining 0.05–5326 Bailey et al. [136] USA Texas Hg-mining 3.8-11 Gray et al. [137] China Wuchuan Hg-mining 2003 0.33 - 320 Qiu et al. [118] China Wanshan Hg-mining 2002 5.1 – 790 Qiu et al. [117] China Xiaoqinling gold mining 0.04-61.2 mean 2.75 Wu et al. [138] Slovenia Idrija Hg-mining 1991- 97 1734-2759 mean 2456 Gnamuš et al. [106] Hg-mining 2000- 01 24- 1055 median 47 Gosar et al. [139] Slovenia Podljubelj Hg-mining 2003- 04 0.35–244 median 3.7 Teršič et al. [140] Slovakia Rudnany Hg-mining 9.1-54.3 Banásová [141] Spain Almaden Hg-mining 6–8889 mean 604 Higueras et al. [142] Spain Almaden Hg-mining 1340-4830 Dago et al. [143] Spain Caunedo old Hg-mining 0.09– 50.0 13.1 mean 13.1 Boente et al. [144] Italy Vallalta old Hg-mining 6-21 Wahsha et al. [145] Turkey Halıköy Hg mining 2004 0.10-33 mean 5.5 Gemici et al. [146] China Tongguan Artisanal gold mining 0.69-23.7 mean 2.91 Xiao et al. [147] China Wanshan Hg-mining, artisanal gold 2012 0.5-187 mean 31.0 Yin et al. [148] China Guizhou Acetic acid 2016 1.09–3.71 Li et al. [102] Italy Mt. Amiata, Hg mining, volcano- geothermal 2.4–68 Chiarantini et al. [149] France Vosges Mountains Chlor-alkali 2002 0.16-3.99 Hissler and Probst [150] France Grenoble Chlor-alkali 1.3-10 Grangeon et al. [151] Kazakhstan Pavlodar Chlor-alkali 0.93-22.3 Ullrich et al. [152] Spain Flix Chlor-alkali 0.04-12.9 mean 0.77 Esbrí et al. [153] Germany Chlor-alkali 0.5-4.2 mean 1.6 Biester et al. [154] Netherlands Chlor-alkali 2004 4.3-1150 Bernaus et al. [155] Portugal Estarreja Chlor-alkali 0.010–91 5.4 mean 5.4 Reis et al. [156] China Huludao Chlor-alkali Zn-smelting 2006- 08 0.05–14.6 Zheng et al. [157] China An Ning Chlor-alkali 0.09-1.30  mean 0.40  Song et al. [158] China Kunming Chlor-alkali polyvinyl chloride 0.15-4.79 Zhu et al. [159] Portugal Caveira sulfide mine 1.1-76.5 Reis et al. [160] China Lianyuan coal main, steel industry 2015 1.20-3601 mean 178 Liang et al. [161] China Zhuzhou Zn/Pb smelter 2012 0.62-2.61 mean 1.54 Wu et al. [162] China Chongqing thermometer factory 0.06–0.88 Wang et al. [163] Czech Rep. Bohemian non-ferrous metal 2017 6.49 Navrátil et al. [164] Switzerland Canton of Valais Industrial region 2017 0.2–390  Osterwalder et al. Table 7 [94] Poland Warsaw thermometer factory 2005 122–393 mean 147 Boszke et al. [165] Pakistan Karachi/Lahore highway 61.5-144 mean 90.7 Khan et al. [166] Germany Rhine-Westph. floodplain soik 2017 31.2 Beckers et al. [167] Italy Etna volcano 0.1-0.4 Bonanno et al. [168] European Union 23000 samples 2009- 12 0-1.59 mean 0.04 Tóth et al. [130] Page 24/24 Page 24/24
https://openalex.org/W2885299355	https://www.nature.com/articles/s41598-018-30095-4.pdf	English	null	All optical dual stage laser wakefield acceleration driven by two-color laser pulses	Scientific reports	2,018	cc-by	6,167	Vishwa Bandhu Pathak 1, Hyung Taek Kim 1,2, J. Vieira 3, L. O. Silva 3 & Chang Hee Nam1,4 We propose an all-optical dual-stage laser wakefield acceleration (LWFA), staged with co-propagating two-color laser pulses in a plasma medium, to enhance the electron bunch energy. After the depletion of the leading fundamental laser pulse that initiates self-injection and sets up the first stage particle acceleration, the subsequent second-harmonic laser pulse takes over the acceleration process and accelerates the electron bunch in the second stage over a significantly longer distance than in the first stage. In this all optical dual-stage LWFA, the electrons can gain 3 times higher energy as compared to the energy gain from the single stage LWFA driven by a single-color laser pulse with equivalent energy. Our multi-dimensional particle-in-cell simulations demonstrate that a 10-GeV electron bunch with 20-pC charge can be obtained by the two-color dual-stage LWFA using total input laser power of 0.6 PW. Laser-plasma electron acceleration1–5 is a promising approach to the next generation of electron accelerators because it can provide an accelerating field that can be more than three orders of magnitude larger than those of conventional devices. The progress of high power laser technologies6,7 and theoretical expectations8,9 have stim- ulated conceptual designs to realize an electron accelerator with energy over 100 GeV based on laser wakefield acceleration (LWFA)10. The radial ponderomotive force associated with a short and intense laser pulse can lead to the cavitation of all plasma electrons from the region where the laser pulse propagates, creating a spherical plasma wave (bubble or blowout11,12). The large accelerating field associated with the blowout can trap and accelerate plasma electrons (self-injection), or positrons if a Laguerre-Gaussian laser beam is used as a driver13.i In LWFA, the acceleration length is one of the important parameters determining the final electron energy. For a self-guided laser pulse14, the effective acceleration length is limited either by the laser etching (depletion) length [Letch ≈ (ω0/ωp)2cτL]9, or by the dephasing length [ = ω ω ω L (4/3)( / ) a c/ d 0 2 p 2 0 p]9, where a0 = eE0/(mecω0) is the normalized vector potential of the laser, c is the speed of light in vacuum, and E0, ω0 and τL are the peak laser electric field, laser central frequency and laser pulse length respectively. In addition, n e m 4 / p 0 2 ω π = is the plasma frequency with n0 being plasma density, and m, e electron rest mass and elementary charge, respectively. www.nature.com/scientificreports www.nature.com/scientificreports www.nature.com/scientificreports All optical dual stage laser wakefield acceleration driven by two-color laser pulses OPEN Received: 11 February 2018 Accepted: 20 July 2018 Published: xx xx xxxx Vishwa Bandhu Pathak 1, Hyung Taek Kim 1,2, J. Vieira 3, L. O. Silva 3 & Chang Hee Nam1,4 Vishwa Bandhu Pathak 1, Hyung Taek Kim 1,2, J. Vieira 3, L. O. Silva 3 & Chang Hee Nam1,4 Vishwa Bandhu Pathak 1, Hyung Taek Kim 1,2, J. Vieira 3, L. O. Silva 3 & Chang Hee Nam1,4 As it is clear from the expressions above, the acceleration length, and thus the energy gain, is higher at lower plasma densities because the ratio ω0/ωp is higher. In contrast, particle trapping is easily met at higher plasma densities, because the required longitudinal electron momentum for self-injection also scales with ω0/ωp. Therefore, the electron injection and the acceleration length are the competing factors in the optimization of plasma density for LWFA in single stage. p y g g In order to solve this problem, dual-stage LWFA10,15–19 has been proposed with a short, high density plasma in the first stage (injector stage) and a long, low density plasma in the second stage (accelerator stage). The dual-staged LWFA can be achieved either by a single laser pulse propagating through two plasma media16,20–22 or by two independent laser pulses focused separately to two plasma media15,18,19. All these methods for dual-staged LWFA require two plasma media with Luo et al.19 showing multi stage coupling of LWFA in curved plasma chan- nel from 2D PIC simulations using OSIRIS23,24.h g Manipulating the laser frequency to realize a multi-stage LWFA is considerably more challenging. There are growing interests on all-optical control of LWFA, by manipulating optical properties of driving laser pulse25–27 or 1Center for Relativistic Laser Science, Institute for Basic Science (IBS), Gwangju, 61005, Korea. 2Advanced Photonics Research Institute, Gwangju Institute of Science and Technology (GIST), Gwangju, 61005, Korea. 3GoLP/Instituto de Plasmas e Fusão Nuclear, Instituto Superior Técnico, Universidade de Lisboa, Lisbon, Portugal. 4Department of Physics and Photon Science, GIST, Gwangju, 61005, Korea. Correspondence and requests for materials should be addressed to H.T.K. (email: htkim@gist.ac.kr) or C.H.N. (email: chnam@gist.ac.kr) ScIentIfIc REPOrts \| (2018) 8:11772 \| DOI:10.1038/s41598-018-30095-4 1 www.nature.com/scientificreports/ by composing multiple laser pulses28–31. While the frequency chirp approach can be employed to control injec- tion and acceleration27, it is severely limited by the maximum chirp that may be introduced in a single laser pulse driver. Here we explore an innovative solution to this limitation by proposing to use a two-color laser pulse to cre- ate the injector and accelerator stage in a single homogeneous plasma column. In our proposed model, the leading pulse, at the fundamental frequency, acts as an injector and the trailing laser pulse, at the second harmonic, works as an accelerator or booster. We derived an analytical expression for the energy scaling of all-optical dual stage LWFA in the matched regime9,12. Results E Energy gain in all optical dual-stage LWFA. The two-color dual-stage LWFA can be understood as following: In the injector stage, the leading lower frequency laser pulse excites a plasma wave with a lower phase velocity, which triggers electron self-injection (injector stage). The second harmonic laser pulse, trailing just behind the leading pulse, is guided within the bubble. The second harmonic laser will drive wakefields once the leading pulse depletes. These wakefields move faster than those of the injector stage, being ideal to boost the energy of the self-injected electrons. This is the acceleration stage. The transition from the injection stage to the acceleration stage is regulated by the temporal delay between the two color laser pulses. We show that, by fine-tuning the time delay between the two-color laser pulses, quasi-mono-energetic features can be obtained in the accelerated electron beam. To estimate the energy gain of the two-color dual-stage LWFA, let us consider a fundamental frequency laser pulse (FL) with normalized vector potential a0( ) 0 ω , frequency ω0, spot-size ω W0( ) 0 and pulse duration τ ω L( ) 0 is prop- agating in a homogeneous plasma with density n0. The FL is followed by a second harmonic laser pulse (SL) with time delay td with respect to FL, normalized vector potential a0(2 ) 0 ω , frequency 2ω0, spot-size W0(2 ) 0 ω and pulse duration L(2 ) 0 τ ω . The length of the first stage of LWFA led by the FL is limited to c L / s1 L( ) 0 2 p 2 0 = τ ω ω ω because of localized etching32, which is also close to the dephasing length in the matched regime9. Assuming that the trailing SL does not significantly influence the bubble evolution in the first stage, the energy gain in the first stage is given by ( ) E mc a / s1 2 3 2 0( ) 0 2 p 2 0 ∆ = ω ω ω . Once the leading laser pulse is absorbed, the SL takes over the wakefield excitation in the accelerator stage characterized by 4 times longer etching length as compared to the first stage. Results E As a result, the two-color laser pulses with Pin = 1 PW, nowadays available from a PW laser facility, can produce 10-GeV electron bunches for α = 0.3. Proof of principle simulation. In order to show the feasibility of the all-optical dual stage LWFA, a set of multi-dimensional PIC simulations are performed using OSIRIS23,24. We consider two-color laser pulses of input laser power 215 TW propagating in a 0.7-cm plasma with density 7.75 × 1018 cm−3. We initialize a simulation box, which moves with the speed of light c, has dimensions of 25 c/ωp × 30 c/ωp × 30 c/ωp and are divided into 3000 × 280 × 280 cells, along z, x and y direction respectively, with 2 particles per cell. The FL with frequency ω0 = 15 ωp is initialized with parameters: a 8 0( ) 0 = ω , = ω ω W 6 c/ 0( ) p 0 and τ = = . ω λ ω − 3 14 L( ) 2 p 1 p 0 , and the SL with frequency 2 ω0 and α = 0.3 is initialized in the simulation box at a distance 4.5c/ωp away from the leading pulse with parameters = ω a 4 0(2 ) 0 , = ω ω W 4 c/ 0(2 ) p 0 and τ = λ ω L(2 ) 2 p 0 . The plasma is initialized just in front of the leading pulse, with an initial density ramp of size 100 c/ωp and then a constant homogeneous plasma density.hh p y p p g p y The 3D PIC simulations demonstrate that two-color laser pulses can realize the dual-stage acceleration. The results obtained from the 3D simulation is shown in Fig. 1, where panels (a)-(e) plot the laser fields for FL and SL, and the corresponding electron density distribution in the y-z simulation plane at various stages of the accelera- tion process, and panels (f)-(j) show the corresponding electron distribution in phase space. The panels (a) and (f) demonstrate the excitation of bubble, self-injection by FL, and the guiding of SL inside the bubble in the first stage. The panels (b) and (g) show the depletion of leading FL and the initial onset of the second stage by the SL. Panels (c) and (h) clearly show the coupling of the electron bunch, from the first stage to the second stage. Results E Since the acceleration length in the second stage is expected to be limited by the dephasing, the acceleration length of the second stage can be given as = ω ω L (8/3) / R s2 0 2 p 2 , where R 2 a0(2 ) 0 = ω 9 is the bubble radius. Hence, the energy gain in the second stage can be written as E (8/3) mc a / s2 2 0(2 ) 0 2 p 2 0 ∆ = ω ω ω and the total energy gain can be esti- mated as, ∆ = ω ω    +    . ω ω ω E 2 3 mc a 1 4 a a (1) 2 0 2 p 2 0( ) 0(2 ) 0( ) 0 0 0 (1) For total input power = + ω ω P P P in 2 0 0 with the FL power, ω P 0, the SL power, P2 0 ω , and the second harmonic conversion efficiency P /P 2 in 0 α = ω , we can re-write Eq. (1) as: For total input power = + ω ω P P P in 2 0 0 with the FL power, ω P 0, the SL power, P2 0 ω , and the second harmonic conversion efficiency P /P 2 in 0 α = ω , we can re-write Eq. (1) as: E G m [GeV] 1 7 P [TW] 100 10 n [cm ] 0 8 [ ] , (2) d in 1 3 18 0 3 2 3 0 4 3 ∆ = .                . λ µ    − (2) where α α = − + G (1 ) 2 d 1/3 4/3 1/3 is the extra gain provided by the dual stage acceleration. For α = 0.05 to 0.9, Gd slowly varies from 2.0 to 3.0, providing a stable gain factor. For α = 0, Gd = 1 and the energy scales same as in Lu et al.9. However, for the same total laser input power our proposed scheme can significantly enhance the acceler- ation gain by a factor in the range between 2 to 3. Vishwa Bandhu Pathak 1, Hyung Taek Kim 1,2, J. Vieira 3, L. O. Silva 3 & Chang Hee Nam1,4 Using PIC simulations with OSIRIS23,24, we demonstrate the proof-of-principle of this concept to achieve quasi-mono-energetic multi-GeV electron bunch in 3D calculations and the generation of 10-GeV electron beam by two-color laser pulses having the input laser power of 0.6 PW in 2D calculations. Results E −   ω ω ω ω ω ω ω ( ) ( ) a a a t a a Min 4 0 5 , 4 Max 1 5 , 4 L p d p L 0( ) 0(2 ) 0( ) (2 ) ( ) 0( ) 0(2 ) 0 0 0 0 0 0 0 (3) Including the effect of time delay, the acceleration length of second stage modifies to = ω L (8/3)( s2 0 2 (4 a 2 a t ) c/ 0( ) 0(2 ) p d p 0 0 − −ω ω ω ω , and thus the energy gain in the second stage can be given in detail Including the effect of time delay, the acceleration length of second stage modifies to = ω ω L (8/3)( / ) s2 0 2 p 2 (4 a 2 a t ) c/ 0( ) 0(2 ) p d p 0 0 − −ω ω ω ω , and thus the energy gain in the second stage can be given in details by ∆ = ω ω −ω ω ω ω ω E (8/3)mc a ( / )(2 a /a t /(2 a )) s2 2 0(2 ) 0 2 p 2 0( ) 0(2 ) p d 0(2 ) . Including the effect of time delay, the acceleration length of second stage modifies to = ω ω L (8/3)( / ) s2 0 2 p 2 (4 a 2 a t ) c/ 0( ) 0(2 ) p d p 0 0 − −ω ω ω ω , and thus the energy gain in the second stage can be given in details by p p 0 0 = ω ω −ω ω ω ω ω E (8/3)mc a ( / )(2 a /a t /(2 a )) s2 2 0(2 ) 0 2 p 2 0( ) 0(2 ) p d 0(2 ) 0 0 0 0 . To verify the role of time delay between the two pulses on the coupling of the electrons from the first stage into the second stage, we further perform a set of 2D simulations with two-color laser pulses of Pin = 210 TW and α = 0.33 propagating in a 1.5-cm plasma with density 4.35 × 1018 cm−3. Results E The simulation box has dimensions of 25 c/ωp × 40 c/ωp and is divided into 4800 × 400 cells, along z and x direction respectively, with 9 particles per cell. We consider FL parameters as: a 5 0( ) 0 = ω , ω0 = 20 ωp, W 7 c/ 0( ) p 0 = ω ω and τ = λ ω L( ) 2 p 0 , and the SL parameters as: a 2 5 0(2 ) 0 = . ω , = ω ω W 5 c/ 0(2 ) p 0 and τ = λ ω L(2 ) 2 p 0 . Though the laser parameters are not exactly in the matched regime9, the robustness of the proposed scheme can be tested. As shown in Fig. 2(a), the maximum energy can be optimized by manipulating the time delay between the two pulses. We observe that for td ≈ 1.8τL the peak energy of 3.6 GeV is obtained after the end of second stage. For smaller time delays our simulations showed no coupling of the accelerated electron from the first stage into the second stage. As we further increase the time delay between the two pulses the effective acceleration length and hence the energy of the trapped electrons in the second stage decreases due to dephasing, as evident from Fig. 2(a) for td 6 p > . ω We also perform a series of 2D PIC simulations to compare the dual stage LWFA driven by two single-frequency pulses (SFTP) with the dual stage driven by the two frequencies two pulse (TFTP). The details are provided in the supplementary material. 10 GeV electron beam driven by 0.6-PW laser. As discussed before, one of the benefits of the two color dual stage LWFA is significant enhancement in the energy of an accelerated electron bunch, with Eq. (2) suggest- ing the realization of 10-GeV electron bunch. Keeping that in mind, we perform 2D PIC simulations using OSIRIS23,24, and the results are summarized in Fig. 3. We consider equally energetic FL and SL with total laser energy of 30 J and each with 50-fs pulse-length, which corresponds to Pin = 0.6 PW. The other FL and SL parame- ters are: = . Results E Thus, the proposed all optical staged LWFA has potential to increase the energy by a factor of 3, and reducing ScIentIfIc REPOrts \| (2018) 8:11772 \| DOI:10.1038/s41598-018-30095-4 ScIentIfIc REPOrts \| (2018) 8:11772 \| DOI:10.1038/s41598-018-30095-4 3 www.nature.com/scientificreports/ beam divergence and emittance, while the beam charge reduces. According to our analytical scaling [Eq. 2], the maximum energy gain for the total combined laser power of 215 TW, 800-nm fundamental wavelength and plasma density of 7.75 × 1018 cm−3, is 1.4 GeV that is very close to the final energy of 1.5 GeV obtained in the 3D PIC simulations, showing 3 times energy enhancement as compared to the energy obtained in the single stage LWFA. As clear from Fig. 1(e), the acceleration length in the second stage is limited by the dephasing of energetic electrons in the wake of the second stage and, as a result, quasi-mono-energetic electrons at 1.4 GeV are produced [Fig. 1: panel (j)]. [ g p (j)] In order to obtain the necessary condition for the coupling of electron beam to the second stage, the case of LWFA in the matched regime9 is considered with spherical and stationary nonlinear wakefields. The group veloc- ity of the leading laser pulse is ≈ − ω ω v c[1 (3/2)( / )] g1 p 2 0 2 because of localized etching9,32, and the following sec- ond harmonic laser pulse propagating within the first-stage bubble has group velocity vg2 ≈ c. As a result, by the end of the first stage the distance between the two pulses reduces to = − τ ω c L ct (3/2) d d L( ) 0 . During this time the leading part of the electron bunch, propagating with velocity c, also covers a distance = τ ω ( )c Lb 3 2 L( ) 0 and, thus, the distance between the SL and the electron bunch stays constant in the first stage. Role of time delay on the acceleration. With these constraints, we can estimate the required time delay between FL and SL for the coupling. Results E An electron beam with 2nC charge in 500–600 MeV energy was injected into the second stage with 12mrad diver- gence and 16 mm mrad emittance, and further accelerated to 1.5 GeV with 0.3 nC charge above 1 GeV, as shown ScIentIfIc REPOrts \| (2018) 8:11772 \| DOI:10.1038/s41598-018-30095-4 2 www.nature.com/scientificreports/ Figure 1. Proof-of-principle 3D PIC simulation. (a–e) Show distribution in the y-z plane at different times, and (f–j) show electrons at respective times. The line in figures (a–e) is the lo The red line in figure (j) represents energy spectrum of the ele 1.4 GeV. Figure 1. Proof-of-principle 3D PIC simulation. (a–e) Show the snapshots of laser fields and electron density distribution in the y-z plane at different times, and (f–j) show the energy-space distribution of accelerated electrons at respective times. The line in figures (a–e) is the longitudinal electric field (Ez) on the laser axis. The red line in figure (j) represents energy spectrum of the electrons showing bunched electron beam around 1.4 GeV. Figure 1. Proof-of-principle 3D PIC simulation. (a–e) Show the snapshots of laser fields and electron density distribution in the y-z plane at different times, and (f–j) show the energy-space distribution of accelerated electrons at respective times. The line in figures (a–e) is the longitudinal electric field (Ez) on the laser axis. The red line in figure (j) represents energy spectrum of the electrons showing bunched electron beam around 1.4 GeV. in panel (j). In electron energy spectrum we observe two quasi-mono-energetic peaks, one at 1.3 Gev with 2.4% full-width-half maximum energy spread, and another peak at 1.5 GeV with 6.7% energy spread. At the same time the beam divergence and emittance reduces to 4 mrad and 10 mm mrad respectively in the second stage. Thus, the proposed all optical staged LWFA has potential to increase the energy by a factor of 3, and reducing in panel (j). In electron energy spectrum we observe two quasi-mono-energetic peaks, one at 1.3 Gev with 2.4% full-width-half maximum energy spread, and another peak at 1.5 GeV with 6.7% energy spread. At the same time the beam divergence and emittance reduces to 4 mrad and 10 mm mrad respectively in the second stage. Results E The lower limit of time delay is limited by two conditions; (i) for proper guiding SL must avoid the interaction with the remnant of the density pile-up created by the FL at its front; thus, ( ) td 3 2 L( ) 0 > τ ω , and (ii) electrons accelerated in the first stage must reach the back of the bubble excited in the second stage, i.e. ≥ ω − ω − L 2(R( ) R(2 )) L b 0 0 d, which gives ≥ − ω ω ω t 4( a a )/ d 0( ) 0(2 ) p 0 0 . The upper limit of time delay again can be obtained from two conditions: (i) at the time of injection in the first stage the electron should not interact with the tail of the SL to avoid degradation in the beam-emittance, i.e. < −τ ω ω ω t (4 a )/ d 0( ) L( ) p 0 0 , and (ii) when coupled into the second stage the electrons should not already fall into the dephasing region of the second-stage-bubble, i.e. t 4( a 0 5 a )/ d 0( ) 0(2 ) p 0 0 < − . ω ω ω . Thus, the suitable condi- tion for choosing time delay for the coupling can be written as, τ ω ω τ   − . −   > >  . Results E ω a 5 38 0( ) 0 , ω0 = 40 ωp, = ω ω ω W 2 a c/ 0( ) 0( ) p 0 0 , τ = λ ω L( ) 2 p 0 , a 3 4 0(2 ) 0 = . ω , W 2 a c/ 0(2 ) 0(2 ) p 0 0 = ω ω ω , L(2 ) 2 p 0 τ = λ ω and td = 4.5 /ωp. To ensure guiding of the SL for a long distance in the second stage, a parabolic plasma channel profile =   +   ∆( ) n x n 1 n 0 n 2 c 0 is considered with n n c / (4/W )( / ) c 0 0(2 ) 4 p 2 0 ∆ = ω ω 2,9 and   n0 = 1 × 1018 cm−3. The acceleration with a single laser pulse of energy 30 J and wavelength 0.8 μm is limited to 5 GeV (the dashed line in Fig. 3) after propagating 2.5-cm plasma due to dephasing. For the two-color dual-stage acceleration, the electron acceleration, after reaching 5 GeV in 2.5-cm plasma, is enhanced to 10 GeV with 20-pC charge in the second stage (the solid line in Fig. 3) within 8-cm plasma.hi g g g p This is a significant improvement as compared to the single-stage LWFA. Lu et al.9 scaling in the matched regime indicates that to generate 10-GeV electron beam in a 10-cm plasma with density 6 × 1017 cm−3 a 10-PW laser pulse is required. On the contrary, in two-color dual-stage LWFA, if the guiding of the second harmonic laser pulse is insured, 10-GeV electron beam is obtained in the same plasma length with an order of magnitude lower laser power. On the other hand, to generate 10-GeV electron bunch in single stage LWFA at lower laser power of 1PW, the laser ScIentIfIc REPOrts \| (2018) 8:11772 \| DOI:10.1038/s41598-018-30095-4 4 www.nature.com/scientificreports/ www.nature.com/scientificreports/ Figure 2. Effect of time-delay between the leading and trailing laser pulse on the electron energy. (a) Variation in the maximum energy with the time delay, and (b) spectrum of accelerated electrons by TFTP with time delay ωptd = 4 (green), 6 (blue), 7 (red), and 10 (black). The results are obtained with 2D PIC simulations. Figure 2. Conclusion In conclusion, we have shown the feasibility of all optical two-color dual-stage LWFA with the help of multi-dimensional PIC simulations to obtain multi-GeV quasi-mono-energetic electron bunches. Two-color dual-stage LWFA yields higher acceleration energies, while keeping the quasi-mono-energetic property of the electron bunch, as compared to the same-energy single-pulse LWFA. We further provide estimates on the effect of the time delay between the two pulses on the acceleration process and show that, by fine-tuning the time delay and thus modifying the dephasing length in the second stage, the bunch energy and its quality can be optimized. We demonstrate that the two-color dual-stage LWFA can produce 10-GeV electron beam with an input power of 0.6 PW, which brings us closer to one of the milestones of LWFA. Results E Such ScIentIfIc REPOrts \| (2018) 8:11772 \| DOI:10.1038/s41598-018-30095-4 5 www.nature.com/scientificreports/ transverse offsets may lead to transverse oscillations of the laser in the plasma channel34,35, which can degrade the electron beam parameters, while it at the same can also lead to compact tunable polarized X-ray sources35. transverse offsets may lead to transverse oscillations of the laser in the plasma channel34,35, which can degrade the electron beam parameters, while it at the same can also lead to compact tunable polarized X-ray sources35. References Nonlinear laser driven donut wakefields for positron and electron acceleration. Phys. Rev. Lett. 215001 (2014). 14. Sprangle, P., Tang, C. M. & Esarey, E. Relativistic self-focusing of short-pulse radiation beams in plasmas. IEEE Trans. Plasma Sci. 15, 145–153 (1987). . Steinke, S. et al. Multistage coupling of independent laser-plasma 6. Kim, H. T. et al. Enhancement of Electron Energy to the Multi-GeV Regime by a Dual-Stage Laser-Wakefield Accelerator Pumped by Petawatt Laser Pulses. Phys. Rev. Lett 111, 165002 (2013). 7. Malka, V., Lifschitz, A., Faure, J. & Glinec, Y. Staged concept of laser-plasma acceleration toward multi-GeV electron beams. Phys Rev. Spec. Top. - Accel. Beams 9, 091301 (2006).i Rev. Spec. Top. - Accel. Beams 9, 091301 (2006). 18. Kaganovich, D. et al. First demonstration of a staged all-optical laser wakefield acceleration First demonstration of a staged all-i p p 18. Kaganovich, D. et al. First demonstration of a staged all-optical laser wakefield acceleration First demonstration of a stage optical laser wakefield acceleration. Phys. Plasmas 12, 100702 (2005).i i y 19. Luo, J. et al. Multistage Coupling of Laser-Wakefield Accelerators with Curved Plasma Channel. Phys. Rev. Lett 120, 154801 (2 19. Luo, J. et al. Multistage Coupling of Laser-Wakefield Accelerators with Curved Plasma Channel. Phys. Rev. Lett 120, 154801 (2018). 20. Pollock, B. B. et al. Demonstration of a narrow energy spread, ~0.5 GeV electron beam from a two-stage laser wakefield accelerator. Phys. Rev. Lett. 107, 045001 (2011). i 20. Pollock, B. B. et al. Demonstration of a narrow energy spread, ~0.5 GeV electron beam from a two-stage laser wakefield accele Phys. Rev. Lett. 107, 045001 (2011). y 21. Buck, A. et al. Shock-Front Injector for High-Quality Laser-Plasma Acceleration. Phys. Rev. Lett 110, 185006 (2013).i 22. Döpp, A. et al. Energy boost in laser wakefield accelerators using sharp density transitions. Phys. Plasmas 23, 056702 (2016) i 23. Fonseca, R. A. et al. In Computational Science–ICCS 2002: International Conference Amsterdam, The Netherlands, April 21–24, 2002 Proceedings, Part III (eds Sloot, P. M. A., Hoekstra, A. G., Tan, C. J. K. & Dongarra, J. J.) 342–351, https://doi.org/10.1007/3-540- 47789-6_36 (Springer Berlin Heidelberg, 2002). g g 4. Fonseca, R. A. et al. One-to-one direct modeling of experiments and astrophysical scenarios: pushing the envelope on kinetic plasma simulations. 50, 124034 (2008).fi 25. Pathak, V. B., Vieira, J., Fonseca, R. A. & Silva, L. O. Effect of the frequency chirp on laser wakefield acceleration. New J. Phys. References 1. Tajima, T. & Dawson, J. M. Laser electron accelerator. Phys. Rev. Lett 4, 267–270 (1979). Tajima, T. & Dawson, J. M. Laser electron accelerator. Phys. Rev. Let 1. Tajima, T. & Dawson, J. M. Laser electron accelerator. Phys. Rev. Lett 4, 267–270 (1979). l b ll f l d l j y 2. Esarey, E., Sprangle, P., Member, S., Krall, J. & Ting, A. Overview of Plasma-Based Accelerator Concepts. IEEE Trans. Plasma Sci. 24 252–288 (1996). ( ) 3. Faure, J. et al. A laser-plasma accelerator producing monoenergetic electron beams. Nature 431, 541–544 (2004).i 4. Geddes, C. G. R. et al. High-quality electron beams from a laser wakefield accelerator using plasma-channel guiding. Nature 431 538–541 (2004).i 5. Faure, J. et al. Controlled injection and acceleration of electrons in plasma wakefields by colliding laser pulses. Nature 444, 737–739 (2006). ( ) 6. Korn, G., LeGarrec, B. & Rus, B. ELI Extreme Light Infrastructure Science and Technology with ultra-intense Lasers. In CLEO: 2013 CTu2D.7, https://doi.org/10.1364/CLEO_SI.2013.CTu2D.7 (2013). p g ( ) 7. Jeong, T. M. et al. Generation of high-contrast, 30 fs, 1.5 PW laser pulses. Opt. Express 20, 10807 (2012).i g g p p p 8. Martins, S. F., Fonseca, R. A., Lu, W., Mori, W. B. & Silva, L. O. Exploring laser-wakefield-accelerator regimes for near-term lasers using particle-in-cell simulation in Lorentz-boosted frames. Nat. Phys. 6, 311–316 (2010).i 8. Martins, S. F., Fonseca, R. A., Lu, W., Mori, W. B. & Silva, L. O. Exploring laser-wakefield-accelerator regimes for ne using particle-in-cell simulation in Lorentz-boosted frames Nat Phys 6, 311–316 (2010) 8. Martins, S. F., Fonseca, R. A., Lu, W., Mori, W. B. & Silva, L. O. Exploring laser-wakefield-acc using particle-in-cell simulation in Lorentz-boosted frames. Nat. Phys. 6, 311–316 (2010).i 9. Lu, W. et al. Generating multi-GeV electron bunches using single stage laser wakefield acceleration in a 3D nonlinear regime. Rev. Spec. Top. - Accel. Beams 10, 061301 (2007). p p 10. Leemans, W. & Esarey, E. Laser-driven plasma-wave electron accelerators. Phys. Today 62, 44–49 (2009).i 11. Pukhov, A. Strong field interaction of laser radiation. Reports Prog. Phys. 65, R1–R55 (2002). h l l h f l l gi p g y 12. Lu, W., Huang, C., Zhou, M., Mori, W. B. & Katsouleas, T. Nonlinear theory for relativistic plasma wakefields in the blowout regime. Phys. Rev. Lett. 96, 165002 (2006).i 13. Vieira, J. & Mendonça, J. T. Results E Effect of time-delay between the leading and trailing laser pulse on the electron energy. (a) Variation in the maximum energy with the time delay, and (b) spectrum of accelerated electrons by TFTP with time delay ωptd = 4 (green), 6 (blue), 7 (red), and 10 (black). The results are obtained with 2D PIC simulations. Figure 3. Time evolution of maximum energy of the accelerating electron bunch with 1PW laser. The solid line shows the energy gain from the TFTP, while the dashed line represents the energy gain from the single frequency single pulse (SFSP). The inset shows the electron spectra for the two cases at the plasma position of the maximum energy (⊙: TETP, ⊡: SFSP). The results are obtained with 2D PIC simulations. Figure 3. Time evolution of maximum energy of the accelerating electron bunch with 1PW laser. The solid line shows the energy gain from the TFTP, while the dashed line represents the energy gain from the single frequency single pulse (SFSP). The inset shows the electron spectra for the two cases at the plasma position of the maximum energy (⊙: TETP, ⊡: SFSP). The results are obtained with 2D PIC simulations. pulse needed to be externally guided through a 1-m long plasma channel with on-axis density of 2 × 1017 cm−3. Similarly, Vay et al.33 have also shown the possibility of obtaining 10-GeV electron bunch from a 40-J laser pulse, externally guided through a 1.5-m-long plasma channel with on-axis 1 × 1017 cm−3 plasma density. In comparison we have shown that two-color dual-stage LWFA can deliver same energies in one-order shorter plasma length. It is worth to mention the concerns on the transverse offset between the two pulse and plasma guiding structure. Such pulse needed to be externally guided through a 1-m long plasma channel with on-axis density of 2 × 1017 cm−3. Similarly, Vay et al.33 have also shown the possibility of obtaining 10-GeV electron bunch from a 40-J laser pulse, externally guided through a 1.5-m-long plasma channel with on-axis 1 × 1017 cm−3 plasma density. In comparison we have shown that two-color dual-stage LWFA can deliver same energies in one-order shorter plasma length. It is worth to mention the concerns on the transverse offset between the two pulse and plasma guiding structure. Acknowledgementsh g This work was supported by the Institute for Basic Science Korea (Project code: IBS-R012-D1) and Korea Institute of Science and Technology Information (Project code: KSC-2015-C2-030). HTK would like to acknowledge the support from GIST Research Institute (GRI) grant 2018. Author Contributions V.B.P. and H.T.K. proposed the concept. V.B.P. developed the theory and performed the simulations. H.T.K. and C.H.N. supervised the physical interpretations. All authors contributed to interpret physical process and write the paper. References 14, 023057 (2012). ( ) 26. Kalmykov, S. Y., Beck, A., Davoine, X., Lefebvre, E. & Shadwick, B. A. Laser plasma acceleration with a negatively chirped pulse: all-optical control over dark current in the blowout regime. New J. Phys. 14, 033025 (2012). p g y 27. Kim, H. T. et al. Stable multi-GeV electron accelerator driven by waveform-controlled PW las 27. Kim, H. T. et al. Stable multi-GeV electron accelerator driven by waveform-controlled PW laser pulses. Sci. Rep. 7, 10203 (201 27. Kim, H. T. et al. Stable multi-GeV electron accelerator driven by waveform-controlled PW laser pulses. Sci. Rep. 7, 10203 (2017). 28. Zeng, M. et al. Multichromatic narrow-energy-spread electron bunches from laser-Wakefield acceleration with dual-color lasers. Phys. Rev. Lett. 114, 084801 (2015). y 9. Xu, X. L. et al. Nanoscale Electron Bunching in Laser-Triggered Ionization Injection in Plasma Accelerators. Phys. Rev. Lett. 117 034801 (2016). 0. Tzoufras, M., Tsung, F. S., Mori, W. B. & Sahai, A. A. Improving the self-guiding of an ultraintense laser by tailoring its longitudina profile. Phys. Rev. Lett. 113, 245001 (2014). i 31. Kalmykov, S. Y., Davoine, X., Lehe, R., Lifschitz, A. F. & Shadwick, B. A. Optical control of electron phase space in plasma accelerators with incoherently stacked laser pulses. Phys. Plasmas 22, 056701 (2015).h y p y 2. Decker, C. D., Mori, W. B., Tzeng, K.-C. & Katsouleas, T. The evolution of ultra-intense, short-pulse lasers in underdense plasmas Phys. Plasmas 3, 2047 (1996). 33. Vay, J. L. et al. Modeling of 10 GeV-1 TeV laser-plasma accelerators using Lorentz boosted simulations. Phys. Plasmas 18, 123103 (2011).i 34. Pathak, V. B., Vieira, J., Silva, L. O. & Nam, C. H. Laser dynamics in transversely inhomogeneous plasma and its relevance to wake fi eld acceleration. Plasma Phys. Control. Fusion 60, 054001 (2018). y 35. Luo, J. et al. A compact tunable polarized X-ray source based on laser-plasma helical undulators. Sci. Rep. 6, 1–8 (2016). 6 ScIentIfIc REPOrts \| (2018) 8:11772 \| DOI:10.1038/s41598-018-30095-4 www.nature.com/scientificreports/ ScIentIfIc REPOrts \| (2018) 8:11772 \| DOI:10.1038/s41598-018-30095-4 Additional Information Supplementary information accompanies this paper at https://doi.org/10.1038/s41598-018-30095-4 Competing Interests: The authors declare no competing interests. Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Cre- ative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not per- mitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. © The Author(s) 2018 ScIentIfIc REPOrts \| (2018) 8:11772 \| DOI:10.1038/s41598-018-30095-4 7
https://openalex.org/W2951793259	https://www.aanda.org/articles/aa/pdf/2018/10/aa32777-18.pdf	English	null	Period spacings in red giants: IV. Toward a complete description of the mixed-mode pattern	null	2,018	cc-by	21,281	1. Introduction limited to stars on the red giant branch (RGB). Two main rea- sons explain this restriction: ﬁrst, the oscillation spectra beneﬁt from a better relative frequency resolution for this evolutionary stage; second, the oscillation spectra remain simple, with rota- tional splittings smaller than period spacings. When stars evolve, these features become intricate, so that confusion is possible. For the most evolved stars, mixed modes are no longer observable (e.g., Baudin et al. 2012; Mosser et al. 2013; Stello et al. 2014). limited to stars on the red giant branch (RGB). Two main rea- sons explain this restriction: ﬁrst, the oscillation spectra beneﬁt from a better relative frequency resolution for this evolutionary stage; second, the oscillation spectra remain simple, with rota- tional splittings smaller than period spacings. When stars evolve, these features become intricate, so that confusion is possible. For the most evolved stars, mixed modes are no longer observable (e.g., Baudin et al. 2012; Mosser et al. 2013; Stello et al. 2014). Probing the cores of stars is diﬃcult since, generally, stellar information arises from their photosphere. Fortunately, astero- seismology of evolved stars reveals stellar interiors in a unique and powerful way: gravity waves that propagate throughout the core couple with pressure waves and construct mixed modes that can be observed (Beck et al. 2011; Bedding et al. 2011; Benomar et al. 2014). The measurement of the global seismic properties of these mixed modes then carries unique information on the core structure (e.g., Montalbán et al. 2013; Lagarde et al. 2016; Bossini et al. 2015, 2017). Observations with the space missions CoRoT and Kepler have provided the measurement of the asymptotic period spacings (Mosser et al. 2012b; Vrard et al. 2016), of the diﬀerential-rotation proﬁle in red giants (Beck et al. 2012; Deheuvels et al. 2014, 2015), and of the core rotation for about 300 stars analyzed by Mosser et al. (2012c). The understanding of any complicated mixed-mode oscil- lation pattern must be based on an unambiguous identiﬁcation of the modes. Up to now, the most eﬃcient method has relied on the use of the asymptotic expansion, completed by a clear description of the inﬂuence of rotation (Mosser et al. 2015). New insights on rotation were provided by the analysis depicted in Gehan et al. (2016), who have developed a methodology to measure rotational splittings in an automated way; Gehan et al. ⋆Full Table 1 is only available at the CDS via anonymous ftp to cdsarc.u-strasbg.fr (130.79.128.5) or via http://cdsarc. u-strasbg.fr/viz-bin/qcat?J/A+A/618/A109 A&A 618, A109 (2018) https://doi.org/10.1051/0004-6361/201832777 c⃝ESO 2018 A&A 618, A109 (2018) https://doi.org/10.1051/0004-6361/201832777 c⃝ESO 2018 Astronomy & Astrophysics Astronomy & Astrophysics A&A 618, A109 (2018) https://doi.org/10.1051/0004-6361/201832777 c⃝ESO 2018 Open Access article, published by EDP Sciences, under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. ABSTRACT Context. Oscillation modes with a mixed character, as observed in evolved low-mass stars, are highly sensitive to the physical properties of the innermost regions. Measuring their properties is therefore extremely important to probe the core, but requires some care, due to the complexity of the mixed-mode pattern. p y p m of this work is to provide a consistent description of the mixed-mode pattern of low-mass stars, based on the asymptoti e also study the variation of the gravity oﬀset εg with stellar evolution. p y g y g Methods. We revisit previous works about mixed modes in red giants and empirically test how period spacings, rotational splittings, mixed-mode widths, and heights can be estimated in a consistent view, based on the properties of the mode inertia ratios. Results. From the asymptotic ﬁt of the mixed-mode pattern of a large set of red giants at various evolutionary stages, we derive unbiased and precise asymptotic parameters. As the asymptotic expansion of gravity modes is veriﬁed with a precision close to the frequency resolution for stars on the red giant branch (10−4 in relative values), we can derive accurate values of the asymptotic parameters. We decipher the complex pattern in a rapidly rotating star, and explain how asymmetrical splittings can be inferred. We also revisit the stellar inclinations in two open clusters, NGC 6819 and NGC 6791: our results show that the stellar inclinations in these clusters do not have privileged orientation in the sky. The variation of the asymptotic gravity oﬀset with stellar evolution is investigated in detail. We also derive generic properties that explain under which conditions mixed modes can be observed. Key words. stars: evolution – stars: interiors – stars: oscillations LESIA, Observatoire de Paris, PSL Research University, CNRS, Université Pierre et Marie Curie, Université Paris Diderot, 92195 Meudon, France e-mail: benoit.mosser@obspm.fr LESIA, Observatoire de Paris, PSL Research University, CNRS, Université Pierre et Marie Curie, Université Paris Diderot, 92195 Meudon, France e-mail: benoit.mosser@obspm.fr Received 6 February 2018 / Accepted 21 July 2018 Received 6 February 2018 / Accepted 21 July 2018 Received 6 February 2018 / Accepted 21 July 2018 Period spacings in red giants⋆ IV. Toward a complete description of the mixed-mode pattern B. Mosser, C. Gehan, K. Belkacem, R. Samadi, E. Michel, and M.-J. Goupil Observatoire de Paris, PSL Research University, CNRS, Université Pierre et Marie Curie, Université Paris Diderot, d F 2.2. Seismic parameters tan θp = q tan θg, (1) (1) tan θp = q tan θg, tan θp = q tan θg, With ζ, we now intend to express the diﬀerent seismic parame- ters. where q is the coupling factor (Mosser et al. 2017b). The phases are related to the large separation ∆ν and the period spacing ∆Π1. The most convenient expressions of the phase refer respectively to the pure1 p and g mode spectra 2.1. Asymptotic expansion The asymptotic expansion of mixed modes is an implicit relation between the phases θp and θg of the pressure- and gravity-wave contributions to the mixed modes, respectively. It reads 2. Mixed-mode parameters ζ = Icore Ienv + Icore , (7) ζ = Icore Ienv + Icore , (7) Following the work of Shibahashi (1979) and Unno et al. (1989), we derived asymptotic expansions of mixed modes for diﬀer- ent seismic parameters: eigenfrequencies (Mosser et al. 2012b), period spacings (Christensen-Dalsgaard 2012), rotational split- tings (Goupil et al. 2013; Deheuvels et al. 2015), and mode widths and mode heights (Grosjean et al. 2014; Belkacem et al. 2015b,a; Mosser et al. 2017a). Here, we intend to revisit all these parameters that depict the mixed-mode spectrum in order to pro- vide a more precise and uniﬁed view. and assuming that the envelope contribution of a mixed mode is similar to the inertia of the closest radial mode (Ienv ≡Inp,0), we ﬁnd that the inertia of the dipole mode with mixed radial order n varies as In,1 = Inp,0 1 −ζ · (8) In,1 = Inp,0 1 −ζ · (8) For the sake of simplicity, we use hereafter the abridged notation In for the inertia of the dipole mixed modes and I0 for the clos- est radial modes, and follow the same convention for the mode heights and widths. 1. Introduction (2017 and 2018) showed how rapid rotation can be addressed eﬃciently. This eﬃciency derives from the use of stretched oscil- lation spectra. Most of the previous studies are based on the measurement and analysis of global seismic parameters, such as the asymp- totic large separation ∆ν and the asymptotic period spacings ∆Π1 (e.g., Miglio et al. 2017). It is now time to access the proper- ties of individual frequencies in red giants. Up to now, most of the studies (e.g., Baudin et al. 2012; Di Mauro et al. 2016) were In this work, we ﬁrst examine in Sect. 2 how the diﬀerent fre- quency spacings in the asymptotic mixed-mode expansion can be expressed as a function of the mode inertia. New expressions are proposed for the mixed-mode spacings and rotational split- tings. Case studies are examined in Sect. 3 to test and validate these expressions. In Sect. 4, we take advantage of the preci- sion of the ﬁts to derive accurate asymptotic period spacings and gravity oﬀsets; for the ﬁrst time, we can exhibit the global evo- A109, page 1 of 25 A&A 618, A109 (2018) where N = ∆ν/(ν2∆Π1) is the density of gravity modes com- pared to pressure modes, in other words the number of mixed modes in a ∆ν-wide interval. Compared to the original form pre- sented in Mosser et al. (2015), the rapidly varying phase θg has been replaced by a function of θp that varies in a smooth way. lution of these gravity oﬀsets as a function of stellar evolution. New insights on the rotational splittings are proposed in Sect. 5; in particular, we show how the asymptotic expansion can be used to provide priors based upon physical assumptions for any ﬁtting code used later in the analysis. Finally, we assess the conditions for observing mixed modes, based on global asymptotic param- eters only (Sect. 6). Section 7 is devoted to our conclusions. p y p y As demonstrated by Goupil et al. (2013) and used by sub- sequent work (Benomar et al. 2014; Deheuvels et al. 2015), the function ζ is connected to the inertia of mixed modes. Introduc- ing the contributions of the envelope and of the core, 2.2.1. Period spacing Following Christensen-Dalsgaard (2012) and Mosser et al. (2015), period spacings can be expressed as θg = π 1 ∆Π1 1 ν −1 νg ! , (2) θp = πν −νp ∆νp , (3) (2) (2) ∆P = Pn −Pn+1 = ζ ∆Π1. (9) (9) (3) This expression is however ambiguous, since ζ may vary signif- icantly between the periods Pn+1 and Pn (> Pn+1). Therefore, we prefer to consider the expression resulting from the integration of Eq. (5) where νp and νg are the asymptotic frequencies of pure pressure and gravity modes, respectively, and ∆νp is the frequency diﬀer- ence between the consecutive pure pressure radial modes with radial orders np and np + 1. In this work, we consider that the radial modes and pure dipole pressure modes obey the universal red giant oscillation pattern (Mosser et al. 2011b) and that the dipole gravity modes follow the asymptotic comb-like pattern ∆P = Pn −Pn+1 = ∆Π1 Z n+1 n ζ(ν) dn = ∆Π1 ⟨ζ⟩n, (10) (10) where we consider that the mixed-mode radial order n is a continuous variable deﬁned by dn = dτ/∆Π1, where τ is the stretched period introduced by Mosser et al. (2015); i.e., 1 νg = (−ng + εg) ∆Π1, (4) 1 νg = (−ng + εg) ∆Π1, (4) dτ = dν ζ ν2 · (11) (11) where ∆Π1 is the period spacing and εg is the gravity oﬀset. where ∆Π1 is the period spacing and εg is the gravity oﬀset. g Mosser et al. (2015) derived that the variation of the oscilla- tion period P with the mixed radial order n writes In fact, n takes consecutive integer values for each mixed mode. In this work, we use an estimate of n = np + ng derived from the pressure and gravity radial orders; np is derived from the univer- sal red giant oscillation pattern (Mosser et al. 2011b), whereas ng is given by In fact, n takes consecutive integer values for each mixed mode. In this work, we use an estimate of n = np + ng derived from the pressure and gravity radial orders; np is derived from the univer- sal red giant oscillation pattern (Mosser et al. 2011b), whereas ng is given by dP dn = ζ ∆Π1. (5) dP dn = ζ ∆Π1. 2.4. Synthetic mixed-mode pattern ⟨ζ⟩n = Z n+1 n ζ dn = Z νn+1,m νn,m dν ∆Π1 ν2 , (17) ⟨ζ⟩m = Z ±1 0 ζ dm = Z νn,±1 νn,0 dν ∆Π1 ν2 , (18) (17) (17) The previous ingredients can be used to depict an oscillation pat- tern. Figure 1 shows the synthetic spectrum of a typical star on the low RGB, based on Eq. (16) for the rotational split- tings, on Eq. (19) for the mode widths, and on Eq. (21) for the mode heights of unresolved modes. This spectrum resem- bles the description derived by Grosjean et al. (2014) from non- adiabatic computations, with a time-dependent treatment of convection which provides the lifetimes of radial and non-radial mixed modes. (18) to account for the period spacings and rotational splittings. The relevance of ⟨ζ⟩n is already proven by Eqs. (10) and (11), whereas the relevance of ⟨ζ⟩m has yet to be demonstrated. If we succeed, we will also have understood the relevance of the use of stretched periods for analyzing the mixed modes (Eq. (11)). 2.2.1. Period spacing (9): using ⟨ζ⟩n is necessarily more accurate than using ζ for computing period spacings. This evident relation justiﬁes the relevance of Eq. (10) instead of Eq. (9): using ⟨ζ⟩n is necessarily more accurate than using ζ for computing period spacings. Fig. 1. Description of the radial order closest to νmax of the oscillation power spectrum of a typical RGB star. Frequencies, widths, and heights are estimated according to the function ζ. Quadupole and radial modes are plotted in gray, dipole mixed modes in dark blue (m = −1), light blue (m = 0), or purple (m = 1), respectively. 2.2.2. Rotational splitting When, for non-resolved mixed modes, the width Γn is less than the frequency resolution δfres, a dilution factor must be considered (Dupret et al. 2009). It expresses δνrot = δνrot,core Z νn,m νn,0 ζ dm = δνrot,core ⟨ζ⟩m, (16) (16) Hn = π 2 Γn δfres H0, (21) (21) where δνrot,core ≡δνrot,g. As for the radial order n in Eq. (10), we consider the azimuthal order m as a continuous variable varying from 0 to ±1. So, we have introduced two mean values of ζ, when radial modes are resolved, which is the common case. 2 Since we consider dipole modes only, we use a simpliﬁed notation νn,m instead of νn,ℓ,m. 2.2.2. Rotational splitting As introduced by Goupil et al. (2013), the function ζ is used to express the mixed-mode rotational splitting as a function of the mean rotational splittings related to pure gravity or pure pressure modes: From this relation, we verify that mixed modes have smaller mode widths than radial modes. However, we recall that a fam- ily of stars behave diﬀerently, when mixed modes are depressed because of an extra damping in the radiative inner region (Mosser et al. 2012a, 2017a; García et al. 2014). δνrot = ζ δνrot,g + (1 −ζ) δνrot,p. (15) (15) As shown by subsequent works (e.g., Deheuvels et al. 2014; Di Mauro et al. 2016; Triana et al. 2017), it is diﬃcult to derive from the observed rotational splittings more than these two mean quantities. As shown by subsequent works (e.g., Deheuvels et al. 2014; Di Mauro et al. 2016; Triana et al. 2017), it is diﬃcult to derive from the observed rotational splittings more than these two mean quantities. ( ) From Belkacem et al. (2015a) we also derive that the ampli- tude of a resolved dipole mixed mode is Again, we have to solve the ambiguity of the meaning of ζ in Eq. (15), since we can either consider the value2 ζ(νn,0), in the framework of the perturbation of the non-rotating frequency νn,0, or ζ(νn,m), considering that the inertia to be considered cor- responds to the actual frequency νn,m. By analogy with the equa- tion dealing with the period spacing, we propose to rewrite the rotational splitting δνrot = νn,m −νn,0, in the limit case where the mean envelope rotation is negligible compared to the mean core rotation, as A2 n = A2 0 (1 −ζ), A2 n = A2 0 (1 −ζ), (20) (20) when the geometrical factor that conducts to a visibility of about 1.54 for red giant dipole modes (Mosser et al. 2012a, 2017a) is omitted. Such amplitudes correspond to similar heights for radial and dipole modes since A2 = πΓH/2. When, for non-resolved mixed modes, the width Γn is less than the frequency resolution δfres, a dilution factor must be considered (Dupret et al. 2009). It expresses when the geometrical factor that conducts to a visibility of about 1.54 for red giant dipole modes (Mosser et al. 2012a, 2017a) is omitted. Such amplitudes correspond to similar heights for radial and dipole modes since A2 = πΓH/2. 2.2.1. Period spacing (5) (5) A convenient way to write the parameter ζ is (Hekker & Christensen-Dalsgaard 2017) A convenient way to write the parameter ζ is (Hekker & Christensen-Dalsgaard 2017) (6) ormed ng = − $ 1 ν ∆Π1 −1 4 % on the RGB, (12) ng = − $ 1 ν ∆Π1 + 1 4 % in the red clump, (13) (12) ζ(ν) = 1 + q N 1 q2 cos2 θp + sin2 θp  −1 , (6) ng = − $ 1 ν ∆Π1 −1 4 % on the RGB, (12) ng = − $ 1 ν ∆Π1 + 1 4 % in the red clump, (13) ζ(ν) = 1 + q N 1 q2 cos2 θp + sin2 θp  −1 , (6) n n (6) (13) 1 Pure p (or g) modes are hypothetical modes that could be formed in the pressure (or gravity) cavity without any coupling with the other cavity. 1 Pure p (or g) modes are hypothetical modes that could be formed in the pressure (or gravity) cavity without any coupling with the other cavity. where the correcting terms ±1/4 that depend on the evolutionary stage are justiﬁed in Sect. 4.4. They diﬀer by 1/2, as depicted A109, page 2 of 25 B. Mosser et al.: Mixed modes Fig. 1. Description of the radial order closest to νmax of the oscillation power spectrum of a typical RGB star. Frequencies, widths, and heights are estimated according to the function ζ. Quadupole and radial modes are plotted in gray, dipole mixed modes in dark blue (m = −1), light blue (m = 0), or purple (m = 1), respectively. by the asymptotic relation (e.g., Tassoul 1980; Benomar et al. 2013). In red giants, the high density N of mixed modes implies that \|ng\| ≫np, so that the mixed-mode orders are negative. by the asymptotic relation (e.g., Tassoul 1980; Benomar et al. 2013). In red giants, the high density N of mixed modes implies that \|ng\| ≫np, so that the mixed-mode orders are negative. \| g\| p g From the deﬁnition of the stretched period, Eq. (10) reduces to g p From the deﬁnition of the stretched period, Eq. (10) reduces to to ∆P = Z νn+1 νn dν ν2 . (14) (14) This evident relation justiﬁes the relevance of Eq. (10) instead of Eq. 3.2. Individual ﬁtting procedure When ﬁtting individually mixed modes, we aim at testing the validity of the asymptotic expression, but not at reaching the ulti- mate precision, which is the role of a dedicated ﬁt of individual modes (e.g., Gaulme et al. 2009). Therefore, in order to simplify the ﬁt, we supposed (and checked a posteriori) that all multiplets can be ﬁtted independently. This is not the case in all red giant spectra, but it is veriﬁed for most stars on the early RGB or in the red clump. where εp is the acoustic oﬀset, nmax = νmax/∆ν −εp, and α = 0.076/nmax. The parameter d01 is function of the large separa- tion, under the form A+B log ∆ν (where ∆ν is expressed in µHz), with A = 0.0553 and B = −0.036, as determined from the large- scale analysis along the RGB conducted by Mosser et al. (2014). The accurate determination of d01 is crucial for the determina- tion of the pure dipole pressure modes, hence for the determina- tion of the minima of the function ζ. In that respect, the small modulation of the radial-mode pattern induced by the sound- speed glitches (Miglio et al. 2010; Vrard et al. 2015) must be considered also. Therefore, we ﬁt the actual position of the radial modes ﬁrst, then use them to reﬁne the pure pressure dipole- mode frequencies according to From the asymptotic ﬁt, we identify in the background- corrected spectrum the power excess associated to each mode. Then, we determine the central frequency of the peak as the barycenter of the power excess. The height H and full width at half maximum Γ are simultaneously derived from the Lorentzian ﬁt of the mode. We use Eqs. (19) and (21) as priors. Modes are ﬁtted individually when the mode density is low, or simultane- ously when the Lorentzians used as priors overlap. νp = νnp,0 + νnp+1,0 /2 + d01 νnp+1,0 −νnp,0 . (23) (23) The ﬁtted spectrum and the seismic parameters of KIC 6144777, used as a ﬁrst study case, are given in Fig. 2 and in Table A.1. We note the large agreement between the observed and asymptotic peaks. As in other stars showing a seismic signal with a high signal-to-noise ratio (S/N), outliers with a height-to- background value R higher than 7 are present. 3.2. Individual ﬁtting procedure Their detection does not invalidate the method presented above: they correspond either to ℓ= 2 or 3 mixed modes, possibly also to ℓ= 4 modes, or to aliases (since the duty cycle is about 93%), or even to noise since the detection of 1 noisy peak with R ≥8 is expected in a 30-µHz frequency range after 4 years of observation, assuming that the noise statistic is a χ2 with two degrees of freedom. The background parameters, derived as in Mosser et al. (2012a), d h l i ib i i h f The background parameters, derived as in Mosser et al. (2012a), are used to correct the granulation contribution in the frequency range around νmax. Hence, mixed modes can be automatically identiﬁed in frequency ranges that have no radial and quadrupole modes when their heights are signiﬁcantly above the back- ground. The automatic selection of the modes relies on a sta- tistical test: the height-to-background ratio of the modes must be higher than a threshold level Rp in order to reject the null hypoth- esis to a low probability p. According to Appourchaux et al. (2006), the relation between Rp and p depends for long-lived modes on the observation duration Tobs and on the width ∆ν of the frequency range where a mode is expected. This relation expresses The quality of the ﬁt is shown by the small residuals between the observed frequencies and the asymptotic ﬁts (Fig. 3); we note that these residuals are comparable to the uncertainties, derived from Libbrecht (1992) or slightly larger when the quality of the ﬁt may be aﬀected by the high mode density. These residuals are of about the frequency resolution. When pressure-dominated mixed modes are excluded, the standard deviation of the asymp- totic ﬁt is 11 nHz. This value represents 1.3 times the frequency resolution δfres, or ∆ν/1000, or a relative precision at νmax of about 10−4. The quality of the ﬁts is based on a small num- ber of parameters: the radial mode frequencies, the mean loca- tion d01 of the expected pure pressure dipole modes, and four asymptotic parameters: the period spacing ∆Π1, the coupling factor q, gravitational oﬀset εg, and the mean core rotational splitting δνrot. Residuals reach maximum values near the Rp ≃ln Tobs∆ν p , (24) (24) when expressed in noise unit. 3. Case studies The work performed by the gas during one oscillation cycle is the same for all modes, associated with surface damping, when the radiative damping in the Brunt-Väisälä cavity is considered as negligible. Hence, Benomar et al. (2014) have estimated that the mode width of the mixed modes writes In this section, we use RGB stars showing clear oscillation spectra as case studies, in order to test the description of the mixed-mode spacings, widths, heights, and rotational splittings, which were previously introduced. The ﬁrst steps consist in identifying their oscillation spectra and in ﬁt- ting as many dipole mixed modes as possible. One of the two stars considered here, KIC 6144777 was already inves- tigated in many previous articles (e.g., Corsaro et al. 2015; García Saravia Ortiz de Montellano et al. 2018). The other one, KIC 3955033, was less studied since it shows a complicated Γn = Γ0 I0 In = Γ0 (1 −ζ). (19) Γn = Γ0 I0 In = Γ0 (1 −ζ). (19) 2 Since we consider dipole modes only, we use a simpliﬁed notation νn,m instead of νn,ℓ,m. A109, page 3 of 25 A109, page 3 of 25 A&A 618, A109 (2018) patternbased onsecure modescanbe usedto searchfor long-lived mixed modes in narrow frequency ranges. We beneﬁt from the fact that the asymptotic ﬁt is precise and enables to search for thin modes in a frequency range ∆ν narrower than 0.1 µHz. Therefore, a threshold of 7 is enough for rejecting the null hypothesis at the 1%-level for these modes whose detection beneﬁts from the infor- mation gained by larger peaks. The thin mode widths (Eq. (19)) are of great use to map the observed spectrum: a thin gravity- dominated mixed mode must be found in the close vicinity, less than 4 times the mode width, of its expected position. For unre- solved peaks, this condition is relaxed to 4 times the frequency resolution. The global seismic parameters of the gravity compo- nent are then derived from the methods described in Vrard et al. (2016) and Mosser et al. (2017b), with a least-square ﬁt between the observed and asymptotic patterns. mixed-mode spectrum; it belongs to the list of red giants with period spacings automatically computed by Vrard et al. (2016). We used data downloaded from the KASOC site3, processed using the Kepler pipeline developed by Jenkins et al. 3. Case studies (2010), and corrected from outliers, occasional jumps, and drifts (see García et al. 2011, for details). 3.1. Identiﬁcation of the mixed modes The location of the mixed modes primarily relies on the ﬁrm identiﬁcation of the pure pressure-mode spectrum. The determi- nation of the large separation ∆ν, ﬁrst derived from the enve- lope autocorrelation function (Mosser & Appourchaux 2009), is based on the universal red giant oscillation pattern. This method provides the eﬃcient identiﬁcation of the radial modes and helps to locate the frequency ranges where mixed modes cannot be mistaken for radial or quadrupole modes. For ℓ= 1 modes, the second-order asymptotic expansion writes At this stage, global seismic parameters are measured and mixed modes are identiﬁed, so that it is possible to measure their individual properties. νp = np + εp + 1 2 + d01 + α 2 [np −nmax]2 ! ∆ν, (22) (22) 3 http://kasoc.phys.au.dk 3.3.1. Period spacings Period spacings were ﬁtted with diﬀerent functions of ζ, accord- ing either to the integrated value ⟨ζ⟩n (Eq. (10)) or to the arith- metical mean ζ = (ζn + ζn+1)/2. The resulting plots are shown in Fig. 4. When ⟨ζ⟩n is not used, one remarks that the ∆P(ζ) relation shows a modulation that results from the concavity of ζ. When ζ is close to unity for gravity-dominated mixed modes, no modulation is seen; in the range [0.7, 0.9], where the function is convex, the period spacings are larger than predicted; below 0.7, where the function is concave, the period spacings are smaller than expected. The relation between ∆P and ⟨ζ⟩n does not show such a modulation. Furthermore, the ﬁt with ⟨ζ⟩n is nearly lin- ear, with residuals two times smaller. From this comparison, we conﬁrm that the use of ⟨ζ⟩n is preferable for ﬁtting the period spacings. Fig. 3. Relative residuals, multiplied by 1000, between the observed and asymptotic mixed-mode frequencies in KIC 6144777. The color of the symbols indicates the azimuthal order: dark blue squares for m = −1, light blue diamonds for m = 0, or purple triangles for m = 1; 1-σ uncer- tainties are also shown. The dashed line corresponds to a perfect ﬁt. The dotted lines show the frequency resolution plus an extra-modulation ∆ν(1 −ζ)/100, which is empirically used to deﬁne the quality of the ﬁt. 3.2. Individual ﬁtting procedure The gray dashed lines indicate the two thresholds used in this work, corresponding to height-to-background ratios of 7 and 10. 3.3. Relationships with ζ Fig. 3. Relative residuals, multiplied by 1000, between the observed and asymptotic mixed-mode frequencies in KIC 6144777. The color of the symbols indicates the azimuthal order: dark blue squares for m = −1, light blue diamonds for m = 0, or purple triangles for m = 1; 1-σ uncer- tainties are also shown. The dashed line corresponds to a perfect ﬁt. The dotted lines show the frequency resolution plus an extra-modulation ∆ν(1 −ζ)/100, which is empirically used to deﬁne the quality of the ﬁt. With the identiﬁcation of the mixed-mode pattern, we aim to verify the relevance of the use of ⟨ζ⟩n for the period spacings, to test the relevance of ⟨ζ⟩m for the rotational splittings, and further test the predictions for the mode widths and heights. 3.2. Individual ﬁtting procedure This situation applies here, since the precise identiﬁcation of the mixed-mode pattern is based on gravity-dominated mixed modes. With 4-year observations and the search of a couple of modes in a frequency range ∆ν = ∆ν/N, thethresholdistypically10forasecureprobabilityrejectionatthe 10−2 level. In practice, mixed modes with a height-to-background ratio higher than 10 are used to initiate the ﬁt. A lower threshold is enough for the ﬁnal agreement, when the synthetic mixed-mode when expressed in noise unit. This situation applies here, since the precise identiﬁcation of the mixed-mode pattern is based on gravity-dominated mixed modes. With 4-year observations and the search of a couple of modes in a frequency range ∆ν = ∆ν/N, thethresholdistypically10forasecureprobabilityrejectionatthe 10−2 level. In practice, mixed modes with a height-to-background ratio higher than 10 are used to initiate the ﬁt. A lower threshold is enough for the ﬁnal agreement, when the synthetic mixed-mode A109, page 4 of 25 B. Mosser et al.: Mixed modes Fig. 2. Fit of the oscillation pattern of the RGB star KIC 6144777, showing the pressure radial orders np from 9 to 12. The power spectrum density has been divided by the ﬁt of the background. Radial and quadrupole modes are highlighted in red and green. The expected locations of dipole mixed modes are labelled with their mixed radial orders. When detected, mixed modes are highlighted in dark blue (m = −1), light blue (m = 0), or purple (m = 1). ℓ= 3 modes, which are also mixed, are located near the abscissa 0.22; extra peaks in the range [−0.2, −0.05] are mixed quadrupole modes. The gray dashed lines indicate the two thresholds used in this work, corresponding to height-to-background ratios of 7 and 10. B. Mosser et al.: Mixed modes B. Mosser et al.: Mixed modes Fig. 2. Fit of the oscillation pattern of the RGB star KIC 6144777, showing the pressure radial orders np from 9 to 12. The power spectrum density has been divided by the ﬁt of the background. Radial and quadrupole modes are highlighted in red and green. The expected locations of dipole mixed modes are labelled with their mixed radial orders. When detected, mixed modes are highlighted in dark blue (m = −1), light blue (m = 0), or purple (m = 1). ℓ= 3 modes, which are also mixed, are located near the abscissa 0.22; extra peaks in the range [−0.2, −0.05] are mixed quadrupole modes. 3.3.2. Rotational splittings Mean rotation splittings of the RGB star KIC 3955033. Top: plot as a function of ζ. Bottom: plot as a function of the mean value ⟨ζ⟩m. The colors code the azimuthal orders; the dashed line indicates the 1:1 relation; 1-σ uncertainties on both the splittings and the mean values of ζ are indicated by vertical and horizontal error bars. variations with either ζ or ⟨ζ⟩m. We therefore performed the ﬁt of the star KIC 3955033 (Fig. A.3), which shows a much more rapid rotation (Fig. 5). From the comparison of δνrot(ζ) and δνrot(⟨ζ⟩m), we derive that this latter expression is more conve- nient since it provides a χ2 ten times smaller than when using ζ, associated with a much more precise estimate of the core rotation: δνrot,core = 765 ± 10 nHz with ⟨ζ⟩m, versus δνrot,core = 730±50 nHz with ζ. From this test, we conclude positively about the relevance of the use of ⟨ζ⟩m for the rotational splittings. As shown by Mosser et al. (2015), Eq. (20) has a strong theo- retical justiﬁcation, since it expresses the conservation of energy: the sum of all the energy distributed in the mixed modes corre- sponds to the energy expected in the single pure pressure mode that should exist in absence of any coupling. So, our result is in line with the ﬁndings of Mosser et al. (2012a), who measured that, except for depressed modes, the observed total visibility of dipole modes matches the theoretical expectations. variations with either ζ or ⟨ζ⟩m. We therefore performed the ﬁt of the star KIC 3955033 (Fig. A.3), which shows a much more rapid rotation (Fig. 5). From the comparison of δνrot(ζ) and δνrot(⟨ζ⟩m), we derive that this latter expression is more conve- nient since it provides a χ2 ten times smaller than when using ζ, associated with a much more precise estimate of the core rotation: δνrot,core = 765 ± 10 nHz with ⟨ζ⟩m, versus δνrot,core = 730±50 nHz with ζ. From this test, we conclude positively about the relevance of the use of ⟨ζ⟩m for the rotational splittings. Due to the stochastic nature of the excitation, the mode heights show a large spread (Fig. 7). Dips in the distributions occur when modes are not resolved. It is however clear in Fig. 7 bottom that the dipole mode heights follow the radial distribu- tion according to the trend of Eq. (21). 3.3.2. Rotational splittings pressure-dominated mixed modes: there, deviations of about ∆ν/200 are observed, to be compared to the amplitudes of pres- sure glitches of about ∆ν/40 (Vrard et al. 2015). We suspect that these residuals are mostly due to the variation of the parameter d01 with frequency. We performed similar test for the rotational splittings. We a pri- ori excluded a dependence on ζ(νn,0), since we clearly observe asymmetrical splittings (see below, Sect. 5.1) that cannot be reproduced with ζ(νn,0). In fact, the rotation rate of KIC 6144777 is not important enough to observe any diﬀerence between the A109, page 5 of 25 A&A 618, A109 (2018) Fig. 5. Mean rotation splittings of the RGB star KIC 3955033. Fig. 5. Mean rotation splittings of the RGB star KIC 3955033. Top: plot as a function of ζ. Bottom: plot as a function of the mean value ⟨ζ⟩m. The colors code the azimuthal orders; the dashed line indicates the 1:1 relation; 1-σ uncertainties on both the splittings and the mean values of ζ are indicated by vertical and horizontal error bars. Fig. 4. Period spacings of the RGB star KIC 6144777. Top: plot as a function of the arithmetical mean value (ζ(νn)+ζ(νn+1))/2. Bottom: plot as a function of the mean value ⟨ζ⟩n. The colors code the azimuthal orders, as in Fig. 2; the dashed line indicates the 1:1 relation; 1-σ uncer- tainties on both the spacings and the mean values of ζ are indicated by vertical and horizontal error bars. Fig. 5. Mean rotation splittings of the RGB star KIC 3955033. Fig. 5. Mean rotation splittings of the RGB star KIC 3955033. Top: plot as a function of ζ. Bottom: plot as a function of the mean value ⟨ζ⟩m. The colors code the azimuthal orders; the dashed line indicates the 1:1 relation; 1-σ uncertainties on both the splittings and the mean values of ζ are indicated by vertical and horizontal error bars. Fig. 4. Period spacings of the RGB star KIC 6144777. Top: plot as a function of the arithmetical mean value (ζ(νn)+ζ(νn+1))/2. Bottom: plot as a function of the mean value ⟨ζ⟩n. The colors code the azimuthal orders, as in Fig. 2; the dashed line indicates the 1:1 relation; 1-σ uncer- tainties on both the spacings and the mean values of ζ are indicated by vertical and horizontal error bars. Fig. 5. Mean rotation splittings of the RGB star KIC 3955033. Fig. 5. 3.3.2. Rotational splittings We note that all mixed modes associated with a given pressure radial order show a sys- tematic behavior. For instance, all mixed modes of KIC 6144777 in the frequency range [137, 143 µHz] associated with the pres- sure mode np = 11 show lower amplitudes than expected from the Gaussian ﬁt of the power excess. Such a behavior recalls us that the excitation of a mixed mode is due to its acoustic component. 3.3.3. Widths, amplitudes, and heights As expected from Eq. (19), the mixed-mode width shows large variations: pressure-dominated mixed modes have widths comparable to those of the radial modes, contrary to gravity- dominated modes that are much thinner (Fig. 6, top panel). Figure 6 also shows the validity of Eq. (19), with the mixed- mode width proportional to (1 −ζ), except for low values where the observations resolution hampers the measurement of very thin widths. The precision of the ﬁt is limited by the stochas- tic excitation, especially for long-lived peaks: the presence or absence of signal in a single frequency bin can modify the width in large proportion. This limit added to the limitation in fre- quency resolution does not allow us to test if small additional radiative damping aﬀects the gravity-dominated mixed modes (Dupret et al. 2009; Grosjean et al. 2014). 3.4. Validation From these two case studies and from other examples shown in the appendix, we can conclude that the asymptotic ﬁts are rel- evant at all evolutionary stages, when the S/N is high enough. A109, page 6 of 25 B. Mosser et al.: Mixed modes Fig. 6. Mode widths as a function of the cyclic frequency (top) or as a function of (1 −ζ)Γ0 (bottom), for the RGB star KIC 6144777. Radial modes are plotted with square symbols and dipole mixed modes with +; 1-σ uncertainties on Γ are indicated by vertical error bars. The value 2δfres/π (Eq. (21)) plotted as a dotted line is proportional to the 4-year long frequency resolution. In the bottom plot, radial modes have been considered too, assuming they have ζ = 0 as pure pressure modes. The dashed line indicates the 1:1 relation. Fig. 7. Mode heights as a function of the cyclic frequency (top) or as a function of the radial mode height, modiﬁed when the modes are not resolved, (bottom), for the RGB star KIC 6144777. Radial modes are plotted with square symbols and dipole mixed modes with +. The Fourier spectrum is plotted in red (green) for emphasizing the radial (quadrupole) modes. The dot-dashed line provides the expected heights of dipole modes, under the assumption that the power excess mimics a Gaussian relation (dotted line). The dashed line indicates the 1:1 rela- tion. Fig. 7. Mode heights as a function of the cyclic frequency (top) or as a function of the radial mode height, modiﬁed when the modes are not resolved, (bottom), for the RGB star KIC 6144777. Radial modes are plotted with square symbols and dipole mixed modes with +. The Fourier spectrum is plotted in red (green) for emphasizing the radial (quadrupole) modes. The dot-dashed line provides the expected heights of dipole modes, under the assumption that the power excess mimics a Gaussian relation (dotted line). The dashed line indicates the 1:1 rela- tion. Fig. 6. Mode widths as a function of the cyclic frequency (top) or as a function of (1 −ζ)Γ0 (bottom), for the RGB star KIC 6144777. Radial modes are plotted with square symbols and dipole mixed modes with +; 1-σ uncertainties on Γ are indicated by vertical error bars. The value 2δfres/π (Eq. (21)) plotted as a dotted line is proportional to the 4-year long frequency resolution. 4. Asymptotic period spacings and gravity offsets In this section, we show how previous ﬁndings can be used to derive accurate period spacings. We also explore the variation of the gravity oﬀsets εg with stellar evolution. These studies rely on the determination of the pure-gravity mode pattern. 3.4. Validation In the bottom plot, radial modes have been considered too, assuming they have ζ = 0 as pure pressure modes. The dashed line indicates the 1:1 relation. So, the equations developed in Sect. 2 allow us to depict the mixed-mode spectrum with a very high accuracy, when the inte- grated values ⟨ζ⟩n and ⟨ζ⟩m are considered for the period spac- ings and the rotational splittings, respectively. Up to now, only red clump stars showing buoyancy glitches cannot be ﬁtted with a single set of parameters. So, the equations developed in Sect. 2 allow us to depict the mixed-mode spectrum with a very high accuracy, when the inte- grated values ⟨ζ⟩n and ⟨ζ⟩m are considered for the period spac- ings and the rotational splittings, respectively. Up to now, only red clump stars showing buoyancy glitches cannot be ﬁtted with a single set of parameters. are mainly based upon the noise level, with Kepler magnitudes brighter than 12 on the low RGB or 14 for more evolved stars. Following the method exposed in Sect. 3.1, we need data with a S/N high enough to allow the identiﬁcation of gravity-dominated mixed modes. When such modes are too few, measurements are impossible. This condition induces a selection bias, speciﬁcally addressed in Sect. 6. The 372 stars that were analyzed are shown in a seismic dia- gram (Fig. 8). We considered stars from the low RGB (Fig. A.1) to more evolved RGB stars (Fig. A.2). The spectrum of the evolved RGB star KIC 2443903 (Fig. A.4) corresponds to a case near the limit of visibility of gravity-dominated mixed modes, with a mode density N ≃22.4 close to the limit value above which the detection is impossible (Sect. 6). The ﬁtting process for red clump stars can be achieved only when the amplitude of the buoyancy glitches remains limited (Fig. A.5); the same limitation appears in the secondary red clump (Fig. A.6). In fact, except for red-clump stars with large buoyancy glitches (Cunha et al. 2015; Mosser et al. 2015), the asymptotic expan- sion provides a relevant ﬁt. We could then obtain precise mea- surements of the asymptotic gravity parameters in Eq. (4) and of their uncertainties for a large number of stars. We must report one exception: KIC 3216736 is the only RGB star of our 4.1. Observations Our analysis was conducted over 372 red giants at vari- ous evolutionary stages, mainly from Mosser et al. (2014) and Vrard et al. (2016), with stars also considered in Beck et al. (2012), Kallinger et al. (2012), Deheuvels et al. (2014), and Corsaro et al. (2015). Data were obtained as for the two stars considered in Sect. 3. When available, eﬀective temperatures are from APOGEE spectra (Albareti et al. 2017). Selection criteria A109, page 7 of 25 A109, page 7 of 25 A&A 618, A109 (2018) Fig. 8. Seismic diagram of the 372 red giants studied in this work, with ν−1 max used as a proxy for the luminosity. The color codes the stellar mass. Stars on the RGB are plotted with + symbols, red clump stars with ⋄. The dotted and dashed lines indicate the limit of the visibility of mixed modes for RGB and clump stars, respectively, as deﬁned by Eq. (31). Evolution tracks on the RGB, computed with MESA for solar metallic- ity (Gehan et al. 2018), are shown for the stellar masses 1.0, 1.3, 1.6, and 1.9 M⊙. The error box indicates the typical 1-σ uncertainties on Teﬀ and νmax. Fig. 9. Stretched-period échelle diagram of KIC 3216736, the only RGB star in our sample showing buoyancy glitches. The spectrum is simple, since only m = 0 dipole mixed modes are present, but shows a large- period modulation instead of the expected vertical alignment. Modes plotted in light blue are pressure dominated; extra peaks that do not follow the global trend are either ℓ= 3 modes or ℓ= 2 mixed modes. Red ﬁgures indicate the radial orders of the radial modes. For clarity, the power spectrum density is also plotted twice, top to tail. Fig. 8. Seismic diagram of the 372 red giants studied in this work, with ν−1 max used as a proxy for the luminosity. The color codes the stellar mass. Stars on the RGB are plotted with + symbols, red clump stars with ⋄. The dotted and dashed lines indicate the limit of the visibility of mixed modes for RGB and clump stars, respectively, as deﬁned by Eq. (31). Evolution tracks on the RGB, computed with MESA for solar metallic- ity (Gehan et al. 2018), are shown for the stellar masses 1.0, 1.3, 1.6, and 1.9 M⊙. The error box indicates the typical 1-σ uncertainties on Teﬀ and νmax. Fig. 8. 4.1. Observations Seismic diagram of the 372 red giants studied in this work, with ν−1 max used as a proxy for the luminosity. The color codes the stellar mass. Stars on the RGB are plotted with + symbols, red clump stars with ⋄. The dotted and dashed lines indicate the limit of the visibility of mixed modes for RGB and clump stars, respectively, as deﬁned by Eq. (31). Evolution tracks on the RGB, computed with MESA for solar metallic- ity (Gehan et al. 2018), are shown for the stellar masses 1.0, 1.3, 1.6, and 1.9 M⊙. The error box indicates the typical 1-σ uncertainties on Teﬀ and νmax. Fig. 9. Stretched-period échelle diagram of KIC 3216736, the only RGB star in our sample showing buoyancy glitches. The spectrum is simple, since only m = 0 dipole mixed modes are present, but shows a large- period modulation instead of the expected vertical alignment. Modes plotted in light blue are pressure dominated; extra peaks that do not follow the global trend are either ℓ= 3 modes or ℓ= 2 mixed modes. Red ﬁgures indicate the radial orders of the radial modes. For clarity, the power spectrum density is also plotted twice, top to tail. sample where we found buoyancy glitches and could not provide a relevant ﬁt of the spectrum, but only an échelle diagram based on stretched periods (Fig. 9). Since we have tested more than 160 stars on the RGB, with a systematic approach, we can con- clude that the most common case on the RGB is the absence of buoyancy glitches, as expected theoretically (Cunha et al. 2015). y y g p y ( ) Characterizing the sample we studied in terms of bias is dif- ﬁcult. Apart from the RGB stars that were already studied in detail in previous works, we have mostly treated the stars with increasing KIC numbers. This systematic method implies that we did not introduce any further bias compared to the Kepler sample of red giants. Considering a high enough S/N, which is almost equivalent to select bright stars in the red giant domain, is not supposed to introduce biases either. Contrary to many previous studies, we are not limited to stars showing rotational splittings smaller than the confusion limit (δνrot ≤ν2 max∆Π1). However, the presence of a strong cutoﬀ(Sect. 6) limits the sample, when gravity-dominated mixed modes disappear. 4.3. Asymptotic period spacings Both cases deserve speciﬁc care beyond the scope of this work. Up to now, measurements of ∆Π1 considering that εg is a free parameter have been obtained for a few stars only (Buysschaert et al. 2016; Hekker et al. 2018, for 3 and 22 observed stars, respectively). The oﬀset εg being arbitrarily ﬁxed, Mosser et al. (2012b and 2014) reported a very high pre- cision for the period spacings, of typically 0.1 s for stars on the RGB and 0.3 s in the red clump. However, owing to the choice of εg = 0, their period spacings were slightly aﬀected by a bias of about a fraction of νmax∆Π2 1. The values reported by Vrard et al. (2016), free of any hypothesis on εg, are not biased but show uncertainties typically ﬁve to ﬁfteen times higher than the new values. Their comparison with our data conﬁrms the absence of systematic oﬀsets (Fig. 10). So, the new method ensures 4.1. Observations Red- clump stars with strong buoyancy glitches are absent in our data set since the ﬁtting process requires then to account for the extra-modulation, which can be quite large (about ∆Π1/10). When mixed modes are depressed, the low height-to-background ratio of the mixed modes allows the measurement of ∆Π1 (Mosser et al. 2017a) but is not enough for ﬁtting the pattern. Both cases deserve speciﬁc care beyond the scope of this work atan correcting term introduces an oﬀset of −∆Π1, which in fact allows to relate the (N + 1) mixed modes in the ∆ν-wide inter- val to N only gravity modes. In order to use all mixed modes, including the \|m\| = 1 components, we corrected ﬁrst the rota- tional splittings, using Eq. (16) in order to obtain ν values that are corrected from the rotational splitting. From the periods of the gravity modes 1/νg, we could then derive the asymptotic parameters ∆Π1 and εg, assuming the ﬁrst- order asymptotic expression for pure gravity modes (Eq. (4)). In practice, a ﬁrst estimate of ∆Π1 derived from the formalism of Mosser et al. (2015) and Vrard et al. (2016) is used in Eq. (25), then iterated with a least-square ﬁt of the linear variation of the gravity modes (Eq. (4)). 5.1. Splitting asymmetry This relation comes from the derivative of Eq. (4). As a result, the median uncertainties are of about 0.08 on the RGB and 0.06 in the red clump. Recently, asymmetries in the rotational splittings were reported by Deheuvels et al. (2017), as the signature of the combined eﬀects of rotation and mode mixing. Using both perturbative and non-perturbative approaches, they computed near-degeneracy eﬀects and could ﬁt the data. In fact, the asymptotic development of mixed modes also describes the combined eﬀects of rotation and mode mixing, so that the rotational splittings based on ⟨ζ⟩m (Eqs. (16) and (18)) are not symmetric. Inversely, the symmetri- cal rotational splitting based on ζ (Eq. (15)) does not reproduce the observed asymmetry. Hence, observing asymmetrical triplets is a way to prove the relevance of the use of ⟨ζ⟩m instead of ζ. p We noticed that the median value of εg on the RGB is in fact close to 1/4, which is the expected asymptotic value in absence of stratiﬁcation below the convection zone (Provost & Berthomieu 1986), derived from the contribution ℓ/2 −εas with ℓ= 1 and εas = 1/4. Hence, we inferred that the degeneracy on the determination of εg is removed. We then noted a slight decrease in εg when stars evolve on the RGB, with an accumulation of values close to 0 for red-clump stars. Hekker et al. (2018) reported values of εg in the range [−0.2, 0.5] for 21 stars on the RGB, but did not identify the accumulation of values in the range [0.20, 0.35] predicted by Takata (2016a) for stars on the low RGB. Our measurements fully conﬁrm this prediction. From a check of their data set, we interpret the diﬀerences in εg as resulting from less precise gravity spacings when large rotational splittings apparently mod- ify the period spacings. As made clear by the recent theoretical developments of the asymptotic expansion (Takata 2016b,a), the Observing the asymmetry is challenging but possible for stars with a rapid rotation rate. As explained by Gehan et al. (2017, 2018), rapid rotation means δνrot ≥ ∆ν/N for seismology. This rotation is however very slow in terms of interior struc- ture, so that the formalism developed by Goupil et al. (2013) and Deheuvels et al. (2014), summarized by Eq. (16), remains rele- vant. It simpliﬁes the study, as shown by Ouazzani et al. 4.2. Pure gravity modes They trans- late, respectively, into 0.06 s on the RGB and 0.22 s in the red clump; Hekker et al. (2018) reach a similar precision. accuracy, in the sense that the measurements of ∆Π1 are now free of any hypothesis on εg and prove that the asymptotic expan- sion for gravity modes (Eq. (4)) is relevant. The relative accu- racy we obtained for the period spacings, assuming Eq. (4), is shown in Fig. 11. The median relative accuracies on the RGB and in the red clump are similar, of about 7 × 10−4. They trans- late, respectively, into 0.06 s on the RGB and 0.22 s in the red clump; Hekker et al. (2018) reach a similar precision. We can study the variation of εg along stellar evolution. On the RGB, the asymptotic expansion predicts εg = 1/4 −ϑ (Provost & Berthomieu 1986), where ϑ is a measure of the strat- iﬁcation just below the convection zone. From this dependence, we can infer that the term ϑ is certainly very small for most stars on the low RGB. Higher values are suspected for evolved RGB stars, but with too few stars to ﬁrmly conclude, whereas lower values are seen for evolved RGB. We checked that the change of regime of εg is not associated with the luminosity bump since it occurs for more evolved stars than our sample (Khan et al. 2018). In the red clump, the ϑ correction seems important, on the order of 0.3, with a larger spread than observed on the RGB. 4.4. Gravity offsets εg We could measure εg for a large set of stars. We however have to face the indetermination of εg modulo 1: we simply assume that εg is in the range [−0.5, 0.5]. The εg values computed for the set of stars presented in the paper is given in Table 1 and plotted in Fig. 12, where an histogram is also given. The complete table is available at the CDS. Uncertainties on εg are small, related to the uncertainties in ∆Π1 by An extended study of εg can now be performed to use this parameter as a probe of the stratiﬁcation occurring in the radia- tive region. This study is however beyond the scope of this work. 4.2. Pure gravity modes The identiﬁcation of the mixed modes depicted in Sect. 3.1 allows us to retrieve the periods of the pure gravity modes and to infer global asymptotic parameters of the gravity components. We compute these periods from the mixed-mode frequencies ν, using Eqs. (1) and (2), 1 νg = 1 ν −∆Π1 π atan tan θp q ! · (25) (25) Close to each radial mode, when θp varies from values less than but close to π/2 to values higher than but close to −π/2, the A109, page 8 of 25 B. Mosser et al.: Mixed modes Fig. 10. Comparison of the asymptotic period spacings with previous values. Light blue triangles show the bias in period spacings computed under the assumption εg = 0 (Mosser et al. 2014), whereas dark blue diamonds are free of this hypothesis (Vrard et al. 2016). Fig. 11. Relative precision of the asymptotic period spacings. Same style as in Fig. 8. t t f th l di d t ∆Π i Fig. 10. Comparison of the asymptotic period spacings with previous values. Light blue triangles show the bias in period spacings computed under the assumption εg = 0 (Mosser et al. 2014), whereas dark blue diamonds are free of this hypothesis (Vrard et al. 2016). Fig. 11. Relative precision of the asymptotic period spacings. Same style as in Fig. 8. Fig. 11. Relative precision of the asymptotic period spacings. Same style as in Fig 8 Fig. 10. Comparison of the asymptotic period spacings with previous values. Light blue triangles show the bias in period spacings computed under the assumption εg = 0 (Mosser et al. 2014), whereas dark blue diamonds are free of this hypothesis (Vrard et al. 2016). Fig. 11. Relative precision of the asymptotic period spacings. Same style as in Fig. 8. accurate measurement of the leading-order term ∆Π1 is neces- sary to provide reliable estimates of εg. accuracy, in the sense that the measurements of ∆Π1 are now free of any hypothesis on εg and prove that the asymptotic expan- sion for gravity modes (Eq. (4)) is relevant. The relative accu- racy we obtained for the period spacings, assuming Eq. (4), is shown in Fig. 11. The median relative accuracies on the RGB and in the red clump are similar, of about 7 × 10−4. 5. Rotation The ﬁts based on the function ζ also allow us to analyze rota- tional splittings in detail. δεg = δ∆Π1 νmax ∆Π2 1 · (26) δεg = δ∆Π1 νmax ∆Π2 1 · (26) 5.3. Stellar inclination From its ability to ﬁt the gravity-dominated modes that carry useful information, the asymptotic ﬁt can be used to derive the stellar inclination i. The amplitude of the m = 0 component of the dipole multiplet is proportional to sin2 i whereas the sum of the amplitudes of the m = ±1 mode is proportional to cos2 i. From Eq. (20), a correction factor of 1/(1 −ζ) should be applied on the amplitudes: its diﬀerential eﬀect is however much below the precision one can get on i. Table 1. Period spacings and gravity oﬀsets. Left: variation of εg with ∆ν, with the same style as Fig. 11. The horizontal dark gray domain corresponds to the expected range predicted for RGB stars by Takata (2016a), whereas the dot-dashed line shows the value εg,as = 1/4 derived from the asymptotic expansion (Provost & Berthomieu 1986). Uncertainties on εg are indicated by vertical lines; uncertainties on ∆ν are smaller than the symbol size. Right: histograms of the distri- butions of εg on the RGB (blue curve) and in the red clump (red curve). The dot-dashed line and the gray domain have the same meaning as indicated above. Fig. 12. Left: variation of εg with ∆ν, with the same style as Fig. 11. The horizontal dark gray domain corresponds to the expected range predicted for RGB stars by Takata (2016a), whereas the dot-dashed line shows the value εg,as = 1/4 derived from the asymptotic expansion (Provost & Berthomieu 1986). Uncertainties on εg are indicated by vertical lines; uncertainties on ∆ν are smaller than the symbol size. Right: histograms of the distri- butions of εg on the RGB (blue curve) and in the red clump (red curve). The dot-dashed line and the gray domain have the same meaning as indicated above. Triana et al. 2017): deriving surface rotation can be achieved for the low RGB only. as ∆ν. We ﬁtted the mixed-mode spectrum of KIC 3955033 with both the symmetrical and asymmetrical splitting. At high radial order np, it is hard to distinguish them. At low orders, when the rotational splittings exceed the mixed-mode spacings, the sym- metrical splittings fail whereas the asymmetrical one provides a consistent solution along the whole spectrum. The radial order np = 8 is shown in Fig. 13, the whole spectrum is shown in Fig. A.3. Table 1. Period spacings and gravity oﬀsets. Table 1. Period spacings and gravity oﬀsets. KIC νmax ∆ν ∆Π1 q εg δνrot ( µHz) ( µHz) (s) (nHz) 1576469 90.60 ± 0.98 7.41 ± 0.04 284.80 ± 0.64 0.23 ± 0.03 −0.102 ± 0.096 67 ± 6 1723700 39.42 ± 0.57 4.48 ± 0.04 323.40 ± 0.17 0.24 ± 0.04 0.066 ± 0.043 57 ± 5 2437976 89.37 ± 1.10 8.22 ± 0.05 74.70 ± 1.00 0.10 ± 0.02 −0.006 ± 0.095 320 ± 30 2443903 66.76 ± 0.90 7.01 ± 0.04 71.10 ± 0.02 0.12 ± 0.02 −0.184 ± 0.078 360 ± 4 3955033 106.10 ± 1.24 9.23 ± 0.05 74.65 ± 0.06 0.13 ± 0.02 0.207 ± 0.115 765 ± 10 5024476 68.66 ± 0.75 5.73 ± 0.04 299.60 ± 1.00 0.27 ± 0.03 −0.199 ± 0.102 63 ± 6 5112373 43.82 ± 0.59 4.63 ± 0.04 240.30 ± 0.14 0.19 ± 0.02 −0.246 ± 0.058 37 ± 3 6144777 128.23 ± 1.50 11.03 ± 0.05 79.05 ± 0.04 0.13 ± 0.02 0.210 ± 0.055 244 ± 5 10272858 341.45 ± 6.16 22.71 ± 0.14 96.90 ± 0.30 0.19 ± 0.02 0.338 ± 0.098 660 ± 20 11353313 127.29 ± 1.46 10.75 ± 0.05 76.95 ± 0.06 0.14 ± 0.02 0.290 ± 0.088 465 ± 7 Notes. The list of the full data set with 372 red giants showing an uncertainty in εg less than 0.15 is available at the CDS. es. The list of the full data set with 372 red giants showing an uncertainty in εg less than 0.15 is available at the CDS. Fig. 12. Left: variation of εg with ∆ν, with the same style as Fig. 11. The horizontal dark gray domain corresponds to the expected range predicted for RGB stars by Takata (2016a), whereas the dot-dashed line shows the value εg,as = 1/4 derived from the asymptotic expansion (Provost & Berthomieu 1986). Uncertainties on εg are indicated by vertical lines; uncertainties on ∆ν are smaller than the symbol size. Right: histograms of the distri- butions of εg on the RGB (blue curve) and in the red clump (red curve). The dot-dashed line and the gray domain have the same meaning as indicated above. as ∆ν. We ﬁtted the mixed-mode spectrum of KIC 3955033 with both the symmetrical and asymmetrical splitting. At high radial Triana et al. 2017): deriving surface rotation can be achieved for the low RGB only. Fig. 12. 5.1. Splitting asymmetry (2013) who treated the case where rotational splittings can be as large A109, page 9 of 25 A&A 618, A109 (2018) 5.2. Surface rotation (b)× symbols indicate the absence of any reliable asymptotic ﬁt for KIC 4937770. (c)Diﬀerent solutions in ∆Π1 are possible for KIC 5112361, which all provide a high inclination. Diﬀerent solutions in δνrot are possible for KIC 5113441, which all provide a low inclina- tion. Typical uncertainties for those stars with low S/N spectra are 0.7% in ∆ν and ∆Π1, 12% in q, and 8% in δνrot. Notes. (a)KIC 4937257 and KIC 4937775 are absent in Corsaro et al. (2017). (b)× symbols indicate the absence of any reliable asymptotic ﬁt for KIC 4937770. (c)Diﬀerent solutions in ∆Π1 are possible for KIC 5112361, which all provide a high inclination. Diﬀerent solutions in δνrot are possible for KIC 5113441, which all provide a low inclina- tion. Typical uncertainties for those stars with low S/N spectra are 0.7% in ∆ν and ∆Π1, 12% in q, and 8% in δνrot. Notes. (a)KIC 4937257 and KIC 4937775 are absent in Corsaro et al. (2017). (b)× symbols indicate the absence of any reliable asymptotic ﬁt for KIC 4937770. (c)Diﬀerent solutions in ∆Π1 are possible for KIC 5112361, which all provide a high inclination. Diﬀerent solutions in δνrot are possible for KIC 5113441, which all provide a low inclina- tion. Typical uncertainties for those stars with low S/N spectra are 0.7% in ∆ν and ∆Π1, 12% in q, and 8% in δνrot. We measured inclinations of red giants in the open clus- ters NGC 6819 observed by Kepler (e.g., Basu et al. 2011; Stello et al. 2011; Miglio et al. 2012). We selected the stars that exhibit mixed modes and could ﬁt 20 mixed-mode patterns with the asymptotic expansion. In one case, the asymptotic ﬁt is impossible, due to a low S/N. In two other cases, diﬀerent pos- sible solutions exist, based either on diﬀerent period spacings, or on diﬀerent rotational splittings, but without any ambiguity for the inclination measurement: when two peaks dominate per period spacing, the inclination is necessarily high, whereas it is low when one single peak only is present. We completed this list with other NGC 6819 members listed in Handberg et al. (2017) and could ﬁt two additional stars, which incidentally show a large inclination. Results for the inclinations and rotational split- tings are given in Table 2. As shown in Fig. 5.2. Surface rotation For stars on the low RGB, surface rotation can be inferred from the rotational splittings (Eq. (15)). The measurement is however diﬃcult, since it results from an extrapolation at ζ = 0, when values are mostly obtained above ζ = 0.6 only (Fig. 5). The highest level of precision, hence the use of ⟨ζ⟩m instead of ζ, is required for deriving a correct estimate of the surface rota- tion. The case of KIC 3955033 is illustrative, with a negative sur- face rotation when using ζ; the use of ⟨ζ⟩m provides a null value (5±20 nHz). This case also conﬁrms the general situation shown by previous works (Goupil et al. 2013; Di Mauro et al. 2016; We tested our results on a set of stars for which the incli- nations measured with other methods have been obtained. We checked that our results are relevant, with a precision lim- ited by the uncertainties on the amplitude measurements. In order to avoid bias, we consider only peaks with a height-to- background ratio larger than 8. Nevertheless, we noted that the stochastic excitation of the modes induces a small bias for large inclinations. Equator-on inclinations, near 90◦, cannot be A109, page 10 of 25 B. Mosser et al.: Mixed modes Table 2. Asymptotic and rotational parameters in NGC 6819. Table 2. Asymptotic and rotational parameters in NGC 6819. Fig. 13. Splitting asymmetry at low frequency in KIC 3955033. Each dipole mixed mode of the spectrum is labelled with its radial and azimuthal orders. The rotational splittings of the radial orders from −142 to −140, plotted with diamonds, do no match the function ζ. Only the multiplet with n = −141 is complete: the m = +1 splitting is much larger than the m = −1 splitting; the colored regions indicate the ranges over which the function ζ is integrated for the components of the mul- tiplet n = −141. The dashed lines indicate height-to-background values of 7 and 10. Table 2. Asymptotic and rotational parameters in NGC 6819. 5.2. Surface rotation KIC ID ∆ν ∆Π1 q δνrot i ( µHz) (s) (nHz) (◦) 4937056 4.76 291.0 0.21 90 60±15 4937257a 4.13 292.1 0.19 27 72±13 4937770b 7.82 161.0 0.18 × × 4937775a 7.33 226.3 0.21 110 75±15 5023953 4.74 293.9 0.24 50 51±28 5024327 4.72 269.5 0.20 55 56±13 5024404 4.78 242.6 0.25 110 80±10 5024414 6.47 283.0 0.30 90 45±20 5024476 5.73 299.5 0.24 56 71±11 5024582 4.76 323.5 0.22 70 55±18 5111718 10.59 88.4 0.12 410 69±21 5111949 4.81 319.0 0.28 35 66±15 5112072 10.08 91.9 0.15 350 72±12 5112361c 6.19 99.0 0.12 350 70±20 5112373 4.63 240.2 0.19 37 47±18 5112387 4.70 267.2 0.28 84 25±17 5112401 4.03 311.0 0.26 50 54±13 5112467 4.75 285.2 0.25 90 61±12 5112491 4.68 324.3 0.30 150 31±16 5112730 4.56 320.0 0.25 45 56±18 5112938 4.73 320.0 0.30 65 45±11 5112950 4.35 319.5 0.38 38 61±18 5112974 4.32 309.6 0.24 60 50±12 5113441c 11.75 89.0 0.13 730 18±18 5200152 4.73 327.2 0.28 50 70±15 Notes. (a)KIC 4937257 and KIC 4937775 are absent in Corsaro et al (2017). (b)× symbols indicate the absence of any reliable asymptoti ﬁt for KIC 4937770. (c)Diﬀerent solutions in ∆Π1 are possible fo KIC 5112361, which all provide a high inclination. Diﬀerent solution in δνrot are possible for KIC 5113441, which all provide a low inclina tion. Typical uncertainties for those stars with low S/N spectra are 0.7% in ∆ν and ∆Π1, 12% in q, and 8% in δνrot. Fig. 13. Splitting asymmetry at low frequency in KIC 3955033. Each dipole mixed mode of the spectrum is labelled with its radial and azimuthal orders. The rotational splittings of the radial orders from −142 to −140, plotted with diamonds, do no match the function ζ. Only the multiplet with n = −141 is complete: the m = +1 splitting is much larger than the m = −1 splitting; the colored regions indicate the ranges over which the function ζ is integrated for the components of the mul- tiplet n = −141. The dashed lines indicate height-to-background values of 7 and 10. retrieved precisely, with measurements reduced toward the range 70–80◦. As a consequence, they are rare in our analysis. How- ever, many stars show inclinations that, according to the uncer- tainties, are compatible with equator-on measurement, so that the bias does not aﬀect the following analysis. Notes. (a)KIC 4937257 and KIC 4937775 are absent in Corsaro et al. (2017). 5.2. Surface rotation The red symbol shows the case where no asymptotic ﬁt could be found, and the green ones to cases without Bayesian ﬁt. The dashed line corresponds to the 1:1 relation. Fig. 17. Fit of the mixed modes corresponding to np = 9 in KIC 2437976 (NGC 6791 member). The color codes the azimuthal order: m = +1 in purple, m = −1 in blue. The location of m = 0 modes is indicated in light blue, but none shows a large height for this star seen equator-on. The gray dashed lines indicate the two thresholds used in this work, corresponding to height-to-background ratios of 7 and 10. Many peaks above the threshold value 5.5 that rejects the null hypothesis at the 5%-level follow the mixed-mode pattern. Fig. 17. Fit of the mixed modes corresponding to np = 9 in KIC 2437976 (NGC 6791 member). The color codes the azimuthal order: m = +1 in purple, m = −1 in blue. The location of m = 0 modes is indicated in light blue, but none shows a large height for this star seen equator-on. The gray dashed lines indicate the two thresholds used in this work, corresponding to height-to-background ratios of 7 and 10. Many peaks above the threshold value 5.5 that rejects the null hypothesis at the 5%-level follow the mixed-mode pattern. Fig. 17. Fit of the mixed modes corresponding to np = 9 in KIC 2437976 (NGC 6791 member). The color codes the azimuthal order: m = +1 in purple, m = −1 in blue. The location of m = 0 modes is indicated in light blue, but none shows a large height for this star seen equator-on. The gray dashed lines indicate the two thresholds used in this work, corresponding to height-to-background ratios of 7 and 10. Many peaks above the threshold value 5.5 that rejects the null hypothesis at the 5%-level follow the mixed-mode pattern. Fig. 15. Comparison of the inclinations of the spin axis of the stars in NGC 6819. Inclinations measured by Corsaro et al. (2017) are plotted on the x-axis, while inclinations derived from the asymptotic analysis are on the y-axis; 1-σ uncertainties are indicated by vertical and hori- zontal error bars. The red symbol shows the case where no asymptotic ﬁt could be found, and the green ones to cases without Bayesian ﬁt. The dashed line corresponds to the 1:1 relation. Fig. 17. 5.2. Surface rotation 14, the distribution of the stellar inclinations mimics the sin i relation expected for ran- dom inclinations, except near 90◦, due to bias mentioned above. A similar test performed on the open cluster NGC 6791 reaches the same conclusion. Fig. 14. Histogram of the inclinations measured in NGC 6819. The dashed line indicates the sin i distribution. Low stellar inclinations in NGC 6819 and 6791 were mea- sured by Corsaro et al. (2017), using a Bayesian analysis, from which aligned spins were inferred. Our measurements how- ever contradict their claim, as shown in Fig. 15 for NGC 6819. In fact, our measurements compared to theirs agree for high inclinations, whereas they mostly disagree for low inclinations. Comparison with the asymptotic ﬁts shows that their Bayesian rotational splittings are most often overestimated and that the related inclinations are most often underestimated. Fig. 14. Histogram of the inclinations measured in NGC 6819. The dashed line indicates the sin i distribution. modes at low pressure radial orders, with frequencies much below νmax, are especially informative, since previous work has shown that their radial mode widths, hence their mixed- mode widths according to Eq. (19), are the thinnest possible (Fig. 5 of Vrard et al. 2018). Figure 1 of the supplementary In fact, the asymptotic ﬁt can be used as a prior for the Bayesian ﬁt. It indicates that the rotational splitting is derived from the thin gravity-dominated mixed modes, with narrow widths (Eq. (19)) and average rotational splittings slightly inferior to the mean core rotation δνrot (Eq. (16)). Mixed A&A 618, A109 (2018) Fig. 15. Comparison of the inclinations of the spin axis of the stars in NGC 6819. Inclinations measured by Corsaro et al. (2017) are plotted on the x-axis, while inclinations derived from the asymptotic analysis are on the y-axis; 1-σ uncertainties are indicated by vertical and hori- zontal error bars. The red symbol shows the case where no asymptotic ﬁt could be found, and the green ones to cases without Bayesian ﬁt. The dashed line corresponds to the 1:1 relation. Fig. 15. Comparison of the inclinations of the spin axis of the stars in NGC 6819. Inclinations measured by Corsaro et al. (2017) are plotted on the x-axis, while inclinations derived from the asymptotic analysis are on the y-axis; 1-σ uncertainties are indicated by vertical and hori- zontal error bars. 5.2. Surface rotation Fit of the mixed modes corresponding to np = 9 in KIC 2437976 (NGC 6791 member). The color codes the azimuthal order: m = +1 in purple, m = −1 in blue. The location of m = 0 modes is indicated in light blue, but none shows a large height for this star seen equator-on. The gray dashed lines indicate the two thresholds used in this work, corresponding to height-to-background ratios of 7 and 10. Many peaks above the threshold value 5.5 that rejects the null hypothesis at the 5%-level follow the mixed-mode pattern. Fig. 16. Fit of the mixed modes corresponding to np = 7 in KIC 5112373 (NGC 6819 member). The color codes the azimuthal order: m = +1 in purple, m = −1 in blue. The gray dashed lines indi- cate the two thresholds used in this work, corresponding to height-to- background ratios of 7 and 10. Contrary to the analysis conducted by Corsaro et al. (2017), modes with m = ±1 are clearly identiﬁed. ensure that close modes do not belong to the same multiplets. All peaks can be explained by the m = ±1 modes. In prac- tice, m = 0 modes are absent, so that this star has necessarily an inclination close to 90◦, whereas Corsaro et al. (2017) mea- sured i ≃0◦. We conclude that some of the low inclinations reported in Corsaro et al. (2017) are incompatible with the anal- ysis presented here. It seems that the diﬀerence is due to a too low range of the linewidth priors in the Bayesian analysis, which favors a solution with a low inclination angle and a high splitting. As a result, stellar spins in old open clusters are neither aligned nor quasi parallel to the line of sight. Our study emphasizes a major role for the asymptotic analysis: providing relevant esti- mates of all features of the mixed-mode pattern, including mode widths. Fig. 16. Fit of the mixed modes corresponding to np = 7 in KIC 5112373 (NGC 6819 member). The color codes the azimuthal order: m = +1 in purple, m = −1 in blue. The gray dashed lines indi- cate the two thresholds used in this work, corresponding to height-to- background ratios of 7 and 10. Contrary to the analysis conducted by Corsaro et al. (2017), modes with m = ±1 are clearly identiﬁed. 6. Observability of the mixed modes All the information derived from mixed modes relies on their observability. The properties of the function ζ can be used to assess under which conditions mixed modes can be actually observed. To achieve this, we investigate ﬁrst the domain where pressure-dominated mixed modes are observed, then the condi- tion for observing gravity-dominated mixed modes. material of Corsaro et al. (2017) provides an explanation of the discrepant Bayesian values. Their ﬁt of the star KIC 5112373 in NGC 6819 provides nearly uniform large mode widths, rel- evant for the pressure-dominated mixed modes but much too high for gravity-dominated modes, in contradiction with the physical variation indicated by Eq. (19). As a consequence, their ﬁt assumes that all the power is concentrated in the m = 0 mode; the resulting stellar inclination is 20 ± 8◦. We show the asymptotic solution of KIC 5112373 in Fig. 16, with thin gravity dominated mixed modes and the clear identiﬁca- tion of triplets. Since m = ±1 modes are observed all along the spectrum, our solution for the inclination is larger, about 47 ± 18◦. 7. Conclusion The asymptotic analysis allows us to depict the whole properties of the mixed-mode spectrum in a consistent way. Period spac- ings, rotational splittings, mode widths, and mode heights, all depend on the mode inertia so that all are related to the parame- ter ζ. We could derive interesting properties: (1 −q2) sin2 θp ≤q N π 2 Γ0 δfres −1 ! −q2. (29) – The asymptotic ﬁt of the mixed modes proves to be precise and unbiased. Its precision for the RGB stars is so high that the asymptotic expansion of gravity modes can be validated when buoyancy glitches are absent. This ensures the delivery of accurate asymptotic parameters ∆Π1, q, and εg. We found only one RGB star with such buoyancy glitches; on the con- trary, buoyancy glitches are often present in red-clump stars. (29) Except at the transition from subgiants to red giants, where mixed modes are unambiguously visible (Benomar et al. 2013; Deheuvels et al. 2014), the terms q2 are negligible, so that modes are clearly visible if – The period spacings and rotational splittings are better esti- mated with integrated values of the function ζ. The use of these mean values ⟨ζ⟩n and ⟨ζ⟩m is useful for evolved RGB stars and is mandatory for stars with intricate splittings and spacings. Using the stretched period (Mosser et al. 2015) is in fact equivalent. \| sin θp\|gm ≤ s q N π 2 Γ0 δfres −1 ! . (30) (30) This condition for observing gravity-dominated mixed modes has many consequences: q – The gravity asymptotic parameters ∆Π1 and εg can now be accurately determined, with typical accuracy of respectively 0.06 s and 0.1 on the RGB, and 0.22 s and 0.08 in the red clump. This opens the way to a fruitful dialogue with the- oretical developments (Takata 2006, 2016b,a) and modeling (e.g., Bossini et al. 2015; Cunha et al. 2015). y q – It can be fulﬁlled only if the deﬁnition of the right term is ensured, which requires a frequency resolution low enough compared to the radial mode width. With Γ0 in the range [100, 150 nHz], the observation must last 50–75 days at least. In fact, mixed modes were observable with CoRoT runs lasting about 150 days (Mosser et al. 2011a), but are hardly observable with K2 80-day time series (Stello et al. 2017). Acknowledgements. We thank the entire Kepler team, whose eﬀorts made these results possible. BM warmly thanks Yvonne Elsworth, James Kuszlewicz and Masao Takata for their comments on the draft submitted for internal review on the website of the Kepler Asteroseismic Science Operations Center. We acknowledge ﬁnancial support from the Programme National de Physique Stel- laire (CNRS/INSU). BM acknowledges the support of the International Space Institute (ISSI) for the program AsteroSTEP (Asteroseismology of STEllar Pop- ulations). 6.1. Pressure-dominated mixed modes The variations in q and N explain the narrowing of the region with pressure-dominated mixed modes when stars evolve on the RGB. An example is shown in Fig. A.4. The expression of δνpm also shows that red-clump stars, with larger q show pressure-dominated mixed modes in a broader region than RGB stars. 6.2. Visible gravity-dominated mixed modes The non-dilution of the mode height expressed by Eq. (21) can be used to deﬁne a criterion of visibility of the gravity-dominated (gm) mixed modes. So, they are clearly visible when they show heights similar to those of the pressure modes (Hn = H0), hence when Γ0(1 −ζ) ≥2δfres/π (Eq. (19)). This condition translates into 6.1. Pressure-dominated mixed modes We can deﬁne the frequency range where mixed modes are pressure-dominated (pm) from the full width at half minimum of the ζ function. So, these modes cover a range, expressed in terms of the pressure phase θp (Eq. (6)), verifying δθp pm = 2q s 1 + 1 Nq (27) (27) under the assumption that q is small, which is veriﬁed for all stars except at the transition between subgiants and red giants (Mosser et al. 2017b). When expressed in frequency and com- pared to the large separation, this condition corresponds to a We provide another example with the star KIC 2437976, a NGC 6791 member. As shown in Fig. 17, rotational splittings are explained in a consistent way with thin unresolved gravity- dominated mixed modes and a rotation rate rapid enough to A109, page 12 of 25 B. Mosser et al.: Mixed modes B. Mosser et al.: Mixed modes frequency range surrounding each pure pressure modes with a width δνpm deﬁned by frequency range surrounding each pure pressure modes with a width δνpm deﬁned by In the conditions of observation of Kepler, with typical parame- ters deﬁned as in Mosser et al. (2017a), this limit corresponds to a mode density N of about 25, for RGB and clump stars, over which no gravity-dominated mixed modes can be identiﬁed. This theoretical estimate is observed in practice, with a few excep- tions with larger N (Fig. 8). On the RGB, observation of mixed modes with Kepler is limited to ∆ν ≥6 µHz, whereas the limit is around 3 µHz for clump stars. As a consequence, visible mixed modes in an oscillation spectrum with ∆ν in the range [3, 6 µHz] most often indicate a red-clump star. Incidentally, the location of the RGB bump was recently identiﬁed by Khan et al. (2018) in the range [5, 6 µHz], depending on the stellar mass and metallic- ity. This means that sounding the bump with mixed modes will be very diﬃcult, if not impossible. δνpm ∆ν = 2q π s 1 + 1 Nq· (28) (28) The variations in q and N explain the narrowing of the region with pressure-dominated mixed modes when stars evolve on the RGB. An example is shown in Fig. A.4. The expression of δνpm also shows that red-clump stars, with larger q show pressure-dominated mixed modes in a broader region than RGB stars. 7. Conclusion – We have made clear that observing mixed modes in evolved red giants requires an observation duration longer than ≃ 100 days. However, gravity-dominated mixed modes are no longer observable when the stars are more evolved than ∆ν ≃6 µHz on the RGB, or ∆ν ≃3 µHz in the red clump. These thresholds are indicative values: the natural spread of the seismic parameters with respect to their mean values explain slight diﬀerences. y – When stars evolve on the RGB, the decrease in q and increase in N contribute to the narrowing of observable modes. Mixed modes are more easily visible in the red clump, owing to larger q values. This criterion is implicitly used by Elsworth et al. (2017) for their determination of the evolu- tionary state of red-giant stars. y g – All mixed modes are clearly visible when the condition expressed by Eq. (30) is always met, that is when N ≤ q(πΓ0/2δfres −1). This condition is met for subgiants, on the lower RGB, and for secondary-clump stars (Mosser et al. 2014). – We have demonstrated the non-alignment of the rotation axis of the stars belonging to the old open clusters NGC 6791 and NGC 6819. These results contradict previous ﬁndings by Corsaro et al. (2017) and illustrate how useful the asymptotic ﬁt will be in the future when used to deﬁne priors to any Bayesian or other type of ﬁt of mixed modes. – No mixed mode can be observed when the condition is so drastic that only pressure-dominated mixed modes can be observed. The combination of the conditions expressed by Eqs. (27) and (30) yields the limit of visibility of gravity- dominated mixed modes, expressed by a condition on the mixed-mode density – No mixed mode can be observed when the condition is so drastic that only pressure-dominated mixed modes can be observed. The combination of the conditions expressed by Eqs. (27) and (30) yields the limit of visibility of gravity- dominated mixed modes, expressed by a condition on the mixed-mode density N ≤1 4q π 2 Γ0 δfres −5 ! . (31) N ≤1 4q π 2 Γ0 δfres −5 ! . (31) A109, page 13 of 25 A109, page 13 of 25 A&A 618, A109 (2018) , Libbrecht, K. G. 1992, ApJ, 387, 712 Libbrecht, K. G. 1992, ApJ, 387, 712 k, P. G., Montalban, J., Kallinger, T., et al. 2012, Nature, 481, 55 Miglio, A., Montalbán, J., Carrier, F., et al. 2010, A&A, 520, L6 Beck, P. G., Montalban, J., Kallinger, T., et al. 2012, Nature, 481, 55 Bedding, T. R., Mosser, B., Huber, D., et al. 2011, Nature, 471, 608 g Bedding, T. R., Mosser, B., Huber, D., et al. 2011, Nature, 471, 608 Miglio, A., Brogaard, K., Stello, D., et al. 2012, MNRAS, 419, 2077 Belkacem, K., Marques, J. P., Goupil, M. J., et al. 2015a, A&A, 579, A31 Miglio, A., Chiappini, C., Mosser, B., et al. 2017, Astron. Nachr., 338, 644 cem, K., Marques, J. P., Goupil, M. J., et al. 2015b, A&A, 579, A3 Montalbán, J., Miglio, A., Noels, A., et al. 2013, ApJ, 766, 118 Belkacem, K., Marques, J. P., Goupil, M. J., et al. 2015b, A&A, 579, A30 Benomar, O., Bedding, T. R., Mosser, B., et al. 2013, ApJ, 767, 158 Benomar, O., Bedding, T. R., Mosser, B., et al. 2013, ApJ, 767, 158 Mosser, B., & Appourchaux, T. 2009, A&A, 508, 877 , , g, , , , , p , , Benomar, O., Belkacem, K., Bedding, T. R., et al. 2014, ApJ, 781, L29 B i i D Mi li A S l i M l 2015 MNRAS 453 2290 Mosser, B., Barban, C., Montalbán, J., et al. 2011a, A&A, 532, A86 Benomar, O., Belkacem, K., Bedding, T. R., et al. 2014, ApJ, 781, L29 Mosser, B., Belkacem, K., Goupil, M., et al. 2011b, A&A, 525, L9 Bossini, D., Miglio, A., Salaris, M., et al. 2015, MNRAS, 453, 2290 Bossini, D., Miglio, A., Salaris, M., et al. 2015, MNRAS, 453, 2290 Bossini, D., Miglio, A., Salaris, M., et al. 2017, MNRAS, 469, 4718 Mosser, B., Elsworth, Y., Hekker, S., et al. 2012a, A&A, 537, A30 Bossini, D., Miglio, A., Salaris, M., et al. 2017, MNRAS, 469, 4718 Mosser, B., Goupil, M. J., Belkacem, K., et al. 2012b, A&A, 540, A143 Buysschaert, B., Beck, P. G., Corsaro, E., et al. 2016, A&A, 588, A82 y Christensen-Dalsgaard, J. 2012, in Progress in Solar/Stellar Physics with Helio- Mosser, B., Goupil, M. J., Belkacem, K., et al. 2012c, A&A, 548, A10 Mosser, B., Dziembowski, W. A., Belkacem, K., et al. 2013, A&A, 559, A137 and Asteroseismology, eds. H. Shibahashi, M. Takata, & A. E. Lynas-Gray, ASP Conf. , Libbrecht, K. G. 1992, ApJ, 387, 712 2016a, PASJ, 68, 109 García Saravia Ortiz de Montellano, A., Hekker, S., & Themeßl, N. 2018, MNRAS, 476, 1470 , Libbrecht, K. G. 1992, ApJ, 387, 712 Ser., 462, 503 Mosser, B., Benomar, O., Belkacem, K., et al. 2014, A&A, 572, L5 Corsaro, E., De Ridder, J., & García, R. A. 2015, A&A, 579, A83 Mosser, B., Vrard, M., Belkacem, K., Deheuvels, S., & Goupil, M. J. 2015, A&A, 584, A50 Corsaro, E., Lee, Y.-N., García, R. A., et al. 2017, Nat. Astron., 1, 0064 Corsaro, E., De Ridder, J., & García, R. A. 2018, A&A, 612, C2 Mosser, B., Belkacem, K., Pinçon, C., et al. 2017a, A&A, 598, A62 Mosser, B., Belkacem, K., Pinçon, C., et al. 2017a, A&A, 598, A62 Cunha, M. S., Stello, D., Avelino, P. P., Christensen-Dalsgaard, J., & Townsend, R. H. D. 2015, ApJ, 805, 127 Mosser, B., Pinçon, C., Belkacem, K., Takata, M., & Vrard, M. 2017b, A&A, 600, A1 Mosser, B., Pinçon, C., Belkacem, K., Takata, M., & Vrard, M. 2017b, A&A, 600, A1 , p , , Deheuvels, S., Do˘gan, G., Goupil, M. J., et al. 2014, A&A, 564, A27 Ouazzani, R.-M., Goupil, M. J., Dupret, M.-A., & Marques, J. P. 2013, A&A, 554, A80 Ouazzani, R.-M., Goupil, M. J., Dupret, M.-A., & Marques, J. P. 2013, A&A, 554, A80 Deheuvels, S., Dogan, G., Goupil, M. J., et al. 2014, A&A, 564, A27 Deheuvels, S., Ballot, J., Beck, P. G., et al. 2015, A&A, 580, A96 Deheuvels, S., Ballot, J., Beck, P. G., et al. 2015, A&A, 580, A96 Deheuvels, S., Ouazzani, R. M., & Basu, S. 2017, A&A, 605, A75 Provost, J., & Berthomieu, G. 1986, A&A, 165, 218 Di Mauro, M. P., Ventura, R., Cardini, D., et al. 2016, ApJ, 817, 65 Shibahashi, H. 1979, PASJ, 31, 87 p Dupret, M., Belkacem, K., Samadi, R., et al. 2009, A&A, 506, 57 Stello, D., Meibom, S., Gilliland, R. L., et al. 2011, ApJ, 739, 13 Stello, D., Meibom, S., Gilliland, R. L., et al. 2011, ApJ, 739, 13 lsworth, Y., Hekker, S., Basu, S., & Davies, G. R. 2017, MNRAS, Stello, D., Compton, D. L., Bedding, T. R., et al. 2014, ApJ, 788, García, R. A., Hekker, S., Stello, D., et al. 2011, MNRAS, 414, L6 Stello, D., Zinn, J., Elsworth, Y., et al. 2017, ApJ, 835, 83 Stello, D., Zinn, J., Elsworth, Y., et al. 2017, ApJ, 835, 8 García, R. A., Pérez Hernández, F., Benomar, O., et al. 2014, A&A, 563, A84 Takata, M. 2006, PASJ, 58, 893 Takata, M. 2016a, PASJ, 68, 109 Takata, M. References Handberg, R., Brogaard, K., Miglio, A., et al. 2017, MNRAS, 472, 979 Hekker, S., & Christensen-Dalsgaard, J. 2017, A&ARv, 25, 1 Albareti, F. D., Allende Prieto, C., Almeida, A., et al. 2017, ApJS, 233, 25 Appourchaux T Berthomieu G Michel E et al 2006 in The CoRoT Missio Albareti, F. D., Allende Prieto, C., Almeida, A., et al. 2017, ApJS, 233, 25 Appourchaux, T., Berthomieu, G., Michel, E., et al. 2006, in The CoRoT Mission Hekker, S., Elsworth, Y., & Angelou, G. C. 2018, A&A, 610, A80 g enkins, J. M., Caldwell, D. A., Chandrasekaran, H., et al. 2010, Ap pp , , , , , , , Pre-Launch Status – Stellar Seismology and Planet Finding, eds. M. Fridlund, p Kallinger, T., Hekker, S., Mosser, B., et al. 2012, A&A, 541, A51 Pre Launch Status Stellar Seismology and Planet Finding, eds. M. Fridlund A. Baglin, J. Lochard, & L. Conroy, ESA SP, 1306, 377 gy g A. Baglin, J. Lochard, & L. Conroy, ESA SP, 1306, 377 A. Baglin, J. Lochard, & L. Conroy, ESA SP, 1306, 377 Basu S Grundahl F Stello D et al 2011 ApJ 729 L10 Khan, S., Hall, O. J., Miglio, A., et al. 2018, ApJ, 859, 156 Basu, S., Grundahl, F., Stello, D., et al. 2011, ApJ, 729, L10 Basu, S., Grundahl, F., Stello, D., et al. 2011, ApJ, 729, L10 Baudin, F., Barban, C., Goupil, M. J., et al. 2012, A&A, 538, A73 B k P G B ddi T R M B t l 2011 S i 332 205 Lagarde, N., Bossini, D., Miglio, A., Vrard, M., & Mosser, B. 2016, MNRAS, 457, L59 Basu, S., Grundahl, F., Stello, D., et al. 2011, ApJ, 729, L10 Baudin, F., Barban, C., Goupil, M. J., et al. 2012, A&A, 538, A73 Baudin, F., Barban, C., Goupil, M. J., et al. 2012, A&A, 538, A73 Beck, P. G., Bedding, T. R., Mosser, B., et al. 2011, Science, 332, 205 A109, page 14 of 25 Appendix A: Seismic parameters detection algorithm that works in a fully blind manner, if we relax their uncertainties that can be as low as δfres/20. We used KIC 6144777 as a case study (Fig. 2). Table A.1 pro- vides the ﬁt of its radial dipole mixed modes. Our results are in agreement with those published by Corsaro et al. (2015) and derive a similar number of modes (about 100), but also show diﬀerences: The potential of the comparison between methods based on diﬀerent principles is very high: coupling the physics of the asymptotic expansion and the power of a pure numerical approach is the next step for delivering duly identiﬁed mixed modes. – The determination of the frequencies in Corsaro et al. (2015) can be as precise as 0.3 nHz. This precision of about δfres/30 was corrected into about δfres/10 in their corrigen- dum (Corsaro et al. 2018), which remains surprisingly good; the frequencies we obtain are given with a precision that is at best about half the frequency resolution (≃4 nHz). The échelle diagrams of the stars mentioned in the main text are also presented: KIC 10272858 lies on the low part of the RGB (Fig. A p – KIC 11353313 is on the RGB (Fig. A.2); – KIC 3955033 is a RGB star with a rapid core rotation (Fig. A.3); its frequencies are given in Table A.2; q y ( ) – Their mode widths are quite diﬀerent and, most often, larger than ours; – KIC 2443903 is more evolved on the RGB, at the limit of detection of mixed modes (Fig. A.4); – Heights also diﬀer, which can come from a diﬀerent treat- ment of the time series. g – KIC 1723700 is in the red clump star (Fig. A.5); C 7 3700 s t e ed c u p sta ( g. .5); – and KIC 1725190 is a secondary red clump star (Fig. A.6). A large agreement is also met with the results obtained by García Saravia Ortiz de Montellano et al. (2018) with a peak Table A.1. Oscillation pattern of the RGB star KIC 6144777. Table A.1. Oscillation pattern of the RGB star KIC 6144777. Appendix A: Seismic parameters np n m ζ νas ν x Γas Γ H R ( µHz) ( µHz) ( µHz) ( µHz) (ppm2µHz−1) Radial modes 8 101.916 101.916 ± 0.014 0.040 0.120 ± 0.023 1390 ± 316 24.1 9 112.612 112.612 ± 0.011 0.010 0.106 ± 0.016 3314 ± 497 85.2 10 123.726 123.622 ± 0.006 0.008 0.054 ± 0.007 23358 ± 3059 338.5 11 134.574 134.534 ± 0.008 −0.003 0.087 ± 0.012 10451 ± 1305 195.7 12 145.842 145.585 ± 0.017 −0.002 0.244 ± 0.057 1491 ± 441 26.6 13 156.846 156.846 ± 0.042 0.019 0.258 ± 0.074 317 ± 121 10.7 Dipole mixed modes 8 −112 1 0.9522 105.827 105.836 ± 0.007 0.396 0.009 0.007 ± 0.004 1185 ± 281 10.1 8 −111 1 0.8506 106.633 106.648 ± 0.007 0.469 0.027 0.011 ± 0.004 2124 ± 440 26.0 8 −110 −1 0.7311 106.957 106.985 ± 0.014 0.500 0.048 0.068 ± 0.017 454 ± 149 14.5 8 −110 0 0.6467 107.134 107.137 ± 0.010 0.514 0.064 0.042 ± 0.009 658 ± 145 13.9 8 −110 1 0.5989 107.278 107.322 ± 0.008 0.530 0.072 0.040 ± 0.007 1497 ± 250 30.6 8 −109 −1 0.6428 107.551 107.604 ± 0.009 0.556 0.064 0.055 ± 0.011 1272 ± 292 17.8 8 −109 0 0.7161 107.706 107.730 ± 0.010 0.567 0.051 0.036 ± 0.008 630 ± 157 11.2 8 −109 1 0.7983 107.899 107.909 ± 0.007 0.584 0.036 0.018 ± 0.005 1646 ± 323 18.0 8 −108 −1 0.8911 108.264 108.263 ± 0.010 0.616 0.020 0.016 ± 0.005 770 ± 254 12.2 8 −108 0 0.9206 108.482 108.484 ± 0.006 0.636 0.014 0.007 ± 0.004 1554 ± 335 14.0 8 −108 1 0.9406 108.710 108.700 ± 0.011 0.655 0.011 0.014 ± 0.005 505 ± 186 8.8 8 −107 −1 0.9617 109.140 109.138 ± 0.007 0.695 0.007 0.006 ± 0.004 1303 ± 313 11.9 8 −107 0 0.9683 109.373 109.375 ± 0.008 0.716 0.006 0.005 ± 0.004 783 ± 323 7.2 8 −106 0 0.9822 110.304 110.274 ± 0.008 0.798 0.003 0.006 ± 0.004 803 ± 175 8.4 8 −104 0 0.9889 112.235 112.235 ± 0.007 −0.024 0.002 0.005 ± 0.004 959 ± 335 9.3 9 −103 −1 0.9889 112.989 113.018 ± 0.012 0.047 0.002 0.006 ± 0.004 739 ± 269 8.1 9 −103 0 0.9889 113.228 113.226 ± 0.006 0.065 0.002 0.005 ± 0.004 1736 ± 480 17.3 9 −102 −1 0.9872 113.999 113.996 ± 0.007 0.135 0.002 0.006 ± 0.004 789 ± 174 8.2 9 −102 1 0.9856 114.477 114.478 ± 0.007 0.179 0.003 0.005 ± 0.004 919 ± 297 9.4 9 −101 0 0.9800 115.262 115.254 ± 0.008 0.249 0.004 0.005 ± 0.004 701 ± 221 7.3 9 −100 −1 0.9672 116.058 116.058 ± 0.006 0.322 0.006 0.006 ± 0.004 1721 ± 254 19.0 9 −100 0 0.9600 116.292 116.285 ± 0.007 0.343 0.007 0.006 ± 0.004 1150 ± 273 12.1 9 −100 1 0.9506 116.523 116.525 ± 0.006 0.364 0.009 0.006 ± 0.004 2571 ± 532 27.2 9 −99 −1 0.9056 117.072 117.076 ± 0.009 0.414 0.017 0.022 ± 0.006 459 ± 134 12.7 9 −99 0 0.8694 117.291 117.318 ± 0.025 0.436 0.023 0.040 ± 0.013 77 ± 34 7.2 9 −99 1 0.8189 117.489 117.520 ± 0.006 0.455 0.033 0.011 ± 0.004 4320 ± 641 54.2 Notes. MNRAS, 476, 1470 Takata, M. 2016b, PASJ, 68, 91 , , Gaulme, P., Appourchaux, T., & Boumier, P. 2009, A&A, 506, 7 Tassoul, M. 1980, ApJS, 43, 469 Gaulme, P., Appourchaux, T., & Boumier, P. 2009, A&A, 506, 7 Gehan, C., Mosser, B., & Michel, E. 2016, ArXiv e-prints [arXiv:1612.05414] Gehan, C., Mosser, B., & Michel, E. 2017, Eur. Phys. J. Web Conf., 160, 04005 Gehan, C., Mosser, B., & Michel, E. 2016, ArXiv e-prints [arXiv: Unno, W., Osaki, Y., Ando, H., Saio, H., & Shibahashi, H. 1989, Nonradial Gehan, C., Mosser, B., & Michel, E. 2017, Eur. Phys. J. Web Conf., 160, 04005 Oscillations of Stars (Tokyo: University of Tokyo Press) Gehan, C., Mosser, B., Michel, E., Samadi, R., & Kallinger, T. 2018, A&A, 616, A24 Vrard, M., Mosser, B., Barban, C., et al. 2015, A&A, 579, A84 Goupil, M. J., Mosser, B., Marques, J. P., et al. 2013, A&A, 549, A75 Grosjean, M., Dupret, M.-A., Belkacem, K., et al. 2014, A&A, 572, A11 Goupil, M. J., Mosser, B., Marques, J. P., et al. 2013, A&A, 549, A75 Vrard, M., Mosser, B., & Samadi, R. 2016, A&A, 588, A87 Vrard, M., Mosser, B., & Samadi, R. 2016, A&A, 588, A87 Vrard, M., Kallinger, T., Mosser, B., et al. 2018, A&A, 616, A94 Grosjean, M., Dupret, M.-A., Belkacem, K., et al. 2014, A&A, 572, A11 Vrard, M., Kallinger, T., Mosser, B., et al. 2018, A&A, 616, A94 A109, page 14 of 25 A109, page 14 of 25 B. Mosser et al.: Mixed modes Appendix A: Seismic parameters ζ is derived from the best asymptotic ﬁt; νas are the asymptotic frequencies, whereas ν correspond to the observed values; x = ν/∆ν−(np−εp is the reduced frequency; Γas are the asymptotic mode widths, whereas Γ correspond to the observed values; H are the observed heights, and R i the height-to-background ratio. Notes. ζ is derived from the best asymptotic ﬁt; νas are the asymptotic frequencies, whereas ν correspond to the observed values; x = ν/∆ν−(np−εp) is the reduced frequency; Γas are the asymptotic mode widths, whereas Γ correspond to the observed values; H are the observed heights, and R is the height-to-background ratio. A109, page 15 of 25 A&A 618, A109 (2018) Table A.1. continued. Appendix A: Seismic parameters np n m ζ νas ν x Γas Γ H R ( µHz) ( µHz) ( µHz) ( µHz) (ppm2µHz−1) 9 −98 −1 0.6256 117.926 117.973 ± 0.007 0.496 0.067 0.043 ± 0.007 2105 ± 327 32.6 9 −98 0 0.5622 118.077 118.138 ± 0.009 0.511 0.079 0.033 ± 0.007 5534 ± 1222 91.7 9 −98 1 0.5461 118.211 118.282 ± 0.009 0.524 0.082 0.034 ± 0.007 3206 ± 652 69.8 9 −97 −1 0.6989 118.597 118.638 ± 0.008 0.556 0.054 0.034 ± 0.007 3294 ± 723 142.5 9 −97 0 0.7744 118.766 118.815 ± 0.007 0.572 0.041 0.013 ± 0.004 3831 ± 566 66.2 9 −97 1 0.8394 118.969 118.994 ± 0.006 0.588 0.029 0.010 ± 0.004 10109 ± 1478 154.7 9 −96 −1 0.9267 119.532 119.539 ± 0.008 0.638 0.013 0.017 ± 0.005 1339 ± 307 23.5 9 −96 0 0.9428 119.758 119.758 ± 0.006 0.658 0.010 0.008 ± 0.004 4038 ± 588 46.3 9 −96 1 0.9539 119.989 119.991 ± 0.007 0.679 0.008 0.008 ± 0.004 2271 ± 444 25.7 9 −95 −1 0.9717 120.623 120.617 ± 0.006 0.735 0.005 0.006 ± 0.004 3853 ± 586 44.1 9 −95 1 0.9778 121.096 121.088 ± 0.006 0.778 0.004 0.006 ± 0.004 6667 ± 971 77.7 9 −94 −1 0.9828 121.763 121.749 ± 0.006 −0.162 0.003 0.006 ± 0.004 5234 ± 769 61.2 9 −93 −1 0.9867 122.933 122.925 ± 0.010 −0.055 0.002 0.007 ± 0.004 2021 ± 647 27.2 10 −92 0 0.9861 124.367 124.355 ± 0.006 0.074 0.002 0.006 ± 0.004 1007 ± 169 13.4 10 −91 −1 0.9833 125.344 125.329 ± 0.007 0.162 0.003 0.005 ± 0.004 2714 ± 740 33.8 10 −91 0 0.9817 125.582 125.568 ± 0.006 0.184 0.003 0.005 ± 0.004 1574 ± 381 19.7 10 −91 1 0.9800 125.819 125.796 ± 0.006 0.205 0.004 0.006 ± 0.004 5347 ± 831 67.2 10 −90 −1 0.9706 126.575 126.561 ± 0.007 0.274 0.005 0.006 ± 0.004 1987 ± 407 25.3 10 −90 0 0.9661 126.810 126.799 ± 0.008 0.296 0.006 0.009 ± 0.004 2797 ± 714 35.8 10 −90 1 0.9594 127.042 127.032 ± 0.006 0.317 0.007 0.006 ± 0.004 5283 ± 1186 67.9 10 −89 −1 0.9167 127.792 127.789 ± 0.006 0.385 0.015 0.010 ± 0.004 2454 ± 424 31.9 10 −89 0 0.8900 128.014 128.018 ± 0.007 0.406 0.020 0.007 ± 0.004 1441 ± 297 18.8 10 −89 1 0.8528 128.220 128.231 ± 0.007 0.425 0.026 0.009 ± 0.004 6599 ± 1274 92.5 10 −88 −1 0.6144 128.822 128.875 ± 0.008 0.484 0.069 0.021 ± 0.005 6832 ± 1038 132.3 10 −88 0 0.5406 128.970 129.032 ± 0.008 0.498 0.083 0.043 ± 0.007 3925 ± 617 91.8 10 −88 1 0.5044 129.095 129.168 ± 0.009 0.510 0.089 0.034 ± 0.007 5624 ± 1181 130.5 10 −87 −1 0.6672 129.569 129.608 ± 0.009 0.550 0.060 0.049 ± 0.010 3066 ± 755 157.3 10 −87 0 0.7439 129.731 129.772 ± 0.007 0.565 0.046 0.014 ± 0.004 12352 ± 1855 257.9 10 −87 1 0.8133 129.927 129.955 ± 0.006 0.582 0.034 0.011 ± 0.004 16357 ± 2408 226.5 10 −86 −1 0.9294 130.677 130.670 ± 0.006 0.647 0.013 0.007 ± 0.004 4084 ± 855 55.9 10 −86 0 0.9433 130.903 130.896 ± 0.006 0.667 0.010 0.007 ± 0.004 7019 ± 1024 96.4 10 −86 1 0.9533 131.134 131.122 ± 0.006 0.688 0.008 0.006 ± 0.004 10575 ± 2196 145.8 10 −85 −1 0.9728 131.986 131.962 ± 0.009 0.764 0.005 0.009 ± 0.004 1419 ± 431 19.8 10 −85 0 0.9756 132.221 132.196 ± 0.009 0.785 0.004 0.006 ± 0.004 530 ± 151 7.4 10 −84 0 0.9833 133.590 133.573 ± 0.006 −0.090 0.003 0.006 ± 0.004 3228 ± 566 46.4 11 −83 0 0.9844 134.993 134.973 ± 0.006 0.037 0.003 0.006 ± 0.004 4520 ± 719 66.5 11 −82 −1 0.9806 136.185 136.161 ± 0.006 0.144 0.004 0.006 ± 0.004 1267 ± 184 20.0 11 −81 −1 0.9639 137.632 137.609 ± 0.006 0.275 0.007 0.006 ± 0.004 1148 ± 170 18.7 11 −81 0 0.9578 137.866 137.837 ± 0.007 0.296 0.008 0.006 ± 0.004 900 ± 151 13.9 11 −81 1 0.9494 138.096 138.075 ± 0.006 0.318 0.009 0.011 ± 0.004 2426 ± 366 39.1 11 −80 −1 0.8644 139.037 139.015 ± 0.007 0.403 0.024 0.012 ± 0.004 2323 ± 388 36.5 11 −80 0 0.8183 139.247 139.230 ± 0.006 0.422 0.033 0.020 ± 0.005 2879 ± 427 50.3 11 −80 1 0.7594 139.431 139.438 ± 0.007 0.441 0.043 0.029 ± 0.006 942 ± 147 18.2 11 −79 −1 0.4672 140.063 140.116 ± 0.009 0.503 0.096 0.078 ± 0.015 2479 ± 552 102.8 11 −79 0 0.4628 140.175 140.185 ± 0.012 0.509 0.097 0.128 ± 0.031 1534 ± 491 74.4 11 −79 1 0.4939 140.295 140.352 ± 0.009 0.524 0.091 0.071 ± 0.012 1831 ± 330 49.4 11 −78 −1 0.8100 140.998 141.019 ± 0.006 0.585 0.034 0.013 ± 0.004 3675 ± 536 61.6 11 −78 0 0.8550 141.194 141.202 ± 0.006 0.601 0.026 0.009 ± 0.004 3583 ± 541 58.3 11 −78 1 0.8894 141.409 141.419 ± 0.009 0.621 0.020 0.027 ± 0.007 1200 ± 332 44.1 11 −77 −1 0.9561 142.436 142.432 ± 0.006 0.713 0.008 0.007 ± 0.004 1831 ± 358 30.4 11 −77 0 0.9617 142.668 142.663 ± 0.006 0.734 0.007 0.006 ± 0.004 1912 ± 278 33.3 11 −77 1 0.9667 142.902 142.895 ± 0.009 0.755 0.006 0.009 ± 0.004 434 ± 133 7.3 12 −74 −1 0.9778 147.312 147.297 ± 0.007 0.154 0.004 0.005 ± 0.004 528 ± 146 9.4 12 −74 0 0.9761 147.549 147.551 ± 0.007 0.177 0.004 0.006 ± 0.004 557 ± 102 11.1 12 −74 1 0.9739 147.785 147.778 ± 0.006 0.197 0.005 0.006 ± 0.004 665 ± 120 12.0 12 −73 −1 0.9511 148.996 148.980 ± 0.009 0.306 0.009 0.010 ± 0.004 747 ± 193 18.1 12 −73 0 0.9422 149.226 149.215 ± 0.007 0.327 0.010 0.009 ± 0.004 756 ± 160 13.9 12 −73 1 0.9306 149.452 149.441 ± 0.006 0.348 0.013 0.011 ± 0.004 726 ± 122 14.3 A109, page 16 of 25 B. Appendix A: Seismic parameters Mosser et al.: Mixed modes Table A.1. continued. np n m ζ νas ν x Γas Γ H R ( µHz) ( µHz) ( µHz) ( µHz) (ppm2µHz−1) 12 −72 −1 0.7456 150.556 150.584 ± 0.008 0.452 0.046 0.041 ± 0.008 545 ± 112 18.2 12 −72 0 0.6672 150.737 150.764 ± 0.011 0.468 0.060 0.051 ± 0.011 298 ± 79 19.8 12 −72 1 0.5911 150.880 150.951 ± 0.011 0.485 0.074 0.060 ± 0.012 444 ± 99 23.6 12 −71 −1 0.4722 151.509 151.598 ± 0.010 0.543 0.095 0.067 ± 0.012 650 ± 126 27.4 12 −71 0 0.5300 151.624 151.720 ± 0.009 0.554 0.085 0.026 ± 0.006 1788 ± 329 66.9 12 −71 1 0.6122 151.772 151.861 ± 0.012 0.567 0.070 0.077 ± 0.018 380 ± 113 20.2 12 −70 1 0.9339 153.268 153.289 ± 0.007 0.697 0.012 0.007 ± 0.004 572 ± 100 12.3 12 −69 0 0.9717 154.828 154.849 ± 0.007 −0.162 0.005 0.005 ± 0.004 587 ± 160 11.7 12 −69 1 0.9733 155.064 155.083 ± 0.007 −0.141 0.005 0.006 ± 0.004 546 ± 97 11.8 A109, page 17 of 25 A&A 618, A109 (2018) Table A.2. Oscillation pattern of the RGB star KIC 3955033. Table A.2. Oscillation pattern of the RGB star KIC 3955033. ble A.2. Oscillation pattern of the RGB star KIC 3955033. Appendix A: Seismic parameters np n m ζ νas ν x Γas Γ H R ( µHz) ( µHz) ( µHz) ( µHz) (ppm2µHz−1) Radial modes 8 84.745 84.745 ± 0.077 0.033 0.198 ± 0.089 1015 ± 656 16.2 9 93.958 93.762 ± 0.012 0.010 0.094 ± 0.017 3985 ± 872 27.3 10 102.911 103.010 ± 0.013 0.012 0.138 ± 0.027 7641 ± 1745 30.2 11 112.215 112.143 ± 0.009 0.002 0.067 ± 0.012 18117 ± 3832 114.7 12 121.869 121.474 ± 0.016 0.013 0.122 ± 0.025 2495 ± 637 19.6 Dipole mixed modes 8 −144 1 0.8744 88.882 88.894 ± 0.006 0.483 0.019 0.007 ± 0.004 5748 ± 1146 15.9 8 −141 −1 0.8051 89.078 89.075 ± 0.009 0.503 0.029 0.020 ± 0.005 2011 ± 512 7.1 8 −142 0 0.7444 89.214 89.205 ± 0.008 0.517 0.038 0.032 ± 0.006 2964 ± 565 10.9 8 −143 1 0.6908 89.338 89.338 ± 0.010 0.531 0.046 0.041 ± 0.009 2226 ± 536 15.9 8 −140 −1 0.6640 89.507 89.496 ± 0.007 0.548 0.050 0.035 ± 0.006 6623 ± 1063 21.4 8 −141 0 0.6955 89.623 89.602 ± 0.011 0.560 0.046 0.036 ± 0.008 1250 ± 297 10.1 8 −142 1 0.7483 89.743 89.725 ± 0.012 0.573 0.038 0.043 ± 0.010 1315 ± 389 10.1 8 −139 −1 0.8432 89.969 89.956 ± 0.009 0.598 0.023 0.024 ± 0.006 2118 ± 541 9.8 8 −141 1 0.9111 90.251 90.238 ± 0.009 0.629 0.013 0.006 ± 0.004 2235 ± 601 7.4 9 −127 −1 0.9047 97.755 97.779 ± 0.011 0.446 0.014 0.023 ± 0.007 1041 ± 358 9.9 9 −126 −1 0.6934 98.346 98.392 ± 0.007 0.512 0.046 0.014 ± 0.004 36038 ± 5534 121.7 9 −127 0 0.6800 98.375 98.395 ± 0.007 0.512 0.048 0.024 ± 0.005 21840 ± 3803 121.7 9 −128 1 0.6615 98.413 98.395 ± 0.007 0.512 0.051 0.027 ± 0.005 19513 ± 3160 121.7 9 −125 −1 0.6969 98.825 98.818 ± 0.007 0.558 0.045 0.014 ± 0.004 10668 ± 1614 47.6 9 −126 0 0.7049 98.843 98.879 ± 0.007 0.565 0.044 0.028 ± 0.006 6104 ± 971 27.3 9 −127 1 0.7220 98.874 98.883 ± 0.007 0.565 0.042 0.031 ± 0.006 5427 ± 941 27.3 9 −124 −1 0.9050 99.430 99.435 ± 0.006 0.625 0.014 0.006 ± 0.004 9343 ± 1572 29.3 9 −126 1 0.9124 99.474 99.470 ± 0.006 0.629 0.013 0.012 ± 0.004 7355 ± 1118 23.3 9 −125 1 0.9647 100.170 100.166 ± 0.007 0.704 0.005 0.006 ± 0.004 2664 ± 543 9.3 10 −120 1 0.9883 103.940 103.940 ± 0.007 0.113 0.002 0.006 ± 0.004 3119 ± 574 11.1 10 −118 −1 0.9880 104.023 104.025 ± 0.010 0.122 0.002 0.007 ± 0.004 2677 ± 836 10.2 10 −118 0 0.9848 104.778 104.780 ± 0.011 0.204 0.002 0.007 ± 0.004 2318 ± 827 9.3 10 −117 −1 0.9844 104.838 104.842 ± 0.009 0.211 0.002 0.006 ± 0.004 4417 ± 1446 14.7 10 −116 −1 0.9748 105.657 105.666 ± 0.006 0.300 0.004 0.005 ± 0.004 4718 ± 1166 15.8 10 −115 −1 0.9438 106.481 106.502 ± 0.022 0.391 0.008 0.019 ± 0.011 1349 ± 1022 16.4 10 −116 1 0.8515 107.096 107.070 ± 0.008 0.452 0.022 0.008 ± 0.004 1890 ± 368 10.1 10 −115 0 0.8294 107.167 107.194 ± 0.011 0.466 0.026 0.041 ± 0.011 1090 ± 354 9.5 10 −114 −1 0.8023 107.238 107.259 ± 0.006 0.473 0.030 0.016 ± 0.004 5430 ± 799 21.0 10 −115 1 0.5811 107.710 107.706 ± 0.007 0.521 0.063 0.019 ± 0.005 22251 ± 3274 99.4 10 −114 0 0.5753 107.757 107.776 ± 0.011 0.529 0.064 0.082 ± 0.021 3573 ± 1204 70.9 10 −113 −1 0.5782 107.819 107.818 ± 0.008 0.533 0.063 0.042 ± 0.009 6466 ± 1487 71.0 10 −114 1 0.7802 108.260 108.253 ± 0.008 0.580 0.033 0.018 ± 0.005 7228 ± 1753 32.4 10 −113 0 0.8149 108.344 108.337 ± 0.009 0.589 0.028 0.027 ± 0.007 3135 ± 809 35.8 10 −112 −1 0.8468 108.441 108.429 ± 0.009 0.599 0.023 0.029 ± 0.007 3843 ± 1006 25.7 10 −113 1 0.9383 109.023 109.019 ± 0.006 0.663 0.009 0.008 ± 0.004 11460 ± 1670 41.2 10 −112 1 0.9728 109.876 109.870 ± 0.006 0.756 0.004 0.006 ± 0.004 6439 ± 938 23.4 11 −104 −1 0.9239 115.807 115.818 ± 0.006 0.400 0.011 0.007 ± 0.004 3908 ± 613 15.8 11 −105 1 0.8432 116.274 116.279 ± 0.006 0.450 0.023 0.009 ± 0.004 6992 ± 1054 26.2 11 −104 0 0.7745 116.468 116.470 ± 0.009 0.471 0.034 0.033 ± 0.008 1247 ± 360 18.8 11 −103 −1 0.6842 116.648 116.670 ± 0.007 0.492 0.047 0.028 ± 0.006 2247 ± 418 28.7 11 −104 1 0.5371 116.961 116.976 ± 0.010 0.525 0.069 0.054 ± 0.011 2255 ± 530 17.6 11 −103 0 0.5477 117.092 117.077 ± 0.015 0.536 0.068 0.082 ± 0.021 1275 ± 415 19.0 11 −102 −1 0.6181 117.248 117.211 ± 0.010 0.551 0.057 0.028 ± 0.007 1479 ± 387 11.9 11 −103 1 0.8015 117.612 117.600 ± 0.007 0.593 0.030 0.017 ± 0.005 2633 ± 464 14.5 11 −101 −1 0.9071 118.072 118.063 ± 0.017 0.643 0.014 0.031 ± 0.013 479 ± 289 10.9 otes. Appendix A: Seismic parameters Radial modes and mixed modes identiﬁed in KIC 3955033 with a height-to-background ratio R larger than 7. Same caption as Table A A109, page 18 of 25 B. Mosser et al.: Mixed modes B. Mosser et al.: Mixed modes Fig. A.1. Fit of the oscillation pattern of the low RGB star KIC 10272858, at the limit of validity of the asymptotic pattern. Owing to the small radial orders, small shifts are seen between observed and asymptotic spectra. Same style as Fig. 2, but ℓ= 3 modes appear near the abscissa 0.28. Fig. A.1. Fit of the oscillation pattern of the low RGB star KIC 10272858, at the limit of validity of the asymptotic pattern. Owing to the smal radial orders small shifts are seen between observed and asymptotic spectra Same style as Fig 2 but ℓ= 3 modes appear near the abscissa 0 28 Fig. A.1. Fit of the oscillation pattern of the low RGB star KIC 10272858, at the limit of validity of the asymptotic pattern. Owing to the small radial orders, small shifts are seen between observed and asymptotic spectra. Same style as Fig. 2, but ℓ= 3 modes appear near the abscissa 0.28. Fig. A.2. Fit of the oscillation pattern of the RGB star KIC 11353313. Same style as Fig. 2. Fig. A.2. Fit of the oscillation pattern of the RGB star KIC 11353313. Same style as Fig. 2. A109, page 19 of 25 A&A 618, A109 (2018) Fig A 3 Fit of the oscillation pattern of the RGB star KIC 3955033 The overlap of mixed modes with diﬀerent mixed mode orders is Fig. A.3. Fit of the oscillation pattern of the RGB star KIC 3955033. The overlap of mixed modes with diﬀerent mixed-mode orders is the signature of the rapid core rotation. The second rotation crossing, where all components of the multiplets overlap (Gehan et al. 2017), occurs at the mixed-order n = −122. Same style as Fig. 2. Fig. A.3. Fit of the oscillation pattern of the RGB star KIC 3955033. The overlap of mixed modes with diﬀerent mixed-mode orders is the signature of the rapid core rotation. The second rotation crossing, where all components of the multiplets overlap (Gehan et al. 2017), occurs at the mixed-order n = −122. Same style as Fig. 2. Fig. A.4. Appendix A: Seismic parameters Fit of the oscillation pattern of the evolved RGB star KIC 2443903, near the limit of capability of identiﬁcation, with a large crowding due to the high mode density. The second rotation crossing, where all m components apparently coincide, occurs at n = −189 (with an abscissa ≃0.1 and np = 9); the third crossing, where \|m\| = 1 components apparently coincide with m = 0 inbetween, occurs at n = −233 (with an abscissa ≃0.25 and np = 7). Same style as Fig. 2. Note that the modes with large heights at an abscissa ≃0.2 are ℓ= 3 modes. Fig. A.4. Fit of the oscillation pattern of the evolved RGB star KIC 2443903, near the limit of capability of identiﬁcation, with a large crowding due to the high mode density. The second rotation crossing, where all m components apparently coincide, occurs at n = −189 (with an abscissa ≃0.1 and np = 9); the third crossing, where \|m\| = 1 components apparently coincide with m = 0 inbetween, occurs at n = −233 (with an abscissa ≃0.25 and np = 7). Same style as Fig. 2. Note that the modes with large heights at an abscissa ≃0.2 are ℓ= 3 modes. A109, page 20 of 25 B. Mosser et al.: Mixed modes B. Mosser et al.: Mixed modes Fig. A.5. Fit of the oscillation pattern of the red-clump star KIC 1723700. Buoyancy glitches explain the small shifts between observed and asymptotic spectra but do no hamper the mode identiﬁcation. Same style as Fig. 2. Fig. A.5. Fit of the oscillation pattern of the red-clump star KIC 1723700. Buoyancy glitches explain the small shifts between observed and asymptotic spectra but do no hamper the mode identiﬁcation Same style as Fig 2 Fig. A.5. Fit of the oscillation pattern of the red-clump star KIC 1723700. Buoyancy glitches explain the small shifts between observed and asymptotic spectra but do no hamper the mode identiﬁcation. Same style as Fig. 2. Fig A 6 Fit of the oscillation pattern of the secondary clump star KIC 1725190 Buoyancy glitches explain the small shifts between observed and Fig. A.6. Fit of the oscillation pattern of the secondary-clump star KIC 1725190. Buoyancy glitches explain the small shifts between observed and asymptotic spectra but do no hamper the mode identiﬁcation. Same style as Fig. 2. Fig. A.6. Fit of the oscillation pattern of the secondary-clump star KIC 1725190. Appendix A: Seismic parameters Buoyancy glitches explain the small shifts between observed and asymptotic spectra but do no hamper the mode identiﬁcation. Same style as Fig. 2. A109, page 21 of 25 A&A 618, A109 (2018) Appendix B: Stars in open clusters is in disagreement with Corsaro et al. (2017) who found an inclination i = 20 ± 7◦. All stars studied by Corsaro et al. (2017) were investigated. The ﬁtting process is challenging, due to the dim magnitudes of such dim stars in open clusters. However, the combination of all pres- sure radial orders near νmax provides in most cases an unambigu- ous ﬁt, and at least a few mixed-mode radial orders provide clear splittings. – Similar conclusions are reached for KIC 2437976 (Fig. B.2), member of the open cluster NGC 6791. Corsaro et al. (2017) found an inclination i = 0 ± 10◦, despite the fact \|m\| = 1 modes are clearly identiﬁed and indicate i = 76 ± 14◦. – Similar conclusions are reached for KIC 2437976 (Fig. B.2), member of the open cluster NGC 6791. Corsaro et al. (2017) found an inclination i = 0 ± 10◦, despite the fact \|m\| = 1 modes are clearly identiﬁed and indicate i = 76 ± 14◦. – Similar conclusions are reached for KIC 2437976 (Fig. B.2), member of the open cluster NGC 6791. Corsaro et al. (2017) found an inclination i = 0 ± 10◦, despite the fact \|m\| = 1 modes are clearly identiﬁed and indicate i = 76 ± 14◦. These stars are representative of the whole data set treated by Corsaro et al. (2017): the inability of the ﬁtting process to iden- tify thin short-lived mixed modes translates into the identiﬁca- tion of a single broad m = 0 peak. In such cases, stellar incli- nations derived from the Bayesian ﬁts are necessarily underesti- mated and biased toward low values. p g – Figure B.1 provides the asymptotic ﬁt of KIC 5024476, member of the open cluster NGC 6819 observed by Kepler. We note that m = ±1 modes are clearly identiﬁed and derive a stellar inclination i = 79 ± 11◦for this star. This result Fig. B.1. Fit of the oscillation pattern of the RGB star KIC 5024476, member of the open cluster NGC 6819. The dim magnitude of the cluster stars explains the low S/N. However, unambiguous doublets are identiﬁed all along the mixed-mode spectrum; m = 0 modes are mostly absent and \|m\| = 1 modes dominate the mixed-mode spectrum, so that a nearly pole-on inclination is not possible. Same style as Fig. 2. Fig. B.1. Appendix B: Stars in open clusters Fit of the oscillation pattern of the RGB star KIC 5024476, member of the open cluster NGC 6819. The dim magnitude of the cluster stars explains the low S/N. However, unambiguous doublets are identiﬁed all along the mixed-mode spectrum; m = 0 modes are mostly absent and \|m\| = 1 modes dominate the mixed-mode spectrum, so that a nearly pole-on inclination is not possible. Same style as Fig. 2. A109, page 22 of 25 B. Mosser et al.: Mixed modes Fig. B.2. Fit of the oscillation pattern of the RGB star KIC 2437976, member of the open cluster NGC 6791. The dim magnitude of the cluster stars explains the low S/N. The identiﬁcation at radial order 9, supported by the radial orders 8 and 10, is unambiguously conclusive: m = 0 modes are mostly absent and \|m\| = 1 modes dominate the mixed-mode spectrum, so that a pole-on inclination is not possible. Same style as Fig. 2. Fig. B.2. Fit of the oscillation pattern of the RGB star KIC 2437976, member of the open cluster NGC 6791. The dim magnitude of the cluster stars explains the low S/N. The identiﬁcation at radial order 9, supported by the radial orders 8 and 10, is unambiguously conclusive: m = 0 modes are mostly absent and \|m\| = 1 modes dominate the mixed-mode spectrum, so that a pole-on inclination is not possible. Same style as Fig. 2. A109, page 23 of 25 A&A 618, A109 (2018) Table B.1. Oscillation pattern of the red-clump star KIC 5024476 in NGC 6819. able B.1. Oscillation pattern of the red-clump star KIC 5024476 in NGC 6819. Appendix B: Stars in open clusters np n m ζ νas ν x Γas Γ H R ( µHz) ( µHz) ( µHz) ( µHz) (ppm2µHz−1) Radial modes 10 63.335 63.096 ± 0.174 −0.058 0.501 ± 0.216 698 ± 429 9.9 11 69.039 69.039 ± 0.055 −0.021 0.350 ± 0.102 1365 ± 536 9.7 12 74.941 74.941 ± 0.039 0.009 0.231 ± 0.067 1158 ± 452 35.6 Dipole mixed modes 10 −43 −1 0.9495 63.873 63.870 ± 0.009 0.077 0.013 0.019 ± 0.006 1887 ± 532 16.0 10 −43 1 0.9481 63.979 63.976 ± 0.007 0.095 0.013 0.009 ± 0.004 3002 ± 632 12.7 10 −42 −1 0.9006 65.035 65.005 ± 0.015 0.275 0.025 0.045 ± 0.012 497 ± 175 11.6 10 −42 0 0.8955 65.086 65.131 ± 0.012 0.297 0.026 0.028 ± 0.008 576 ± 191 7.6 10 −42 1 0.8903 65.136 65.131 ± 0.011 0.297 0.027 0.027 ± 0.008 589 ± 186 7.6 10 −41 −1 0.6178 66.050 66.050 ± 0.013 0.457 0.096 0.095 ± 0.020 1859 ± 503 20.0 10 −41 0 0.6011 66.085 66.061 ± 0.013 0.459 0.100 0.099 ± 0.022 1959 ± 563 20.0 10 −41 1 0.5866 66.117 66.083 ± 0.014 0.463 0.103 0.106 ± 0.025 1783 ± 534 18.7 10 −40 −1 0.6453 66.784 66.788 ± 0.017 0.586 0.089 0.085 ± 0.020 800 ± 240 8.9 10 −40 1 0.6794 66.858 66.897 ± 0.020 0.605 0.080 0.097 ± 0.023 632 ± 197 14.4 10 −39 1 0.9132 67.985 68.003 ± 0.007 0.798 0.022 0.009 ± 0.004 4805 ± 844 23.5 11 −37 −1 0.9084 70.563 70.562 ± 0.007 0.244 0.023 0.009 ± 0.004 2295 ± 521 11.4 11 −37 1 0.9003 70.665 70.672 ± 0.006 0.264 0.025 0.008 ± 0.004 2474 ± 398 14.8 11 −36 −1 0.6140 71.776 71.786 ± 0.015 0.458 0.097 0.103 ± 0.022 1228 ± 333 18.4 11 −36 1 0.5811 71.843 71.884 ± 0.019 0.475 0.105 0.127 ± 0.030 854 ± 264 12.1 11 −35 −1 0.6100 72.588 72.615 ± 0.010 0.603 0.097 0.072 ± 0.013 2715 ± 560 29.9 11 −35 1 0.6455 72.658 72.625 ± 0.011 0.605 0.089 0.078 ± 0.015 2287 ± 527 29.9 11 −34 −1 0.9051 73.870 73.881 ± 0.007 −0.176 0.024 0.016 ± 0.005 2320 ± 417 12.5 11 −34 1 0.9111 73.972 73.977 ± 0.007 −0.160 0.022 0.008 ± 0.004 4124 ± 750 22.2 12 −33 −1 0.9314 75.420 75.422 ± 0.006 0.093 0.017 0.007 ± 0.004 5610 ± 987 31.2 12 −33 1 0.9296 75.524 75.530 ± 0.008 0.112 0.018 0.008 ± 0.004 1498 ± 379 8.3 12 −32 −1 0.8053 76.972 76.954 ± 0.008 0.360 0.049 0.024 ± 0.006 1836 ± 391 14.5 12 −32 1 0.7828 77.061 77.060 ± 0.009 0.379 0.054 0.034 ± 0.007 1103 ± 222 13.2 12 −31 −1 0.4338 78.016 78.029 ± 0.017 0.548 0.142 0.156 ± 0.035 1362 ± 387 32.1 12 −31 0 0.4373 78.041 78.045 ± 0.018 0.550 0.141 0.160 ± 0.036 1326 ± 379 16.4 12 −31 1 0.4416 78.066 78.062 ± 0.017 0.553 0.140 0.154 ± 0.034 1354 ± 382 14.3 12 −30 −1 0.8397 79.185 79.226 ± 0.008 0.757 0.040 0.013 ± 0.004 2127 ± 377 12.8 12 −30 1 0.8538 79.280 79.310 ± 0.009 0.771 0.037 0.027 ± 0.006 977 ± 236 9.7 otes. Appendix B: Stars in open clusters Same caption as Table A.1. Notes. Same caption as Table A.1. A109, page 24 of 25 A109, page 24 of 25 B. Mosser et al.: Mixed modes Table B.2. Oscillation pattern of the RGB star KIC 2437976 in NGC 6791. Table B.2. Oscillation pattern of the RGB star KIC 2437976 in NGC 6791. Notes. Same caption as Table A.1. Appendix B: Stars in open clusters np n m ζ νas ν x Γas Γ H R ( µHz) ( µHz) ( µHz) ( µHz) (ppm2µHz−1) Radial modes 8 75.045 75.045 ± 0.041 0.013 0.106 ± 0.042 6520 ± 3676 7.8 9 83.284 83.172 ± 0.024 0.001 0.153 ± 0.036 15645 ± 4765 31.3 10 91.382 91.331 ± 0.013 −0.006 0.100 ± 0.017 22777 ± 4443 31.1 Dipole mixed modes 8 −161 −1 0.8533 78.887 78.893 ± 0.008 0.481 0.026 0.022 ± 0.006 10581 ± 2376 12.3 8 −160 −1 0.6438 79.234 79.170 ± 0.010 0.514 0.064 0.038 ± 0.007 14647 ± 2841 11.9 8 −161 1 0.6511 79.343 79.276 ± 0.012 0.527 0.063 0.031 ± 0.007 10930 ± 2616 9.6 8 −159 −1 0.7884 79.561 79.519 ± 0.008 0.557 0.038 0.026 ± 0.006 14986 ± 2978 15.1 8 −160 1 0.8566 79.691 79.671 ± 0.009 0.575 0.026 0.023 ± 0.006 6508 ± 1802 9.8 9 −146 −1 0.9726 86.115 86.141 ± 0.010 0.362 0.005 0.009 ± 0.004 16564 ± 5717 7.6 9 −144 −1 0.7541 87.141 87.185 ± 0.010 0.489 0.044 0.054 ± 0.013 20608 ± 6137 33.9 9 −145 1 0.7180 87.193 87.191 ± 0.009 0.490 0.051 0.049 ± 0.011 23464 ± 6034 33.9 9 −143 −1 0.6110 87.503 87.500 ± 0.011 0.528 0.070 0.067 ± 0.014 19703 ± 4780 29.1 9 −144 1 0.6404 87.550 87.517 ± 0.009 0.530 0.065 0.050 ± 0.009 26627 ± 5525 29.1 9 −142 −1 0.8736 87.940 87.943 ± 0.007 0.582 0.023 0.009 ± 0.004 21699 ± 5272 10.0 9 −143 1 0.8929 88.001 87.989 ± 0.006 0.587 0.019 0.009 ± 0.004 29926 ± 5102 13.8 9 −141 −1 0.9588 88.486 88.498 ± 0.007 0.649 0.007 0.007 ± 0.004 17180 ± 3838 8.1 9 −142 1 0.9622 88.537 88.540 ± 0.006 0.654 0.007 0.006 ± 0.004 38360 ± 5733 18.3 10 −132 1 0.9463 94.663 94.689 ± 0.007 0.402 0.010 0.006 ± 0.004 19058 ± 3790 9.7 10 −131 −1 0.9440 94.692 94.708 ± 0.008 0.405 0.010 0.005 ± 0.004 15841 ± 4952 7.6 10 −131 1 0.7969 95.259 95.216 ± 0.009 0.466 0.037 0.034 ± 0.008 21679 ± 5266 18.4 10 −130 −1 0.7827 95.284 95.257 ± 0.009 0.471 0.039 0.033 ± 0.008 19865 ± 4877 20.4 10 −130 1 0.5518 95.692 95.630 ± 0.008 0.517 0.081 0.025 ± 0.005 42908 ± 7274 26.4 10 −129 −1 0.5572 95.709 95.633 ± 0.008 0.517 0.080 0.027 ± 0.006 39045 ± 6386 26.4 10 −129 1 0.8453 96.161 96.162 ± 0.007 0.582 0.028 0.009 ± 0.004 29045 ± 5946 21.6 10 −125 0 0.9891 98.575 98.551 ± 0.008 −0.128 0.002 0.004 ± 0.004 15436 ± 11044 7.5 11 −118 −1 0.8082 103.469 103.439 ± 0.010 0.467 0.034 0.023 ± 0.006 6844 ± 1785 9.8 12 −108 1 0.7958 111.880 111.821 ± 0.012 0.487 0.037 0.022 ± 0.006 5310 ± 1445 7.9 12 −105 −1 0.9174 113.342 113.329 ± 0.007 0.670 0.015 0.006 ± 0.004 16746 ± 4518 8.6 Notes Same caption as Table A 1 A109, page 25 of 25 A109, page 25 of 25
https://openalex.org/W2945901734	https://hal.archives-ouvertes.fr/hal-02379667/file/PhysRevLett.123.063601.pdf	English	null	Entanglement between a Diamond Spin Qubit and a Photonic Time-Bin Qubit at Telecom Wavelength	Physical review letters	2,019	cc-by	5,442	Entanglement between a Diamond Spin Qubit and a Photonic Time-Bin Qubit at Telecom Wavelength Anna Tchebotareva, Sophie L. N. Hermans, Peter C. Humphreys, Dirk Voigt, Peter J. Harmsma, Lun K. Cheng, Ad L. Verlaan, Niels Dijkhuizen, Wim de Jong, Anaïs Dréau, et al. To cite this version: Anna Tchebotareva, Sophie L. N. Hermans, Peter C. Humphreys, Dirk Voigt, Peter J. Harmsma, et al.. Entanglement between a Diamond Spin Qubit and a Photonic Time-Bin Qubit at Telecom Wavelength. Physical Review Letters, 2019, 123 (6), pp.063601. 10.1103/PhysRevLett.123.063601. hal-02379667 Distributed under a Creative Commons Attribution 4.0 International License HAL Id: hal-02379667 https://hal.science/hal-02379667v1 Submitted on 2 Jun 2021 L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés. HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. Distributed under a Creative Commons Attribution 4.0 International License Delft University of Technology Entanglement between a Diamond Spin Qubit and a Photonic Time-Bin Qubit at Telecom Wavelength DOI: 10.1103/PhysRevLett.123.063601 DOI: 10.1103/PhysRevLett.123.063601 Quantum networks connecting and entangling long-lived qubits via photonic channels [1] may enable new experi- ments in quantum science as well as a range of applications such as secure information exchange between multiple nodes, distributed quantum computing, clock synchroniza- tion, and quantum sensor networks [2–10]. A key building block for long-distance entanglement distribution via opti- cal fibers is the generation of entanglement between a long- lived qubit and a photonic telecom-wavelength qubit [see Fig. 1(a)]. Such building blocks are now actively explored for various qubit platforms [11–17]. photon polarization after frequency conversion has recently been demonstrated [13–16,26,27], no experiments using NV centers and no experiments using time-bin photonic qubits have so far been reported. Recently, we have realized the conversion of 637 nm NV photons to 1588 nm (in the telecom L band) using a difference frequency generation (DFG) process and shown that the intrinsic single-photon character is maintained during this process [28]. However, for entanglement distribution an additional critical require- ment is that the quantum information encoded by the photon is preserved during the frequency conversion. The nitrogen-vacancy (NV) center in diamond is a promising candidate to act as a node in such quantum networks thanks to a combination of long spin coherence and spin-selective optical transitions that allow for high- fidelity initialization and single-shot readout [18]. Moreover, memory qubits are provided in the form of surrounding carbon-13 nuclear spins. These have been employed for demonstrations of quantum error correction [19–21] and entanglement distillation [22]. Heralded entanglement between separate NV center spin qubits has been achieved by generating spin-photon entangled states followed by a joint measurement on the photons [23]. Here we demonstrate entanglement between a NV center spin qubit and a time-bin encoded frequency-converted photonic qubit at telecom wavelength. The concept of our experiment is depicted in Fig. 1(b). We first generate spin- photon entanglement, then convert the photonic qubit to the telecom band, and finally characterize the resulting state through spin-photon correlation measurements in different bases. We use two of the NV center electron spin-1 sublevels as our qubit subspace. We denote the ms ¼ 0 and ms ¼ −1 ground states as j0i and j1i, respectively. To generate the desired spin-photon entangled state, we first initialize the spin in j0i and prepare the balanced superposition jψi ¼ ð1= ﬃﬃﬃ 2 p Þðj0i þ j1iÞ using a microwave (MW) π=2 pulse. Entanglement between a Diamond Spin Qubit and a Photonic Time-Bin Qubit at Telecom Wavelength Entanglement between a Diamond Spin Qubit and a Photonic Time-Bin Qubit at Telecom Wavelength Tchebotareva, Anna; Hermans, Sophie L.N.; Humphreys, Peter C.; Voigt, Dirk; Harmsma, Peter J.; Cheng, Lun K.; Verlaan, Ad L.; Dijkhuizen, Niels; De Jong, Wim; Dréau, Anaïs DOI 10.1103/PhysRevLett.123.063601 Publication date 2019 Document Version Final published version Published in Physical Review Letters Citation (APA) Tchebotareva, A., Hermans, S. L. N., Humphreys, P. C., Voigt, D., Harmsma, P. J., Cheng, L. K., Verlaan, A. L., Dijkhuizen, N., De Jong, W., Dréau, A., & Hanson, R. (2019). Entanglement between a Diamond Spin Qubit and a Photonic Time-Bin Qubit at Telecom Wavelength. Physical Review Letters, 123(6), [063601]. https://doi.org/10.1103/PhysRevLett.123.063601 Citation (APA) Tchebotareva, A., Hermans, S. L. N., Humphreys, P. C., Voigt, D., Harmsma, P. J., Cheng, L. K., Verlaan, A. L., Dijkhuizen, N., De Jong, W., Dréau, A., & Hanson, R. (2019). Entanglement between a Diamond Spin Qubit and a Photonic Time-Bin Qubit at Telecom Wavelength. Physical Review Letters, 123(6), [063601]. https://doi.org/10.1103/PhysRevLett.123.063601 Important note To cite this publication, please use the final published version (if applicable). Please check the document version above. Copyright Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons. ght than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the conse author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons. Takedown policy Please contact us and provide details if you believe this document breaches copyrights. We will remove access to the work immediately and investigate your claim. This work is downloaded from Delft University of Technology. For technical reasons the number of authors shown on this cover page is limited to a maximum of 10. PHYSICAL REVIEW LETTERS 123, 063601 (2019) (Received 21 May 2019; published 5 August 2019) We report on the realization and verification of quantum entanglement between a nitrogen-vacancy electron spin qubit and a telecom-band photonic qubit. First we generate entanglement between the spin qubit and a 637 nm photonic time-bin qubit, followed by photonic quantum frequency conversion that transfers the entanglement to a 1588 nm photon. We characterize the resulting state by correlation measurements in different bases and find a lower bound to the Bell state fidelity of ≥0.77 0.03. This result presents an important step towards extending quantum networks via optical fiber infrastructure. Entanglement between a Diamond Spin Qubit and a Photonic Time-Bin Qubit at Telecom Wavelength Anna Tchebotareva,1,2,* Sophie L. N. Hermans,1,3,* Peter C. Humphreys,1,3 Dirk Voigt,1,2 Peter J. Harmsma,1,2 Lun K. Cheng,1,2 Ad L. Verlaan,1,2 Niels Dijkhuizen,1,2 Wim de Jong,1,2 Anaïs Dr´eau,1,3,4 and Ronald Hanson1,3,† 1QuTech, Delft University of Technology, P.O. Box 5046, 2600 GA Delft, Netherlands 2Netherlands Organisation for Applied Scientific Research (TNO), P.O. Box 155, 2600 AD Delft, Netherlands 3Kavli Institute of Nanoscience, Delft University of Technology, P.O. Box 5046, 2600 GA Delft, Netherlands 4Laboratoire Charles Coulomb, Universit´e de Montpellier and CNRS, 34095 Montpellier, France Anna Tchebotareva,1,2,* Sophie L. N. Hermans,1,3,* Peter C. Humphreys,1,3 Dirk Voigt,1,2 Peter J. Harmsma,1,2 Lun K. Cheng,1,2 Ad L. Verlaan,1,2 Niels Dijkhuizen,1,2 Wim de Jong,1,2 Anaïs Dr´eau,1,3,4 and Ronald Hanson1,3,† 1QuTech, Delft University of Technology, P.O. Box 5046, 2600 GA Delft, Netherlands 2Netherlands Organisation for Applied Scientific Research (TNO), P.O. Box 155, 2600 AD Delft, Netherlands 3Kavli Institute of Nanoscience, Delft University of Technology, P.O. Box 5046, 2600 GA Delft, Netherlands 4Laboratoire Charles Coulomb, Universit´e de Montpellier and CNRS, 34095 Montpellier, France DOI: 10.1103/PhysRevLett.123.063601 Then we apply a spin-selective optical π pulse, such that the j0i state will be excited, followed by photon emission (lifetime of 12 ns). Next, we flip the spin state using a MW π pulse and apply the optical excitation for a second time. This generates the following spin-photon entangled state: Extending such entanglement distribution over long distances is severely hindered by photon loss in the fibers. The wavelength at which the NV center emits resonant photons, the so-called zero-phonon line at 637 nm, exhibits high attenuation in optical glass fibers. Quantum-coherent frequency conversion to the telecom band can mitigate these losses by roughly 7 orders of magnitude for a distance of 10 km [24,25] and would enable the quantum network to optimally benefit from the existing telecom fiber infra- structure. While entanglement between matter systems and jNV spin; photoni ¼ ð1= ﬃﬃﬃ 2 p Þj1; Ei þ ð1= ﬃﬃﬃ 2 p Þj0; Li; ð1Þ ð1Þ jNV spin; photoni ¼ ð1= ﬃﬃﬃ 2 p Þj1; Ei þ ð1= ﬃﬃﬃ 2 p Þj0; Li; ð1Þ 063601-1 © 2019 American Physical Society © 2019 American Physical Society 0031-9007=19=123(6)=063601(6) PHYSICAL REVIEW LETTERS 123, 063601 (2019) FIG. 1. (a) In a long-distance quantum network, heralded entanglement between the nodes is generated by interference on a beam splitter and subsequent measurement of telecom photonic qubits. (b) Concept of the experiment. A NV spin qubit– photon qubit entangled state is generated. The photonic qubit is converted to a telecom wavelength by the difference frequency generation (DFG) process [28]. For measuring the photonic qubit in the X and Y bases, an imbalanced interferometer is used. Inset: Simplified level scheme of the NV center. (c) DFG process. A 637 nm photon is converted to a wavelength of 1588 nm using late time bins. To access other photonic qubit bases we use an imbalanced interferometer [29] with a tunable phase difference Δϕ between the two arms. For each photonic qubit basis, we read out the spin state in the basis where maximum correlation is expected. From the measured correlations in three orthogonal bases, we find the fidelity to the desired maximally entangled state. The diamond sample containing the NV center is cooled to ≈4 K. The optical setup is schematically depicted in Fig. 2(a). Laser light at 637 nm is used to apply the optical π pulses. DOI: 10.1103/PhysRevLett.123.063601 In the photon detection path, the emitted 637 nm photons are separated from reflected excitation light using a cross-polarization configuration and time filtering. The off- resonant phonon sideband emission is separated by dichroic filtering and sent to a detector (D1) for spin readout. The 637 nm photons are combined with a strong pump laser (emission wavelength of 1064 nm) and directed into the PPLN crystal for the DFG process. Afterwards, the remain- ing pump laser light is filtered out by a prism, a long-pass dielectric filter and a narrow band fiber Bragg grating. The total conversion efficiency of the DFG setup is ηc ≈17% [28]. To ensure the frequency and phase stability of the converted photons, both the NV excitation laser and the pump laser are locked to an external reference cavity (Stable Laser Systems). Figure 2(b) shows the experimental sequence used in the experiments. Our protocol starts with checking whether the NV center is in the desired charge state and on resonance with the control lasers [30]. Once this test is passed, the spin-photon entangled state is generated. If a photon is detected, we read out the spin state in the appropriate basis and restart the protocol. In case no photon is detected, we reinitialize the spin and again generate an entangled state. After 250 failed attempts to detect a photon, we restart the protocol. FIG. 1. (a) In a long-distance quantum network, heralded entanglement between the nodes is generated by interference on a beam splitter and subsequent measurement of telecom photonic qubits. (b) Concept of the experiment. A NV spin qubit– photon qubit entangled state is generated. The photonic qubit is converted to a telecom wavelength by the difference frequency generation (DFG) process [28]. For measuring the photonic qubit in the X and Y bases, an imbalanced interferometer is used. Inset: Simplified level scheme of the NV center. (c) DFG process. A 637 nm photon is converted to a wavelength of 1588 nm using a nonlinear PPLN crystal. Left inset: Electron microscope image of the diamond device. Right inset: Image of the PPLN crystal with ridge waveguides (NTT/NEL). We first measure spin-photon correlations in the ZZ basis. DOI: 10.1103/PhysRevLett.123.063601 (a) Experimental setup for the spin-telecom photon entangled state generation. Emitted 637 nm photons are combined with the pump laser (1064 nm) in the difference frequency generation setup (DFG1). The two lasers are frequency locked to an external reference cavity. Tomography in the Z basis: the frequency-converted photons are detected using a superconducting nanowire detector (D2) discriminating the early and late time bins. (b) Experimental protocol for generating and detecting spin-telecom photon entangled states (see main text). (c) Results for correlations measured in the Z basis for both the red and the frequency-converted photons at telecom wavelength. A key requirement for this experiment is that the interferometer is stable with respect to the frequency of the down-converted photons; any instabilities in the inter- ferometer will reduce the interference contrast and prevent us from accessing the true spin-photon correlations. For this reason, the interferometer is thermally and vibration- ally isolated. Furthermore, we split the experiment into cycles of 1 s [see Fig. 4(a)], of which the first 100 ms is used to actively stabilize the phase setpoint of the inter- ferometer. Within this 100 ms, we feed metrology light into the interferometer in the reverse direction via shutter S and a circulator. This metrology light is generated by a second DFG setup, using input from the excitation and pump lasers, thus ensuring a fixed frequency relation between the metrology light and the frequency-converted photons. By comparing the light intensities on detectors PD2 and PD3 with the values corresponding to the desired Δϕ setpoint as determined from a visibility fringe (calibrated every 100 s), an error signal is computed and feedback is applied to the fiber piezostretcher (FPS). After this adjustment the light intensities are measured again. A histogram of the mea- sured phases during the experiments relative to the set- points is plotted in Fig. 4(b). We note that one could also measure the spin-photon correlations at the second output of the interferometer, which for symmetric states as Eq. (1) would yield the same correlations but with opposite sign; however, in the current experiment the slow (≈1 s) recov- ery of the detector after being blinded due to metrology light leakage through this output port prevented us from using the second output. In the remaining 900 ms of each cycle, spin-photon correlations are measured using the same protocol as for the ZZ basis [see Fig. 2(b)]. DOI: 10.1103/PhysRevLett.123.063601 To measure the photon in the Z basis, we send the frequency-converted photons directly to a superconducting nanowire detector (D2) that projects the photonic qubit in the time-bin basis, and, upon photon detection, we read out the spin qubit in the corresponding Z basis. Figure 2(c) shows the observed correlation data. The probability to measure the spin in j0i is plotted for photon detection events in the early and late time bins. We have performed this measurement for both the 637 nm photons (red) and the frequency-converted photons at 1588 nm (purple). For the unconverted photons we measure correlations that are perfect within measurement uncertainty [contrast of EZ ¼ jPEðj0iÞ −PLðj0iÞj ¼ 0.997 0.018]. For the frequency- converted photons we measure PEðj0iÞ ¼ 0.09 0.05 for the early time bin and PLðj0iÞ ¼ 0.95 0.05 for the late time bin, yielding a contrast of EZ ¼ 0.86 0.07. All data in this work are corrected for spin readout infidelity and dark counts of the detectors, both of which are determined independently. where the basis states for the photonic qubit are given by the early (jEi) and late (jLi) time bins, which are separated in the experiment by 190 ns, limited by the state prepa- ration time. Next, the photon is converted to the telecom wavelength of 1588 nm using a difference frequency generation process, by combining it with a strong pump laser inside a periodically poled lithium niobate (PPLN) crystal wave- guide [Fig. 1(c)] [28]. The resulting spin-telecom photon state is characterized via correlation measurements. We read out the photonic qubit in the Z basis using time- resolved detection that discriminates between the early and 063601-2 PHYSICAL REVIEW LETTERS 123, 063601 (2019) PHYSICAL REVIEW LETTERS 123, 063601 (2019) FIG. 2. (a) Experimental setup for the spin-telecom photon entangled state generation. Emitted 637 nm photons are combined with the pump laser (1064 nm) in the difference frequency generation setup (DFG1). The two lasers are frequency locked to an external reference cavity. Tomography in the Z basis: the frequency-converted photons are detected using a superconducting nanowire detector (D2) discriminating the early and late time bins. (b) Experimental protocol for generating and detecting spin-telecom photon entangled states (see main text). (c) Results for correlations measured in the Z basis for both the red and the frequency-converted photons at telecom wavelength. , FIG. 2. PHYSICAL REVIEW LETTERS 123, 063601 (2019) Light intensities measured by photodiodes PD2 and PD3 are used to generate a feedback signal to the fiber piezostretcher (FPS) to maintain the target phase Δϕ. (b) Bloch sphere with the selected photon qubit readout bases indicated on it, and the corresponding phase setpoints of the imbalanced interferometer. FIG. 3. (a) Polarization-maintaining fiber-based imbalanced interferometer used for the photon state readout in X and Y bases. The frequency-converted single photons are directed into the interferometer. One output port of the interferometer is connected to a superconducting nanowire detector (detector D3). Every second the phase of the interferometer is stabilized. Classical frequency-converted light created by a second DFG setup (DFG2) is sent into the interferometer via a shutter S and a circulator. Light intensities measured by photodiodes PD2 and PD3 are used to generate a feedback signal to the fiber piezostretcher (FPS) to maintain the target phase Δϕ. (b) Bloch sphere with the selected photon qubit readout bases indicated on it, and the corresponding phase setpoints of the imbalanced interferometer. FIG. 4. (a) Experimental protocol for measurements in the photon X and Y bases. (b) Measured phase difference Δϕ just before stabilization (orange, with 900 ms free evolution time) and directly after stabilization (blue) for the two setpoints ΔϕX ¼ π=4 and ΔϕY ¼ 3π=4. From the standard deviations in these data, we estimate a residual phase drift of 0.05 and 0.01 rad=s for the X and Y photon qubit readout bases, respectively. (c) Results for the correlations in the X and Y bases in purple. The top panel shows ideal correlations. In total we have measured 1595 photon detection events. out the NV spin state in the appropriate rotated basis, the eigenstates jXi (jYi) and the j −Xi (j −Yi) are mapped onto the j0i and j1i states, respectively, by applying an appropriate MW pulse before optical readout. Figure 4(c) shows the measured spin-photon correlations in the X and Y bases (bottom), along with expected correlations for the ideal state (top). The letters indicate the spin and photon bases, respectively; for example, −XX indicates that the NV spin is measured along the −X axis on the Bloch sphere, while the photon is projected on þX. The measured contrast between the correlations and anti- correlations in the X basis is EX ¼ 0.52 0.07 and EY ¼ 0.69 0.07 in the Y basis. PHYSICAL REVIEW LETTERS 123, 063601 (2019) FIG. 4. (a) Experimental protocol for measurements in the photon X and Y bases. (b) Measured phase difference Δϕ just before stabilization (orange, with 900 ms free evolution time) and directly after stabilization (blue) for the two setpoints ΔϕX ¼ π=4 and ΔϕY ¼ 3π=4. From the standard deviations in these data, we estimate a residual phase drift of 0.05 and 0.01 rad=s for the X and Y photon qubit readout bases, respectively. (c) Results for the correlations in the X and Y bases in purple. The top panel shows ideal correlations. In total we have measured 1595 photon detection events. TERS 123, 063601 (2019) out the NV spin state in the appropriate rotated basis the FIG. 3. (a) Polarization-maintaining fiber-based imbalanced interferometer used for the photon state readout in X and Y bases. The frequency-converted single photons are directed into the interferometer. One output port of the interferometer is connected to a superconducting nanowire detector (detector D3). Every second the phase of the interferometer is stabilized. Classical frequency-converted light created by a second DFG setup (DFG2) is sent into the interferometer via a shutter S and a circulator. Light intensities measured by photodiodes PD2 and PD3 are used to generate a feedback signal to the fiber piezostretcher (FPS) to maintain the target phase Δϕ. (b) Bloch sphere with the selected photon qubit readout bases indicated on it, and the corresponding phase setpoints of the imbalanced interferometer. FIG. 4. (a) Experimental protocol for measurements in the photon X and Y bases. (b) Measured phase difference Δϕ just before stabilization (orange, with 900 ms free evolution time) and directly after stabilization (blue) for the two setpoints ΔϕX ¼ π=4 and ΔϕY ¼ 3π=4. From the standard deviations in these data, we estimate a residual phase drift of 0 05 and 0 01 rad=s for the X PHYSICAL REVIEW LETTERS 123, 063601 (2019) FIG. 3. (a) Polarization-maintaining fiber-based imbalanced interferometer used for the photon state readout in X and Y bases. The frequency-converted single photons are directed into the interferometer. One output port of the interferometer is connected to a superconducting nanowire detector (detector D3). Every second the phase of the interferometer is stabilized. Classical frequency-converted light created by a second DFG setup (DFG2) is sent into the interferometer via a shutter S and a circulator. DOI: 10.1103/PhysRevLett.123.063601 To read The contrast for the telecom photons is lowered by noise coming from spontaneous parametric down-converted pho- tons and Raman scattering induced by the strong pump laser [28,31]. We characterize this noise contribution separately by blocking the incoming 637 nm path and find an expected signal-to-noise ratio (SNR) between 4.8 and 7.3. This SNR bounds the maximum observable contrast for the ZZ correlations to SNR=ðSNR þ 1Þ ¼ 0.85 0.03, and thus fully explains our data. We use this SNR later to determine the different noise contributions for the correlation data in the other bases. Additionally, we conclude from the relative number of detection events in the early and late time bin (659 versus 642 events) that the amplitudes of the two parts of the spin-photon entangled state are well balanced. To verify the spin-photon entanglement, we measure spin-photon correlations in two other spin-photon bases by sending the frequency-converted photons into the imbal- anced fiber interferometer [see Fig. 3(a)]. The fiber arm length difference is ≈40 m, which corresponds to a photon travel time difference of 190 ns between the two arms. In this way the early time bin taking the long arm overlaps at the second beam splitter with the late time bin taking the short arm, thus allowing us to access the phase relation between the two. To access a specific photon qubit basis, we introduce a tunable phase difference Δϕ between the long and short arms of the interferometer. In particular, detection of a photon by the detector D3 projects the spin into the state jNViD3 ¼ ð1= ﬃﬃﬃ 2 p Þðj0i þ eiðΔϕ−π=4Þj1iÞ: ð2Þ ð2Þ We use two orthogonal set points, labeled X and Y, with Δϕ ¼ π=4 and Δϕ ¼ 3π=4, respectively, as indicated in Fig. 3(b). 063601-3 PHYSICAL REVIEW LETTERS 123, 063601 (2019) PHYSICAL REVIEW LETTERS 123, 063601 (2019) PHYSICAL REVIEW LETTERS 123, 063601 (2019) PHYSICAL REVIEW LETTERS 123, 063601 (2019) [4] Liang Jiang, J. M. Taylor, Kae Nemoto, W. J. Munro, Rodney Van Meter, and M. D. Lukin, Phys. Rev. A 79, 032325 (2009). contrast of the XX and YY correlations, but not that of the ZZ correlations. Second, spontaneous parametric down- conversion and Raman scattered photons, produced during the frequency conversion process, add noise to the state as described above and reduce correlations in all bases. Based on these factors, we expect a state fidelity in the range 0.82–0.87. [5] A. Ekert and R. Renner, Nature (London) 507, 443 (2014). [6] J. I. Cirac, A. K. Ekert, S. F. Huelga, and C. Macchiavello, Phys. Rev. A 59, 4249 (1999). [7] D. Gottesman, T. Jennewein, and S. Croke, Phys. Rev. Lett. 109, 070503 (2012). The slight difference between the expected and measured state fidelity could be due to the inaccuracies and fluctua- tions in setting the interferometer phase setpoint. Imperfect interferometer settings result in measurement bases that slightly deviate from the expected X and Y bases, reducing the maximally observable correlations. Therefore, the obtained F ≥0.77 0.03 sets a lower bound on the true entangled state fidelity. [8] N. H. Nickerson, J. F. Fitzsimons, and S. C. Benjamin, Phys. Rev. X 4, 041041 (2014). [9] P. Kómár, E. M. Kessler, M. Bishof, L. Jiang, A. S. Sørensen, J. Ye, and M. D. Lukin, Nat. Phys. 10, 582 (2014). [10] J.-D. Bancal, S. Pironio, A. Acn, Y.-C. Liang, V. Scarani, and N. Gisin, Nat. Phys. 8, 867 (2012). [11] P. Farrera, G. Heinze, and H. de Riedmatten, Phys. Rev. Lett. 120, 100501 (2018). g y In conclusion, we demonstrated entanglement between a NV center spin qubit and a time-bin encoded photonic qubit at telecom wavelength, which is an essential step towards long-distance quantum networks based on remote entanglement between NV center nodes. The current frequency conversion combined with recently demon- strated entangling protocols [35] would lead to a spin- spin entangling rate of a few hertz over 30 km of fiber (ignoring heralding signals). In future experiments the observed state fidelity can be further increased in several ways. A more narrow band frequency filtering after the DFG1 setup would reduce the added noise in the fre- quency conversion, as the current narrow band filter has a linewidth ∼10 times larger than the NV-emitted resonant zero-phonon line photons. *A. T. and S. L. N. H. contributed equally to this work. †R.Hanson@tudelft.nl [1] S. Wehner, D. Elkouss, and R. Hanson, Science 362, eaam9288 (2018). [2] M. Ben-Or, C. Crepeau, D. Gottesman, A. Hassidim, and A. Smith,inProceedingsofthe47thAnnualIEEESymposium on Foundations of Computer Science (IEEE, New York, 2006). [3] A. Broadbent, J. Fitzsimons, and E. Kashefi, in Proceedings of the 50th Annual IEEE Symposium on Foundations of Computer Science (IEEE, New York, 2009). PHYSICAL REVIEW LETTERS 123, 063601 (2019) The signal could be increased by improving the conversion efficiency. Finally, the emission rate of resonant photons and the collection efficiency can be increased by placing the NV center in an optical cavity [36–41], which could also increase the entangling rate by up to 3 orders of magnitude. [12] T. Walker, K. Miyanishi, R. Ikuta, H. Takahashi, S. Vartabi Kashanian, Y. Tsujimoto, K. Hayasaka, T. Yamamoto, N. Imoto, and M. Keller, Phys. Rev. Lett. 120, 203601 (2018). [13] V. Krutyanskiy, M. Meraner, J. Schupp, V. Krcmarsky, H. Hainzer, and B. P. Lanyon, arXiv:1901.06317. [14] Y. O. Dudin, A. G. Radnaev, R. Zhao, J. Z. Blumoff, T. A. B. Kennedy, and A. Kuzmich, Phys. Rev. Lett. 105, 260502 (2010). [15] M. Bock, P. Eich, S. Kucera, M. Kreis, A. Lenhard, C. Becher, and J. Eschner, Nat. Commun. 9, 1998 (2018). [16] K. De Greve, L. Yu, P. L. McMahon, J. S. Pelc, C. M. Natarajan, N. Y. Kim, E. Abe, S. Maier, C. Schneider, M. Kamp, S. Höfling, R. H. Hadfied, A. Forchel, M. M. Fejer, and Y. Yamamoto, Nature (London) 491, 421 (2012). [17] J. H. Weber, B. Kambs, J. Kettler, S. Kern, J. Maisch, H. Vural, M. Jetter, S. L. Portalupi, C. Becher, P. Michler, Nat. Nanotechnol. 14, 23 (2019). [18] D. D. Awschalom, R. Hanson, J. Wrachtrup, and B. B. Zhou, Nat. Photonics 12, 516 (2018). [19] G. Waldherr, Y. Wang, S. Zaiser, M. Jamali, T. Schulte- Herbrüggen, H. Abe, T. Ohshima, J. Isoya, J. F. Du, P. Neumann, and J. Wrachtrup, Nature (London) 506, 204 (2014). We thank L. Schriek, E. Nieuwkoop, J. Lugtenburg, and W. Peterse for experimental assistance and M. J. A. de Dood and C. Osorio Tamayo for useful discussions. We acknowledge financial support from the Netherlands Organisation for Scientific Research (NWO) through a VICI grant and the Zwaartekracht Grant Quantum Software Consortium and the European Research Council through an ERC Consolidator Grant. [20] T. H. Taminiau, J. Cramer, T. van der Sar, V. V. Dobrovitski, and R. Hanson, Nat. Nanotechnol. 9, 171 (2014). [21] J. Cramer, N. Kalb, M. A. Rol, B. Hensen, M. S. Blok, M. Markham, D. J. Twitchen, R. Hanson, and T. H. Taminiau, Nat. Commun. 7, 11526 (2016). [22] N. Kalb, A. A. Reiserer, P. C. Humphreys, J. J. W. Bakermans, S. J. Kamerling, N. H. Nickerson, S. C. Benjamin, D. J. Twitchen, M. Markham, and R. PHYSICAL REVIEW LETTERS 123, 063601 (2019) yielding a fidelity of F ¼ 0.77 0.03. This value exceeds the classical boundary of 0.5 by more than 8 standard deviations, proving the generation of entanglement between the NV spin qubit and the frequency-converted photonic qubit. For comparison, reported fidelities for unconverted NV spin-photon entangled states range from ≈0.7 [32,33] to more than 0.9 (estimated from an observed spin-spin entangled state fidelity of ≈0.9 [34]). All data show clear (anti)correlation between the NV spin qubit and the telecom photonic qubit. With the contrast data from all three orthogonal photon readout bases, we calculate the fidelity F of our produced state (conditioned on photon detection) to the maximally entangled state of Eq. (1) as The observed fidelity is reduced compared to the ideal value of 1 due to several factors. First, the initial spin-photon entangled state has imperfections, for instance, due to photon emission and reexcitation of the NV center during the optical π pulse [35] and small frequency shifts due to spectral diffusion. In addition, the remaining frequency variations of the two locked lasers (∼200 kHz) lead to phase uncertainty between the two terms in Eq. (1). All these effects reduce the F ¼ 1 4 ð1 þ EX þ EY þ EZÞ; ð3Þ ð3Þ 063601-4 PHYSICAL REVIEW LETTERS 123, 063601 (2019) Hanson, Science 356, 928 (2017). [23] H. Bernien, B. Hensen, W. Pfaff, G. Koolstra, M. S. Blok, L. Robledo, T. H. Taminiau, M. Markham, D. J. Twitchen, L. Childress, and R. Hanson, Nature (London) 497, 86 (2013). [24] S. Wehner, D. Elkouss, and R. Hanson, Science 362, eaam9288 (2002). [25] T. Miya, Y. Terunuma, T. Hosaka, and T. Miyashita, Electron. Lett. 15, 106 (1979). [26] R. Ikuta, T. Kobayashi, T. Kawakami, S. Miki, M. Yabuno, T. Yamashita, H. Terai, M. Koashi, T. Mukai, T. Yamamoto, and N. Imoto, Nat. Commun. 9, 1997 (2018). 063601-5 PHYSICAL REVIEW LETTERS 123, 063601 (2019) PHYSICAL REVIEW LETTERS 123, 063601 (2019) [36] A. Faraon, C. Santori, Z. Huang, V. M. Acosta, and R. G. Beausoleil, Phys. Rev. Lett. 109, 033604 (2012). [27] Y. Yu, F. Ma, X. Luo, B. Jing, P. Sun, R. Fang, C. Yang, H. Liu, M. Zheng, X. Xie, W. Zhang, L. You, Z. Wang, T. Chen, Q. Zhang, X. Bao, and J. Pan, arXiv:1903.11284. [37] S. Johnson, P. R. Dolan, T. Grange, A. A. P. Trichet, G. Hornecker, Y. C. Chen, L. Weng, G. M. Hughes, A. A. R. Watt, A. Auff`eves, and J. M. Smith, New J. Phys. 17, 122003 (2015). [28] A. Dr´eau, A. Tchebotareva, A. E. Mahdaoui, C. Bonato, and R. Hanson, Phys. Rev. Applied 9, 064031 (2018). [29] J. D. Franson, Phys. Rev. Lett. 62, 2205 (1989). [38] H. Kaupp, T. Hümmer, M. Mader, B. Schlederer, J. Benedikter, P. Haeusser, H.-Ch. Chang, H. Fedder, T. W. Hänsch, and D. Hunger, Phys. Rev. Applied 6, 054010 (2016). [30] L. Robledo, L. Childress, H. Bernien, B. Hensen, P. F. A. Alkemade, and R. Hanson, Nature (London) 477, 574 (2011). [31] J. S. Pelc, C. Langrock, Q. Zhang, and M. M. Fejer, Opt. Lett. 35, 2804 (2010). [32] E. Togan et al., Nature (London) 466, 730 (2010). [39] D. Riedel, I. Söllner, B. J. Shields, S. Starosielec, P. Appel, E. Neu, P. Maletinsky, and R. J. Warburton, Phys. Rev. X 7, 031040 (2017). [33] R. Vasconcelos et al., arXiv:1812.10338. [34] B. Hensen, H. Bernien, A. E. Dr´eau, A. Reiserer, N. Kalb, M. S.Blok,J.Ruitenberg,R. F. L.Vermeulen,R. N.Schouten, C. Abellán, W. Amaya, V. Pruneri, M. W. Mitchell, M. Markham, D. J. Twitchen, D. Elkouss, S. Wehner, T. H. Taminiau, and R. Hanson, Nature (London) 526, 682 (2015). [40] S. Bogdanovic, S. B. van Dam, C. Bonato, L. C. Coenen, A.-M. J. Zwerver, B. Hensen, M. S. Z. Liddy, T. Fink, A. Reiserer, M. Loncar, and R. Hanson, Appl. Phys. Lett. 110, 171103 (2017). [41] N. H. Wan, B. J. Shields, D. Kim, S. Mouradian, B. Lienhard, M.Walsh, H. Bakhru, T. Schrder, and D. Englund, Nano Lett. 18, 2787 (2018). [35] P. C. Humphreys, N. Kalb, J. P. J. Morits, R. N. Schouten, R. F. L. Vermeulen, D. J. Twitchen, M. Markham, and R. Hanson, Nature (London) 558, 268 (2018). 063601-6
https://openalex.org/W2160871951	https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0015481&type=printable	English	null	An Integrative Approach to the Identification of Arabidopsis and Rice Genes Involved in Xylan and Secondary Wall Development	PloS one	2,010	cc-by	16,219	Abstract Xylans constitute the major non-cellulosic component of plant biomass. Xylan biosynthesis is particularly pronounced in cells with secondary walls, implying that the synthesis network consists of a set of highly expressed genes in such cells. To improve the understanding of xylan biosynthesis, we performed a comparative analysis of co-expression networks between Arabidopsis and rice as reference species with different wall types. Many co-expressed genes were represented by orthologs in both species, which implies common biological features, while some gene families were only found in one of the species, and therefore likely to be related to differences in their cell walls. To predict the subcellular location of the identified proteins, we developed a new method, PFANTOM (plant protein family information-based predictor for endomembrane), which was shown to perform better for proteins in the endomembrane system than other available prediction methods. Based on the combined approach of co-expression and predicted cellular localization, we propose a model for Arabidopsis and rice xylan synthesis in the Golgi apparatus and signaling from plasma membrane to nucleus for secondary cell wall differentiation. As an experimental validation of the model, we show that an Arabidopsis mutant in the PGSIP1 gene encoding one of the Golgi localized candidate proteins has a highly decreased content of glucuronic acid in secondary cell walls and substantially reduced xylan glucuronosyltransferase activity. Citation: Oikawa A, Joshi HJ, Rennie EA, Ebert B, Manisseri C, et al. (2010) An Integrative Approach to the Identification of Arabidopsis and Rice Genes Involved in Xylan and Secondary Wall Development. PLoS ONE 5(11): e15481. doi:10.1371/journal.pone.0015481 ation: Oikawa A, Joshi HJ, Rennie EA, Ebert B, Manisseri C, et al. (2010) An Integrative Approach to the Identification of Arabidopsis and an and Secondary Wall Development. PLoS ONE 5(11): e15481. doi:10.1371/journal.pone.0015481 Editor: Samuel P. Hazen, University of Massachusetts Amherst, United States of America Editor: Samuel P. Hazen, University of Massachusetts Amherst, United States of America Received August 20, 2010; Accepted September 24, 2010; Published November 23, 2010 This is an open-access article distributed under the terms of the Creative Commons Public Domain declaration which stipulates that, once placed in the public domain, this work may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. Funding: This work was supported by the US Department of Energy, Office of Science, Office of Biological and Environmental Research, through contract DE- AC02-05CH11231 with Lawrence Berkeley National Laboratory. A.O. Abstract was additionally supported by funds from Japanese Yamada Science Foundation, and E.A.R. was supported by a National Institutes of Health (NIH) Pre-doctoral Training Grant. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. * E-mail: HScheller@lbl.gov An Integrative Approach to the Identification of Arabidopsis and Rice Genes Involved in Xylan and Secondary Wall Development Hiren J. Joshi1,2, Emilie A. Rennie1,3, Berit Ebert1,2, Chithra Manisseri1,2, Joshua L. , Henrik Vibe Scheller1,2,3* Ai Oikawa1,2, Hiren J. Joshi1,2, Emilie A. Rennie1,3, Berit Ebert1,2, Chithra Manisseri1,2, Joshua L. Heazlewood1,2, Henrik Vibe Scheller1,2,3* 1 Feedstocks Division, Joint BioEnergy Institute, Emeryville, California, United States of America, 2 Physical Biosciences Division, Lawrence Berkeley National Laboratory, Berkeley, California, United States of America, 3 Department of Plant & Microbial Biology, University of California, Berkeley, California, United States of America Co-expression of three irx genes encoding Arabidopsis xylan synthase y y In Arabidopsis, genes IRX9 (At2g37090), IRX14 (At4g36890), and IRX10 (At1g27440) are members of the GT43A, GT43B and GT47D subfamilies. The nomenclature used here for the different clades in GT43 and GT47 is according to Arabidopsis [28] and Brachypodium [29] studies and differs from that used to designate poplar genes [30]. These IRX genes are all expressed in tissues with secondary wall growth and are involved in xylan backbone synthesis [13,14]. To view the correlation of their expression patterns across many microarray experiments, we performed a scatter plot analysis using CoexViewer based on the 237 data sets related to developmental series in the ATTED-II database (Figure S1) [31]. Scatter plots of pairwise combinations of the three genes showed very similar patterns and strongly correlated expression, whereas the negative control RALF gene (At4g15800), which is mainly expressed in tissues with primary wall growth (e.g. rosette leaf), did not show any correlation with IRX9 expression. To identify other candidate genes likely to be involved in xylan accumulation, we used the CoexSearch tool available at ATTED- II, which uses MR for evaluation of the correlation between two different gene expressions. Table S1 shows the 300 most highly co- expressed Arabidopsis genes obtained for each of the three baits, IRX9, IRX14, and IRX10. Each co-expression list included the three bait genes as strongly co-expressed genes (low MR), and many genes were shared between the three sets of 300 genes, with a total of 124 genes (ca. 23%) shared amongst all three data sets (Table S2; Figure 1A). Co-expression analysis of genes is a method to identify candidate proteins involved in the same biological process, including proteins that function together in a complex. Along with the accumulation of microarray datasets, transcriptome co- expression analysis has proven to be a powerful tool for identifying regulatory relationships in the transcriptional networks of model organisms, including Escherichia coli [16], yeast [17] and Arabi- dopsis [18]. While Arabidopsis is well established as the primary model species in plant biology, rice is quickly gaining popularity as a model organism. In addition to the availability of substantial genetic, molecular, and genomic resources, two features make rice attractive as a reference species: it represents distinct monocots and is a crop species. In an important transcriptional study, Mitchell et al. [19] compared EST data available for members of the Poaceae with transcriptional data for dicots. Co-expression of three irx genes encoding Arabidopsis xylan synthase Based on this data they proposed candidates of GT families involved in grass xylan synthesis. Recently, high-density Affymetrix array data for rice has become publicly available, thereby enabling more sensitive co- expression profiling analysis for rice [20]. The significance of this tight linkage amongst the three genes was further examined by analyzing other members of the Arabidopsis GT43 (IRX9-L and IRX14-L) and GT47D (F8H, FRA8 and IRX10-L) families [32]. The maximum MR range for the 300th gene had weaker values ranging from 400 to 600 (except FRA8) when compared to co-expression sets for IRX9, IRX10 and IRX14 where the maximum MR was less than 400 (Table 1). Surprisingly, Arabidopsis FRA8, which appears to be involved in forming the oligosaccharide at the reducing end of xylan, did not tightly co-express with IRX9, IRX10 or IRX14. FRA8 produced a tight network with an MR,192.8 for the 300th most highly co- expressed genes (Table 1), but this network did not overlap considerably with the network defined by xylan backbone synthesis genes IRX9, IRX10 and IRX14. Furthermore, the FRA8 network did not include the two other known genes implicated in synthesis of the oligosaccharide, i.e. IRX8 and PARVUS [4,13] while both these genes were co-expressed with IRX9/IRX10/ IRX14 genes. p p g y [ ] A number of online tools are available for plant co-expression analysis [21]. Among them, GeneCAT and ATTED-II are databases available for both Arabidopsis and rice co-expression data [22,23]. ATTED-II currently uses array data from 1388 and 208 GeneChip slides for Arabidopsis and rice, respectively, and genes co-expressed with bait genes are listed according to ‘Mutual Rank’ (MR), which performs significantly better than Pearson’s correlation coefficient value [24]. In addition to the co-expression analysis, information regarding subcellular localization can also assist in determining functional associations between proteins [25]. A number of methods have been developed to predict the subcellular location of eukaryotic proteins. These methods can be broadly classified into methods utilizing sorting signals, experi- mental annotations, and amino acid composition [26]. While these approaches have been used to predict protein localizations in a variety of eukaryotic organelles, they have had limited success when applied to compartments of the endomembrane system [27]. In this study, we performed a comparative analysis of co- expression networks between Arabidopsis and rice, focusing on xylan biosynthesis. Introduction In Arabidopsis, several genes involved in the formation of the secondary wall have been identified by screening for irregular xylem (irx) mutants and analysis of genes co-expressed with genes already shown to be involved in secondary wall PLoS ONE \| www.plosone.org November 2010 \| Volume 5 \| Issue 11 \| e15481 1 PLoS ONE \| www.plosone.org Xylan Biosynthesis Network in Arabidopsis and Rice formation [11,12]. Several of the irx mutants are affected in genes encoding Type II membrane GTs that appear to be involved in xylan biosynthesis. These genes include members of families GT8 (PARVUS, IRX8), GT43 (IRX9 and IRX14), and GT47 (FRA8 also known as IRX7, and IRX10) [4,13,14]. The corresponding mutants have decreased xylan in stems, and an increase in the proportion of 4-O-Me-GlcA side branches relative to the non- methylated GlcA [13,15]. Further biochemical analysis of the xylan reducing end structure and xylan chain length suggested that FRA8, IRX8 and PARVUS are involved in the synthesis of the reducing end structure, whereas IRX9, IRX10 and IRX14 may function in xylan backbone chain elongation [4,13–15]. None of these proteins have had their biochemical activity demonstrated, but nevertheless the evidence that they are somehow involved in xylan biosynthesis is strong. The b-1,4-linked backbone of xylan led many to expect that CSL proteins would be responsible for synthesis of the backbone, but this seems highly unlikely as there is no candidate CSL family available for such an activity. formation [11,12]. Several of the irx mutants are affected in genes encoding Type II membrane GTs that appear to be involved in xylan biosynthesis. These genes include members of families GT8 (PARVUS, IRX8), GT43 (IRX9 and IRX14), and GT47 (FRA8 also known as IRX7, and IRX10) [4,13,14]. The corresponding mutants have decreased xylan in stems, and an increase in the proportion of 4-O-Me-GlcA side branches relative to the non- methylated GlcA [13,15]. Further biochemical analysis of the xylan reducing end structure and xylan chain length suggested that FRA8, IRX8 and PARVUS are involved in the synthesis of the reducing end structure, whereas IRX9, IRX10 and IRX14 may function in xylan backbone chain elongation [4,13–15]. None of these proteins have had their biochemical activity demonstrated, but nevertheless the evidence that they are somehow involved in xylan biosynthesis is strong. Introduction The b-1,4-linked backbone of xylan led many to expect that CSL proteins would be responsible for synthesis of the backbone, but this seems highly unlikely as there is no candidate CSL family available for such an activity. predictions of subcellular locations, we propose a model of Arabidopsis and rice xylan synthesis and conserved signaling components for secondary cell wall development. Introduction including Arabidopsis, have been shown to contain the complex structure b-D-Xyl-(1R4)-b-D-Xyl-(1R3)-a-L-Rha-(1R2)-a-D-GalA -(1R4)-D-Xyl at the reducing end [2,3,4]. Such structures are yet to be reported in grasses. Plant cell walls are complex structures, predominantly com- posed of polysaccharides. Secondary walls develop in some cell types after the termination of cell expansion, and these walls usually contain lignin in addition to polysaccharides. The polysaccharides in secondary walls are largely represented by cellulose and hemicelluloses, particularly xylans. Pectin and other hemicelluloses, e.g. mannans and xyloglucans are much less abundant in secondary walls. For a recent review of hemicellulose structure and function, see Scheller and Ulvskov [1]. Xylans have a backbone of 1,4-linked b-xylosyl residues, some of which are substituted with single glucuronosyl (GlcA), 4-O-methyl-GlcA, and arabinofuranosyl residues. Furthermore, the xylose residues can be acetylated at O-2 and/or O-3, and in Poales the arabinofuranosyl residues can be feruloylated at O-5. More complex side chains can also be present, and the structural patterns vary both between species and tissues. Secondary walls in angiosperms contain xylan as the major hemicellulose, and this xylan generally has little or no arabinose and a high acetate content. Grass xylans tend to have more arabinose, and no arabinose has been detected in xylan from Arabidopsis. Xylans from different dicot and gymnosperm species, p g Pectin and hemicelluloses are synthesized in Golgi vesicles by glycosyltransferases (GTs) which use nucleotide sugars as donor substrates. The understanding of this biosynthesis is still rather limited, but multi-membrane-spanning enzymes belonging the Cellulose Synthase Like (CSL) family of proteins have been shown to synthesize b-1,4-linked backbones of mannans and glucoman- nans and be involved in biosynthesis of mixed linkage glucans and xyloglucan backbones. In contrast, the backbone of pectic homogalacturonan and sidechains of hemicelluloses and pectins seem to be synthesized by other families of GTs that are Type II membrane proteins. Recent reviews describe biosynthesis of hemicelluloses and pectin [1,5–10]. Despite the abundance of xylans and their importance in wood, animal feed and food, little was known until recently about the genes required for xylan biosynthesis. PLoS ONE \| www.plosone.org Co-expression of three irx genes encoding Arabidopsis xylan synthase From a list comprising 1146 co-expressed genes from Arabidopsis and rice using the ATTED-II database, we identified novel candidates involved in signal transduction, regulation and substrate transport, as well as enzymes directly involved in secondary wall biosynthesis. Furthermore, to predict their subcellular localization, we developed a new algorithm employing a Pfam-based method with experimental data from Arabidopsis. Based on the co-expression analysis and the A number of online tools are available for plant co-expression analysis [21]. Among them, GeneCAT and ATTED-II are databases available for both Arabidopsis and rice co-expression data [22,23]. ATTED-II currently uses array data from 1388 and 208 GeneChip slides for Arabidopsis and rice, respectively, and genes co-expressed with bait genes are listed according to ‘Mutual Rank’ (MR), which performs significantly better than Pearson’s correlation coefficient value [24]. In addition to the co-expression analysis, information regarding subcellular localization can also assist in determining functional associations between proteins [25]. A number of methods have been developed to predict the subcellular location of eukaryotic proteins. These methods can be broadly classified into methods utilizing sorting signals, experi- mental annotations, and amino acid composition [26]. While these approaches have been used to predict protein localizations in a variety of eukaryotic organelles, they have had limited success when applied to compartments of the endomembrane system [27]. Co-expression analysis of GT43 and GT47D genes in rice Co-expression analysis of GT43 and GT47D genes in rice To gain a better understanding of the similarities and difference between xylan gene networks in Arabidopsis and in grasses, we also investigated gene networks in rice. Rice has ten and seven genes belonging to the GT43 and GT47D families, respectively (Table 1). Phylogenic analysis clearly separated the ten GT43 and seven GT47D genes into distinct clades, with six genes in the IRX10/IRX10-L clade, one gene in the FRA8/F8H clade, eight genes in the IRX9/IRX9-L clade, and two genes in the IRX14/ IRX14-L clade (Figure 2A). We examined the expression of the rice GT43 and GT47D genes in different developmental stages In this study, we performed a comparative analysis of co- expression networks between Arabidopsis and rice, focusing on xylan biosynthesis. From a list comprising 1146 co-expressed genes from Arabidopsis and rice using the ATTED-II database, we identified novel candidates involved in signal transduction, regulation and substrate transport, as well as enzymes directly involved in secondary wall biosynthesis. Furthermore, to predict their subcellular localization, we developed a new algorithm employing a Pfam-based method with experimental data from Arabidopsis. Based on the co-expression analysis and the PLoS ONE \| www.plosone.org November 2010 \| Volume 5 \| Issue 11 \| e15481 2 Xylan Biosynthesis Network in Arabidopsis and Rice Figure 1. Comparative co-expression analysis between Arabi- dopsis and rice. (A, B) Venn diagrams of the co-expressed genes with each of the three individual baits from Arabidopsis (A) and rice (B). (C, D) Expression profiles of the 25 genes that most closely match with the baits (thick lines). To confirm co-regulation, the transcriptional expression pattern of the top 25 shared genes of the 3-way intersection from Arabidopsis (A) and rice (B) is plotted. The y-axes show relative gene expression values in base-2 logarithm against the average expression levels of each gene. doi:10.1371/journal.pone.0015481.g001 selected the IRX9 ortholog designated ‘OsGT43A’, IRX14 ortholog designated ‘OsGT43B’, and IRX10 ortholog designated ‘OsGT47D’ and used them as baits for co-expression analysis at ATTED-II (Table 1, Table S3). A large number of shared genes were observed in the pairwise combinations (Figure 1B) with OsGT47D- OsGT43B (165 genes), OsGT47D-OsGT43A (121 genes), and OsGT43A-OsGT43B (123 genes). Combining the three genes, 83 (ca. 14%) of the 300 highest ranked genes were shared (Table S4). Development of a Pfam-based predictor for plant endomembrane localization Knowledge of both co-expression of genes and sub-cellular localization of the corresponding proteins contribute to our understanding of protein function and putative interactions. To date, existing prediction algorithms have been unable to reliably predict localization to the endomembrane system in plants. For other eukaryotes, the pTARGET database employing a genome wide prediction method based on location-specific functional domains currently provides the best prediction for subcellular location in the endomembrane systems [36]. Inspired by this technique we developed a predictor that utilizes Protein functional domain information (Pfam) [37] to predict plant sub-cellular localization. The prediction algorithm required a Pfam training set to establish the baseline distribution of the Pfam domains across multiple subcellular locations. The most comprehensive source of subcellular localization data was obtained from the AmiGO database (http://amigo.geneontology.org). To further improve the robustness of the training set, only 2740 experimentally evidenced (i.e. associated with the ‘IDA’ tag) entries were selected from the 5077 Arabidopsis proteins found in this database. This experi- mental AmiGO data set was further segmented into groups based upon the subcellular localization associated with the annotation, and Pfam domain information retrieved for each protein. This analysis resulted in data sets that map any given Pfam domain to an experimentally observed subcellular localization. In contrast to pTARGET, the developed prediction algorithm was modified to allow for more than one Pfam domain to contribute to the determination of localization. The pTARGET algorithm bases predictions upon Pfam domains that are uniquely located in specific subcellular compartments. In Arabidopsis, such a method would lead to, at best, 50% of the proteins being correctly localized. By accepting domains that are distributed across different localizations, all loci with Pfam domains can be identified. The trade-off to this method is that the algorithmic detection can become overly broad, and the specificity of the algorithm is lowered. The efficacy of the algorithm is dictated by the size of the training set, and the number of individual Pfam domains that are found in each subcellular compartment (Table S5). Table S6 can be used for the Pfam-based prediction tool. Subcellular localization prediction using Table S6 are performed by finding Pfam domain of a given target protein. For example, a bHLH protein (AT5G48560) has a PF00010 domain and the highest prediction score for PF00010 is 80.9% for nuclear localization. On the other hand, a LRR protein (AT1G67510) contains three Pfam domains, PF000560, PF00069, and PF08263. Development of a Pfam-based predictor for plant endomembrane localization The highest prediction scores for these Pfam domains are 82.4%, 70.3% and 87.8% for plasma membrane localization, respectively. The final prediction score for this LRR protein for plasma Figure 1. Comparative co-expression analysis between Arabi- dopsis and rice. (A, B) Venn diagrams of the co-expressed genes with each of the three individual baits from Arabidopsis (A) and rice (B). (C, D) Expression profiles of the 25 genes that most closely match with the baits (thick lines). To confirm co-regulation, the transcriptional expression pattern of the top 25 shared genes of the 3-way intersection from Arabidopsis (A) and rice (B) is plotted. The y-axes show relative gene expression values in base-2 logarithm against the average expression levels of each gene. doi:10.1371/journal.pone.0015481.g001 p g doi:10.1371/journal.pone.0015481.g001 using rice Affymetrix DNA array GSE6893 data [33]. Interest- ingly, the expression patterns could be clearly defined into two distinct groups (Figure 2B). One type of expression profile was strongly dependent on tissue development stage and had high expression levels in tissues associated with secondary wall deposition. The other type of expression profile had relatively constant expression levels. For simplicity, we designate these two patterns ‘mountain type’ and ‘flat type’ expression, based on the appearance in Figure 2B. Most of the genes showing ‘mountain type’ had low range of maximum MR for the 300th gene (Figure 2B, Table 1). The high expression level in tissues with secondary wall formation and the strong co-expression indicate that the ‘mountain type’ genes are the likely homologs implicated in xylan biosynthesis in secondary walls, and hence the functional orthologs of the three Arabidopsis IRX genes used for the analysis above. using rice Affymetrix DNA array GSE6893 data [33]. Interest- ingly, the expression patterns could be clearly defined into two distinct groups (Figure 2B). One type of expression profile was strongly dependent on tissue development stage and had high expression levels in tissues associated with secondary wall deposition. The other type of expression profile had relatively constant expression levels. For simplicity, we designate these two patterns ‘mountain type’ and ‘flat type’ expression, based on the appearance in Figure 2B. Most of the genes showing ‘mountain type’ had low range of maximum MR for the 300th gene (Figure 2B, Table 1). Co-expression analysis of GT43 and GT47D genes in rice These genes included well-known genes such as BC1 encoding COBL4 and cellulose synthase genes OsCesA4, OsCesA7 and OsCesA9, which are involved in secondary wall synthesis [34,35]. Figure 1 (C and D) illustrates the transcriptional co-regulation of the top 25 shared genes for both species. PLoS ONE \| www.plosone.org Development of a Pfam-based predictor for plant endomembrane localization The MRs for all genes are listed in Table S1 and Table S3. doi:10.1371/journal.pone.0015481.t001 The values signify the MR of each of the 300th co-expressed genes with GT43 and GT47D members. Thus, the other 299 co-expressed genes have lower MR values. Low values signify highly correlated expression patterns. The genes shown in bold were used as baits for the final comparative co-expression analysis. The MRs for all genes are listed in Table S1 and Table S3. doi:10.1371/journal.pone.0015481.t001 membrane is then calculated as the geometric mean, i.e. the cubic root of (0.824 * 0.703 * 0.878) = 0.80. locations, but shows a lower sensitivity rate compared with Pfam- based prediciton. This improvement in localization capability is likely due to the various predictors focusing on properties and datasets not well tailored to the endomembrane system. Based on the improvement in the endomembrane prediction, we named this prediction method PFANTOM (plant protein family information- based predictor for endomembrane) and used Table S6 as the PFANTOM tool. Development of a Pfam-based predictor for plant endomembrane localization The high expression level in tissues with secondary wall formation and the strong co-expression indicate that the ‘mountain type’ genes are the likely homologs implicated in xylan biosynthesis in secondary walls, and hence the functional orthologs of the three Arabidopsis IRX genes used for the analysis above. To select the best rice candidate genes for co-expression analysis we identified the IRX9, IRX10 and IRX14 homologs that fulfilled the following criteria: 1) ‘mountain-type’ expression profile (Figure 2B), 2) lowest maximum MR in the 300 most highly co- expressed genes and 3) maximum number of shared genes. We PLoS ONE \| www.plosone.org November 2010 \| Volume 5 \| Issue 11 \| e15481 3 Xylan Biosynthesis Network in Arabidopsis and Rice Table 1. Mutual Rank (MR) of the 300th co-expressed gene to each of the GT43 and GT47D family members. GT family AGI (name) MR of the 300th RAP (Defined name) MR of the 300th GT47D At1g27440 (IRX10) 356.9 Os01g0926400 (OsGT47D) 264.9 At2g28110 (FRA8) 192.8 Os01g0926600 264.9 At5g22940 (F8H) 407.3 Os01g0926700 - At5g61840 (IRX10-L) 456.1 Os03g0107900 449.5 Os02g0520750 - Os04g0398600 326.2 Os10g0180000 226.7 GT43 At1g27600 (IRX9-L) 580.8 Os04g0103100 - At2g37090 (IRX9) 295.6 Os01g0157700 - At4g36890 (IRX14) 380.2 Os01g0675500 244.8 At5g67230 (IRX14-L) 682.0 Os03g0287800 - Os06g0687900 364.7 Os04g0650300 (OsGT43A) 201.1 Os05g0123100 (OsGT43B) 221.3 Os05g0559600 400.2 Os07g0694400 274.4 Os10g0205300 285.6 The values signify the MR of each of the 300th co-expressed genes with GT43 and GT47D members. Thus, the other 299 co-expressed genes have lower MR values. Low values signify highly correlated expression patterns. The genes shown in bold were used as baits for the final comparative co-expression analysis. The MRs for all genes are listed in Table S1 and Table S3. doi:10.1371/journal.pone.0015481.t001 Table 1. Mutual Rank (MR) of the 300th co-expressed gene to each of the GT43 and GT47D family members. ch of the 300th co-expressed genes with GT43 and GT47D members. Thus, the other 299 co-expressed genes have lower MR values. Low expression patterns. The genes shown in bold were used as baits for the final comparative co-expression analysis. The MRs for all gene e S3. The values signify the MR of each of the 300th co-expressed genes with GT43 and GT47D members. Thus, the other 299 co-expressed genes have lower MR values. Low values signify highly correlated expression patterns. The genes shown in bold were used as baits for the final comparative co-expression analysis. Characterization of the Pfam-based prediction performance p Characterization of the Pfam-based predictor was carried out by calculating the sensitivity and specificity of the predictor upon a non-independent set of proteins. Since the algorithm is highly dependent upon the number and uniqueness of Pfam domains to determine localization, the training set was used as a benchmark set to understand how well the algorithm would work in a best case scenario, since all the Pfam domains have already been seen in the training set. Sensitivity across subcellular localizations ranges from 65% (Vacuole) to 85% (Nucleus), while specificity drops down to only 76% (Table S7). To illustrate the threshold-dependency of the algorithm performance, a receiver operating characteristic (ROC) plot was used (Figure S2). Across the different thresholds for including Pfam data in the final calculation, the localization algorithm performs much better than random localization as shown as dotted line on the ROC plots. This algorithm fills a specific need for the predicted localization of endomembrane system proteins. The performance of this algorithm can be readily compared to the best performing predictors as outlined in SUBA, the Arabidopsis subcellular database (Table S7) [38]. SUBA contains pre-calculated localization scores for all Arabidopsis proteins, and the sensitivity and specificity was calculated for the members of the training set for each of the different predictors. SUBA also makes an ‘all predictors’ call using a winner-takes-all localization call to combine the results from multiple individual predictors. A comparison of this integrated ‘all predictors’ call with the Pfam-based predictor indicates that the most significant improvements in sensitivity over current predictors are found for the Golgi, plasma membrane, and vacuole (Figure 3A). In fact only the WoLFPSORT algorithm provides any prediction for these PLoS ONE \| www.plosone.org erimental validation of localization predictions To validate the predictions given by PFANTOM, we selected 16 Arabidopsis proteins predicted by PFANTOM to be located in nucleus, Golgi apparatus, and plasma membrane (Table 2). Thirteen of the Arabidopsis proteins have not previously been examined for their intracellular distribution, whereas the remain- ing three proteins (ROP7, RIC2, RIC4) have been reported in fluorescent fusion protein experiments to be located in plasma membrane [39,40] and were included as positive controls. For all 16 proteins the intracellular localization was determined by transiently expressing YFP-fusion proteins in Nicotiana benthamiana (Table 2, Figure 3). For comparison we also predicted localization with publically available web-based algorithms (Table 2). As predicted by PFANTOM, two transcription factors, bHLH protein (AT5G48560; Figure 3B) and SND1 (AT1G32770; Figure 3C) showed the YFP signal in nucleus. AT3G18660 (PGSIP1; Figure 3G), AT4G33330 (PGSIP3; Figure 3H), AT5G01360 (DUF231 protein; Figure 3I), and AT2G38320 (DUF231 protein; Figure 3J) showed the YFP signal in small, moving, and oval dots very similar to what was seen with the Golgi marker (ST-tmd-GFP; Figure 3D), and clearly different from the ER marker (GFP-HDEL; Figure 3F), which showed typical network pattern. We selected PGSIP1 and PGSIP3, which were both predicted by PFANTOM to be Golgi localized, although PGSIP1 has been reported to be a chloroplast protein [41] and the PLoS ONE \| www.plosone.org November 2010 \| Volume 5 \| Issue 11 \| e15481 4 Xylan Biosynthesis Network in Arabidopsis and Rice Figure 2. Rice GT43A, GT43B and GT47D genes. (A) Phylogenic relationship of rice and Arabidopsis genes. Numbers at branches indicate bootstrap values from 500 trials. Phylogenetic tree was built by neighboring-joining method using ClustalW. The IRX10 and FRA8 genes in GT47D, the IRX9 and IRX14 genes in GT43, and their rice orthologues are clearly separated into distinct clades. (B) The ‘Mountain type’ expression pattern of the genes showing MR,280 for the 300 most highly co-expressed genes (upper panel). The ‘Flat type’ expression pattern of the genes which showed Figure 2. Rice GT43A, GT43B and GT47D genes. (A) Phylogenic relationship of rice and Arabidopsis genes. Numbers at branches indicate bootstrap values from 500 trials. Phylogenetic tree was built by neighboring-joining method using ClustalW. The IRX10 and FRA8 genes in GT47D, the IRX9 and IRX14 genes in GT43, and their rice orthologues are clearly separated into distinct clades. erimental validation of localization predictions (B) The ‘Mountain type’ expression pattern of the genes showing MR,280 for the 300 most highly co-expressed genes (upper panel). The ‘Flat type’ expression pattern of the genes which showed MR.280 or had no co-expression data available in ATTED-II (lower panel). Os01g0926700 is included in the mountain type because it has a similar profile, although no co-expression data are available for this gene in ATTED-II. The red and blue arrowheads show the genes used as baits for the final comparative co-expression analysis. The y-axes show raw expression values from rice Affymetrix DNA array GSE6893 data [33]. The x-axes show tissue type: R; Root_7d_seedling, ML; Mature_leaf, YL; Young_leaf, P1; Young_inflorescence_P1, P2-P6, Inflorescence stage P2 to P6; S1-S5, Seed stage S1 to S5. doi:10.1371/journal.pone.0015481.g002 Figure 2. Rice GT43A, GT43B and GT47D genes. (A) Phylogenic relationship of rice and Arabidopsis genes. Numbers at branches indicate bootstrap values from 500 trials. Phylogenetic tree was built by neighboring-joining method using ClustalW. The IRX10 and FRA8 genes in GT47D, the IRX9 and IRX14 genes in GT43, and their rice orthologues are clearly separated into distinct clades. (B) The ‘Mountain type’ expression pattern of the genes showing MR,280 for the 300 most highly co-expressed genes (upper panel). The ‘Flat type’ expression pattern of the genes which showed MR.280 or had no co-expression data available in ATTED-II (lower panel). Os01g0926700 is included in the mountain type because it has a similar profile, although no co-expression data are available for this gene in ATTED-II. The red and blue arrowheads show the genes used as baits for the final comparative co-expression analysis. The y-axes show raw expression values from rice Affymetrix DNA array GSE6893 data [33]. The x-axes show tissue type: R; Root_7d_seedling, ML; Mature_leaf, YL; Young_leaf, P1; Young_inflorescence_P1, P2-P6, Inflorescence stage P2 to P6; S1-S5, Seed stage S1 to S5. doi:10.1371/journal.pone.0015481.g002 doi:10.1371/journal.pone.0015481.g002 PLoS ONE \| www.plosone.org November 2010 \| Volume 5 \| Issue 11 \| e15481 5 Xylan Biosynthesis Network in Arabidopsis and Rice Figure 3. Validation of the PFANTOM method for plant subcellular localizations. (A) Comparison of the prediction performance of Pfam based prediction (PFANTOM), all predictors by SUBA database, and WoLF PSORT. NC, nucleus; MT, mitochondrion; V, vacuole; PX, peroxisome; E Figure 3. Validation of the PFANTOM method for plant subcellular localizations. (A) Comparison of the prediction perform based prediction (PFANTOM), all predictors by SUBA database, and WoLF PSORT. PLoS ONE \| www.plosone.org Intracellular network for xylan formation in Arabidopsis and rice In an attempt to integrate the localization predictions based on Pfam and the co-expression information from Arabidopsis and rice, we examined the Pfam domain information from the intersection sets identified in the previous sections, comprising 124 and 83 genes for Arabidopsis and rice, respectively (Figure S3A, Table S2 and S4). Table S8 outlines the Pfam annotations in the high-ranking co-expression sets showing an average MR of less than 70. Interestingly, many of the Pfam domains were identified in both co-expression sets, suggesting functional components present in both Arabidopsis and rice. Furthermore, several sets of Pfam annotations were unique to either Arabidopsis or rice and likely relate to distinct features of their cell walls. Lastly, to obtain an insight into the putative functional interaction at the subcellular level, the localization of high-ranking co-expressed components in Table S8 were predicted by the PFANTOM (Figure S3B, Table S6). Table 3 outlines the putative intracellular distribution of each Pfam domain from the co-expression set. The majority of proteins from this collection of tightly co-expressed genes could be assigned to three distinct intracellular compartments, the Golgi apparatus Table 2. Validation of the Pfam prediction of Arabidopsis subcellular localization. PFANTOM Other Prediction Programs Experimental Location Common Name AGI Pfam ID Score (%) Location iPSORT MitoProtII Predotar TargetP WoLF PSORT PM LRR protein AT1G67510 PF00560, PF00069, PF08263 80 PM ER EX PM PM NCRK AT2G28250 PF00069 70 PM ER EX unclear PM ROP7 AT5G45970 PF00071 78 PM MT CT PM RIC2 AT1G27380 PF00786 100 PM MT NC PM RIC4 AT5G16490 PF00786 100 PM MT ER EX PM PM ROPGEF4 AT2G45890 PF03759 100 PM PL NC PM COBL4 (IRX6) AT5G15630 PF04833 100 PM MT ER EX V PM FLA11 AT5G03170 PF02469 82 PM MT ER EX CT PM FLA12 AT5G60490 PF02469 82 PM MT ER EX PM PM CTL2 AT3G16920 PF00182 67 PM ER EX EX NC bHLH AT5G48560 PF00010 81 NC NC NC SND1 AT1G32770 PF02365 94 NC MT NC GO PGSIP1 AT3G18660 PF01501 80 GO CT MT CT unclear GO PGSIP3 AT4G33330 PF01501 80 GO CT MT EX CT GO (PM) DUF231 (TBL3) AT5G01360 PF03005 50/50 V/PL EX CT GO DUF231unknown AT2G38320 PF03005 50/50 V/PL MT MT EX unclear Sixteen co-expressed genes were selected and intracellular localization was predicted by iPSORT [77], MitoProt II [78], Predotar [79], TargetP.1 [80], and WoLFPSORT [81]. erimental validation of localization predictions Furthermore, we also DUF231 proteins, which were ambiguously pred vacuolar, but which belong to a large family of pro of which are known to play a role in cell wall structu the Plasma membrane and/or extracellular localiz YFP fusion proteins of GH19 family and GPI ancho such as AT3G16920 (CTL2; Figure 3K), AT5G03 Figure 3L), AT5G60490 (FLA12; Figure 3M), (COBL4, IRX6; Figure 3N), AT1G67510 (LR Figure 3O), AT2G28250 (NCRK; Figure 3P), (RIC2; Figure 3Q), AT5G16490 (RIC4; F AT5G45970 (ROP7; Figure 3S) showed the YFP single layer surrounding the cytoplasm identical t observed with a plasma membrane marker (pm-rk; Interestingly, CTL2 and FLA11 show oval dots in ad plasma membrane signal. These dots were larger th and we are uncertain what they represent. The Y AT2G45890 (RopGEF4; Figure 3T) belonging t family, which is recruited to lipid rafts for sm activation [43,44] was not uniformly distributed b as large dots associated within the plasma membran an interaction with endogenous membrane proteins show that for 11 of the 13 proteins (16 proteins e three positive controls), i.e. all the nuclear and plasm Table 2. Validation of the Pfam prediction of Ar Experimental Location Common Name AGI Pfam PM LRR protein AT1G67510 PF0056 PF0006 PF0826 PM NCRK AT2G28250 PF0006 PM ROP7 AT5G45970 PF0007 PM RIC2 AT1G27380 PF0078 PM RIC4 AT5G16490 PF0078 PM ROPGEF4 AT2G45890 PF0375 PM COBL4 (IRX6) AT5G15630 PF0483 PM FLA11 AT5G03170 PF0246 PM FLA12 AT5G60490 PF0246 PM CTL2 AT3G16920 PF0018 NC bHLH AT5G48560 PF0001 NC SND1 AT1G32770 PF0236 GO PGSIP1 AT3G18660 PF0150 GO PGSIP3 AT4G33330 PF0150 GO (PM) DUF231 (TBL3) AT5G01360 PF0300 GO DUF231unknown AT2G38320 PF0300 Sixteen co-expressed genes were selected and intracellular local Pfam was the prediction method described in this study, and ex endoplasmic reticulum; GO, Golgi apparatus; CT, cytosol; PM, p doi:10.1371/journal.pone.0015481.t002 proteins and for the PGSIP proteins, there was agreement between the predicted location by PFANTOM and the observed localization (Table 2). Arabidopsis protein does have an N-terminal sequence that appears to fulfill the characteristics of a transit peptide according to the TargetP predictor. Furthermore, we also selected two DUF231 proteins, which were ambiguously predicted to be vacuolar, but which belong to a large family of proteins, several of which are known to play a role in cell wall structure [42]. erimental validation of localization predictions For the Plasma membrane and/or extracellular localized proteins, YFP fusion proteins of GH19 family and GPI anchored proteins such as AT3G16920 (CTL2; Figure 3K), AT5G03170 (FLA11; Figure 3L), AT5G60490 (FLA12; Figure 3M), AT5G15630 (COBL4, IRX6; Figure 3N), AT1G67510 (LRR protein; Figure 3O), AT2G28250 (NCRK; Figure 3P), AT1G27380 (RIC2; Figure 3Q), AT5G16490 (RIC4; Figure 3R), AT5G45970 (ROP7; Figure 3S) showed the YFP signals as a single layer surrounding the cytoplasm identical to what was observed with a plasma membrane marker (pm-rk; Figure 3E). Interestingly, CTL2 and FLA11 show oval dots in addition to the plasma membrane signal. These dots were larger than for Golgi and we are uncertain what they represent. The YFP signal of AT2G45890 (RopGEF4; Figure 3T) belonging to the GEF family, which is recruited to lipid rafts for small GTPase activation [43,44] was not uniformly distributed but observed as large dots associated within the plasma membrane, suggesting an interaction with endogenous membrane proteins. The results show that for 11 of the 13 proteins (16 proteins excluding the three positive controls), i.e. all the nuclear and plasma membrane protein (AT2G38320, belonging to DUF231); (K) CTL2 (AT3G16920); (L) FLA11 (AT5G03170); (M) FLA12 (AT5G60490); (N) COBL4 (AT5G15630, IRX6); (O) LRR protein (AT1G67510); (P) NCRK (AT2G28250); (Q) RIC2 (AT1G27380); (R) RIC4 (AT5G16490); (S) ROP7 (AT5G45970); (T) ROPGEF4 (AT2G45890). A summary of the localization experiments is shown in Table 2. Scale bar= 20 mm. doi:10.1371/journal.pone.0015481.g003 erimental validation of localization predictions NC, nucleus; MT, mitochondrion; V, vacuole; PX, p endoplasmic reticulum; GO, Golgi apparatus; CT, cytosol; PM, plasma membrane; PL, plastid; EX, extracellular. (B-T) Subcellular localizatio expressed YFP-fusion proteins in N. benthamiana. (B) bHLH (AT5G48560); (C) SND1 (AT1G32770); (D) Golgi marker (STtmd-GFP); (E) ER HDEL); (F) Plasma membrane marker (pm-rk); (G) PGSIP1 (AT3G18660); (H) PGSIP3 (AT4G33330); (I) TBL3 (AT5G01360, belonging to DUF231 Figure 3. Validation of the PFANTOM method for plant subcellular localizations. (A) Comparison of the prediction perform based prediction (PFANTOM), all predictors by SUBA database, and WoLF PSORT. NC, nucleus; MT, mitochondrion; V, vacuole; PX, pe endoplasmic reticulum; GO, Golgi apparatus; CT, cytosol; PM, plasma membrane; PL, plastid; EX, extracellular. (B-T) Subcellular localization expressed YFP-fusion proteins in N. benthamiana. (B) bHLH (AT5G48560); (C) SND1 (AT1G32770); (D) Golgi marker (STtmd-GFP); (E) ER HDEL); (F) Plasma membrane marker (pm-rk); (G) PGSIP1 (AT3G18660); (H) PGSIP3 (AT4G33330); (I) TBL3 (AT5G01360, belonging to DUF231 Figure 3. Validation of the PFANTOM method for plant subcellular localizations. (A) Comparison of the prediction performance of Pfam- based prediction (PFANTOM), all predictors by SUBA database, and WoLF PSORT. NC, nucleus; MT, mitochondrion; V, vacuole; PX, peroxisome; ER, endoplasmic reticulum; GO, Golgi apparatus; CT, cytosol; PM, plasma membrane; PL, plastid; EX, extracellular. (B-T) Subcellular localization of transiently expressed YFP-fusion proteins in N. benthamiana. (B) bHLH (AT5G48560); (C) SND1 (AT1G32770); (D) Golgi marker (STtmd-GFP); (E) ER marker (GFP- HDEL); (F) Plasma membrane marker (pm-rk); (G) PGSIP1 (AT3G18660); (H) PGSIP3 (AT4G33330); (I) TBL3 (AT5G01360, belonging to DUF231); (J) Unknown November 2010 \| Volume 5 \| Issue 11 \| e15481 PLoS ONE \| www.plosone.org 6 Xylan Biosynthesis Network in Arabidopsis and Rice protein (AT2G38320, belonging to DUF231); (K) CTL2 (AT3G16920); (L) FLA11 (AT5G03170); (M) FLA12 (AT5G60490); (N) COBL4 (AT5G15630, IRX6); (O) LRR protein (AT1G67510); (P) NCRK (AT2G28250); (Q) RIC2 (AT1G27380); (R) RIC4 (AT5G16490); (S) ROP7 (AT5G45970); (T) ROPGEF4 (AT2G45890). A summary of the localization experiments is shown in Table 2. Scale bar= 20 mm. doi:10.1371/journal.pone.0015481.g003 Arabidopsis protein does have an N-terminal se appears to fulfill the characteristics of a transit pepti to the TargetP predictor. Intracellular network for xylan formation in Arabidopsis and rice Pfam was the prediction method described in this study, and experimental data are shown in Figure 5. NC, nucleus; MT, mitochondrion; V, vacuole; PX, peroxisome; ER, endoplasmic reticulum; GO, Golgi apparatus; CT, cytosol; PM, plasma membrane; PL, plastid; EX, extracellular. doi:10.1371/journal.pone.0015481.t002 Table 2. Validation of the Pfam prediction of Arabidopsis subcellular localization. November 2010 \| Volume 5 \| Issue 11 \| e15481 PLoS ONE \| www.plosone.org 7 7 Xylan Biosynthesis Network in Arabidopsis and Rice Table 3. Analysis of the high-ranking co-expression sets from Arabidopsis and rice using the Pfam-based predictor. ble 3. PLoS ONE \| www.plosone.org Intracellular network for xylan formation in Arabidopsis and rice Analysis of the high-ranking co-expression sets from Arabidopsis and rice using the Pfam-based y g g p p g p Pfam ID Predicted Score (%) Number of genes Common protein name Name used in working model Related function Os At Golgi apparatus PF03016 GO 100 4 6 GT47 GT47 Glycosyltransferase PF03360 GO 100 4 2 GT43 GT43 Glycosyltransferase PF01501 GO 80 3 4 GT8, GAUT/GATL IRX8; PARVUS Glycosyltransferase PF01501 GO 80 0 2 GT8, PGSIP PGSIP Glycosyltransferase PF01762 GO 67 3 1 GT31 GT31 Glycosyltransferase PF04577 no data 0 a) 3 0 GT61 GT61 Glycosyltransferase PF03214 GO 100 2 0 GT75 RGP UDP-arabinose mutase PF03141 GO 92 2 1 Putative methyltransferase DUF248 methylation PF04669 no data 0 a) 1 5 DUF579 DUF579 unknown PF03138 GO 55 2 2 DUF246 DUF246 unknown PF03005 V/PL 50/50 a) 8 5 DUF231 DUF231 acetylation PF07779 no data 0 a) 4 1 O-acetyltransferase-related Cas1P acetylation PF03151 GO 38 a) 6 9 Triose-phosphate transporter TPT NDP-sugar transport PF01370 GO 13 a) 4 5 epimerase UXS epimerase Plasma Membrane (extracellular) PF00071 PM 78 5 7 ROP, RAB GTPase Rac signaling/vesicle PF00025 PM 57 2 1 ADP-ribosylation GTPase unknown/vesicle PF00612 PM 71 3 7 IQ protein IQ signaling PF00069 PM 70 11 10 protein kinase signaling PF00560, PF08263, PF00069 PM 80 3 5 LRR family LRR signaling PF00786 PM 100 1 2 ROP interactive CRIB RIC signaling PF02469 PM 82 9 4 Fasciclin-like AGP FAS signaling PF00182 EX 67 1 2 GH19; chitinase-like CTL Glycosyl hydrolase PF00295 PM 50 3 5 GH28; Polygalacturonase Glycosyl hydrolase PF07983 PM 100 2 3 GH17; b-(1;3)-glucanase Glycosyl hydrolase PF00759 PM 80 3 3 GH9; cellulase Glycosyl hydrolase PF07731, PF00394, PF07732 PM 57 4 6 putative laccase lignin formation PF03552 PM 60 8 3 cellulose synthase Glycosyltransferase PF04833 PM 100 2 1 COBRA GPI anchored PF02298 PM 78 6 4 plastocyanin-like GPI anchored PF00097 PM 51 12 8 zinc finger unknown PF06749 PM 100 2 2 DUF1218 unknown PF00190 EX 75 2 3 Germin-like unknown PF07058 PM 67 0 1 Myosin HC-like unknown PF00786, PF00620 PM 71 0 1 RhoGAP signaling PF07320 PM 78 10 0 Hairpin-induced unknown Nucleus PF00249 NC 96 6 10 MYB MYB Transcription PF00010 NC 81 1 7 bHLH (MYC) MYC Transcription PF02365 NC 94 4 9 NAC NAC Transcription PF00642 NC 83 2 3 dTIS unknown PF04640 no data 0 1 0 DUF597 unknown PF00514 NC 54 3 1 armadillo/beta-catenin repeat unknown PF00719 NC 50 2 0 PRLI-interacting factor unknown PF04852 NC 100 0 4 LSH unknown PLoS ONE \| www.plosone.org PLoS November 2010 \| Volume 5 \| Issue 11 \| e15481 PLoS ONE \| www.plosone.org 8 Xylan Biosynthesis Network in Arabidopsis and Rice Pfam ID Predicted Score (%) Number of genes Common protein name Name used in working model Related function Os At Other components PF00141 V 38 11 3 peroxidase lignin formation PF00026 EX/PM 33/33 1 1 aspartyl protease unknown PF01419 NC/PL 40/40 1 1 jacalin lectin unknown PF03999 Cytoskeleton 100 3 1 MAP65-8 unknown PF04784 PL 67 1 1 DUF547 unknown PF00240, PF02179 - 7 1 Ubiquitin domain unknown PF00657 V 64 2 1 Lipase, GDSL domain lignin formation PF06814 V 50 3 1 Transmembrane receptor signaling Groups of protein with the same Pfam domains were found in co-expression dataset. Intracellular network for xylan formation in Arabidopsis and rice The table shows groups, where at least one gene exhibited an average MR of less than 70. The number of genes corresponding to each Pfam is from the entire co-expression set in Arabidopsis (Table S1) and rice (Table S3). The AGI and RAP codes are listed in Table S8. The predicted location and score are shown according to the Pfam-based predictor. Abbreviations for subcellular compartments are the same as in Table 2. a)PF01370, PF03005, PF04669, PF03151, PF4577, and PF07779 proteins have been observed in Golgi apparatus (unpublished data) [82]. Based on the references they were categorized into Golgi apparatus, although they showed a low score by PFANTOM. doi:10.1371/journal.pone.0015481.t003 Table 3. cont. Groups of protein with the same Pfam domains were found in co-expression dataset. The table shows groups, where at least one gene exhibited an average MR of less than 70. The number of genes corresponding to each Pfam is from the entire co-expression set in Arabidopsis (Table S1) and rice (Table S3). The AGI and RAP codes are listed in Table S8. The predicted location and score are shown according to the Pfam-based predictor. Abbreviations for subcellular compartments are the same as in Table 2. a)PF01370, PF03005, PF04669, PF03151, PF4577, and PF07779 proteins have been observed in Golgi apparatus (unpublished data) [82]. Based on the references they were categorized into Golgi apparatus, although they showed a low score by PFANTOM. doi:10.1371/journal.pone.0015481.t003 (13 Pfam domains), the plasma membrane (21 Pfam domains), and the nucleus (8 Pfam domains). This information assisted modeling of potential interactions in the context of shared subcellular localization. An intracellular working model was constructed outlining common and unique machinery in both Arabidopsis and rice (Figure S3C, Figure 4). stems revealed a highly significant 66% reduction in the content of GlcA compared to the wild type (Figure 5c). None of the other monosaccharides showed a difference. We furthermore tested the xylan GlcA transferase activity in microsomes isolated from stems, using an assay with exogenous xylohexaose as acceptor. The results showed that the GlcA transferase activity in pgsip1 was only about 50% of the wild type level (Figure 5D). This data strongly supports the hypothesis that PGSIP1 (and likely PGSIP3 as well) is a xylan a-glucuronosyltransferase. Obviously, it will be necessary to substantiate this hypothesis by analysis of an independent allele or complementation of the mutant. PGSIP proteins are putative glucuronyltransferases involved in glucuronoxylan synthesis The co-expression analysis identified a number of GTs located in the Golgi apparatus (Table 3, Figure 4). Most of these were the IRX genes already known to be involved in xylan biosynthesis. The four additional GT groups identified were RGP (GT75) and GT61 in rice, PGSIP1 and PGSIP3 in Arabidopsis, and GT31 proteins in both species but most highly co-expressed in rice. The rice RGPs (UAM1 and UAM3) have been shown to be UDP- arabinose mutases [45]. Since arabinose is abundant in rice xylan but has not been detected in Arabidopsis xylan, the data suggest that GT61 could be xylan arabinosyltransferases, in agreement with earlier speculations [1,46]. The GT31 proteins do not have an obvious suggested function, but they may be involved in synthesis of arabinogalactan proteins, e.g. the FLA11 and FLA12 proteins that are also seen in the co-expressed data sets. PGSIP1 and PGSIP3 proteins belong to GT8 family which in contrast to GT61 and GT31 contains retaining enzymes [47]. PGSIP proteins are only distantly related to the PARVUS and IRX8 proteins. Therefore, the most obvious function of PGSIP1 and PGSIP3 would be as xylan a-glucuronosyltransferases, given that a major difference between rice and Arabidopsis secondary walls is the 10- fold higher GlcA/Xyl ratio in Arabidopsis (data not shown). To test this hypothesis, we analyzed an Arabidopsis mutant in the PGSIP1 gene, which is more highly expressed in stems than PGSIP3. The pgsip1 mutant has a T-DNA insertion in the coding region and plants carrying the homozygous insertion were selected by PCR. No functional transcript could be detected in plants homozygous for the insertion (Figure 5). Although no morpho- logical or irx phenotype was observed for the pgsip1 mutant line [11], the monosaccharide composition of cell walls from pgsip1 Discussion Proteins with a predicted location score more than 0.65 with common functional domain group in mammals and plants were selected as the components on plasma membrane. For the nucleus, proteins with functional domain typically annotated as transcription factor and with a predicted location score greater than 0 8 are shown in this model Figure 4. Working model for xylan synthesis and regulatory components for secondary wall development. (A) Xylan synthesis candidates in Golgi apparatus. The components in this model correspond to Table S3. In general only the larger gene family name is shown, except for GT8 and epimerases, which have many members of different molecular function and include PGSIP, IRX8, PARVUS, and UXS. Three gray color components, GT61, RGP and PGSIP proteins, were identified as co-expressed genes unique to Arabidopsis or rice. (B) Signaling and regulatory components on the plasma membrane to nucleus in mammals, Arabidopsis and rice. The components in this model correspond to genes outlined in Table 3. Proteins with a predicted location score more than 0.65 with common functional domain group in mammals and plants were selected as the components on plasma membrane. For the nucleus, proteins with functional domain typically annotated as transcription factor and with a predicted location score greater than 0.8 are shown in this model. doi:10.1371/journal.pone.0015481.g004 Figure 4. Working model for xylan synthesis and regulatory components for secondary wall development. (A) Xylan synthesis candidates in Golgi apparatus. The components in this model correspond to Table S3. In general only the larger gene family name is shown, except for GT8 and epimerases, which have many members of different molecular function and include PGSIP, IRX8, PARVUS, and UXS. Three gray color components, GT61, RGP and PGSIP proteins, were identified as co-expressed genes unique to Arabidopsis or rice. (B) Signaling and regulatory components on the plasma membrane to nucleus in mammals, Arabidopsis and rice. The components in this model correspond to genes outlined in Table 3. Proteins with a predicted location score more than 0.65 with common functional domain group in mammals and plants were selected as the components on plasma membrane. For the nucleus, proteins with functional domain typically annotated as transcription factor and with a predicted location score greater than 0.8 are shown in this model. doi:10.1371/journal.pone.0015481.g004 ylase (UXS), UDP-glucose 6-dehydrogenase and several glycoside hydrolases (GH) (Table S9). Discussion Two in silico approaches; co-expression and localization To obtain a better insight into the biosynthesis and regulation of xylan across species we extended the strategy of co-expression analysis to both Arabidopsis and rice. Our comparative co- expression analysis used three bait genes in each species and identified both known and novel candidate genes involved in signal transduction, regulation and substrate transport, as well as enzymes directly involved in secondary wall biosynthesis. Previous studies have identified co-expressed genes related to secondary wall formation in Arabidopsis using different transcriptional profiling methods [11,12,48,49]. Persson et al. [12] used regression analysis, Brown et al. [11] analyzed the slope profile using five selected tissue types, Ko et al. [48] identified genes highly expressed in stem, and Mutwil et al. [49] used mutual rank-based correlation matrices (cut-off of 30) for a co-expression network with the secondary wall CESA genes. Almost all the components identified in the previous studies are also included in the most highly co-expressed genes in our study (i.e. the 124 Arabidopsis genes in Table S2 include 83% of the genes published in any of the four references. The entire Arabidopsis co-expression list in Table S1 includes 93% of the genes published in any of the four references). Hence, it is clear that it does not make much difference whether the analysis is done with CESA genes as in the previous studies or with xylan synthesis IRX genes as in our study. Nevertheless, our study led to the identification of many additional PLoS ONE \| www.plosone.org November 2010 \| Volume 5 \| Issue 11 \| e15481 9 Xylan Biosynthesis Network in Arabidopsis and Rice Figure 4. Working model for xylan synthesis and regulatory components for secondary wall development. (A) Xylan synthesis candidates in Golgi apparatus. The components in this model correspond to Table S3. In general only the larger gene family name is shown, except for GT8 and epimerases, which have many members of different molecular function and include PGSIP, IRX8, PARVUS, and UXS. Three gray color components, GT61, RGP and PGSIP proteins, were identified as co-expressed genes unique to Arabidopsis or rice. (B) Signaling and regulatory components on the plasma membrane to nucleus in mammals, Arabidopsis and rice. The components in this model correspond to genes outlined in Table 3. November 2010 \| Volume 5 \| Issue 11 \| e15481 Xylan Biosynthesis Network in Arabidopsis and Rice Xylan Biosynthesis Network in Arabidopsis and Rice Figure 5. Analysis of an Arabidopsis pgsip1 mutant. (A) A T-DNA insertion mutant carrying an insertion in the third exon of PGSIP1 (AT3G18660) was obtained. White arrows indicate primer-annealing sites used for genotyping of the plants, and black arrows indicate primer-annealing sites used for RT-PCR. (B) No functional transcript could be detected by RT-PCR in homozygous individuals. (C) Cell wall material isolated from the first to the third internode of inflorescence stems showed significantly lower contents of GlcA compared to wild type (Col-0) plants (five biological replicates, values are mean 6 SD, significantly different at p,1026 (t-test)). (D) Xylan glucuronosyltrans- ferase activity was determined in microsomes prepared from the second internode of inflorescence stems. Data shown are mean 6 SD with three biological replicates, each consisting of stem internodes from four plants. The activity is shown in dpm, where the maximal activity determined for Col-0 is about 400 dpm corresponding to 0.15 nmol GlcA incorporated per mg of protein. doi:10.1371/journal.pone.0015481.g005 Obviously, localization for proteins that do not contain Pfam motifs with localization information cannot be predicted at present. In the lists of co-expressed genes (Table S1 and S3), 85% of Arabidopsis and 84% of rice genes have Pfam information, and 79% of Arabidopsis genes and 76% of rice genes with Pfam domain information can be predicted with PFANTOM. Out of these components, 66% and 75% in Arabidopsis and rice, respectively, are predicted by PFANTOM to be located in the endomembrane compartments. Coverage is still limited for the Golgi apparatus in the Pfam-based predictor, although it exhibits an improved performance over the other available predictors for the Golgi apparatus (Figure 3A). For example, 1) DUF231 proteins, which are localized in Golgi apparatus as shown in Figure 4, were predicted to be localized to the vacuole and the plastid, 2) GT61, DUF579, and Cas1p proteins could not be predicted by the Pfam-based predictor because no published experimental data is available for the associated Pfams (Table 3). More comprehensive studies of the Golgi proteome and/or combining with other prediction algorithm using hydrophobicity [50] would improve the usability of our prediction method. Our new method for the prediction of subcellular locations of plant proteins is robust enough for genome-wide predictions since it does not rely on the presence of signal or target peptides. Xylan Biosynthesis Network in Arabidopsis and Rice Therefore, we were able to predict localization for the Arabidopsis and rice co-expression sets to gain further insights into putative functional interactions (Table 3). Interestingly, many of our co- expressed components in secondary wall formation were predicted to be located in the endomembrane system, especially at the plasma membrane and Golgi apparatus, consistent with a role in signal transduction and cell wall formation. The available Golgi proteomic data is quite limited, and this also prevented us from testing the predictor with an independent test set. However, the experimental validation of 11 out of 13 tested proteins was very encouraging. Very recently, a method for subcellular prediction using machine learning and homology has been published and claimed to be efficient at predicting Golgi localization [51]. However, this method cannot predict ER or vacuole, but more importantly, their best classifier could only correctly predict the localization of CTL2, whereas the other 15 proteins we investigated experimentally either had no prediction at all or were incorrectly predicted. We are therefore convinced that our PFANTOM method, in spite of its limitations and simplicity, is better, at least for the analysis of the co-expressed data sets in this study, where proteins located in the endomembrane compart- ments are highly represented. Figure 5. Analysis of an Arabidopsis pgsip1 mutant. (A) A T-DNA insertion mutant carrying an insertion in the third exon of PGSIP1 (AT3G18660) was obtained. White arrows indicate primer-annealing sites used for genotyping of the plants, and black arrows indicate primer-annealing sites used for RT-PCR. (B) No functional transcript could be detected by RT-PCR in homozygous individuals. (C) Cell wall material isolated from the first to the third internode of inflorescence stems showed significantly lower contents of GlcA compared to wild type (Col-0) plants (five biological replicates, values are mean 6 SD, significantly different at p,1026 (t-test)). (D) Xylan glucuronosyltrans- ferase activity was determined in microsomes prepared from the second internode of inflorescence stems. Data shown are mean 6 SD with three biological replicates, each consisting of stem internodes from four plants. The activity is shown in dpm, where the maximal activity determined for Col-0 is about 400 dpm corresponding to 0.15 nmol GlcA incorporated per mg of protein. doi:10.1371/journal.pone.0015481.g005 Figure 5. Analysis of an Arabidopsis pgsip1 mutant. (A) A T-DNA insertion mutant carrying an insertion in the third exon of PGSIP1 (AT3G18660) was obtained. Discussion These novel genes were probably identified by this study because our analysis by the MR includes genes that were not identified in the previously published studies, including additional kinases, calmodulin binding proteins, MYC and MYB transcription factors, UDP-glucuronic acid decarbox- November 2010 \| Volume 5 \| Issue 11 \| e15481 PLoS ONE \| www.plosone.org PLoS ONE \| www.plosone.org 10 Xylan Biosynthesis Network in Arabidopsis and Rice White arrows indicate primer-annealing sites used for genotyping of the plants, and black arrows indicate primer-annealing sites used for RT-PCR. (B) No functional transcript could be detected by RT-PCR in homozygous individuals. (C) Cell wall material isolated from the first to the third internode of inflorescence stems showed significantly lower contents of GlcA compared to wild type (Col-0) plants (five biological replicates, values are mean 6 SD, significantly different at p,1026 (t-test)). (D) Xylan glucuronosyltrans- ferase activity was determined in microsomes prepared from the second internode of inflorescence stems. Data shown are mean 6 SD with three biological replicates, each consisting of stem internodes from four plants. The activity is shown in dpm, where the maximal activity determined for Col-0 is about 400 dpm corresponding to 0.15 nmol GlcA incorporated per mg of protein. doi:10.1371/journal.pone.0015481.g005 Xylan Biosynthesis Network in Arabidopsis and Rice belonging to the GATL clade, the same as PARVUS, and Os3g0211800/Os3g0413400 in the GAUT clade with IRX8. Since IRX8 and PARVUS are apparently involved in forming the reducing-end structure, our findings raise the possibility that rice also has the reducing-end structure. Other unknown proteins specific to plants such as DUF231, DUF246, and DUF579 were also included in co-expressed components in both Arabidopsis and rice. PGSIP1 (At3g18660) and PGSIP3 (At4g33330) and the DUF579 genes (At3g50220, At1g09610, At1g33800, At5g67120, At4g09990) showed stronger co-regulation in Arabidopsis, imply- ing more important roles in Arabidopsis than in rice. PGSIP1 has previously been reported to be located in the plastid and be involved in starch biosynthesis [41]. We have shown here that PGSIP1 and PGSIP3 are clearly located in the Golgi and that PGSIP1 appears to be xylan glucuronosyltransferase (Figure 3G- H, Figure 5). The proteins are unlikely to have a direct function in starch biosynthesis, and in fact we could not observe any difference in starch content in the pgsip1 mutant by iodine staining of leaves (data not shown). PGSIP1 is a good example how the comparison of wall structure and co-expression patterns between rice and Arabidopsis enabled us to predict a function for PGSIP1, which was in turn experimentally confirmed. Since the genes identified in this study are related to secondary wall formation and not specifically to xylan biosynthesis, we could not a priori assume that e.g. PGSIP1 would have a role in xylan biosynthesis. However, the differences between the two species were most consistent with a role in xylan biosynthesis. We therefore believe that the comparative analysis in the present study is a very powerful tool to form hypotheses that can be tested and yield much more information than analysis of only Arabidopsis as in previous studies. proteins are good candidates for xylan arabinosyltransferases [1,46]. Mitchell et al. [19] identified members of the BAHD acyltransferases as candidate feruloyl transferases, and recent work supports their involvement in rice xylan feruloylation [53]. The BAHD proteins are cytoplasmic and this is consistent with a role in feruloylation of a cytoplasmic intermediate and not a direct feruloylation of xylan [1,54]. We do not find BAHD family genes co-expressed with all three baits in rice, but Os06g0595800 and Os02g0483500 are BAHD family genes that are co-expressed with two of the baits, OsGT43B and OsGT47D (Table S3). Highlights on signaling and regulatory components g g g g g y p The co-expression analysis revealed a large number of highly co-expressed plasma membrane associated proteins and transcrip- tion factors (Table 3). In general, orthologous or very similar proteins were found in both the rice and Arabidopsis co-expressed gene sets. Many of the identified proteins belong to protein families well known to participate in signal transduction. Notably, co-expression of entire gene sets related to GTPase signal cascade were conserved in both Arabidopsis and rice. GTPase signal cascade such as LRR receptor kinase, Rop/Rac GTPase, RICs, RopGEFs, and IQ domain protein resemble components known from other signal transduction pathways in mammals where such pathways are understood in more detail than in plants. As an example we have shown components of the mammalian TLR2 signaling cascade in Figure 4B. TLR2 contains an extracellular LRR domain that is critical for transmitting the peptideglycan, lipopeptide, and chitin signal across the cell membrane to initiate innate immunity response against pathogens [55,56]. An adaptor molecule, MyD 88, associated with the toll/interleukin-1 receptor (TIR) intracellular domain of TLR2, recruits PI3K kinase, which is also regulated by Rho/Rac GTPase via RhoGEF protein [39,55,57]. The TLR complex, consisting of MyD88, PI3K kinase, and Rho/Rac GTPase activates a MAP kinase cascade, which leads to the activation of transcription factors including NF-kB [55]. Recent studies show that TLR2 and another receptor, TLR4, could also percept endogenous ligands and lead to not only immune response but also tissue remodeling especially for neurogenesis [58]. Similar to TLRs, the extracellular LRR domain of plant LRR kinase may recognize small molecules such as peptides and saccharides, while the intracellular kinase domain of the LRR protein transduces the signal to kinase cascades when activated by Rop/Rac GTPase proteins. The Rop/Rac signaling activity is regulated by RICs and RopGEFs [59,60]. AtRAC2/ ROP7 is specifically expressed during late stages of xylem differentiation in Arabidopsis [39]. This signaling by AtRAC2/ ROP7 might be mediated by co-expressed IQ domain proteins in our list based on the report that human IQGAP protein interacts with Rho/Rac GTPase [61]. This result suggests that they are key regulatory pathways during secondary wall development and can be crucial for the signaling perception. The localization of the Cas1p-like protein, RWA1 (At5G46340), could not be predicted based on AmiGO data, but the RWA2 protein (At3g06550) has been found in Golgi preparations (H. Parsons and J. Heazlewood, unpublished data). Xylan Biosynthesis Network in Arabidopsis and Rice Hence, these two BAHD members are good candidates for feruloyltransferases involved in xylan biosynthesis. Candidate genes for xylan formation in Golgi apparatus Candidate genes for xylan formation in Golgi apparatus A working model was developed for proteins identified by co- expression with predicted locations in the endomembrane system (Figure 4A). Genes encoding xylan biosynthesis components were expected to be co-expressed with UDP-GlcA decarboxylase (UXS). Importantly, we identified AtUXS3 and AtUXS6 in Arabidopsis and their two orthologs in rice as tightly co-regulated with xylan synthase genes (Table S2, S4, and S9). These proteins had been missed in previously published studies although it is evident that UDP-GlcA decarboxylase must play an important role in secondary wall biosynthesis. Hence, the fact that we find these proteins further supports the relevance of the candidates identified in this study. Further candidates involved in xylan synthesis in the Golgi apparatus, that we identified are nucleotide sugar transporters (TPT) and members of GT8 family (IRX8, PARVUS, PGSIP1, PGSIP3). Although the reducing end structure -Xyl-Xyl-Rha-GalA-Xyl has not been identified in grasses, our rice co-expression profiling identified Os3g0300900 co-expressed genes that show low PCC value [22,24]. Many of the newly identified genes showed conserved functional domains in both the rice and the Arabidopsis based sets (see below), lending support to the relevance of these genes and our method to identify them. As a further novel strategy in this study, we predicted the subcellular location of the co-expressed proteins. The currently available predictors either could not identify endomembrane proteins or had very low sensitivities for this subcellular location. Consequently we developed a Pfam-based algorithm, PFANTOM that improved sensitivity and specificity (Table S7, Figure S2, Figure 3). However, it should be noted that the ROC plots represent a characterization of a hypothetical best-case scenario based on the distribution of Pfam domains for which experimental data on subcellular location is already available. PLoS ONE \| www.plosone.org PLoS ONE \| www.plosone.org November 2010 \| Volume 5 \| Issue 11 \| e15481 PLoS ONE \| www.plosone.org 11 Xylan Biosynthesis Network in Arabidopsis and Rice Co-expression analysis and assessment of the gene function The binary vectors for expression of the N-terminal YFP fusion proteins under the control of 35S promoter were constructed via LR-reaction using the corresponding Entry clones. The full-length genes were cloned into the destination vectors pEarleyGate 101 [73]. Marker proteins for ER (GFP-HDEL), Golgi (STtmd-GFP), and plasma membrane (pm-rk) have been described previously [74,75]. The gene encoding p19 protein from tomato bushy stunt virus was used to suppress gene silencing. All vectors were used to transform Agrobacterium tumefaciens strain C58-1 pGV3850. Prior to leaf infiltration the bacteria were resuspended in AS-medium (10 mM MgCl2, 150 mM acetosyringone, 10 mM MES pH 5.7) to OD600 0.5. Agrobacterium strains containing the YFP constructs and the p19 silencing plasmid were mixed 1:1 and co-infiltrated into leaves of 3–4 week old N. benthamiana plants. Abaxial epidermis of infiltrated leaves was assayed for fluorescence by confocal laser- scanning microscopy 2–3 d post infiltration. Co-expression information was obtained from the ATTED-II database (http://atted.jp). Source of GeneChip data in ATTED-II version 5.5 are the 1388 array slides from the 58 experiments on each developing stage, biotic and abiotic treatment. Scatter plots of co-expression of two genes were made with CoexViewer available at the ATTED-II database [31]. ATTED-II provides the top 300 genes co-expressed with bait genes in both Arabidopsis and rice. We used three bait genes for each species, obtained the MR for each gene, and calculated the average MR as the geometric mean of the three individual MR. Transcript level information during developmental stage in Arabidopsis and rice were obtained from Arabidopsis Affymetrix DNA array data available from AtGeneExpress at TAIR (http://www.arabidopsis.org) and rice Affymetrix DNA array data GSE6893 available from Rice array database (http://www.ricearray.org) [20]. The Pfam database (ver. 24.0) has a collection of 7677 unique protein functional domains based on Hidden Markov Models (http://pfam.wustl.edu) [37]. Pfam domain information of whole genome in Arabidopsis was downloaded from TAIR and for rice from the rice genome annotation project (http://rice.plantbiology.msu.edu)[72]. Highlights on signaling and regulatory components RWA proteins are involved in polysaccharide acetylation (Y. Manabe and H.V. Scheller, unpublished) and have sequence similarity with the C- terminal multimembrane-spanning domain of Cas1p from fungi and animals. Interestingly, recent sequence analysis has shown that the N-terminal domain of Cas1p has similarities with esterases and with DUF231 proteins, while the C-terminal domain has similarity with acetyltransferases [52]. This suggests that DUF231 and RWA proteins in plants exist together in protein complexes, which are likely to catalyze glycan acetylation. We propose that the RWA proteins (4 in Arabidopsis and 3 in rice) are unspecific whereas the DUF231 proteins (47 in Arabidopsis and 59 in rice) confer the specificity for particular polysaccharides. The presence of RWA1 and DUF231 proteins in the co-expressed sets may suggest their involvement in xylan acetylation, but they may have other roles in secondary walls as suggested by analysis of DUF231 mutants that are deficient in cellulose [42]. The co-expressed genes also include GT31 proteins (Table 3, Figure 4), which may be involved in arabinogalactan biosynthesis. Interestingly, many of the rice genes in the families of galactosyltransferase (GT31), putative methyltransferase (DUF248) and Golgi unknown protein (DUF246) were identified as components showing strong co-regulation, compared with Arabidopsis, suggesting a more important role in rice cell walls and potentially related to the structural differences between rice and Arabidopsis walls. Furthermore, rice co-expression profiling showed that 18 genes comprising 5 Pfams are specific to rice, with no orthologs in the Arabidopsis co-regulation network. They include RGPs (UAM1 and UAM3), which have UDP-arabinose mutase activity [45], and several GT61 proteins. The GT61 An important question concerns the actual signals that trigger the pathway. Strongly co-expressed components such as fasciclin- like arabinogalactan proteins (FLA11, FLA12) and chitinase-like protein (CTL2) were located to plasma membrane (Table 2). In support of a role of these proteins in secondary wall development, high expression levels of CTL and FLA genes were also found in development of poplar tension wood and cotton fiber [62,63,64,65]. Two similar protein families, namely the fasciclin domain containing protein TGFBI (big-H3), and chitinase-like proteins, CHI3L1 and CHI3L2, are present in mammals, and recent Massively Parallel Signature Sequencing (MPSS) analysis November 2010 \| Volume 5 \| Issue 11 \| e15481 12 PLoS ONE \| www.plosone.org Xylan Biosynthesis Network in Arabidopsis and Rice show similar expression pattern in brain cancer cells [66]. Cloning and transient expression of proteins All clones used in this study were constructed using GatewayTM technology (Invitrogen). The Entry clones were obtained via BP- reaction in pDONR-Zeo (for Golgi proteins) or through TOPO- reaction using the pENTR/D-TOPO vector (for plasma mem- brane and nuclear proteins). The genes were cloned using cDNA from Arabidopsis stem as template. The reverse primers contained no stop codon to enable C-terminal fusions. Sequences of forward and reverse primers can be sent on request. All Entry clones were verified by restriction analysis and sequencing. Pfam domain profiling of Arabidopsis proteins Pfam domain profiling of Arabidopsis proteins Pfam domain profiling of Arabidopsis proteins p g p p To develop the Pfam domain-based algorithm in plants, we downloaded Arabidopsis gene product information from the AmiGO database (http://amigo.geneontology.org) for the follow- ing localization terms; GO:0005634 (3012 proteins; nucleus), GO:0005739 (1310 proteins; mitochondrion), GO:0005773 (621 proteins; vacuole), GO:0005777 (201 proteins; peroxisome), GO:0005783 (407 proteins; endoplasmic reticulum), GO:0005794 (238 proteins; Golgi apparatus), GO:0005829 (669 proteins; cytosol), GO:0005886 (2236 proteins; plasma membrane), GO:0009504 (18 proteins; cell plate), GO:0009536 (3724 proteins; plastid), and GO:0048046 (333 proteins; extracellular). To remove uncertain localization annotations such as ‘by similarity’ or ‘probable’, we restricted the gene products to 4422 Arabidopsis genes with the IDA evidence code, which indicates that the annotation is derived from experimental data. Since a protein can have multiple Pfam domains, this set of 4422 genes encoded a total of 6141 Pfam domain annotations consisting of 1781 different Pfam domains (Table S5). In order to make predictions about the location of un-localized proteins, a reference data set was established from the AmiGO-derived data, which captured the distribution of Pfam annotations across the different localization GO terms. Since it is possible in the training set for a single Pfam to be annotated to more than one subcellular localization, we define a localization ratio for a single Pfam as the percentage of time the Pfam annotation is seen in a given localization. For a new protein, a score (valued from 0– 100%) for a single localization can be obtained by calculating the geometric mean of the localization ratio for each of the Pfam domains that it is annotated with. By calculating this score for all localizations, and selecting the localization with the highest score, it is possible to suggest the localization for a protein. Conclusion By the combined in silico approaches of expression profiling and localization prediction, we identified putative components of the intracellular network related to xylan synthesis and secondary wall development and proposed models for their function and interactions. Many of the components are identified in both Arabidopsis and rice, giving confidence that they have important roles in the functional network. The analysis enabled us to hypothesize a function of PGSIP proteins as xylan glucuronosyl- transferases that was subsequently experimentally verified. To obtain direct evidence of the role of the other candidate genes in secondary wall formation, future work will involve confirmation of protein-protein interactions, and determination of enzymatic activity of the biosynthetic enzymes. Highlights on signaling and regulatory components Another mammalian fasciclin domain containing protein, Stabilin-1, is receptor protein and has been reported to interact with chitinase- like protein SI-CLP [67]. Both SI-CLP and the plant chitinase-like proteins including CTL1 (At3g16920) and CTL2 (At1g05850) lack a chitin biding domain and catalytic residues involved in chitin hydrolysis and appear to have no chitinase activities [30,67,68]. Based on these conserved characteristics of the chitinase-like proteins, i.e. lacking chitin-binding domain and chitinase activity, and transcriptional co-regulation with fasciclin domain proteins, chitinase-like proteins could bind with fasciclin domain proteins in plants as well as mammals and might lead to ligand-receptor signaling for the GTPase cascade [67] (Figure 5b). As the final components in the signal transduction pathways, we find several transcription factors such as MYB, MYC, and NAC, which may be activated by the kinase cascades and/or the calcium signaling and turn on downstream target genes, e.g. genes involved in cytoskeleton organization and encoding cellulose, lignin and xylan biosynthetic enzymes [69,70,71]. The simple model outlined in Figure 4B will clearly need modification as additional components in the signal transduction pathways are identified and as the hypothesized interactions are experimentally tested. Nevertheless, such models are useful frameworks for developing hypotheses that can be tested. Table S9 Comparison of the co-expressed genes pre- sented in this study and previously published studies. (XLS) Table S9 Comparison of the co-expressed genes pre- sented in this study and previously published studies. (XLS) Figure S2 Comparison of the prediction performance of different subcellular locations using ROC plots. Dotted line shows a random assignment. (TIF) Figure S2 Comparison of the prediction performance of different subcellular locations using ROC plots. Dotted line shows a random assignment. (TIF) We would like to thank Mr. Andrew Carroll for help with the confocal microscope, Dr. Peijian Cao for helpful advice about rice bioinformatic resources, and Ms. Sherry Chan for taking care of plants. We would like to thank Mr. Andrew Carroll for help with the confocal microscope, Dr. Peijian Cao for helpful advice about rice bioinformatic resources, and Ms. Sherry Chan for taking care of plants. Figure S3 Flowchart of data processing and analysis in this study. (A) Co-expressed gene information from Arabidopsis and rice to the Pfam functional domain information. (B) Arabidopsis gene product information with subcellular localization data to the Pfam functional domain information. (C) Integration of the conserved functions across the species from (A) and subcellular Author Contributions Conceived and designed the experiments: AO HJJ EAR BE CM JLH HVS. Performed the experiments: AO HJJ EAR BE CM. Analyzed the data: AO HJJ EAR BE CM JLH HVS. Contributed reagents/materials/ analysis tools: CM. Wrote the paper: AO HJJ EAR BE JLH HVS. Conceived and designed the experiments: AO HJJ EAR BE CM JLH HVS. Performed the experiments: AO HJJ EAR BE CM. Analyzed the data: AO HJJ EAR BE CM JLH HVS. Contributed reagents/materials/ analysis tools: CM. Wrote the paper: AO HJJ EAR BE JLH HVS. Table S8 The Pfam information of the high-ranking co- expression sets. (XLS) Table S8 The Pfam information of the high-ranking co- expression sets. (XLS) Table S2 The 124 shared Arabidopsis genes. (XLS) Xylan glucuronosyltransferase activity in microsomes prepared from stems was determined essentially as described [10], using 3.7 mM UDP-14C-D-GlcA (740 Bq per reaction, MP Biomedicals, Solon, Ohio), 50 mM unlabeled UDP-D-GlcA, and 6 mg xylohex- aose (Megazyme, Bray, Ireland) as acceptor in a 30 ml reaction volume. Products were separated by paper chromatography and analyzed by liquid scintillation counting according to Lee et al. [10]. Xylan glucuronosyltransferase activity in microsomes prepared from stems was determined essentially as described [10], using 3.7 mM UDP-14C-D-GlcA (740 Bq per reaction, MP Biomedicals, Solon, Ohio), 50 mM unlabeled UDP-D-GlcA, and 6 mg xylohex- aose (Megazyme, Bray, Ireland) as acceptor in a 30 ml reaction volume. Products were separated by paper chromatography and analyzed by liquid scintillation counting according to Lee et al. [10]. Table S3 300 co-expressed rice genes listed by three baits, OsGT43A (Os05g0123100), OsGT43B (Os04g0650300) and OsGT47D (Os01g0926400). (XLS) Table S4 The 83 shared rice genes. (XLS) Table S5 Distribution of Pfam domains across sub- cellular localizations. (XLS) Table S6 Localization scores used for PFANTOM tool. (XLS) Table S7 The specificities and sensitivities of each predictor in ten subcellular compartments. (XLS) Supporting Information Figure S1 Expression correlation among the three IRX genes. (A–D) Scatter plot analysis between IRX9 and IRX10 (A), IRX9 and IRX14 (B) IRX10 and IRX14 (C), IRX9 and a non-co- expressed gene (At4g15800) as negative control (D). (TIF) 6. Harholt J, Suttangkakul A, Scheller HV (2010) Biosynthesis of pectin. Plant Physiol 153: 384–395. 7. Caffall KH, Mohnen D (2009) The structure, function, and biosynthesis of plant cell wall pectic polysaccharides. Carbohydr Res 344: 1879–1900. 8. Fincher GB (2009) Revolutionary times in our understanding of cell wall biosynthesis and remodeling in the grasses. Plant Physiol 149: 27–37. 9. Pauly M, Keegstra K (2008) Cell-wall carbohydrates and their modification as a resource for biofuels. Plant J 54: 559–568. 10. York WS, O’Neill MA (2008) Biochemical control of xylan biosynthesis - which end is up? Curr Opin Plant Biol 11: 258–265. 5. Liepman AH, Wightman R, Geshi N, Turner SR, Scheller HV (2010) Arabidopsis - a powerful model system for plant cell wall research. Plant J 61: 1107–1121. Table S2 The 124 shared Arabidopsis genes. (XLS) Table S2 The 124 shared Arabidopsis genes. (XLS) p p y [ ] Xylan glucuronosyltransferase activity in microsomes prepared from stems was determined essentially as described [10], using 3.7 mM UDP-14C-D-GlcA (740 Bq per reaction, MP Biomedicals, Solon, Ohio), 50 mM unlabeled UDP-D-GlcA, and 6 mg xylohex- aose (Megazyme, Bray, Ireland) as acceptor in a 30 ml reaction volume. Products were separated by paper chromatography and analyzed by liquid scintillation counting according to Lee et al. [10]. For RT-PCR, total RNA was isolated from frozen stem tissue using the Plant RNeasy Mini kit (Qiagen) according to the manufacturer’s instructions. First-strand DNA synthesis was per- formed with oligo(dT) anchor primer and Superscript III reverse transcriptase (Invitrogen) according to the manufacturer’s instruc- tions. Two mL were used as template for PCR using the primers 59- GTTTACGTCTGCGGTGCAAT-39 and 59-AATTATTGCGT- CACAAGTTATGG-39 to amplify PGSIP1 cDNA and 59-CTCA- AAGACCAGCTCTTCCATC-39 and 59-GCCTTTGATCTTGA- GAGCTTAG–39 to amplify ACT2 cDNA. The PCR program consisted of 2 min at 95uC, followed by 30 cycles of 20 s at 95uC, 30 s at 49uC, and 1 min 15 s at 72uC, with a final extension step of 10 min at 72uC. PCR products were visualized on 0.8% agarose gels. p p y [ ] Xylan glucuronosyltransferase activity in microsomes prepared from stems was determined essentially as described [10], using 3.7 mM UDP-14C-D-GlcA (740 Bq per reaction, MP Biomedicals, Solon, Ohio), 50 mM unlabeled UDP-D-GlcA, and 6 mg xylohex- aose (Megazyme, Bray, Ireland) as acceptor in a 30 ml reaction volume. Products were separated by paper chromatography and analyzed by liquid scintillation counting according to Lee et al. [10]. For RT-PCR, total RNA was isolated from frozen stem tissue using the Plant RNeasy Mini kit (Qiagen) according to the manufacturer’s instructions. First-strand DNA synthesis was per- formed with oligo(dT) anchor primer and Superscript III reverse transcriptase (Invitrogen) according to the manufacturer’s instruc- tions. Two mL were used as template for PCR using the primers 59- GTTTACGTCTGCGGTGCAAT-39 and 59-AATTATTGCGT- CACAAGTTATGG-39 to amplify PGSIP1 cDNA and 59-CTCA- AAGACCAGCTCTTCCATC-39 and 59-GCCTTTGATCTTGA- GAGCTTAG–39 to amplify ACT2 cDNA. The PCR program consisted of 2 min at 95uC, followed by 30 cycles of 20 s at 95uC, 30 s at 49uC, and 1 min 15 s at 72uC, with a final extension step of 10 min at 72uC. PCR products were visualized on 0.8% agarose gels. 10. York WS, O’Neill MA (2008) Biochemical control of xylan biosynthesis - which end is up? Curr Opin Plant Biol 11: 258–265. Mutant analysis and glycosyltransferase assay Mutant analysis and glycosyltransferase assay The T-DNA insertion mutant in the PGSIP1 gene (SALK_063763) was obtained from the Arabidopsis Biologial Resource Center, Ohio. Plants were grown under short day conditions (8 h photoperiod) in growth chambers for 6 weeks before they were transferred to a growth room with a long-day regime (16 h photoperiod). After 14 days growth under long-day conditions the first to third internodes were harvested and cell walls prepared, hydrolyzed with TFA and subsequently analyzed by HPAEC for monosaccharide composition as previously described [76]. Mutant analysis and glycosyltransferase assay The T-DNA insertion mutant in the PGSIP1 gene (SALK_063763) was obtained from the Arabidopsis Biologial Resource Center, Ohio. Plants were grown under short day conditions (8 h photoperiod) in growth chambers for 6 weeks before they were transferred to a growth room with a long-day regime (16 h photoperiod). After 14 days growth under long-day conditions the first to third internodes were harvested and cell walls prepared, hydrolyzed with TFA and subsequently analyzed by HPAEC for monosaccharide composition as previously described [76]. Table S1 300 co-expressed Arabidopsis genes listed by three baits, IRX9 (At2g37090), IRX10 (At1g27440) and IRX14 (At4g36890). (XLS) Xylan Biosynthesis Network in Arabidopsis and Rice localization information form (B) with the Pfam domains. The integrated information leads to the intracellular working model across the species. The working model was validated by the fluorescence protein experiments, knock-out mutant analysis, and/ or enzyme assay. (TIF) nm laser lines for excitation; signals were detected between 500 and 530 nm. Images were processed using the Leica Confocal Software (Leica Microsystems) and Adobe Photoshop 7.0. Confocal Microscopy A Leica confocal microscope (Leica Microsystems) was used for confocal laser-scanning microscopy. All images were obtained with 636 magnification and a glycerol-immersion objective. GFP and YFP channels were acquired by simultaneous scanning using 488- PLoS ONE \| www.plosone.org November 2010 \| Volume 5 \| Issue 11 \| e15481 PLoS ONE \| www.plosone.org 13 Xylan Biosynthesis Network in Arabidopsis and Rice p p y y 8. Fincher GB (2009) Revolutionary times in our understanding of cell wall biosynthesis and remodeling in the grasses. Plant Physiol 149: 27–37. 9. Pauly M, Keegstra K (2008) Cell-wall carbohydrates and their modification as a resource for biofuels. Plant J 54: 559–568. 1. Scheller HV, Ulvskov P (2010) Hemicelluloses. Annu Rev Plant Biol 61: 263–289. 2. Andersson SI, Samuelson O, Ishihara M, Shimizu K (1983) Structure of the reducing end-groups in spruce xylan. Carbohydr Res 111: 283–288. 3. Johansson MH, Samuelson O (1977) Reducing end groups in birch xylan and their alkaline degradation. Wood Sci Technol 11: 251–263. 4. Pen˜a MJ, Zhong R, Zhou G-K, Richardson EA, O’Neill MA, et al. (2007) Arabidopsis irregular xylem8 and irregular xylem9: Implications for the complexity of glucuronoxylan biosynthesis. Plant Cell 19: 549–563. 5. Liepman AH, Wightman R, Geshi N, Turner SR, Scheller HV (2010) Arabidopsis - a powerful model system for plant cell wall research. Plant J 61: 1107–1121. g 4. Pen˜a MJ, Zhong R, Zhou G-K, Richardson EA, O’Neill MA, et al. (2007) Arabidopsis irregular xylem8 and irregular xylem9: Implications for the complexity of glucuronoxylan biosynthesis. Plant Cell 19: 549–563. 7. Caffall KH, Mohnen D (2009) The structure, function, and biosynthesis of plant cell wall pectic polysaccharides. Carbohydr Res 344: 1879–1900. 6. Harholt J, Suttangkakul A, Scheller HV (2010) Biosynthesis of pectin. Plant Physiol 153: 384–395. Xylan Biosynthesis Network in Arabidopsis and Rice Xylan Biosynthesis Network in Arabidopsis and Rice 11. Brown DM, Zeef LA, Ellis J, Goodacre R, Turner SR (2005) Identification of novel genes in Arabidopsis involved in secondary cell wall formation using expression profiling and reverse genetics. Plant Cell 17: 2281–2295. 39. Brembu T, Winge P, Bones AM (2005) The small GTPase AtRAC2/ROP7 is specifically expressed during late stages of xylem differentiation in Arabidopsis. J Exp Bot 419: 2465–2476. p 40. Gu Y, Fu Y, Dowd P, Li S, Vernoud V, et al. (2005) A Rho family GTPase controls actin dynamics and tip growth via two counteracting downstream pathways in pollen tubes. J Cell Biol 169: 127–138. g p p g g 12. Persson S, Wei H, Milne J, Pageand GP, Somerville CR (2005) Identification of genes required for cellulose synthesis by regression analysis of public microarray data sets. Proc Nat Acad Sci USA 102: 8633–8638. 41. Chatterjeea M, Berbezya P, Vyasa D, Coatesb S, Barsbya T (2005) Reduced expression of a protein homologous to glycogenin leads to reduction of starch content in Arabidopsis leaves. Plant Sci 168: 501–509. 13. Brown DM, Goubet F, Wong VW, Goodacre R, Stephens E, Dupree P, Turner SR (2007) Comparison of five xylan synthesis mutants reveals new insight into the mechanisms of xylan synthesis. Plant J 52: 1154– 1168. 42. Bischoff V, Nita S, Neumetzler L, Schindelasch D, Urbain A, et al. (2010) TRICHOME BIREFRINGENCE and its homolog At5g01360 encode plant specific DUF231 proteins required for cellulose biosynthesis in Arabidopsis. Plant Physiol 153: 590–602. 14. Brown DM, Zhang ZN, Stephens E, Dupree P, Turner SR (2009) Characterization of IRX10 and IRX10-like reveals an essential role in glucuronoxylan biosynthesis in Arabidopsis. Plant J 57: 732–746. y 43. Prag S, Parsons M, Keppler MD, Ameer-Beg SM, Barber P, et al. (2007) Activated ezrin promotes cell migration through recruitment of the GEF Dbl to lipid rafts and preferential downstream activation of Cdc42. Mol Biol Cell 18: 2935–2948. 15. Lee C, O’Neill MA, Tsumuraya Y, Darvill AG, Ye ZH (2007) The irregular xylem9 mutant is deficient in xylan xylosyltransferase activity. Plant Cell Physiol 48: 1624–1634. 16. Balazsi G, Barabashi AL, Oltvai ZN (2005) Topological units of environmental signal processing in the transcriptional-regulatory network of Escherichia coli. Proc Nat Acad Sci USA 102: 7841–7846. 44. Yang T, Chaudhuri S, Yang L, Chen Y, Poovaiah BW (2004) Calcium/ calmodulin up-regulates a cytoplasmic receptor-like kinase in plants. Xylan Biosynthesis Network in Arabidopsis and Rice Obayashi T, Kinoshita K (2009) Rank of correlation coefficient as a comparable measure for biological significance of gene coexpression. DNA Res 16: 249–260. p y 52. Anantharaman V, Aravind L (2010) Novel eukaryotic enzymes modifying cell- surface biopolymers. Biol Direct 5: 1. 53. Piston F, Uauy C, Fu L, Langston J, Labavitch J, et al. (2010) Down-regulation of four putative arabinoxylan feruloyltransferase genes from family PF02458 reduces ester-linked ferulate content in rice cell walls. Planta 231: 677–691. 25. Geisler-Lee J, O’Toole N, Ammar R, Provart NJ, Millar AH, Geisler M (2007) A predicted interactome for Arabidopsis. Plant Physiol 145: 317–329. 26. Rastogi S, Rost B (2010) Bioinformatics predictions of localization and targeting. In Protein Secretion. In: Economou A, ed. New York: Humana Press. pp 285–305. 54. Buanafina MMO (2009) Feruloylation in grasses: current and future perspec- tives. Mol Plant 2: 861–872. 55. Ruse M, Knaus UG (2006) New players in TLR-mediated innate immunity: PI3K and small Rho GTPases. Immunol Res 34: 33–48. 27. Wrzeszczynski KO, Rost B (2004) Annotating proteins from endoplasmic reticulum and Golgi apparatus in eukaryotic proteomes. Cell Mol Life Sci 61: 1341–1353. 56. Da Silva CA, Hartl D, Liu W, Lee CG, Elias JA (2008) TLR-2 and IL-17A in chitin-induced macrophage activation and acute inflammation. J Immunol 181: 4279–4286. 28. Li X, Cordero I, Caplan J, Molhoj M, Reiter WD (2004) Molecular analysis of 10 coding regions from Arabidopsis that are homologous to the MUR3 xyloglucan galactosyltransferase. Plant Physiol 134: 940–950. 57. Shibolet O, Giallourakis C, Rosenberg I, Mueller T, Xavier RJ, et al. (2007) AKAP13, a RhoA GTPase-specific guanine exchange factor, is a novel regulator of tlr2 signaling. J Biol Chem 282: 35308–35317. 29. The International Brachypodium Initiative (2010) Genome sequencing and analysis of the model grass Brachypodium distachyon. Nature 463: 763–768. g g J 58. Rolls A, Shechter R, London A, Ziv Y, Ronen A, et al. (2007) Toll-like receptors modulate adult hippocampal neurogenesis. Nature Cell Biol 9: 1081–1088. 30. Aspeborg H, Schrader J, Coutinho PM, Stam M, Kallas A, et al. (2005) Carbohydrate-active enzymes involved in the secondary cell wall biogenesis in hybrid aspen. Plant Physiol 137: 983–997. 59. Wu G, Gu Y, Li S, Yang Z (2001) A genome-wide analysis of Arabidopsis Rop- interactive CRIB motif-containing proteins that act as Rop GTPase targets. Plant Cell 13: 2841–2856. 31. Xylan Biosynthesis Network in Arabidopsis and Rice Kinoshita K, Obayashi T (2009) Multi-dimensional correlations for gene coexpression and application to the large-scale data of Arabidopsis. Bioinfor- matics 20: 2677–2684. 60. Nibau C, Wu HM, Cheung AY (2006) RAC/ROP GTPases: ‘Hubs’ for signal integration and diversification in plants. Trends Plant Sci 11: 309–315. 32. Wu AM, Hornblad E, Voxeur A, Gerber L, Rihouey C, et al. (2010) Analysis of the Arabidopsis IRX9/IRX9-L and IRX14/IRX14-L pairs of glycosyltransferase genes reveals critical contributions to biosynthesis of the hemicellulose glucuronoxylan. Plant Physiol 153: 542–554. 61. Brill S, Li S, Lyman CW, Church DM, Wasmuth JJ, et al. (1996) The Ras GTPase-activating-protein-related human protein IQGAP2 harbors a potential actin binding domain and interacts with calmodulin and Rho family GTPases. Mol Cell Biol 16: 4869–4878. 33. Jain M, Nijhawan A, Arora R, Agarwal P, Ray S, et al. (2007) F-box proteins in rice. Genome-wide analysis, classification, temporal and spatial gene expression during panicle and seed development, and regulation by light and abiotic stress. Plant Physiol 143: 1467–1483. 62. Lafarguette F, Leple´ JC, De´jardin A, Laurans F, Costa C, et al. (2004) Poplar genes encoding fasciclin-like arabinogalactan proteins are highly expressed in tension wood. New Phytologist 164: 107–121. 63. Geisler-Lee J, Geisler M, Coutinho PM, Segerman B, Nishikubo N, et al. (2006) Poplar carbohydrate-active enzymes. Gene identification and expression analyses. Plant Physiol 140: 946–962. 34. Li Y, Qian Q, Zhou Y, Yan M, Sun L, et al. (2003) BRITTLE CULM1, which encodes a COBRA-like Protein, affects the mechanical properties of rice plants. Plant Cell 15: 2020–2031. y y 64. Liu D, Tu L, Li Y, Wang L, Zhu L, et al. (2008) Genes encoding fasciclin-like arabinogalactan proteins are specifically expressed during cotton fiber development. Plant Mol Biol Rep 26: 98–113. 35. Tanaka K, Murata K, Yamazaki M, Onosato K, Miyao A, et al. (2003) Three distinct rice cellulose synthase catalytic subunit genes required for cellulose synthesis in the secondary wall. Plant Physiol 133: 73–83. 65. Zhang D, Hrmova M, Wan CH, Wu C, Balzen J, et al. (2004) Members of a new group of chitinase-like genes are expressed preferentially in cotton cells with secondary walls. Plant Mol Biol 54: 353–372. 36. Guda C, Subramaniam S (2005) pTARGET: A new method for predicting protein sub-cellular localization in eukaryotes. Bioinformatics 21: 3963–3969. 37. Bateman A, Coin L, Durbin R, Finn RD, Hollich V, et al. (2004) The Pfam protein families database. Nucleic Acids Res 32: D138–D141. Xylan Biosynthesis Network in Arabidopsis and Rice J Biol Chem 279: 42552–42559. 17. Ihmels J, Levy R, Barkai N (2004) Principles of transcriptional control in the metabolic network of Saccharomyces cerevisiae. Nat Biotechnol 22: 86–92. 45. Konishi T, Takeda T, Miyazaki Y, Ohnishi-Kameyama M, Hayashi T, et al. (2007) A plant mutase that interconverts UDP-arabinofuranose and UDP- arabinopyranose. Glycobiology 17: 345–354. 18. Hirai MY, Sugiyama K, Sawada Y, Tohge T, Obayashi T, et al. (2007) Omics-based identification of Arabidopsis Myb transcription factors regulating aliphatic glucosinolate biosynthesis. Proc Nat Acad Sci USA 104: 6478– 6483. 46. Toole GA, Le Gall G, Colquhoun IJ, Nemeth C, Saulnier L, et al. (2010) Temporal and spatial changes in cell wall composition in developing grains of wheat cv. Hereward. Planta 232: 677–689. 19. Mitchell RAC, Dupree P, Shewry PR (2007) A novel bioinformatics approach identifies candidate genes for the synthesis and feruloylation of arabinoxylan. Plant Physiol 144: 43–53. 47. Yin Y, Chen H, Hahn MG, Mohnen D, Xu Y (2010) Evolution and function of the plant cell wall synthesis-related glycosyltransferase family 8. Plant Physiol 153: 1729–1746. y 20. Jung KH, Lee J, Dardick C, Seo YS, Cao P, et al. (2008) Identification and functional analysis of light-responsive unique genes and gene family members in rice. PLoS Genet 4: e1000164. 48. Ko JH, Beers EP, Han KH (2006) Global comparative transcriptome analysis identifies gene network regulating secondary xylem development in Arabidopsis thaliana. Mol Gen Genomics 276: 517–531. 21. Aoki K, Ogata Y, Shibata D (2007) Approaches for extracting practical information from gene co-expression networks in plant biology. Plant Cell Physiol 48: 381–390. 49. Mutwil M, Ruprecht C, Giorgi FM, Bringmann M, Usadel B, et al. (2009) Transcriptional wiring of cell wall-related genes in Arabidopsis. Mol Plant 2: 1015–1024. 22. Mutwil M, Obro J, Willats WG, Persson S (2008) GeneCAT—novel webtools that combine BLAST and co-expression analyses. Nucleic Acids Res 36: W320–326. 50. van Dijk AD, Bosch D, ter Braak CJ, van der Krol AR, van Ham RC (2008) Predicting sub-Golgi localization of type II membrane proteins. Bioinformatics 24: 1779–1786. 23. Obayashi T, Hayashi S, Saeki M, Ohta H, Kinoshita K (2009) ATTED-II provides coexpressed gene networks for Arabidopsis. Nucleic Acids Res 37: D987–991. 51. Kaundal R, Saini R, Zhao PX (2010) Combining machine learning and homology-based approaches to accurately predict subcellular localization in Arabidopsis. Plant Physiol 154: 36–54. 24. References 1. Scheller HV, Ulvskov P (2010) Hemicelluloses. Annu Rev Plant Biol 61: 263–289. 2. Andersson SI, Samuelson O, Ishihara M, Shimizu K (1983) Structure of the reducing end-groups in spruce xylan. Carbohydr Res 111: 283–288. 3. Johansson MH, Samuelson O (1977) Reducing end groups in birch xylan and their alkaline degradation. Wood Sci Technol 11: 251–263. 4. Pen˜a MJ, Zhong R, Zhou G-K, Richardson EA, O’Neill MA, et al. (2007) Arabidopsis irregular xylem8 and irregular xylem9: Implications for the complexity of glucuronoxylan biosynthesis. Plant Cell 19: 549–563. PLoS ONE \| www.plosone.org November 2010 \| Volume 5 \| Issue 11 \| e15481 14 Xylan Biosynthesis Network in Arabidopsis and Rice in alternatively activated macrophages and secreted via lysosomal pathway. Blood 107: 3221–3228. 75. Nelson BK, Cai X, Nebenfu¨hr A (2007) A multicolored set of in vivo organelle markers for co-localization studies in Arabidopsis and other plants. Plant J 51: 1126–1136. 68. Hermans C, Porco S, Verbruggen N, Bush DR (2010) Chitinase-like protein CTL1 plays a role in altering root system architecture in response to multiple environmental conditions. Plant Physiol 152: 904–917. 76. Harholt J, Jensen JK, Sorensen SO, Orfila C, Pauly M, et al. (2006) ARABINAN DEFICIENT 1 is a putative arabinosyltransferase involved in biosynthesis of pectic arabinan in Arabidopsis. Plant Physiol 140: 49–58. biosynthesis of pectic arabinan in Arabidopsis. Plant Physiol 140: 4 69. Mitsuda N, Iwase A, Yamamoto H, Yoshida M, Seki M, et al. (2007) NAC transcription factors, NST1 and NST3, are key regulators of the formation of secondary walls in woody tissues of Arabidopsis. Plant Cell 19: 270–280. 77. Bannai H, Tamada Y, Maruyama O, Nakai K, Miyano S (2002) Extensive feature detection of N-terminal protein sorting signals. Bioinformatics 18: 298–305. secondary walls in woody tissues of Arabidopsis. Plant Cell 19 70. Wightman R, Turner SR (2008) The roles of the cytoskeleton during cellulose deposition at the secondary cell wall. Plant J 54: 794–805. 70. Wightman R, Turner SR (2008) The roles of the cytoskele 78. Claros MG, Vicens P (1996) Computational method to predict mitochondrially imported proteins and their targeting sequences. Eur J Biochem 241: 779–786. deposition at the secondary cell wall. Plant J 54: 794–805. 71. Zhong R, Ye ZH (2007) Regulation of cell wall biosynthesis. Curr Opin Plant Biol 10: 564–572. 79. Small I, Peeters N, Legeai F, Lurin C (2004) Predotar: a tool for rapidly screening proteomes for N-terminal targeting sequences. Proteomics 4: 1581–1590. 72. Ouyang S, Zhu W, Hamilton J, Lin H, Campbell M, et al. (2007) The TIGR Rice Genome Annotation Resource: improvements and new features. Nucleic Acids Res 35: D883–887. 80. Emanuelsson O, Nielsen H, Brunak S, von Heijne G (2000) Predicting subcellular localization of proteins based on their N-terminal amino acid sequence. J Mol Biol 300: 1005–1016. 73. Earley KW, Haag JR, Pontes O, Opper K, Juehne T, et al. (2006) Gateway- compatible vectors for plant functional genomics and proteomics. Plant J 45: 616–629. 81. Horton P, Park KJ, Obayashi T, Fujita N, Harad H, et al. (2007) WoLF PSORT: protein localization predictor. Nucleic Acids Res 35: W585–W587. Xylan Biosynthesis Network in Arabidopsis and Rice y 66. Lin B, Madan A, Yoon JG, Fang X, Yan X, et al. (2010) Massively parallel signature sequencing and bioinformatics analysis identifies up-regulation of TGFBI and SOX4 in human glioblastoma. PLoS ONE 5: e10210. 38. Heazlewood JL, Verboom RE, Tonti-Filippini J, Small I, Millar AH (2007) SUBA: the Arabidopsis Subcellular Database. Nucleic Acids Res 35: D213–D218. 67. Kzhyshkowska J, Mamidi S, Gratchev A, Kremmer E, Schmuttermaier C, et al. (2006) Novel stabilin-1 interacting chitinase-like protein (SI-CLP) is up-regulated PLoS ONE \| www.plosone.org November 2010 \| Volume 5 \| Issue 11 \| e15481 15 Xylan Biosynthesis Network in Arabidopsis and Rice 74. Boevink P, Santa Cruz S, Hawes C, Harris N, Oparka KJ (1996) Virus-mediated delivery of the green fluorescent protein to the endoplasmic reticulum of plant cells. Plant J 10: 935–941. p p 82. Dunkley TP, Watson R, Griffin JL, Dupree P, Lilley KS (2004) Localization of organelle proteins by isotope tagging (LOPIT). Mol Cell Proteomics 3: 1128–1134. PLoS ONE \| www.plosone.org November 2010 \| Volume 5 \| Issue 11 \| e15481 16
https://openalex.org/W3153092360	https://www.scielo.br/j/reeusp/a/yhBhQ9XpLJYrDdLWB9gzVrs/?lang=en&format=pdf	English	null	The role of VES-13 to identify limited life expectancy in older adults in primary healthcare settings	Revista da Escola de Enfermagem da USP	2,021	cc-by	6,058	ORIGINAL ARTICLE https://doi.org/10.1590/S1980-220X2020003603743 The role of VES-13 to identify limited life expectancy in older adults in primary healthcare settings* O potencial do VES-13 na identificação da expectativa de vida limitada em idosos na atenção primária à saúde Potencial del VES-13 para identificar la esperanza de vida limitada de adultos mayores en centros de atención primaria How to cite this article: How to cite this article: Assis DL, Chagas VO, Saulo H, Suemoto CK, Santana ANC. The role of VES-13 to identify limited life expectancy in older adults in primary healthcare settings. Rev Esc Enferm USP. 2021;55:e03743. https://doi.org/10.1590/S1980-220X2020003603743 Danilo Lopes Assis1 Virgínia Oliveira Chagas1 Helton Saulo2 Claudia Kimie Suemoto3 Alfredo Nicodemos Cruz Santana4 ORIGINAL ARTICLE ABSTRACT Objective: To investigate the potential role of the Vulnerable Elders Survey to identify older adults with limited life expectancy in primary healthcare settings. Method: This cross-sectional study was performed in all (nine) healthcare units in Jatai, Goiás (Brazil) from July to December 2018. A sample size of 407 older adults was obtained considering an older population (≥ 60 years old). Participants answered a questionnaire about sociodemographic and clinical characteristics, including the Vulnerable Elders Survey and the Suemoto index. We tested the association between limited life expectancy and the Vulnerable Elders Survey using multiple logistic regression analysis. Results: The mean age was 68.9 ± 6.6 yo, and 58.0% were women. The mean score of the Vulnerable Elders Survey was 2.0 ± 2.2, the mean score of Suemoto index was 31.5 ± 21.1%, and 17.2% had limited life expectancy. The Vulnerable Elders Survey was associated with limited life expectancy (OR = 1.57; p = < 0.0001). Conclusion: The Vulnerable Elders Survey was able to identify older adults with limited life expectancy in primary healthcare settings and can play a role in detecting older adults who would not benefit from screening and strict control of chronic diseases. * Extracted from the dissertation: “Risco de mortalidade em 10 anos e vulnerabilidade em saúde em pessoas idosas na Estratégia Saúde da Família em um município de Goiás”, Escola Superior Ciências da Saúde, 2019. 1 Universidade Federal de Jataí, Unidade Acadêmica Especial Ciências da Saúde, Jataí, GO, Brazil. 2 Universidade de Brasília, Departamento de Estatística, Brasília, DF, Brazil. Aged; Frailty; Mass Screening; Sensitivity and Specificity; Primary Health Care. www.scielo.br/reeusp Scenario Health vulnerability can be evaluated through a spe- cific questionnaire, called the Vulnerable Elders Survey (VES-13)(3). In a validation study, the VES-13 detected 32% of older adults with health vulnerability who showed an excessive risk of functional decline and mortality over a 2-year period(3). In other studies, the VES-13 also identified older adults who more frequently used health services, those with complications and mortality after traumatic injuries and after emergency abdominal surgery, as well as older adults with functional decline and increased risk of 5-year mortality(4). A study more recently showed the potential role of VES-13 to detect poor quality of life among older adults in Primary Healthcare (PHC) settings(5). The study was conducted in all PHC units (9 units) in the municipality of Jataí, in the state of Goiás, Brazil. Sample definitionh The sample size for this study was calculated considering the total older adult population (≥ 60 years old) in Jataí (n = 10,853), a prevalence of 50%, a margin of error of 5%, and a confidence level of 95%, as similarly described in previous studies(5,15). The estimated sample size consequently consisted of 370 individuals. Finally, 10% was added to compensate for possible missing information in the questionnaires. Therefore, the final sample size calculation comprised 407 older adults. Each participating PHC unit included the same number of individuals(16). h Another important issue in PHC is estimating the long- term risk of mortality. Older adults with limited life expec- tancy (LLE), defined by a high risk of mortality (≥ 50%) in 10 years, may not benefit from cancer screening (colorec- tal, breast and prostate), or be more prone to adverse out- comes from strict diabetes control(6-11). Thus, it is important to detect these patients in PHC in order to discuss this topic and possibly avoid unnecessary screening and aggressive treatment procedures(12). The following inclusion criteria were considered: adults aged 60 years or older attended by PHC for more than 12 months, residing in the urban area and having no cogni- tive impairment [defined as a score > 5 in the 10-point cognitive screener (10-CS)]. These inclusion criteria were chosen because we considered it is important to have been adequately attended at a PHC center for a certain period of time to assure the collection of reliable information; sig- nificant cognitive impairment can cause problems in self- reported data; also, rural and urban areas have different characteristics, and approximately 85% of older adults live in urban areas(17-19). Accordingly, prognostic indexes can help to identify older adults with LLE. In this scenario, the Suemoto index can have a potential role considering that it was developed using data from 35,367 older people from five cohorts, including information from Brazil and Mexico(6). Unfortunately, PHC centers in Brazil do not routinely calculate the Suemoto index. On the other hand, the VES-13 is routinely used in PHC centers, since its use is recommended by the Brazilian Ministry of Health(13). 2 Rev Esc Enferm USP · 2021;55:e03743 Type of studyh This is a cross-sectional study. www.scielo.br/reeusp INTRODUCTION INTRODUCTION Population aging is a phenomenon which represents a big challenge to healthcare systems, especially to settings with limited economic resources. According to the Statute for the Elderly in Brazil, people are considered to be elderly/older adults at 60 years of age(1). The last Brazilian census con- ducted in 2010 shows that this age group already corre- sponds to more than 11% of the total population, accounting for more than 21 million people aged 60 and over(2). h There are still few studies in the literature which address the issue of vulnerability in the health of older adults, espe- cially those that provide information and guidance on plan- ning healthcare services in providing care for older adults. Therefore, this study is relevant to guide healthcare ser- vice planning(14). The aging of our population makes it essential to dis- cuss practices in the healthcare system. Thus, the concept of health vulnerability arises, which covers aspects related to the aging and illness process, and which involve specific and multidimensional aspects(3). In view of the above, the objective of this study was to investigate the potential role of the Vulnerable Elders Survey to identify older adults with limited life expectancy in primary healthcare settings. Health vulnerability refers to people’s susceptibility to problems and damage to health, and older adults who are at increased risk of functional decline or death over a 2-year period(3). Due to its peculiarities, aging implies an increased risk for vulnerabilities which increase the chances of becom- ing ill and favor the occurrence of adverse clinical outcomes, impaired well-being, hospitalizations, institutionalization and death(3). DESCRIPTORS DESCRIPTORS Aged; Frailty; Mass Screening; Sensitivity and Specificity; Primary Health Care. Received: 03/04/2020 Approved: 12/03/2020 Corresponding author: Danilo Lopes Assis Rua Luzia Miranda, 38 – Setor Hermosa CEP 75803-315 – Jataí, GO, Brazil nilomed.dla@gmail.com 1 Rev Esc Enferm USP · 2021;55:e03743 Rev Esc Enferm USP · 2021;55:e03743 www.scielo.br/reeusp The role of VES-13 to identify limited life expectancy in older adults in primary healthcare settings The role of VES-13 to identify limited life expectancy in older adults in primary healthcare settings to the complexity of the factors that influence the aging process(3). Thus, it is important to know the processes which involve aging and detect vulnerable groups in the older adult population, as they provide indicators for allocating resources to this group, reducing risks of unfavorable outcomes such as reduced functionality and death(5). h Data collectionh The data collection occurred from July to December 2018. Five researchers were trained by the main researcher for one month before starting the data collection at the units. These researchers contacted the patients while they were waiting for medical consultation, vaccination, wound care, or other health services in the PHC unit(5,15). Interviews were In this scenario, it is essential to perform adequate screening, diagnostic, and therapeutic approaches to reduce health vulnerability. However, this situation of vulnerability is not always identified by health service professionals due www.scielo.br/reeusp Assis DL, Chagas VO, Saulo H, Suemoto CK, Santana ANC carried out in a standardized manner using a questionnaire based on previous elderly cohorts(20,21). Regarding sociode- mographic data, we collected information on age, gender, education level (years of formal education), ethnicity (white, black, and others), and living arrangements (living alone or not). Regarding health history, we collected information on self-reported hypertension, depression, and number of medi- cations. In addition, cognitive impairment was evaluated by the 10-CS, a validated tool for Brazilian older adults(22). Mann-Whitney test for numerical variables. Logistic regres- sion analysis adjusted for age, sex, ethnicity, and education was used to test the association between limited life expectancy and health vulnerability measured by the VES-13. We cal- culated the area under the Receiver Operating Characteristic (Auroc) curve to determine the discrimination of the VES-13 in identifying individuals with LLE. Additionally, we deter- mined the best VES-13 cut-off with the best accuracy to identify these individuals using the Youden index(24). The Stata 15 software program (College Station, TX: StataCorp LLC) was used for the statistical analysis. The alpha level was set at the 5% level. y , Health vulnerability was measured using the VES-13 version validated for Brazilian Portuguese(3,23). The VES-13 has 13 items with specific scores: age (1 point for 75-84 years old; 3 points for ≥ 85 years old), self-rated health sta- tus evaluation (1 point for bad or regular), physical activity limitations [1 point for too much difficulty (or unable) in performing each item, with a maximum score of 2 points. RESULTS The independent variable was the VES-13 score (continuous variable). We compared participants according to the presence of LLE using the chi- squared test for categorical variables and unpaired t-test or Table 1 – Sample characteristics by limited life expectancy sta- tus* – Jataí, GO, Brazil, 2018. Limited Life Expectancy (n = 407) p-value No n = 337 Yes n = 70 Age (years), mean (SD)† 67.0 (4.8) 77.8 (6.3) < 0.0001 Women, %‡ 60.5 45.7 0.0200 Education (years), mean (SD)§ 4.1 (3.8) 1.4 (2.5) < 0.0001 Ethnicity, %‡ 0.1400 White 39.2 41.4 Black 58.2 51.4 Other 2.7 7.1 continue… Table 1 – Sample characteristics by limited life expectancy sta- tus* – Jataí, GO, Brazil, 2018. Table 1 – Sample characteristics by limited life expectancy sta- tus* – Jataí, GO, Brazil, 2018. p y p y tus* – Jataí, GO, Brazil, 2018. Limited Life Expectancy (n = 407) p-value No n = 337 Yes n = 70 Age (years), mean (SD)† 67.0 (4.8) 77.8 (6.3) < 0.0001 Women, %‡ 60.5 45.7 0.0200 Education (years), mean (SD)§ 4.1 (3.8) 1.4 (2.5) < 0.0001 Ethnicity, %‡ 0.1400 White 39.2 41.4 Black 58.2 51.4 Other 2.7 7.1 continue… www.scielo.br/reeusp Data collectionh The items were bending, crouching or kneeling; lifting or carrying objects weighing approximately 5 kilograms; raising or extending arms above shoulder level; handling and hold- ing small objects; walking 400 meters; doing heavy house work such as scrubbing floors or cleaning windows], limi- tations for basic and instrumental activities of daily life (4 points for difficulty in performing one or more item, with a maximum score of 4 points. The items were buying personal items; dealing with money; walking through the room; per- forming light housework; taking a shower or bath). Thus, the possible VES-13 scores vary from 0 to 10, with higher scores associated with more health vulnerability. Older adults with scores of 3 or more are classically classified as vulnerable(8). However, we used the VES-13 as a continuous variable (from 0 to 10) in this study.h Ethical aspects The research project was approved by the Ethics and Research Committee with Human Beings of the Fundação de Ensino e Pesquisa em Ciências da Saúde/Secretaria de Saúde/Distrito Federal (Fepecs/SES/DF), under number 2.804.385/2018. Patients received information about this study, and those who agreed to participate and met the eli- gibility criteria signed the informed consent form. RESULTS A total of 450 older adults were assessed to enroll in this study. However, eight individuals refused to participate, and 35 were excluded due to low scores on the 10-CS. Thus, the study included 407 individuals, in agreement with the sample size calculation. The mean age was 68.9 ± 6.6 yo, 58.0% were women, 60.0% were non-Caucasian, the mean level of education was 3.6 ± 3.7 years, 19.0% lived alone, 71.0% had hypertension, 29.0% had depression, and the mean number of medications was 3.2 ± 2.1. The mean score on the VES-13 was 2.0 ± 2.2. The mean predicted l0-year mortality by the Suemoto index was 31.5 ± 21.1%, and 17.2% had LLE (Suemoto index ≥ 50%). y The estimated limited life expectancy in 10 years was based on Suemoto index. This index predicts the mortality risk in 10 years for community-dwelling older adults. It uses information on age, gender, self-reported diabetes, heart disease, lung disease, cancer, smoking status, alcohol use, body mass index, physical activity, difficulty with bathing or showering, difficulty with walking several blocks, correctly reporting today’s date, and self-reported health status(6). This index reports a predicted mortality in 10 years which varies from 0 to 100% based on the effect size of each of these variables and mortality in the test cohort. We used the online calculator for the Suemoto index available at e-prognosis (https://eprognosis.ucsf.edu/suemoto.php). We considered that LLE was present when the predicted 10-year mortality risk was 50% or more. A cut-off of 50% was used, as previ- ously used in another study related to cancer screening in older adults(12). Sociodemographic and clinical data were compared according to LLE (Table 1). Participants with LLE accord- ing to the Suemoto index were older, more frequently males, and had fewer years of formal education than individuals without LLE. 3 www.scielo.br/reeusp Rev Esc Enferm USP · 2021;55:e03743 calculator for the Suemoto index available at e-prognosis (https://eprognosis.ucsf.edu/suemoto.php). We considered that LLE was present when the predicted 10-year mortality risk was 50% or more. A cut-off of 50% was used, as previ- ously used in another study related to cancer screening in older adults(12). Data analysis Numerical variables in the statistical analysis are presented as mean and standard deviations (SD), while categorical vari- ables are expressed as absolute numbers and percentages. The dependent variable was LLE (predicted mortality risk ≥ 50% calculated by the Suemoto index). Rev Esc Enferm USP · 2021;55:e03743 DISCUSSION The main finding of this study was to show that VES- 13 was associated with LLE in older adults in PHC set- tings for the first time. In addition, the VES-13 showed good discrimination to detect older adults with LLE according to the Auroc, and a cut-off point ≥ 2 showed the best accuracy. Detecting patients with possible LLE is important, because these older adults should addition- ally be evaluated through the Suemoto index to confirm the LLE (10-year mortality risk ≥ 50%) and then would possibly be advised to discontinue screening (such as colorectal, breast and prostate cancer) and to avoid strict control of chronic diseases (such as diabetes) due to the limited possibility of increasing survival and a high risk of short-term complications(7-11). These screenings and treat- ments only present potential benefits after many years of medical follow-up.h Limited life expectancy defined by 10-year mortality risk ≥ 50% calculated by the Suemoto Index (J Gerontol A Biol Sci Med Sci. 2017;72(3):410-416). †unpaired t-test; ‡chi-squared test; §Mann-Whitney test. †unpaired t-test; ‡chi-squared test; §Mann-Whitney test. Health vulnerability measured by the VES-13 was associated with LLE predicted by the Suemoto index (OR = 1.57, 95% CI = 1.31-1.64, p < 0.0001) (Table 2). The VES-13 is already routinely used in the PHC set- ting in Brazil(13). Therefore, primary healthcare profession- als could promptly use the VES-13 to detect older adults with LLE. Considering the specificity of the VES-13 score ≥ 2 points was 57% to detect LLE, individuals with these VES-13 scores should be further evaluated by the Suemoto index to confirm LLE. At this point, it is impor- tant to emphasize that the use of two sequential test (as we propose the consecutive use of the VES-13 and Suemoto index to detect LLE) is frequently used in different health scenarios, such as D-dimer testing and computed tomo- graphic pulmonary angiography for pulmonary embo- lism detection(25). However, future studies should use the VES-13 at baseline and investigate the ability of this tool to predict mortality after 10 years of follow-up. Table 2 – Association between limited life expectancy and health vulnerability measured by the VES-13 – Jataí, GO, Brazil, 2018. OR 95% CI p Simple 1.47 1.31-1.64 < 0.0001 Multiple† 1.57 1.28-1.92 < 0.0001 Limited life expectancy defined by 10-year mortality risk ≥ 50% calculated by the Suemoto Index (J Gerontol A Biol Sci Med Sci. 2017;72(3):410-416). Data analysis Numerical variables in the statistical analysis are presented as mean and standard deviations (SD), while categorical vari- ables are expressed as absolute numbers and percentages. The dependent variable was LLE (predicted mortality risk ≥ 50% calculated by the Suemoto index). The independent variable was the VES-13 score (continuous variable). We compared participants according to the presence of LLE using the chi- squared test for categorical variables and unpaired t-test or 3 Rev Esc Enferm USP · 2021;55:e03743 The role of VES-13 to identify limited life expectancy in older adults in prim …continuation Limited Life Expectancy (n = 407) p-value No n = 337 Yes n = 70 Living alone, %‡ 18.1 25.7 0.1400 Hypertension, %‡ 70.5 72.9 0.7000 Depression, %‡ 29.4 28.6 0.8900 Number of medications, mean (SD)† 3.1 (2.5) 3.6 (2.4) 0.1400 10-CS score, mean (SD)† 7.9 (1.3) 7.8 (1.3) 0.6500 Limited life expectancy defined by 10-year mortality risk ≥ 50% calculated by the Suemoto Index (J Gerontol A Biol Sci Med Sci. 2017;72(3):410-416). †unpaired t-test; ‡chi-squared test; §Mann-Whitney test. The role of VES-13 to identify limited life expectancy in older adults in prim …continuation Limited Life Expectancy (n = 407) p-value No n = 337 Yes n = 70 Living alone, %‡ 18.1 25.7 0.1400 Hypertension, %‡ 70.5 72.9 0.7000 Depression, %‡ 29.4 28.6 0.8900 Number of medications, mean (SD)† 3.1 (2.5) 3.6 (2.4) 0.1400 10-CS score, mean (SD)† 7.9 (1.3) 7.8 (1.3) 0.6500 Limited life expectancy defined by 10-year mortality risk ≥ 50% calculated by the Suemoto Index (J Gerontol A Biol Sci Med Sci. 2017;72(3):410-416). †unpaired t-test; ‡chi-squared test; §Mann-Whitney test. Rev Esc Enferm USP · 2021;55:e03743 www.scielo.br/reeusp DISCUSSION †Logistic regression model adjusted for age, gender, ethnicity, and educational level. Note: (n = 407) Table 2 – Association between limited life expectancy and health vulnerability measured by the VES-13 – Jataí, GO, Brazil, 2018. Table 2 – Association between limited life expectancy* and health vulnerability measured by the VES-13 – Jataí, GO, Brazil, 2018. Regarding the unnecessary cancer screening in older adults with LLE, one should consider the case of colorec- tal cancer (CRC). CRC is one of the leading causes of cancer mortality, and screening has proven to reduce CRC mortality. Although CRC screening can bring benefits, older individuals should not have LLE, since this type of screening requires 10 years before 1 death related to CRC can be prevented among 1,000 screened individu- als(12). However, approximately 20-30% of individuals with LLE inappropriately undergo CRC screening(12,26). Thus, it is very important to provide health professionals with prognosis information (such as LLE), so they can make the best clinical decisions. Furthermore, it is important to keep in mind that CRC screening is associated with a combined adverse event (gastrointestinal perforation or bleeding, cardiovascular events) rate of up to 12.3 per 1,000 screened individuals(27).h The VES-13 showed good discrimination to iden- tify older adults with LLE (AUROC curve = 0.764, 95% CI = 0.708-0.820) (Figure 1). The best cut-off of the VES-13 according to the Youden index regarding the accu- racy of the VES-13 to detect LLE was ≥ 2, with a sensitivity of 81% and specificity of 57%. 1.00 1.00 0.75 0.75 0.50 0.50 0.25 0.25 0.00 0.00 AUROC curve = 0.764 (95%Cl = 0.708-0.820) 1 – Specificity Sensitivity Note: (n = 407) Figure 1 – Area under the receiver operator curve for the VES-13 to detect limited life expectancy. 1.00 1.00 0.75 0.75 0.50 0.50 0.25 0.25 0.00 0.00 1 – Specificity Sensitivity Sensitivity The personalized action plan for older adults with LLE in 10 years should go beyond the discussion about the unnecessary cancer screening and strict glycemic con- trol. Primary healthcare professionals should develop col- laborative work with gerontologists and geriatricians when treating patients with LLE. All health professionals (i.e. nurses, physicians) working in primary healthcare can increase the sensitivity of screening vulnerable older adults. Note: (n = 407) Figure 1 – Area under the receiver operator curve for the VES-13 to detect limited life expectancy. Objetivo: Investigar o potencial do instrumento Vulnerable Elders Survey para identificar idosos com expectativa de vida limitada, em ambientes de atenção primária à saúde. Método: Estudo transversal realizado em todas as (nove) unidades de saúde de Jataí, Goiás (Brasil), no período de julho a dezembro de 2018. Obteve-se uma amostra de 407 idosos, considerando uma população ≥ 60 anos. Os participantes responderam a um questionário sobre características sociodemográficas e clínicas, incluindo o Vulnerable Elders Survey e o índice de Suemoto. Testamos a associação entre a expectativa de vida limitada e o Vulnerable Elders Survey usando análise de regressão logística múltipla. Resultados: A idade média foi de 68,9 ± 6,6 anos, e 58,0% dos participantes eram mulheres. A pontuação média do Vulnerable Elders Survey foi de 2,0 ± 2,2, a pontuação média do índice de Suemoto foi de 31,5 ± 21,1%, e 17,2% dos participantes tinham expectativa de vida limitada. O Vulnerable Elders Survey foi associado a uma expectativa de vida limitada (OR = 1,57; p = < 0,0001). Conclusão: O instrumento Vulnerable Elders Survey foi capaz de identificar idosos com expectativa de vida limitada em ambientes de atenção primária à saúde, além de poder auxiliar na detecção de idosos que não se beneficiariam com a triagem e o controle estrito de doenças crônicas. RESUMO Conclusão: O instrumento Vulnerable Elders Survey foi capaz de identificar idosos com expectativa de vida limitada em ambientes de atenção primária à saúde, além de poder auxiliar na detecção de idosos que não se beneficiariam com a triagem e o controle estrito de doenças crônicas. DISCUSSION 4 Rev Esc Enferm USP · 2021;55:e03743 www.scielo.br/reeusp Assis DL, Chagas VO, Saulo H, Suemoto CK, Santana ANC These professionals consider the needs of older adults in the overall health assessment, and can contribute to screening these older adults and refer those who are vulnerable to high complexity to gerontology/geriatric outpatient clinics. This collaborative care would analyze aspects which can be improved to maximize functional independence, quality of life and patient autonomy(5). estimation, and not the actual 10-year mortality measured during a 10-year period. However, estimated mortality or estimated adverse event rate is frequently used in clinical practice to decide treatments, as in studies that used the Schonberg index(12). Another questionable point is the cut- off of 50% of mortality risk to define LLE. However, this cut-off was previously used to define LLE and to decide about cancer screening in older adults(12). Another important point in this study is to emphasize the use of VES-13 as a continuous variable, and not as a categorical variable, as previously used (≥ 3 as vulnerable; < 3 as non-vulnerable). In three studies (including this one), mortality risk increased with each 1-point increment in the VES-13 score, corroborating the use of this score as a continuous variable(8,28). In addition, scores ≥ 2 on the VES-13 also detected poor quality of life in older adults in PHC(5). CONCLUSION The Vulnerable Elders Survey was able to screen older adults with limited life expectancy in primary healthcare settings. Older adults with VES-13 scores ≥ 2 should addi- tionally be evaluated through the Suemoto index to con- firm the possibility of LLE. The need to use the acquired contributions is emphasized, using them as a subsidy to guide and plan actions which can recognize vulnerable older adults. Thus, this study can help health professionals working in primary healthcare to prioritize the referral of older adults for gerontology/geriatric evaluation and assist managers in planning health services and policies focused on older adults.h Considering the external validity of this study, the char- acteristics of the older adults included in the present work were similar to those of previous studies(5,15,29). Other studies included older adults with similar characteristics regarding age, gender, education level, and hypertension(5,15). The pres- ent study was also similar to another study using the VES-13 regarding age, gender, ethnicity, living arrangements, hyper- tension, and number of chronic medications(5). Finally, there were similarities regarding the frequencies of gender and depression, and in the age distribution in the study also carried out with older adults in PHC(29). The fact that this study only addresses older adults who were in the PHC settings may also have excluded vulnerable older adults who are unable to reach the PHC. As any older adults with cognitive impairment were excluded, this may have resulted in underestimating the prevalence of vulner- able older adults and of LLE. However, further longitudinal studies are needed to attain a better assessment of the role of VES-13 in detecting LLE over 10 years. This study presents some limitations which should be considered. First, it has a cross-sectional design, and the VES-13 was consequently associated with an LLE www.scielo.br/reeusp Idoso; Fragilidade; Programas de Rastreamento; Sensibilidade e Especificidade; Atenção Primária à Saúde. Idoso; Fragilidade; Programas de Rastreamento; Sensibilidade e Especificidade; Atenção Primária à Saúde. Idoso; Fragilidade; Programas de Rastreamento; Sensibilidade e Especificidade; Atenção Primária à Saúde. RESUMO Objetivo: Investigar o potencial do instrumento Vulnerable Elders Survey para identificar idosos com expectativa de vida limitada, em ambientes de atenção primária à saúde. Método: Estudo transversal realizado em todas as (nove) unidades de saúde de Jataí, Goiás (Brasil), no período de julho a dezembro de 2018. Obteve-se uma amostra de 407 idosos, considerando uma população ≥ 60 anos. Os participantes responderam a um questionário sobre características sociodemográficas e clínicas, incluindo o Vulnerable Elders Survey e o índice de Suemoto. Testamos a associação entre a expectativa de vida limitada e o Vulnerable Elders Survey usando análise de regressão logística múltipla. Resultados: A idade média foi de 68,9 ± 6,6 anos, e 58,0% dos participantes eram mulheres. A pontuação média do Vulnerable Elders Survey foi de 2,0 ± 2,2, a pontuação média do índice de Suemoto foi de 31,5 ± 21,1%, e 17,2% dos participantes tinham expectativa de vida limitada. O Vulnerable Elders Survey foi associado a uma expectativa de vida limitada (OR = 1,57; p = < 0,0001). Conclusão: O instrumento Vulnerable Elders Survey foi capaz de identificar idosos com expectativa de vida limitada em ambientes de atenção primária à saúde, além de poder auxiliar na detecção de idosos que não se beneficiariam com a triagem e o controle estrito de doenças crônicas. Objetivo: Investigar o potencial do instrumento Vulnerable Elders Survey para identificar idosos com expectativa de vida limitada, em ambientes de atenção primária à saúde. Método: Estudo transversal realizado em todas as (nove) unidades de saúde de Jataí, Goiás (Brasil), no período de julho a dezembro de 2018. Obteve-se uma amostra de 407 idosos, considerando uma população ≥ 60 anos. Os participantes responderam a um questionário sobre características sociodemográficas e clínicas, incluindo o Vulnerable Elders Survey e o índice de Suemoto. Testamos a associação entre a expectativa de vida limitada e o Vulnerable Elders Survey usando análise de regressão logística múltipla. Resultados: A idade média foi de 68,9 ± 6,6 anos, e 58,0% dos participantes eram mulheres. A pontuação média do Vulnerable Elders Survey foi de 2,0 ± 2,2, a pontuação média do índice de Suemoto foi de 31,5 ± 21,1%, e 17,2% dos participantes tinham expectativa de vida limitada. O Vulnerable Elders Survey foi associado a uma expectativa de vida limitada (OR = 1,57; p = < 0,0001). REFERENCES 1. Brasil. Ministério da Saúde. Portaria n. 2.528, de 19 de outubro de 2006. Aprova a Política Nacional de Saúde da Pessoa Idosa. Diário Oficial da União, Brasília, 20 out. 2006. Seção 1, p. 142. 2. Instituto Brasileiro de Geografia e Estatística. Pesquisa Nacional por Amostra de Domicílios: síntese de indicadores [Internet] . Rio de Janeiro: Instituto Brasileiro de Geografia e Estatística; 2015 [cited 2019 Sep 10]. Available from: https://biblioteca.ibge.gov.br/visualizacao/ livros/liv98887.pdf 3. Saliba D, Elliott M, Rubenstein LZ, Solomon DH, Young RT, Kamberg CJ, et al. The Vulnerable Elders Survey: a tool for identifying vulnerable older people in the community. J Am Geriatr Soc. 2001;49(12):1691-9. https://doi.org/10.1046/j.1532-5415.2001.49281.x 4. Min L, Yoon W, Mariano J, Wenger NS, Elliott MN, Kamberg C, et al. The vulnerable elders-13 survey predicts 5-year functional decline and mortality outcomes in older ambulatory care patients. J Am Geriatr Soc. 2009;57(11):2070-6. https://doi.org/10.1111/j.1532- 5415.2009.02497.x 5. Silva SM, Santana AN, Silva NN, Novaes MR. VES-13 and WHOQOL-bref cutoff points to detect quality of life in older adults in primary health care. Rev Saude Publica. 2019;53:26. https://doi.org/10.11606/S1518-8787.2019053000802 6. Suemoto CK, Ueda P, Beltrán-Sánchez H, Lebrão ML, Duarte YA, Wong R, et al. Development and validation of a 10-year mortality prediction model: meta-analysis of individual participant data from five cohorts of older adults in developed and developing countries. J Gerontol A Biol Sci Med Sci. 2017;72(3):410-6. https://doi.org/10.1093/gerona/glw166 7. Qaseem A, Denberg TD, Hopkins RH Jr, Humphrey LL, Levine J, Sweet DE, et al. Screening for colorectal cancer: a the American College of Physicians. Ann Intern Med. 2012;156(5):378-86. https://doi.org/10.7326/0003-4819- rg TD, Hopkins RH Jr, Humphrey LL, Levine J, Sweet DE, et al. Screening for colorectal cancer: a guidance statement fro ege of Physicians. Ann Intern Med. 2012;156(5):378-86. https://doi.org/10.7326/0003-4819-156-5-201203060-000 8. Qaseem A, Barry MJ, Denberg TD, Owens DK, Shekelle P. Screening for prostate cancer: a guidance statement from the Clinical Guidelines Committee of the American College of Physicians. Ann Intern Med. 2013;158(10):761-9. https://doi.org/10.7326/0003-4819-158-10- 201305210-00633 9. American Diabetes Association. 11. Older adults: Standards of Medical Care in Diabetes-2018. Diabetes Care. 2018;41 Suppl 1:S119-25. https://doi.org/10.2337/dc18-S011 10. Oeffinger KC, Fontham ET, Etzioni R, Herzig A, Michaelson JS, Shih YC, et al. Breast cancer screening for women at average risk: 2015 guideline update from the American Cancer Society. JAMA. 2015;314(15):1599-614. https://doi.org/10.1001/jama.2015.12783 11. Abdollah F, Sun M, Sammon JD, Choueiri TK, Menon M, Weissman JS, et al. REFERENCES Prevalence of nonrecommended screening for prostate cancer and breast cancer in the United States: a nationwide survey analysis. JAMA Oncol. 2016;2(4):543-5. https://doi.org/10.1001/ jamaoncol.2015.5871 12. Schonberg MA, Breslau ES, Hamel MB, Bellizzi KM, McCarthy EP. Colon cancer screening in U.S. adults aged 65 and older according to life expectancy and age. J Am Geriatr Soc. 2015;63(4):750-6. https://doi.org/10.1111/jgs.13335 13. Brasil. Ministério da Saúde. Secretaria de Atenção à Saúde. Departamento de Ações Programáticas Estratégicas. Caderneta de saúde da pessoa idosa. 4a ed. Brasília: Ministério da Saúde; 2017. 14. Mariano PP, Baldissera VD, Martins JT, Carreira L. Nursing work organization in long-stay institutions for the elderl and suffering at work. Texto Contexto Enferm. 2015;24(3):756-65. https://doi.org/10.1590/0104-070720150-11 15. Silva PA, Soares SM, Santos JF, Silva LB. Cut-off point for WHOQOL-bref as a measure of quality of life of older a 2014;48(3):390-7. https://doi.org/10.1590/S0034-8910.2014048004912 16. Moreschi C, Rempel C, Siqueira DF, Backes DS, Pissaia LF, Grave MT. Family Health Strategies: profile/quality of life of people with diabetes. Rev Bras Enferm. 2018;71(6):2899-906. https://doi.org/10.1590/0034-7167-2018-0037 17. Silva LB, Soares SM, Silva PA, Santos JF, Miranda LC, Santos RM. Assessment of the quality of primary care for the elderly according to the Chronic Care Model. Rev Lat Am Enfermagem. 2018;26(0):e2987. https://doi.org/10.1590/1518-8345.2331.2987 18. Silva CS, Barbosa MM, Pinho L, Figueiredo MF, Amaral CO, Cunha FO, et al. Family health strategy: relevance to the functional capacity of older people. Rev Bras Enferm. 2018;71Suppl 2:740-6. https://doi.org/10.1590/0034-7167-2017-0078 19. Garbaccio JL, Tonaco LA, Estêvão WG, Barcelos BJ. Aging and quality of life of elderly people in rural areas. Rev Bras Enferm. 2018;71 Suppl 2:724-32. https://doi.org/10.1590/0034-7167-2017-0149 20. Lebrão ML, Laurenti R. Saúde, bem-estar e envelhecimento: o estudo SABE no Município de São Paulo. Rev Bras Epidemiol. 2005;8(2):127-41. https://doi.org/10.1590/S1415-790X2005000200005 21. Lima-Costa MF, Andrade FB, Souza Junior PR, Neri AL, Duarte YA, Castro-Costa E, et al. The Brazilian Longitudinal Study of Aging (ELSI-Brazil): objectives and design. Am J Epidemiol. 2018;187(7):1345-53. https://doi.org/10.1093/aje/kwx387 22. Apolinario D, Lichtenthaler DG, Magaldi RM, Soares AT, Busse AL, Amaral JR, et al. Using temporal orientation, category fluency, and word recall for detecting cognitive impairment: the 10-point cognitive screener (10-CS). Int J Geriatr Psychiatry. 2016 Jan;31(1):4-12. https://doi.org/10.1002/gps.4282 23. Maia FO, Duarte YA, Secoli SR, Santos JL, Lebrão ML. Cross-cultural adaptation of the Vulnerable Elders Survey-13 (VES-13): helping in the identification of vulnerable older people. Rev Esc Enferm USP. 2012;46(SpecNo):116-22. https://doi.org/10.1590/S0080- 62342012000700017 24. Ruopp MD, Perkins NJ, Whitcomb BW, Schisterman EF. RESUMEN Objetivo: Investigar el potencial del instrumento Vulnerable Elders Survey para identificar adultos mayores con esperanza de vida limitada en centros de atención primaria. Método: Se trata de un estudio transversal realizado en todas las (nueve) unidades sanitarias de Jataí, Goiás (Brasil) de julio a diciembre de 2018. Se consideró una población de ≥ 60 años, de la cual se obtuvo una muestra de 407 adultos mayores. Los participantes respondieron un cuestionario sobre características sociodemográficas y clínicas, incluyendo el Vulnerable Elders Survey y el índice de Suemoto. Se comprobó la asociación entre la esperanza de vida limitada y el Vulnerable Elders Survey, mediante el análisis de regresión logística múltiple. Resultados: La edad promedio era de 68,9 ± 6,6 años y el 58,0% de los participantes pertenecía al sexo femenino. La puntuación media del Vulnerable Elders Survey resultó en 2,0 ± 2,2; la puntuación media del índice de Suemoto, 31,5 ± 21,1% y el 17,2% de los participantes tenía una esperanza de vida limitada. El Vulnerable Elders Survey estaba asociado a una esperanza de vida limitada (OR = 1,57; p = < 0,0001). Conclusión: El instrumento Vulnerable Elders Survey ha sido capaz de identificar a los adultos mayores con una esperanza de vida limitada en los centros de atención primaria, además de ayudar en la detección de aquellos adultos mayores que no se beneficiarían con el triaje y el control estricto de las enfermedades crónicas. DESCRIPTORES Anciano; Fragilidad; Tamizaje Masivo; Sensibilidad y Especificidad; Atención Primaria de Salud. 5 Rev Esc Enferm USP · 2021;55:e03743 www.scielo.br/reeusp www.scielo.br/reeusp The role of VES-13 to identify limited life expectancy in older adults in primary healthcare settings 28. Wang J, Lin W, Chang LH. The linear relationship between the Vulnerable Elders Survey-13 score and mortality in an Asian population of community-dwelling older persons. Arch Gerontol Geriatr. 2018;74:32-8. https://doi.org/10.1016/j.archger.2017.09.005 29. Nogueira EL, Rubin LL, Giacobbo SS, Gomes I, Cataldo Neto A. Screening for depressive symptoms in older adults in the Family Health Strategy, Porto Alegre, Brazil. Rev Saude Publica. 2014;48(3):368-77. https://doi.org/10.1590/S0034-8910.2014048004660 27. Warren JL, Klabunde CN, Mariotto AB, Meekins A, Topor M, Brown ML, et al. Adverse events after outpatient colonoscopy in the Medicare population. Ann Intern Med. 2009;150(12):849-57. https://doi.org/10.7326/0003-4819-150-12-200906160-00008 Assis DL, Chagas VO, Saulo H, Suemoto CK, Santana ANC 27. Warren JL, Klabunde CN, Mariotto AB, Meekins A, Topor M, Brown ML, et al. Adverse events after outpatient colonoscopy in the Medicare population. Ann Intern Med. 2009;150(12):849-57. https://doi.org/10.7326/0003-4819-150-12-200906160-00008 28. Wang J, Lin W, Chang LH. The linear relationship between the Vulnerable Elders Survey-13 score and mortality in an Asian population of community-dwelling older persons. Arch Gerontol Geriatr. 2018;74:32-8. https://doi.org/10.1016/j.archger.2017.09.005 29. Nogueira EL, Rubin LL, Giacobbo SS, Gomes I, Cataldo Neto A. Screening for depressive symptoms in older adults in the Family Health Strategy, Porto Alegre, Brazil. Rev Saude Publica. 2014;48(3):368-77. https://doi.org/10.1590/S0034-8910.2014048004660 REFERENCES Youden Index and optimal cut-point estimated from observations affected by a lower limit of detection. Biom J. 2008;50(3):419-30. https://doi.org/10.1002/bimj.200710415 25. Konstantinides SV, Torbicki A, Agnelli G, Danchin N, Fitzmaurice D, Galiè N, et al. 2014 ESC guidelines on the diagnosis and management of acute pulmonary embolism. Eur Heart J. 2014;35(43):3033-69. https://doi.org/10.1093/eurheartj/ehu283 26. Mittal S, Lin YL, Tan A, Kuo YF, El-Serag HB, Goodwin JS. Limited life expectancy among a subgroup of medicare beneficiaries receiving screening colonoscopies. Clin Gastroenterol Hepatol. 2014;12(3):443-450.e1. https://doi.org/10.1016/j.cgh.2013.08.021 6 Rev Esc Enferm USP · 2021;55:e03743 www.scielo.br/reeusp Assis DL, Chagas VO, Saulo H, Suemoto CK, Santana ANC Assis DL, Chagas VO, Saulo H, Suemoto CK, Santana ANC This is an open-access article distributed under the terms of the Creative Commons Attribution License. 7 Rev Esc Enferm USP · 2021;55:e03743 Rev Esc Enferm USP · 2021;55:e03743 www.scielo.br/reeusp
https://openalex.org/W2094937572	https://europepmc.org/articles/pmc4423234?pdf=render	English	null	Observation of finite-wavelength screening in high-energy-density matter	Nature communications	2,015	cc-by	9,114	ARTICLE Received 3 Feb 2014 \| Accepted 26 Feb 2015 \| Published 23 Apr 2015 Received 3 Feb 2014 \| Accepted 26 Feb 2015 \| Published 23 Apr 2015 Observation of ﬁnite-wavelength screening in high-energy-density matter D.A. Chapman1,2, J. Vorberger3, L.B. Fletcher4, R.A. Baggott2, L. Divol5, T. Do¨ppner5, R.W. Falcone6, S.H. Glenzer4, G. Gregori7, T.M. Guymer1, A.L. Kritcher5, O.L. Landen5, T. Ma5, A.E. Pak5 & D.O. Gericke2 A key component for the description of charged particle systems is the screening of the Coulomb interaction between charge carriers. First investigated in the 1920s by Debye and Hu¨ckel for electrolytes, charge screening is important for determining the structural and transport properties of matter as diverse as astrophysical and laboratory plasmas, nuclear matter such as quark-gluon plasmas, electrons in solids, planetary cores and charged macromolecules. For systems with negligible dynamics, screening is still mostly described using a Debye–Hu¨ckel-type approach. Here, we report the novel observation of a signiﬁcant departure from the Debye–Hu¨ckel-type model in high-energy-density matter by probing laser-driven, shock-compressed plastic with high-energy X-rays. We use spectrally resolved X-ray scattering in a geometry that enables direct investigation of the screening cloud, and demonstrate that the observed elastic scattering amplitude is only well described within a more general approach. 1 AWE plc, Radiation Physics Department, Aldermaston, Reading RG7 4PR, UK. 2 Centre for Fusion, Space and Astrophysics, Department of Physics, University of Warwick, Coventry CV4 7AL, UK. 3 Max-Planck-Institut fu¨r die Physik komplexer Systeme, Dresden 01187, Germany. 4 High-Energy-Density Science Division, SLAC National Accelerator Laboratory, Menlo Park, California 94025, USA. 5 National Ignition Facility and Photon Science Directorate, Lawrence Livermore National Laboratory, Livermore, California 94550, USA. 6 Physics Department, University of California, Berkeley, California 94720, USA. 7 Department of Physics, University of Oxford, Parks Road, Oxford OX1 3PU, UK. Correspondence and requests for materials should be addressed to D.A.C. (email: david.chapman@awe.co.uk). 1 NATURE COMMUNICATIONS \| 6:6839 \| DOI: 10.1038/ncomms7839 \| www.nature.com/naturecommunications & 2015 Macmillan Publishers Limited. All rights reserved. The latter requirement is only fulﬁlled for processes with a wavelength much larger than the screening length, leading to a wavenumber k effectively approaching zero. In contrast, we will refer to static screening models, which are not similarly restricted, as ﬁnite-wavelength screening. In contrast, the elastic Rayleigh feature is the result of scattering from electrons that follow the low-frequency density ﬂuctuations of the ions and, thus, is an excellent measure for the static ionic structure and the associated screening properties22–25. The elastic scattering strength in WDM with multiple ion species is given by26 To investigate static screening beyond the long-wavelength limit, processes involving large momentum transfers need to be considered. Large-angle collisions are a typical process where ﬁnite-wavelength screening could be observed. Although such collisions are highly unlikely in ideal, low-density plasmas, strong scattering is known to modify transport and relaxation properties in dense plasmas10,11. Another possibility to investigate deviations from Debye-like screening is the interaction of X-rays with dense matter under large scattering angles12. Indeed, spectrally resolved X-ray Thomson scattering (XRTS)13 is particularly suited for these investigations, as it simultaneously allows for the determination of the plasma conditions and the study of the screening cloud from a single spectrum. WR k ð Þ ¼ X a;b ﬃﬃﬃﬃﬃﬃﬃﬃﬃ xaxb p fa k ð Þ þ qa k ð Þ ½ fb k ð Þ þ qb k ð Þ ½ Sab k ð Þ; ð2Þ ð2Þ where xa ¼ na= P a na is the concentration of ion species a, with na its density, fa(k) is the form factor of bound electrons and qa(k) is the screening cloud13. The latter quantities are Fourier transforms of the respective electron densities around ions of species a. Sab(k) denotes the partial ion–ion structure factors. The form factors for bound electrons are readily obtained from ﬁrst-principles methods for the tightly bound K-shell electrons in carbon. In the following, we report observations of ﬁnite-wavelength screening in dense matter probed via spectrally resolved XRTS on laser-driven, shock-compressed plastic (CH) capsules. The strength of the elastic Rayleigh feature is used to further constrain simultaneous measurements of the electron density, temperature and mean ionization obtained from the inelastic Compton feature. We show that agreement between modelled and measured values for the Rayleigh amplitude can only be obtained if ﬁnite-wavelength screening is considered. & 2015 Macmillan Publishers Limited. All rights reserved. & 2015 Macmillan Publishers Limited. All rights reserved. ARTICLE NATURE COMMUNICATIONS \| DOI: 10.1038/ncomms7839 T he Debye–Hu¨ckel theory of charge screening1 is one of the seminal results of electrolyte and plasma physics, whereupon the long-range Coulomb force between a pair of charge carriers is replaced by an exponentially decaying, short-range potential to account for the interaction with the surrounding medium. The extension of this classical description to degenerate electrons, Thomas–Fermi screening2,3, plays an important role in the description of electrons in solids4 and warm dense matter (WDM)5. Moreover, the concept of Debye-like screening is also applied to describe a large range of systems usually not associated with particles interacting via Coulomb forces, including quark-gluon matter6,7, ultra-cold systems in traps8 or chemical and biological systems9. For all these cases, the structural, thermodynamic, transport or relaxation properties are determined by effective interactions between localized charges. T are o ¼ oi–os and k ¼ ki–ks, respectively. Stot ee ðk; oÞ is the total dynamic structure factor containing information on the spatio- temporal correlations of the fully coupled electron system. For a composite plasma probed with high-energy X-rays, the latter can be decomposed into distinct contributions from elastic and inelastic scattering15. Both terms depend on the temperature, density and mean charge state of the ions. Thus, in principle, the thermodynamic state of the target may be determined by ﬁtting the measured spectrum with theoretical calculations16–18. p It is well known that robust estimates of the electron density and temperature of the samples probed with XRTS can be made by matching to the inelastic Compton scattering feature of the spectrum19,20. Moreover, the relative contributions of free and bound electrons enable the ionization state, and therefore material density, to be measured21. Uncertainties in the models describing this feature are effectively constrained by probing at large k (see Methods). In this work, we obtain conditions suggestive of a WDM state with free-electron densities of B1024 cm 3 and temperatures of t10 eV. Furthermore, we ﬁnd the mean charge state of the carbon ions to be Zf C ¼ 4, consistent with estimates of continuum lowering in WDM. y g Although Debye-like screening has been applied for many applications, it contains a number of inherent restrictions: weakly interacting particles in the screening cloud, negligible dynamic evolution and the long-wavelength limit. & 2015 Macmillan Publishers Limited. All rights reserved. For most conditions, the largest theoretical uncertainty in WR is related to the ionic structure factors (see, for example, ref. 27). In the present experiment, the ambiguity with respect to the ionic structure is circumvented by the large scattering angle applied as shown in Fig. 1a. At the resulting large k values (shaded vertical band), all spatial correlations have decayed and we have unity for Saa and negligible values for Sab. This result is independent of the method applied to obtain the ion structure (see Methods). Thus, the only unknown remaining in the description of the elastic scattering strength is the form of the screening cloud qa(k), which can be determined by our measurements. NATURE COMMUNICATIONS \| 6:6839 \| DOI: 10.1038/ncomms7839 \| www.nature.c 2 NATURE COMMUNICATIONS \| 6:6839 \| DOI: 10.1038/ncomms7839 \| www.nature.com/naturecommunications & 2015 Macmillan Publishers Limited. All rights reserved. \| \| / \| / & 2015 Macmillan Publishers Limited. All rights reserved. & 2015 Macmillan Publishers Limited. All rights reserved. Theoretical description of the screening cloud. In general, the response of the electrons to the ions may be derived within a quantum statistical framework from the ratio of dynamic struc- ture factors, thereby rigorously incorporating strong interactions. Theoretical description of the screening cloud. In general, the response of the electrons to the ions may be derived within a quantum statistical framework from the ratio of dynamic struc- ture factors, thereby rigorously incorporating strong interactions. As it is associated with the ions, the frequency dependence of the screening cloud is negligible since the electrons react to the ion distribution almost instantaneously, that is, electronic screening can be treated in the static limit o-0. For the conditions of interest, electron–ion correlations are expected to be weak by virtue of the high Fermi energy, justifying a linear response approach28 As it is associated with the ions, the frequency dependence of the screening cloud is negligible since the electrons react to the ion distribution almost instantaneously, that is, electronic screening can be treated in the static limit o-0. For the conditions of interest, electron–ion correlations are expected to be weak by virtue of the high Fermi energy, justifying a linear response approach28 The spectrum of radiation scattered by a plasma is propor- tional to the double-differential scattering cross-section14 @2s k; o ð Þ @O@os ¼ sT os oi 2 Stot ee k; o ð Þ; ð1Þ qFWS a k ð Þ ¼ w0 ee k; 0 ð ÞVea k ð Þ eee k; 0 ð Þ ; ð3Þ qFWS a k ð Þ ¼ w0 ee k; 0 ð ÞVea k ð Þ eee k; 0 ð Þ ; ð3Þ ð1Þ ð3Þ where sT ¼ 6.65 10 25 cm2 is the Thomson cross-section. The frequency and wavenumber shifts associated with the scattering in which the screening is provided by the dielectric function eee k; 0 ð Þ ¼ 1 Vee k ð Þw0 ee k; 0 ð Þ. Strong correlations between NATURE COMMUNICATIONS \| 6:6839 \| DOI: 10.1038/ncomms7839 \| www.nature.com/naturecommunications Results X-ray scattering in high-energy-density matter. Investigation of the screening cloud at large wavenumbers requires highly com- pressed and moderately heated matter to be probed by high- energy X-rays. These target conditions were created at the Omega laser facility by compressing thin CH shells with multiple coa- lescent shocks driven by intense laser beams. He-a radiation at B9 keV from a laser-produced Zn plasma is used as the probe. Measuring the intensity of the X-ray scatter for angles between 120 and 150 accesses the response at the wavenumbers of interest (see Methods). Theoretical description of the screening cloud. In general, the response of the electrons to the ions may be derived within a quantum statistical framework from the ratio of dynamic struc- ture factors, thereby rigorously incorporating strong interactions. & 2015 Macmillan Publishers Limited. All rights reserved. (a) Partial ion–ion structure factors Sab(k) for CH with Zf H ¼ 1, Zf C ¼ 4 at r ¼ 5.84 g cm 3 (ne ¼ 1.3 1024 cm 3) and T ¼ 10 eV applying multicomponent classical hypernetted-chain calculations (HNC—solid curves) with screened Coulomb interactions between ions, and density functional molecular dynamics (DFT-MD—dotted curves with circles). (b) Screening clouds qa(k) around the H and C ions for the same conditions as a calculated with Debye-like (DH—solid curves) and ﬁnite-wavelength screening (FWS—dashed curves) approaches. Corrections to the RPA due to strong static correlations (dotted curve with squares) are negligible for both components at large k. The form factors of the bound electrons around the C ions based on the density functional theory (DFT—dot-dashed curve) and screened hydrogenic wavefunctions (SH—dotted curve with diamonds) are also shown for comparison. (c) Behaviour of the charge-normalized screening cloud qa Zf a for various temperatures at ne ¼ 1024 cm 3, comparing the FWS (solid curves) and DH (dashed curves) models. The different scaling in the limits k-0 and k-N are demonstrated with thin grey lines. The range of wavenumbers probed by in this experiment is shown by the vertical shaded band in all panels. electrons can be accounted for by modifying the potential with local ﬁeld corrections Vee k ð Þ ! 1 Gee k ð Þ ½ Vee k ð Þ. However, a weak coupling approximation, Gee(k)E0, may be assumed for the conditions created. The non-interacting density response is then given by the random phase approximation (RPA)5 equation (5), as can be seen in Fig. 1b. Under non-degenerate conditions, however, the Debye-like behaviour is recovered for intermediate k before changing to scale as 1/k4 in the limit k-N (Fig. 1c). Note that the ﬁnite-wavelength results tend to a constant for large k as the electron density close to the ions becomes independent of the temperature due to quantum degeneracy. w0 ee k; 0 ð Þ ¼ neb ﬃﬃﬃp p kDe Z1 0 dx x ln 1 þ eZe x þ k ð Þ2 1 þ eZe x k ð Þ2 " # ; ð4Þ ð4Þ Furthermore, this strong 1/k4 decay directly leads to the well- known Friedel oscillations of the real-space electron density distribution around the ions31, which have been observed in cold solids32. & 2015 Macmillan Publishers Limited. All rights reserved. NATURE COMMUNICATIONS \| DOI: 10.1038/ncomms7839 ARTICLE –0.4 –0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Sab(k) 0 1 2 3 4 5 6 k [aB –1] k [aB –1] k [aB –1] = 120°–150° H-H C-C H-C 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 fa(k), qa(k) 0 1 2 3 4 5 6 qa: DH qa: FWS qa: Incl. LFCs fC: DFT qH qC fC 10–6 10–5 10–4 10–3 10–2 10–1 1 10 qa(k)/Z f a 10−1 1 10 102 T = 10 eV T = 100 eV T = 1,000 eV qa ∝ k –2 qa ∝ k –4 FWS DH ne = 1024 cm–3 EF = 36.5 eV a b c fC: SH (Z b C = 2) DFT-MD HNC-Y (Z f C = 4) – Figure 1 \| Elastic scattering contributions and limiting behaviour of the screening cloud. (a) Partial ion–ion structure factors Sab(k) for CH with Zf H ¼ 1, Zf C ¼ 4 at r ¼ 5.84 g cm 3 (ne ¼ 1.3 1024 cm 3) and T ¼ 10 eV applying multicomponent classical hypernetted-chain calculations (HNC—solid curves) with screened Coulomb interactions between ions, and density functional molecular dynamics (DFT-MD—dotted curves with circles). (b) Screening clouds qa(k) around the H and C ions for the same conditions as a calculated with Debye-like (DH—solid curves) and ﬁnite-wavelength screening (FWS—dashed curves) approaches. Corrections to the RPA due to strong static correlations (dotted curve with squares) are negligible for both components at large k. The form factors of the bound electrons around the C ions based on the density functional theory (DFT—dot-dashed curve) and screened hydrogenic wavefunctions (SH—dotted curve with diamonds) are also shown for comparison. (c) Behaviour of the charge-normalized screening cloud qa Zf a for various temperatures at ne ¼ 1024 cm 3, comparing the FWS (solid curves) and DH (dashed curves) models. The different scaling in the limits k-0 and k-N are demonstrated with thin grey lines. The range of wavenumbers probed by in this experiment is shown by the vertical shaded band in all panels. k [aB –1] 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 fa(k), qa(k) 0 1 2 3 4 5 6 qa: DH qa: FWS qa: Incl. & 2015 Macmillan Publishers Limited. All rights reserved. 1a). Moreover, we have performed self- consistent calculations of the ionization equilibrium5 that include continuum lowering34 and use temperature and density proﬁles given by radiation hydrodynamics simulations of the target. Figure 2 clearly shows that the mean charge state in the bulk of the target is dominated by contributions from helium-like carbon. Data analysis. To determine the plasma conditions from the measured XRTS spectra, we apply a statistical analysis with respect to the probability density corresponding to w2 statistics (see, for example, ref. 33) over a wide range of density– temperature space. The best ﬁt is deﬁned by minimizing the mean square deviation between data and model, that is w2, whereas the quality and robustness of the ﬁt is estimated by the values of w2 in the density–temperature region around the best ﬁt. g g p Taking the electron–ion interaction to be the Coulomb potential, one may assume Vea ¼ Zf aVee and subsequently ﬁnd the well-known expression for the screening cloud For these calculations, the mean charge state of carbon is held constant at Zf C ¼ 4 for all parameters considered. This estimate is supported by previous observations of continuum lowering21 based on the shape of the inelastic Compton feature, and also by the excellent agreement we ﬁnd comparing ab initio simulations based on density functional theory (DFT)-molecular dynamics to the predictions of the classical ﬂuid theory for the partial structure factors (see Fig. 1a). Moreover, we have performed self- consistent calculations of the ionization equilibrium5 that include continuum lowering34 and use temperature and density proﬁles given by radiation hydrodynamics simulations of the target. Figure 2 clearly shows that the mean charge state in the bulk of the target is dominated by contributions from helium-like carbon. qDH a k ð Þ ¼ Zf a k2 e k2 þ k2 e : ð5Þ ð5Þ Reviewing equation (5) reveals that the ratio k/ke determines whether the long-wavelength limit is applicable. For collective scattering29,30, the wavelength of density ﬂuctuations is larger than the screening length: a ¼ ke/k41. Accordingly, the long- wavelength limit is valid. Conversely, we have ao1 in the non- collective scattering regime, and the screening function arising from combining equations (3) and (4) should be used. & 2015 Macmillan Publishers Limited. All rights reserved. LFCs fC: DFT qH qC fC (k)/Z f b c fC: SH (Z b C = 2) –0.4 –0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Sab(k) 0 1 2 3 4 5 6 k [aB –1] = 120°–150° H-H C-C H-C a DFT-MD HNC-Y (Z f C = 4) – b k [aB –1] 10–6 10–5 10–4 10–3 10–2 10–1 1 10 qa(k)/Z f a 10−1 1 10 102 T = 10 eV T = 100 eV T = 1,000 eV qa ∝ k –2 qa ∝ k –4 FWS DH ne = 1024 cm–3 EF = 36.5 eV c a Figure 1 \| Elastic scattering contributions and limiting behaviour of the screening cloud. (a) Partial ion–ion structure factors Sab(k) for CH with Zf H ¼ 1, Zf C ¼ 4 at r ¼ 5.84 g cm 3 (ne ¼ 1.3 1024 cm 3) and T ¼ 10 eV applying multicomponent classical hypernetted-chain calculations (HNC—solid curves) with screened Coulomb interactions between ions, and density functional molecular dynamics (DFT-MD—dotted curves with circles). (b) Screening clouds qa(k) around the H and C ions for the same conditions as a calculated with Debye-like (DH—solid curves) and ﬁnite-wavelength screening (FWS—dashed curves) approaches. Corrections to the RPA due to strong static correlations (dotted curve with squares) are negligible for both components at large k. The form factors of the bound electrons around the C ions based on the density functional theory (DFT—dot-dashed curve) and screened hydrogenic wavefunctions (SH—dotted curve with diamonds) are also shown for comparison. (c) Behaviour of the charge-normalized screening cloud qa Zf a for various temperatures at ne ¼ 1024 cm 3, comparing the FWS (solid curves) and DH (dashed curves) models. The different scaling in the limits k-0 and k-N are demonstrated with thin grey lines. The range of wavenumbers probed by in this experiment is shown by the vertical shaded band in all panels. Figure 1 \| Elastic scattering contributions and limiting behaviour of the screening cloud. (a) Partial ion–ion structure factors Sab(k) for CH with Zf H ¼ 1, Figure 1 \| Elastic scattering contributions and limiting behaviour of the screening cloud. & 2015 Macmillan Publishers Limited. All rights reserved. & 2015 Macmillan Publishers Limited. All rights reserved. This effect has thus far been hidden in XRTS experiments by probing higher temperatures, use of smaller scattering angles or weak ionization. In our experiments, the large scattering angle combined with strongly driven WDM states results in an intermediate scattering regime with aB0.3, where the screening cloud exhibits changes due to ﬁnite-wavelength effects and also gives a signiﬁcant contribution to the elastic Rayleigh feature. where we have deﬁned k ¼ ‘k ﬃﬃﬃ 8 p pe, pe ¼ ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ me=b p and b ¼ 1/kBT. Furthermore, De ¼ neL3 e is the degeneracy parameter, with Le ¼ ﬃﬃﬃﬃﬃ 2p p ‘=pe being the thermal de Broglie wavelength, and Ze ¼ bme is the dimensionless chemical potential. Note that equation (4) is valid for arbitrary k, making it suitable to describe ﬁnite-wavelength screening. While the idea of such a generalized approach to screening is not new, until recently the vast majority of XRTS data has been evaluated on the basis of the Debye–Hu¨ckel or Thomas–Fermi theories, wherein the dielectric function has the form eDH ee k; 0 ð Þ ¼ 1 þ ke=k ð Þ2. The inverse screening length ke is chosen to be consistent with the appropriate limit, but may also be derived for arbitrary degeneracies5. One may readily obtain this form of the dielectric function by taking the long-wavelength (k-0) limit of equation (4). Thus, the Debye-like (Debye–Hu¨ckel and Thomas–Fermi) theories are restricted to small momentum transfer events, such as long-range interactions between particles. Data analysis. To determine the plasma conditions from the measured XRTS spectra, we apply a statistical analysis with respect to the probability density corresponding to w2 statistics (see, for example, ref. 33) over a wide range of density– temperature space. The best ﬁt is deﬁned by minimizing the mean square deviation between data and model, that is w2, whereas the quality and robustness of the ﬁt is estimated by the values of w2 in the density–temperature region around the best ﬁt. For these calculations, the mean charge state of carbon is held constant at Zf C ¼ 4 for all parameters considered. This estimate is supported by previous observations of continuum lowering21 based on the shape of the inelastic Compton feature, and also by the excellent agreement we ﬁnd comparing ab initio simulations based on density functional theory (DFT)-molecular dynamics to the predictions of the classical ﬂuid theory for the partial structure factors (see Fig. NATURE COMMUNICATIONS \| 6:6839 \| DOI: 10.1038/ncomms7839 \| www.nature.com/naturecommunications NATURE COMMUNICATIONS \| 6:6839 \| DOI: 10.1038/ncomms7839 \| www.nature.com/naturecommunications & 2015 Macmillan Publishers Limited. All rights reserved. \| \| / \| / & 2015 Macmillan Publishers Limited. All rights reserved. & 2015 Macmillan Publishers Limited. All rights reserved. Figure 4 compares the elastic Rayleigh strength obtained using the three approaches presented versus the ideal electron pressure Pid e ¼ 2nekBTF 3=2 Ze ð Þ=De (combining electron density and temperature into a single parameter). Here, we take the plasma conditions obtained from ﬁtting with equation (6) as the reference states, and the corresponding errors are propagated through the calculation of WR for each model. We ﬁnd signiﬁcant deviations between the results using Debye-like screening and the extracted (experimental) values for all time delays. On the other hand, ﬁnite-wavelength screening is in much better agreement for 3.4 and 3.5 ns. For 3.6 ns, the plasma has cooled considerably and one expects strong correlations within the electron gas. The reduction of accuracy found for ﬁnite- wavelength screening points to an inadequate description within the RPA as applied here. experimental data. This is achieved by decomposing the total scattered power into elastic and inelastic contributions, as per the generalized Chihara formula26, and rearranging to obtain Wexp R k ð Þ ¼ RWC k; opk WC k; 0 ð Þ 0 ð Þ R opk : ð6Þ ð6Þ Here, R is the ratio of the amplitudes for elastic (o ¼ 0) and inelastic (o ¼ opkE:k2/2me) scattering in the measured spectra, S(o) is shape of the X-ray source function and WC(k, o) is the calculated total inelastic contribution to the scattered power. Using equation (6) to analyse the data, the observed shape and amplitude of the Rayleigh peak is perfectly reproduced. Since it does not depend on any models for the elastic contributions, this approach yields the experimentally determined value for the strength of the Rayleigh peak Wexp R . The resulting conditions of the best ﬁt are then largely determined by the sensitivity of the free-electron response to the density and temperature19. Under partially ionized conditions, one may additionally infer the existence or absence of contributions from particular bound states from the shape of the Compton peak; in particular, our data show the absence of L-shell states for the carbon ions21. Up to now the free electrons have been assumed to respond to the ions via a screened Coulomb-like potential determined by the effective charge of the ions. Clearly, this is justiﬁed for the fully ionized hydrogen component. However, the electrons still bound to the carbon ions are known to modify the electron–ion interactions. & 2015 Macmillan Publishers Limited. All rights reserved. 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 xCj = nCj / Σj nCj 0 1 2 3 4 5 6 24 26 28 30 32 34 36 38 40 42 44 46 r [μm] Z f C = 1 Z f C = 2 Z f C = 3 Z f C = 4 Z f C = 5 Z f C = 6 Z f C = Σj xCj Z f Cj – Z f C – Figure 2 \| Radial proﬁle of the ionization equilibrium of carbon. the inelastic scattering spectra due to such errors is shown in Fig. 3d. g Well-deﬁned conditions are also found when the strength of the Rayleigh feature, WR, is calculated using the Debye-like model, as well as for the ﬁnite-wavelength approaches for the screening cloud (see Fig. 3b,c). We note that in both cases the shape of the contours indicate roughly twice the density sensitivity and substantially improved accuracy with respect to the temperature as compared with the approach determining Wexp R . This stems from the additional constraints due to the temperature and density dependence of the screening cloud. However, the different screening models yield quite different conditions. For example, at the time t ¼ 3.4 ns, the best ﬁt using ﬁnite-wavelength screening is found at ne ¼ 1.52(±0.1) 1024 cm 3 and kBT ¼ 9.3(±2.4) eV, whereas for the Debye-like model, the best ﬁt occurs for ne ¼ 4.8(±0.76) 1023 cm 3 and kBT ¼ 21(±2) eV. Again, cooling and decompression is predicted for later times. r [μm] r [μm] Figure 2 \| Radial proﬁle of the ionization equilibrium of carbon. Relative populations of carbon charge states in the target based on a degeneracy-corrected Saha–Boltzmann equation6. Continuum lowering is self-consistently accounted for according to the Stewart–Pyatt model35. The mass density and temperature proﬁles are given by radiation hydrodynamics simulations at t ¼ 3.4 ns. The vertical grey-shaded region represents the location of the bulk of the mass in the imploding shell, which dominates the scattering signal. Applied to the same conditions, the elastic scattering strength predicted by different screening models signiﬁcantly differs. In general, we ﬁnd that the Rayleigh peak is too large if the k-0 (Debye-like) limit is applied, whereas the amplitude given by ﬁnite-wavelength screening agrees with the measurements within the error bars (Fig. 3e). NATURE COMMUNICATIONS \| 6:6839 \| DOI: 10.1038/ncomms7839 \| www.nature.com/naturecommunications & 2015 Macmillan Publishers Limited. All rights reserved. NATURE COMMUNICATIONS \| 6:6839 \| DOI: 10.1038/ncomms7839 \| www.nature.com/naturecommunications & 2015 Macmillan Publishers Limited. All rights reserved. & 2015 Macmillan Publishers Limited. All rights reserved. In this regime, the tail of the screening cloud decays signiﬁcantly faster than the 1/k2-scaling predicted by the Debye-like theories Reviewing equation (5) reveals that the ratio k/ke determines whether the long-wavelength limit is applicable. For collective scattering29,30, the wavelength of density ﬂuctuations is larger than the screening length: a ¼ ke/k41. Accordingly, the long- wavelength limit is valid. Conversely, we have ao1 in the non- collective scattering regime, and the screening function arising from combining equations (3) and (4) should be used. In this regime, the tail of the screening cloud decays signiﬁcantly faster than the 1/k2-scaling predicted by the Debye-like theories Before we analyse the data using the screening models, we note that it is possible to obtain the Rayleigh weight directly from the 3 ARTICLE NATURE COMMUNICATIONS \| DOI: 10.1038/ncomms7839 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 xCj = nCj / Σj nCj 0 1 2 3 4 5 6 24 26 28 30 32 34 36 38 40 42 44 46 r [μm] Z f C = 1 Z f C = 2 Z f C = 3 Z f C = 4 Z f C = 5 Z f C = 6 Z f C = Σj xCj Z f Cj – Z f C – 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 xCj = nCj / Σj nCj 0 1 2 3 4 5 6 24 26 28 30 32 34 36 38 40 42 44 46 r [μm] Z f C = 1 Z f C = 2 Z f C = 3 Z f C = 4 Z f C = 5 Z f C = 6 Z f C = Σj xCj Z f Cj – Z f C – Figure 2 \| Radial proﬁle of the ionization equilibrium of carbon. Relative populations of carbon charge states in the target based on a degeneracy-corrected Saha–Boltzmann equation6. Continuum lowering is self-consistently accounted for according to the Stewart–Pyatt model35. The mass density and temperature proﬁles are given by radiation hydrodynamics simulations at t ¼ 3.4 ns. The vertical grey-shaded region represents the location of the bulk of the mass in the imploding shell, which dominates the scattering signal. \| \| \| & 2015 Macmillan Publishers Limited. All rights reserved. NATURE COMMUNICATIONS \| DOI: 10.1038/ncomms7839 NATURE COMMUNICATIONS \| DOI: 10.1038/ncomms7839 0.5 1.0 1.5 2.0 2.5 ne [1024 cm–3] kBT [eV] kBT [eV] kBT [eV] a Γee = 1 De = 1 0.00 0.08 0.16 0.24 0.32 b 0.00 0.07 0.14 0.21 0.29 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 c 0.00 0.07 0.13 0.20 0.26 p(χ2) ∝ e–2 p(χ2) ∝ e–2 p(χ2) ∝ e–2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 2Ps /Ωs [a.u.] 8.4 8.5 8.6 8.7 8.8 8.9 9.0 Exp. data ne0, T0 ne0, T = 0 ne0, T0 + T + ne0 – n– e ,T0 ne0 + n+ e,T0 d 0.0 0.4 0.8 1.2 1.6 2.0 2.4 8.7 8.8 8.9 9.0 9.1 9.2 Exp. data W exp R [k] FWS DH e hs [keV] hs [keV] 2Ps /Ωs [a.u.] ΓCC = 10 Γee = 1 De = 1 ΓCC = 10 Γee = 1 De = 1 ΓCC = 10 Figure 3 \| Statistical analysis comparing calculations and experimental data. (a–c) Colour maps of the probability density function for the w2 statistic, p(w2), comparing experimental data and theoretical calculations. (a) Results using the extracted Rayleigh amplitude according to equation (6), (b) using the ﬁnite-wavelength form given by equations (3) and (4) and (c) using the usual Debye-like approach equation (5). The 1s conﬁdence intervals for each plot are marked by the dashed black curves. Contours delineating regimes in which degeneracy and strong coupling become important are further shown by the green curves. Note that Debye-like screening predicted weakly coupled and weakly degenerate states, in contrast to the other approaches. (d) Sensitivity of the Compton feature within the errors given by equation (6). (e) Comparison of ideal and calculated Rayleigh features at the best ﬁt conditions from a, including largest errors at 1s. All panels relate to a pump–probe delay of t ¼ 3.4 ns. NATURE COMMUNICATIONS \| DOI: 10.1038/ncomms7839 (a) Results using the extracted Rayleigh amplitude according to equation (6), (b) using the ﬁnite-wavelength form given by equations (3) and (4) and (c) using the usual Debye-like approach equation (5). The 1s conﬁdence intervals for each plot are marked by the dashed black curves. Contours delineating regimes in which degeneracy and strong coupling become important are further shown by the green curves. Note that Debye-like screening predicted weakly coupled and weakly degenerate states, in contrast to the other approaches. (d) Sensitivity of the Compton feature within the errors given by equation (6). (e) Comparison of ideal and calculated Rayleigh features at the best ﬁt conditions from a, including largest errors at 1s. All panels relate to a pump–probe delay of t ¼ 3.4 ns. 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 WR[k] 10 20 30 40 50 60 70 Pid e [Mbar] Exp. value FWS + Coul. DH + Coul. DH + pseudo. 3.6 ns 3.5 ns 3.4 ns Figure 4 \| Results for the elastic scattering amplitudes using different models. Rayleigh weight WR for various pump–probe delays: extracted values from experimental data (red squares), results using ﬁnite wavelength of a Coulomb electron–ion interaction (blue diamonds), Debye–Hu¨ckel screening of a Coulomb potential (magenta circles) and a soft-core pseudopotential (orange triangles). The horizontal errors are propagated from the ﬁtting errors based on Wexp R , while the vertical errors also contain contributions from estimated modelling errors, as discussed in the Methods section. dynamics36, the nucleation times37 and relaxation properties10 of such states. Thus, our understanding of the particle and energy transport in stars or large planets as well as our modelling capabilities for inertial conﬁnement fusion experiments must be improved with respect to better screening models. As the concept of Debye–Hu¨ckel-like screening is also used to describe the electron dynamics in solids38, quark-gluon matter6,7 and charged macromolecules8, the application of our ﬁnding goes far beyond the physics of high-energy-density matter. 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 WR[k] 10 20 30 40 50 60 70 Pid e [Mbar] Exp. value FWS + Coul. DH + Coul. DH + pseudo. 3.6 ns 3.5 ns 3.4 ns & 2015 Macmillan Publishers Limited. All rights reserved. Here, we estimate this effect using a soft-core pseudopotential similar to the well-known empty-core approach for metals near room temperature28. The cutoff radius of the ionic core (rc ¼ 29 pm) is taken from the effective size of C þ 4 ions. The resulting weaker electron–ion interactions lead to oscillations in Fourier space, and negative values of the screening cloud in the range of wavenumbers under study. This effect is exacerbated by the weak decay of the Debye–Hu¨ckel/Thomas– Fermi dielectric function in k-space. Indeed, within this pseudopotential approach, we do not ﬁnd a distinct peak in the probability density. As demonstrated by Fig. 3, the ﬁtting procedure results in a well-deﬁned state with a clear peak in the probability density. The latter has a highly elongated shape along the temperature axis indicating much greater sensitivity to the electron density, as expected for partially degenerate conditions. For a pump–probe delay of 3.4 ns, the best ﬁt occurs, for example, at ne ¼ 1.31(±0.37) 1024 cm 3 and kBT ¼ 10.5( þ 6.5) eV. Extracted conditions for delays of 3.5 and 3.6 ns yield decreasing densities and temperatures consistent with cooling and expanding plasmas in the release states after shock compression. This trend agrees with the predictions of radiation hydrodynamics simulations. The XRTS data and their analyses clearly show the inapplic- ability of the long-wavelength approximation inherent to the Debye–Hu¨ckel and Thomas–Fermi theories of screening for the high-density matter under investigation. In contrast, the full k-dependence of the screening function, which shows a stronger decay for larger wavenumbers and Friedel oscillations in real space near the ion, provides better agreement with measurements under weakly coupled conditions. This ﬁnding has important implications when modelling processes with large momentum transfers. The errors on our measurements are quantiﬁed by the contour delineating the region in density–temperature space containing 68.3% of the total probability, that is, the 1-sigma conﬁdence interval. Using equation (6) to obtain Wexp R , we ﬁnd typical error bars of ±20–30% for the density, whereas only an upper limit, typically around 60%, can be rigorously deduced for the temperature due to the temperature insensitivity of the theoretical models for colder, strongly degenerate states. The sensitivity of Presently, Debye-like screening builds the basis for theoretical investigations of, for example, the phase diagram35, the ion 4 4 & 2015 Macmillan Publishers Limited. All rights reserved. ARTICLE NATURE COMMUNICATIONS \| DOI: 10.1038/ncomms7839 0.5 1.0 1.5 2.0 2.5 ne [1024 cm–3] kBT [eV] kBT [eV] kBT [eV] a Γee = 1 De = 1 0.00 0.08 0.16 0.24 0.32 b 0.00 0.07 0.14 0.21 0.29 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 c 0.00 0.07 0.13 0.20 0.26 p(χ2) ∝ e–2 p(χ2) ∝ e–2 p(χ2) ∝ e–2 ΓCC = 10 Γee = 1 De = 1 ΓCC = 10 Γee = 1 De = 1 ΓCC = 10 0.0 0.4 0.8 1.2 1.6 2.0 2.4 8.7 8.8 8.9 9.0 9.1 9.2 Exp. data W exp R [k] FWS DH e hs [keV] 2Ps /Ωs [a.u.] 0.0 0.2 0.4 0.6 0.8 1.0 1.2 2Ps /Ωs [a.u.] 8.4 8.5 8.6 8.7 8.8 8.9 9.0 Exp. data ne0, T0 ne0, T = 0 ne0, T0 + T + ne0 – n– e ,T0 ne0 + n+ e,T0 d hs [keV] d comparing calculations and experimental data. (a–c) Colour maps of the probability density function for the w2 i t l d t d th ti l l l ti ( ) R lt i th t t d R l i h lit d di t ti Figure 3 \| Statistical analysis comparing calculations and experimental data. (a–c) Colour maps of the probability density function for the w2 statistic, p(w2), comparing experimental data and theoretical calculations. (a) Results using the extracted Rayleigh amplitude according to equation (6), (b) using the ﬁnite-wavelength form given by equations (3) and (4) and (c) using the usual Debye-like approach equation (5). The 1s conﬁdence intervals for each plot are marked by the dashed black curves. Contours delineating regimes in which degeneracy and strong coupling become important are further shown by the green curves. Note that Debye-like screening predicted weakly coupled and weakly degenerate states, in contrast to the other approaches. (d) Sensitivity of the Compton feature within the errors given by equation (6). (e) Comparison of ideal and calculated Rayleigh features at the best ﬁt conditions from a, including largest errors at 1s. All panels relate to a pump–probe delay of t ¼ 3.4 ns. Figure 3 \| Statistical analysis comparing calculations and experimental data. (a–c) Colour maps of the probability density function for the w2 statistic, p(w2), comparing experimental data and theoretical calculations. References The physics of charge inversion in chemical and biological systems. Rev. Mod. Phys. 74, 329–345 (2002). Finally, bound-free transitions have been treated within the impulse approximation13 due to the large k values considered and result in a low amplitude tail on the red wing of the inelastic feature. Fortunately, there is only a small overlap between the bound-free and free-free components of the inelastic feature for Zf C ¼ 4. Thus, the ﬁtting procedure is not strongly sensitive to the bound-free contribution. 10. Gericke, D. O., Murillo, M. S. & Schlanges, M. Dense plasma temperature equilibration in the binary collision approximation. Phys. Rev. E 65, 036418 (2002). 11. Gericke, D. O. & Schlanges, M. Beam-plasma coupling effects on the stopping power of dense plasmas. Phys. Rev. E 60, 904–910 (1999). 12. Glenzer, S. H. et al. Demonstration of spectrally resolved x-ray scattering in dense plasmas. Phys. Rev. Lett. 90, 175002 (2003). Elastic scattering contributions. Besides the screening function, the modelling of elastic scattering requires the form factor for electrons bound to the carbon ions fC(k) and the partial ion–ion structure factors Sab(k). The form factor can be obtained with high precision using DFT calculations, and is also well described (see Fig. 1b) using hydrogenic wavefunctions with tabulated effective screening constants13. 13. Glenzer, S. H. & Redmer, R. X-ray Thomson scattering in high energy density plasmas. Rev. Mod. Phys. 81, 1625–1663 (2009). 14. Crowley, B. J. B. & Gregori, G. X-ray scattering in many-particle systems. New J. Phys. 15, 015014 (2013). y 15. Chihara, J. Interaction of photons with plasmas and liquid metals— photoabsorption and scattering. J. Phys. Condens. Matter 12, 231–247 (2000). The partial static structure factors of the ion subsystem were calculated from both DFT coupled to molecular dynamics26 and via a multicomponent generalization of the hypernetted-chain equations from the classical ﬂuid theory27. The structure predicted from DFT-molecular dynamics is well reproduced if the ion–ion potential is approximated with a Yukawa-like pseudopotential. We ﬁnd that ﬁnite-wavelength screening of the effective ion–ion interaction is not important, as the inter-ionic separation is sufﬁciently large at the densities considered for the k-0 approximation to be reasonable. Further considerations such as the additional short-range repulsion resulting from the electrons bound to the carbon ions can be included using a simple extension25. The latter yields negligible improvements owing to the small size of the effective core radius in helium-like carbon. References We ﬁnd very small deviations from the ideal gas values (unity or zero) for the k values probed by our experiment, making our analysis very robust against uncertainties in the ionic structure. Thus, the only signiﬁcant uncertainty lies in the description of the screening clouds. 16. Garcia-Saiz, E. et al. Probing warm dense lithium by inelastic x-ray scattering. Nat. Phys. 4, 940–944 (2008). y 17. Lee, H. J. et al. X-ray Thomson scattering measurements of density and temperature in shock-compressed beryllium. Phys. Rev. Lett. 102, 115001 (2009). 18. Regan, S. P. et al. Inelastic x-ray scattering from shocked liquid deuterium. Phys. Rev. Lett. 109, 265003 (2012). 19. Kritcher, A. L. et al. In-ﬂight measurements of capsule shell adiabats in laser- driven implosions. Phys. Rev. Lett. 107, 015002 (2011). 20. Fortmann, C. et al. Measurement of the adiabatic index in Be compressed by counterpropagating shocks. Phys. Rev. Lett. 108, 175006 (2012). 21. Fletcher, L. B. et al. Observations of continuum depression in warm dense matter with x-ray Thomson scattering. Phys. Rev. Lett. 112, 145004 (2014). y g y , ( ) 22. Riley, D. et al. Direct observation of strong coupling in a dense plasma. Phys. Rev. E 66, 046408 (2002). 23. Barbrel, B. et al. Measurement of short-range correlations in shock-compressed plastic by short-pulse x-ray scattering. Phys. Rev. Lett. 102, 165004 (2009). Evaluation of modelling uncertainties. As implied by equation (6), the value of Wexp R extracted from the experimental data is sensitive to the shape of the normalized X-ray source function S(o). We have considered time-dependent proﬁles extracted from the data21, and also models such as Gaussian, Lorenztian and Voigtian functions. We ﬁnd that the latter all require stronger elastic scattering to reproduce the amplitude of the observed peak, yielding typical uncertainties of B þ 10%. Moreover, we ﬁnd that the asymmetric shape of the extracted source function, which implicitly contains the depth-broadening effect of the HOPG crystal40, is crucial to simultaneously ﬁtting all spectral features of the data. 24. Kritcher, A. L. et al. Measurements of ionic structure in shock-compressed lithium hydride from ultrafast x-ray Thomson scattering. Phys. Rev. Lett. 103, 245004 (2009). 25. Ma, T. et al. X-ray scattering measurements of strong ion-ion correlations in shock-compressed aluminium. Phys. Rev. Lett. 110, 065001 (2013). 26. Wu¨nsch, K., Vorberger, J., Gregori, G. & Gericke, D. O. X-ray scattering as a probe for warm dense mixtures and high-pressure miscibility. References 1. Debye, P. & Hu¨ckel, E. Zur Theorie der Elektrolyte. I. 1. Debye, P. & Hu¨ckel, E. Zur Theorie der Elektrolyte. I. Gefrierpunktserniedrigung und verwandte Erscheinungen. Phys. Zeitschr. 24, 1. Debye, P. & Hu¨ckel, E. Zur Theorie der Elektrolyte. I. Gefrierpunktserniedrigung und verwandte Erscheinungen. Phys. Zeitschr. 24, 185–206 (1923). Gefrierpunktserniedrigung und verwandte Erscheinungen. Phys. Zeitschr. 24, 185–206 (1923). 2. Thomas, L. H. The calculation of atomic ﬁelds. Proc. Camb. Philos. Soc. 23, 542–548 (1927). Inelastic scattering contributions. Following the semi-classical approach of Chihara15, the dynamic structure factor is decomposed into three different contributions: the scattering from electrons following the ion motion (see equation (2)), scattering from free electrons in the continuum and scattering related to bound-free transitions driven by the incident X-rays13. The free-free contribution to the total inelastic scattering is given by the ﬂuctuation-dissipation theorem, which has been evaluated within the RPA5. 3. Fermi, E. A statistical method for the determination of some atomic properties and the application of this method to the theory of the periodic system of elements. Z. Phys. 48, 73–79 (1928). y 4. Ashcroft, N. W. & Mermin, N. D. Solid State Physics (Holt, Rinehart & Winston, 1976). Considerations beyond the RPA, such as nonlinear coupling between density ﬂuctuations and collisions with the ions, have also been investigated. We estimate strong coupling in the electronic subsystem using static local ﬁeld corrections5 and appeal to the Born–Mermin ansatz13 to include screened electron–ion collisions in ﬁrst Born approximation. We ﬁnd that nonideal effects beyond RPA yield only minor changes to our ﬁtting procedure. Collisions do not noticeably modify the Compton feature owing to strong Pauli blocking of electron–ion-scattering channels under the degenerate conditions of interest. As shown in Fig. 1b, local ﬁeld effects do not signiﬁcantly inﬂuence the free-electron response at large k. ll b d f h b d h h l 5. Kremp, D., Schlanges, M. & Kraeft, W. D. Quantum Statistics of Nonideal Plasmas (Springer, 2005). 6. Braaten, E. & Nieto, A. Next-to-leading order Debye mass for the quark-gluon plasma. Phys. Rev. Lett. 73, 2402–2404 (1994). 7. Heinz, U. Quark-gluon transport theory I. The classical theory. Ann. Phys. 161, 48–80 (1985). 8. Bannasch, G., Killian, T. C. & Pohl, T. Strongly coupled plasmas via Rydberg blockade of cold atoms. Phys. Rev. Lett. 110, 253003 (2013). 9. Yu., A., Grosberg, T. T., Nguyen & Shklovskii, B. I. NATURE COMMUNICATIONS \| DOI: 10.1038/ncomms7839 NATURE COMMUNICATIONS \| DOI: 10.1038/ncomms7839 a ¼ ke/kB0.3. Thus, for the relatively large k values probed, we expect non- collective scattering and signiﬁcant contributions from the screening cloud. a ¼ ke/kB0.3. Thus, for the relatively large k values probed, we expect non- collective scattering and signiﬁcant contributions from the screening cloud. References EPL 94, 25001 (2011). The strong angular divergence of the source can be accounted for by averaging calculations over a weighted distribution of scattering angles. Such a consideration has been shown to be important in the collective scattering regime29. Here, the scattering is non-collective for all scattering angles considered, and the net effect on the spectrum is a small degree of additional blurring in frequency space. A small impact on the Rayleigh weight stems from enhanced inelastic weight at o ¼ 0 and the explicit k-dependence of the terms of equation (2). We ﬁnd that changes to the functional form and width of the weighting distribution yield uncertainties that are well within an accuracy of approximately 5%. 27. Wu¨nsch, K., Hilse, K. P., Schlanges, M. & Gericke, D. O. Structure of strongly coupled multicomponent plasmas. Phys. Rev. E 77, 056404 (2008). 28. Gericke, D. O., Vorberger, J., Wu¨nsch, K. & Gregori, G. Screening of ionic cores in partially ionized plasma within linear response. Phys. Rev. E 81, 065401(R) (2010). 29. Glenzer, S. H. et al. Observation of plasmons in warm dense matter. Phys. Rev. Lett. 98, 065002 (2007). 30. Neumeyer, P. et al. Plasmons in strongly coupled and shock-compressed matter. Phys. Rev. Lett. 105, 075003 (2007). Finally, recent work41 for highly compressed plastic targets points to attenuation of the probe X-rays through the target as another potential source of error in WR. In the present experiment, the much smaller size and lesser degree of compression of the target at the time of measurement ensures that negligible attenuation. Furthermore, radiation hydrodynamics simulations of the implosion suggest that the driven shell of material yields a fairly uniform density distribution in the region that dominates the scattering21. 31. Friedel, J. Electronic structure of primary solid solutions in metals. Adv. Phys. 3, 446–507 (1954). 32. Ono, M. et al. Electrostatic potential screened by a two-dimensional electron system: a real-space observation by scanning-tunnelling spectroscopy. Phys. Rev. Lett. 96, 016801 (2006). 33. Bevington, P. R. & Robinson, D. K. Data Reduction and Error Analysis for the Physical Sciences (McGraw Hill, 2003). In summary, estimates of error sources have been determined to be on the order of emodB þ 10%/ 5%. These errors are combined with the statistical errors arising from the ﬁtting procedure by summing in quadrature, e2 ¼ e2 fit þ e2 mod. 34. Stewart, J. C. & Pyatt, Jr. K. D. Methods Experimental details. The experiments were performed at the Omega Laser Facility at LLE in Rochester, USA, which provides multiple high-energy beams of 351-nm ultraviolet laser light39. Eight laser beams with a total energy of 13.5 kJ were used to launch multiple shocks into a 70-mm-thin CH shell. The conditions created by the coalescent shocks were then probed with different pump–probe delays21. Figure 4 \| Results for the elastic scattering amplitudes using different models. Rayleigh weight WR for various pump–probe delays: extracted values from experimental data (red squares), results using ﬁnite wavelength of a Coulomb electron–ion interaction (blue diamonds), Debye–Hu¨ckel screening of a Coulomb potential (magenta circles) and a soft-core pseudopotential (orange triangles). The horizontal errors are propagated from the ﬁtting errors based on Wexp R , while the vertical errors also contain contributions from estimated modelling errors, as discussed in the Methods section. Figure 4 \| Results for the elastic scattering amplitudes using different models. Rayleigh weight WR for various pump–probe delays: extracted values from experimental data (red squares), results using ﬁnite wavelength of a Coulomb electron–ion interaction (blue diamonds), Debye–Hu¨ckel screening of a Coulomb potential (magenta circles) and a soft-core pseudopotential (orange triangles). The horizontal errors are propagated from the ﬁtting errors based on Wexp R , while the vertical errors also contain contributions from estimated modelling errors, as discussed in the Methods section. y Another set of eight laser beams illuminated a Zn foil, creating a hot plasma that emits He-a radiation with a strong peak around 9 keV. These X-rays were used to probe the compressed CH and recorded using a time-gated microchannel plate coupled to a high-resolution highly-oriented pyrolytic graphite (HOPG) crystal spectrometer under scattering angles of 135±15. A gold cone protected the spectrometer from direct illumination from the Zn plasma. For non-relativistic probe energies, the momentum transfer from the photons to the plasma, for example, the k values probed, is given by the incident X-ray energy Ei and the scattering angle y via kE(2Ei/:c)sin(y/2) (ref. 13). In the present experiment, we have k ¼ 8.4–9.2 Å 1. In comparison, the inverse screening length in the plasma is keB2.5 Å 1 for the densities and temperatures achieved in the target, yielding 5 & 2015 Macmillan Publishers Limited. All rights reserved. Competing ﬁnancial interests: The authors declare no competing ﬁnancial interests. Competing ﬁnancial interests: The authors declare no competing ﬁnancial interests. 40. Pak, A. et al. X-ray line measurements with high-efﬁciency Bragg crystals. Rev. Sci. Instrum. 75, 3747–3749 (2004). Reprints and permission information is available online at http://npg.nature.com/ reprintsandpermissions/ 41. Chapman, D. A. et al. Simulating x-ray Thomson scattering signals from high-density, millimetre scale plasmas at the National Ignition Facility. Phys. Plasmas 21, 082709 (2014). How to cite this article: Chapman, D.A. et al. Observation of ﬁnite-wavelength screening in high-energy-density matter. Nat. Commun. 6:6839 doi: 10.1038/ ncomms7839 (2015). ARTICLE ARTICLE NATURE COMMUNICATIONS \| DOI: 10.1038/ncomms7839 acknowledges support from the NLUF Grant DE-NA0000910 and DOE NNSA award DE-FG52-10 NA29649. 35. Hamaguchi, S., Faroukia, R. T. & Dunin, D. H. E. Phase diagram of Yukawa systems near the one-component-plasma limit revisited. J. Chem. Phys. 105, 7641–7647 (1996). Author contributions 36. Vorberger, J., Donko, Z., Tkachenko, I. M. & Gericke, D. O. Dynamic ion structure factor of warm dense matter. Phys. Rev. Lett. 109, 225001 (2012). D.A.C., J.V., R.A.B. and D.O.G. developed and applied the theory presented. J.V. performed the DFT-MD simulations. L.B.F., L.D., T.D., R.W.F., S.H.G., G.G., A.L.K., T.M. and A.E.P. carried out the experiment and the data were analysed by D.A.C., L.B.F. and T.M.G. D.A.C. and D.O.G. wrote the paper. S.H.G., G.G., O.L.L. and D.O.G. provided additional support for the experiment, data analysis and interpretation. 37. Daligault, J. Crystal nucleation in the one-component plasma. Phys. Rev. E 73, 056407 (2006). 38. Kaiser, A., Rethfeld, B., Vicanek, M. & Simon, G. Microscopic processes in 38. Kaiser, A., Rethfeld, B., Vicanek, M. & Simon, G. Microscopic processes in dielectrics under irradiation by subpicosecond laser pulses. Phys. Rev. B 61, 11437 (2000). p p dielectrics under irradiation by subpicosecond laser pulses. Phys. Rev. B 61, 11437 (2000). NATURE COMMUNICATIONS \| 6:6839 \| DOI: 10.1038/ncomms7839 \| www.nature.com/naturecommunications & 2015 Macmillan Publishers Limited. All rights reserved. Additional information 39. Boehly, T. R. et al. Initial performance results of the OMEGA laser system. Opt. Commun. 133, 495–506 (1997). References Lowering of ionization potentials in plasmas. Astrophys. J. 144, 1203–1211 (1966). NATURE COMMUNICATIONS \| 6:6839 \| DOI: 10.1038/ncomms7839 \| www.nature.com/naturecommunications NATURE COMMUNICATIONS \| 6:6839 \| DOI: 10.1038/ncomms7839 \| www.nature.com 6 & 2015 Macmillan Publishers Limited. All rights reserved. Acknowledgements This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/ D.A.C., R.A.B. and D.O.G. thank the UK’s EPSRC for support. Part of this work was performed by the assistance of Lawrence Livermore National Laboratory under Contract DE-AC52-07 NA27344 and supported by LDRD grant 11-ER-050, as well as the NLUF Grants DE-FG52-07 NA28057 and DE-FG52-09 NA29035. The work by SLAC HED was supported by DOE Ofﬁce of Science, Fusion Energy Science under FWP 100182. R.W.F. D.A.C., R.A.B. and D.O.G. thank the UK’s EPSRC for support. Part of this work was performed by the assistance of Lawrence Livermore National Laboratory under Contract DE-AC52-07 NA27344 and supported by LDRD grant 11-ER-050, as well as the NLUF Grants DE-FG52-07 NA28057 and DE-FG52-09 NA29035. The work by SLAC HED was supported by DOE Ofﬁce of Science, Fusion Energy Science under FWP 100182. R.W.F. D.A.C., R.A.B. and D.O.G. thank the UK’s EPSRC for support. Part of this work was performed by the assistance of Lawrence Livermore National Laboratory under Contract DE-AC52-07 NA27344 and supported by LDRD grant 11-ER-050, as well as the NLUF Grants DE-FG52-07 NA28057 and DE-FG52-09 NA29035. The work by SLAC HED was supported by DOE Ofﬁce of Science, Fusion Energy Science under FWP 100182. R.W.F. 7
https://openalex.org/W4380090290	https://www.researchsquare.com/article/rs-2978274/latest.pdf	English	null	Using Tree-Based Models to Identify Factors Contributing to Trait Negative Affect in Adults With and Without Major Depression	Research Square (Research Square)	2,023	cc-by	8,067	Using Tree-Based Models to Identify Factors Contributing to Trait Negative Affect in Adults With and Without Major Depression Catalina Canizares (  ccani007@fiu.edu ) Florida International University Yvonne Gomez Universidad de Los Andes Eugenio Ferro Instituto Colombiano del Sistema Nervioso Clínica Montserrat Carlos Arturo Torres Universidad de Los Andes Diana Maria Agudelo Universidad de Los Andes Gabriel Odom Florida International University Catalina Canizares (  ccani007@fiu.edu ) Florida International University Results The findings revealed that the cognitive schemas of disconnection and rejection and impaired autonomy had a significant impact on the likelihood of higher scores on the State Depression Inventory (IDER) test (p < 0.001), as indicated by both beta regression and regression tree analyses. Additionally, childhood adversity emerged as a crucial factor in determining high levels of NA. The regression tree model achieved strong performance metrics, including an R-squared value of 0.77. Objective Determine whether maladaptive cognitive schemas, attributional style, childhood adversity, and lifestyle factors (including alcohol and drug use and physical activity) could effectively predict negative affect (NA) in adults. Conclusions This study represents a significant step forward in the understanding of NA, as it considers a broad range of individual factors, such as cognitive schemas, lifestyle, and demographics, to predict its impact on NA, with potential implications for prevention programs aimed at reducing NA. Methods A secondary data analysis was performed on a sample of 342 depressed and non-depressed adults. Beta regression and regression tree analyses were conducted to identify the principal risk factors and their interactions. The regression tree model was trained with 5-fold cross-validation on 75% of the sample, with 25% of observations held for testing. Background Individuals with high levels of negative affect (NA) are at an increased risk of experiencing distress and negative self-views. Theoretical models suggest that NA plays a critical role in psychopathology, particularly in Major Depressive Disorder (MDD), and is linked to cognitive-perceptual and affective regulation issues. Keywords: License:   This work is licensed under a Creative Commons Attribution 4.0 International License. Read Full License Page 1/20 Page 1/20 Page 1/20 Page 1/20 Background Page 2/20 Page 2/20 Major Depressive Disorder (MDD) is a highly prevalent mental disorder worldwide, affecting nearly 280 million people globally and causing significant disability [1]. In Colombia, the most recent data from the National Survey of Mental Health [2] reports a prevalence rate of 4.3% for MDD, 1% for minor depression, and 0.5% for dysthymia, underscoring its prominence in the country. According to classic theoretical frameworks [3] and research [4], Negative Affect (NA) is a central component of MDD. NA refers to the tendency to experience unpleasant emotions, such as anger, guilt, worry, sadness, and disgust [3]. Individuals with high levels of NA are more likely to experience distress and hold a negative self-view, in contrast to those with Positive Affect (PA), who generally exhibit contentment and a positive self-perception [5]. NA has also been linked to specific personality dimensions [6]. Neuroticism is associated with high NA, while extraversion is associated with high PA [6]. Considering NA and PA as part of the dimensional understanding of personality allows for conceptualizing them in terms of both state and trait [6–8]. Trait refers to a relatively stable predisposition to respond in a certain way to situations, while state refers to emotional and mental conditions that are transitory [6]. Therefore, the interaction between state and trait characteristics enables a dimensional understanding of psychopathology, rather than simply considering the frequency of occurrence of symptoms [7, 9]. Therefore, the relationship between NA and psychopathology has been extensively studied, with numerous studies demonstrating a positive relationship between NA and various mental disorders [4, 10]. Moreover, research has established a correlation between NA and a range of cognitive-perceptual and affective regulation issues [4]. These issues include unconventional beliefs, distorted assessments of stimuli, and impaired decision-making [11], for instance, there is close association between NA and early maladaptive schemas. Early maladaptative schemas are rigid and long-standing negative beliefs about oneself and the world, which can develop early in life and can be perpetuated by ongoing negative experiences [12]. These negative beliefs can lead to negative affect, including feelings of sadness, anxiety, and anger and NA can in turn reinforce and perpetuate these negative beliefs, creating a cycle of negative thinking [12]. Given the significant impact and relevance of NA in psychopathology, particularly in MDD, several instruments have been developed to measure its presence. Background Among these, the IDER questionnaire -gets its name from its Spanish title, “Inventario de Depresión Estado Rasgo”- stands out as a valuable tool for evaluating the severity and variability of MDD symptoms, as well as the presence of NA [7, 9]. What sets the IDER questionnaire apart is its ability to measure both state and trait aspects of MDD, with a high level of sensitivity for detecting the presence of NA [7, 9]. It has proven useful in both clinical and research settings, particularly in areas such as personality, emotion, neuropsychology, and cognition [7]. Given the significant impact and relevance of NA in psychopathology, particularly in MDD, several instruments have been developed to measure its presence. Among these, the IDER questionnaire -gets its name from its Spanish title, “Inventario de Depresión Estado Rasgo”- stands out as a valuable tool for evaluating the severity and variability of MDD symptoms, as well as the presence of NA [7, 9]. What sets the IDER questionnaire apart is its ability to measure both state and trait aspects of MDD, with a high level of sensitivity for detecting the presence of NA [7, 9]. It has proven useful in both clinical and research settings, particularly in areas such as personality, emotion, neuropsychology, and cognition [7]. Importantly, the IDER questionnaire focuses on affective symptoms of depression, rather than somatic and physical aspects, making it a measure to evaluate the presence of NA rather than a diagnostic tool [9]. Additionally, the instrument has been tested for its psychometric properties in Colombian samples, demonstrating high levels of reliability (.71 to .86) and confirming the bifactorial structure (dysthymia and euthymia) in exploratory and confirmatory factorial analyses [6]. Importantly, the IDER questionnaire focuses on affective symptoms of depression, rather than somatic and physical aspects, making it a measure to evaluate the presence of NA rather than a diagnostic tool [9]. Additionally, the instrument has been tested for its psychometric properties in Colombian samples, demonstrating high levels of reliability (.71 to .86) and confirming the bifactorial structure (dysthymia and euthymia) in exploratory and confirmatory factorial analyses [6]. Page 3/20 Page 3/20 Considering that theoretical models suggest that NA plays a critical role in psychopathology especially in MDD and its correlation with cognitive schemas it is essential to identify factors that increase an individual’s likelihood of having higher levels of trait NA. Background Therefore, in this study, we aim to investigate the extent to which various factors, such as cognitive schemas, attributional style, childhood adversity, and lifestyle factors, including alcohol, drug use, and physical activity, can accurately predict trait NA in a sample of both depressed and non-depressed adults in Colombia using tree-based models. Recently, there have been applications of decision tree methods in mental health research given it automatizes detection of main effects and interactions [13]. The advantage of using tree-based models in this study is that they can handle complex relationships between predictors and outcomes, such as intricate interaction effects [14]. Additionally, tree-based models provide an interpretable and intuitive visualization method that can be easily communicated to clinicians [14]. By identifying factors that increase an individual’s likelihood of having higher levels of trait NA, this study may lead to more targeted interventions aimed at reducing NA and improving mental health outcomes in both depressed and non- depressed adults. Methods This study is a secondary data analysis utilizing data collected from a previous study conducted by Lattig and Collegues [15] that investigated the relationship between depression, genetics, and psychosocial variables. By leveraging the existing data, we aim to gain new insights into the factors contributing to NA scores, providing a complementary perspective to the original study. Participants The study recruited a total of 342 individuals, out of which 171 were diagnosed with Major Depressive Disorder (MDD) and recruited from two psychiatric hospitals in Bogota. The diagnosis was confirmed using the DSM IV-TR [16] criteria. To be included in the depressive group, participants had to be inpatients with MDD, over 18 years of age, and have completed at least basic primary education. Those with comorbid substance abuse or dependence, psychotic disorders, or dementia, as well as those with bipolar depression or delirium were excluded. The other 171 were healthy individuals from the general population who had no history of MDD, were over 18 years of age, and had completed basic primary education or higher. Participants were excluded from the control group if they had past or present mental disorders or a family relationship with a case subject. These diagnoses were also confirmed by the Mini-International Neuropsychiatric Interview M.I.N.I. [17]. Measures The research team designed a questionnaire to gather information on demographics, substance use, physical exercise and exposure to childhood adversity. Demographics. Childhood adversity. The personal questionnaire seeked information on the participant’s history of childhood abuse, specifically asking if they have experienced any form of physical and psychologival abuse and the age range in which it occurred. Attributional style. The Life Experiences Survey (LES) was the tool used to measure an individual’s attributional style in the face of stressful situations. The LES [18] is a 66-item self-report questionnaire designed to evaluate a broad range of stress-inducing life events that an individual has experienced in the past two years. The questionnaire assesses the frequency and perceived stress level of each event, with a rating scale ranging from 0 (not at all stressful) to 4 (highly stressful). In addition, the LES also evaluates the perceived positivity/negativity, expectation, and control of each event, providing a comprehensive understanding of an individual’s life experiences. The reliability of the LES has been confirmed with a Cronbach’s alpha of 0.82, and it has been shown to have normative data and validity through various studies [18]. Physical Exercise. The personal questionnaire gathered information on the participants’ exercise routines through questions on their weekly physical activity or sports participation. Demographics. Page 4/20 The personal questionnaire instrument gathered participants’ sex (male, female) and age. Substance use. The personal questionnaire also collected data on an individual’s alcohol, psychoactive substance, and cigarette usage. Participants were asked to report any usage in the past 30 days and the frequency of their daily consumption. Cognitive schemas. To measure early maladpatative schemas The Young Schema Questionnaire-Short Form (YSQ-SF) was used. The YSQ-SF comprises of 75 items and evaluates the presence of 15 schemas proposed by Jeffrey Young. Participants are asked to rate each of the 5 items on a six-point Likert scale, ranging from 1 (completely untrue of me) to 6 (describes me perfectly). A study conducted in a Colombian population confirmed the presence of the 15 schemas and demonstrated good psychometric properties, with reported Cronbach’s alpha values ranging from 0.73 to 0.88 for the different schemas [19]. Procedure The original study was approved by the ethics committees of the participating institutions and an independent research ethics committee. Written informed consent was obtained from those who were interested in participating. The M.I.N.I. structured interview [17] was administered by a trained psychologist on the research team to confirm inclusion criteria and rule out exclusion criteria. Eligible participants then completed a battery of questionnaires, administered by one of the trained psychologists on the research team. For this study the de-identified dataset was used. This dataset was previously curated to include only the psychosocial and cognitive variables. IDER. The IDER questionnaire is a tool used to assess NA for depression both in terms of its frequency (trait) and intensity (state) at the time of evaluation [7]. It consists of 20 items divided into two scales: Trait and State, each with 10 items, with 5 items for dysthymia and 5 items for euthymia. The subject’s responses Page 5/20 are scored based on their chosen answer option (1, 2, 3, or 4), with the score being assigned to the items related to dysthymia for both scales. For the items related to euthymia, the score is reversed. The final score of a scale is obtained by summing the results of the two sub-scales, with a range from 20 to 80. The reliability levels of the test are reported to be high, ranging from 0.71 to 0.92 for the different scales in the general population, with Cronbach’s alphas ranging from 0.71 to 0.86 in the Colombian population [6]. In this study, the trait scale was chosen as the outcome measure, low scores range from 20 to 30 and high scores range from 30 to 40. Data analysis where N is the sample size and s is a constant between 0 and 1. the s acts as if we were taking a prior into account so we decided to choose 0.5 conidering a Bayesian standpoint [21]. The beta regression was fitted using the betareg function from the betareg package in R 4.2.0. The regression tree was fitted using the tidymodels package in R 4.2.0. In preparation for modeling, the data was divided into training (75%) and testing sets (25%) using the initial_split function from the rsample. The division of the data was done to ensure that the model was trained on the training set and its performance could be accurately evaluated on the test set. The regression tree model was specified using the r_part engine in regression mode. To avoid overfitting the model, 5-cross-validation was performed using the vfold_cv function from the rsample package. To optimize the model, a grid search was performed to find the best cost complexity parameter (CP). The tune_grid function from the tune package was used to specify a grid of possible cost complexity parameters to test, and the fit function was used to fit the model for each value of the cost CP. The performance metrics, Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE), were used to compare the results for each value of the CP and select the best model. Finally, the best CP was selected and the final regression tree model was built using the entire training data and was then evaluated on the validation set. All code and analysis scripts are available in the authors GitHub repository. The results of the model fit is displayed in a tree structure (Fig. 1) and summarized using performance metrics. Data analysis To explore the factors associated with NA, we performed beta regression and regression tree analysis (CART) on the entire dataset without distinguishing between the depressed and non-depressed groups. By analyzing the whole sample as a single group, we were able to capture the full range of possible scores on the IDER scale and examine NA as a construct present in both groups. Considering that NA is a construct that is related to, yet different from, depression given that individuals may experience NA whether or not they are depressed [20]. Both Beta regression and CART were fitted to compare the results using different statistical techniques. The beta regression model provides a standard approach for mental health audiences to interpret the regression results, while the regression tree output is a useful tool for clinical decision-making as it presents decision heuristics and pathways that clinicians can use when working with clients who have high levels of NA. The beta regression model is suitable for identifying the most significant predictors of NA, while the regression tree model can help clinicians tailor their treatment plans based on individual client needs. Therefore, the two models provide complementary insights that can be used to enhance the understanding and management of NA in mental health settings. We chose to fit the data in a beta regression considering that the IDER score is conditionally distributed rather than Gaussian [21], meaning it is a continuous variable bounded between 10 and 40. To proceed the analysis we transformed the IDER scores to be between 0 and 1 using the following formula Page 6/20 x1 = x1 −lower upper −lower x1 = x1 −lower upper −lower where is a specific score, lower is the minimum score in the range of scores, in this case 10 and upper is the highest score meaning 40. x1 Given that the data contains values that are at the upper and lower bound it was necessary to apply the lemon squeezer [21] transformation to squeeze the data that lies in [0,1] to be in (0,1). The tranformation applied was x′ = x (N −1) + s N N where N is the sample size and s is a constant between 0 and 1. the s acts as if we were taking a prior into account so we decided to choose 0.5 conidering a Bayesian standpoint [21]. Sample Characteristics Sample Characteristics Table 1 presents the demographic and life style characteristics of the sample. The sample comprised 249 women and 93 men, with an average age of 35.4 years (median = 33; SD = 11.9). A total of 38.6% of participants reported experiencing childhood adversity. The majority of participants reported low levels of physical exercise (39.8%), cigarette consumption (19.9%), alcohol use (14.9%), and psychoactive drug use (3.2%). Page 7/20 Page 7/20 Table 1 Demographics and Life Style Variables for 342 Depressed and Non-depressed Participants Overall (N = 342) Sex Female 249 (72.8%) Male 93 (27.2%) Age Mean (SD) 35.4 (11.9) Median [Min, Max] 33.0 [18.0, 86.0] Childhood Adversity 132 (38.6%) Physical Excercise 136 (39.8%) Smoking Cigarettes 68 (19.9%) Alcohol Use 51 (14.9%) Psychoactive Substance Use 11 (3.2%) Table 1 Demographics and Life Style Variables for 342 Depressed and Non-depressed Participants Overall (N = 342) Sex Female 249 (72.8%) Male 93 (27.2%) Age Mean (SD) 35.4 (11.9) Median [Min, Max] 33.0 [18.0, 86.0] Childhood Adversity 132 (38.6%) Physical Excercise 136 (39.8%) Smoking Cigarettes 68 (19.9%) Alcohol Use 51 (14.9%) Psychoactive Substance Use 11 (3.2%) Table 2 presents the IDER trait score, scores for the five cognitive domains, and attributional styles. The IDER score for the sample had a mean score of 21.7 (median = 19; SD = 8.6). For the five cognitive domains, the results showed mean scores of 13.7 (SD = 6.11) for disconnection and rejection, 12.1 (SD = 5.49) for impaired autonomy, 15.2 (SD = 5.72) for impaired limits, 16.3 (SD = 5.63) for Other-Directedness, and 17.6 (SD = 4.74) for Over-Vigilance/Inhibition. Regarding attributional styles, more than half of participants (57.6%) reported Negative Attribution to stressful events, 50.3% reported Unexpected Attribution, and 21.3% reported Out of Control Attribution. Table 2 presents the IDER trait score, scores for the five cognitive domains, and attributional styles. The IDER score for the sample had a mean score of 21.7 (median = 19; SD = 8.6). For the five cognitive domains, the results showed mean scores of 13.7 (SD = 6.11) for disconnection and rejection, 12.1 (SD = 5.49) for impaired autonomy, 15.2 (SD = 5.72) for impaired limits, 16.3 (SD = 5.63) for Other-Directedness, and 17.6 (SD = 4.74) for Over-Vigilance/Inhibition. Regarding attributional styles, more than half of participants (57.6%) reported Negative Attribution to stressful events, 50.3% reported Unexpected Attribution, and 21.3% reported Out of Control Attribution. Sample Characteristics Page 8/20 Table 2 IDER, Maladpatative Cognitive Schemas and Cognitive Attibution Scores for 342 Depressed and Non- depressed Participants Overall (N = 342) IDER Score Mean (SD) 21.7 (8.66) Median [Min, Max] 19.0 [10.0, 40.0] Disconnection and Rejection Mean (SD) 13.7 (6.11) Median [Min, Max] 13.0 [5.00, 29.0] Impaired Autonomy Mean (SD) 12.1 (5.49) Median [Min, Max] 11.0 [5.00, 26.0] Impaired Limits Mean (SD) 15.2 (5.72) Median [Min, Max] 15.0 [5.00, 30.0] Other-Directedness Mean (SD) 16.3 (5.63) Median [Min, Max] 16.0 [5.00, 30.0] Over-Vigilance/Inhibition Mean (SD) 17.6 (4.74) Median [Min, Max] 18.0 [6.00, 30.0] Negative Attribution 197 (57.6%) Unexpected Attribution 172 (50.3%) Out of Control Attribution 73 (21.3%) Higher Scores for the Cognitive Schemas reflect more significant maladaptive schemas. IDER scores: 24–40 High, 17–22 Average, 16 or less Low Beta Regression The beta regression model results are presented in Table 3 To verify the model’s assumptions we visually IDER, Maladpatative Cognitive Schemas and Cognitive Attibution Scores for 342 Depressed and Non- depressed Participants Overall (N = 342) IDER Score Mean (SD) 21.7 (8.66) Median [Min, Max] 19.0 [10.0, 40.0] Disconnection and Rejection Mean (SD) 13.7 (6.11) Median [Min, Max] 13.0 [5.00, 29.0] Impaired Autonomy Mean (SD) 12.1 (5.49) Median [Min, Max] 11.0 [5.00, 26.0] Impaired Limits Mean (SD) 15.2 (5.72) Median [Min, Max] 15.0 [5.00, 30.0] Other-Directedness Mean (SD) 16.3 (5.63) Median [Min, Max] 16.0 [5.00, 30.0] Over-Vigilance/Inhibition Mean (SD) 17.6 (4.74) Median [Min, Max] 18.0 [6.00, 30.0] Negative Attribution 197 (57.6%) Unexpected Attribution 172 (50.3%) Out of Control Attribution 73 (21.3%) Higher Scores for the Cognitive Schemas reflect more significant maladaptive schemas. IDER scores: 24–40 High, 17–22 Average, 16 or less Low Beta Regression Page 9/20 The beta regression model results are presented in Table 3.To verify the model’s assumptions, we visually inspected the beta regression using R’s plot function. The generated graphics revealed that all The beta regression model results are presented in Table 3.To verify the model’s assumptions, we visually inspected the beta regression using R’s plot function. The generated graphics revealed that all assumptions were well within the acceptable range, the residuals appeared to be randomly distributed around zero, and the Cook’s distance plot did not indicate the presence of influential observations. Therefore, the beta regression model is valid for interpreting our data. The plots are available at the authors GitHub repository and in the appendix. Sample Characteristics assumptions were well within the acceptable range, the residuals appeared to be randomly distributed around zero, and the Cook’s distance plot did not indicate the presence of influential observations. Therefore, the beta regression model is valid for interpreting our data. The plots are available at the authors GitHub repository and in the appendix. We also conducted a variance inflation factor (VIF) analysis to detect multicollinearity in the beta regression model. The results indicated that most predictor variables had VIF values below 3.9, indicating a moderate degree of multicollinearity. However, one variable had a VIF of 4.9, which could suggest a stronger correlation with other predictor variables in the model. Nevertheless, since none of the VIF values exceeded the recommended threshold of 5 or 10 [22], we concluded that the collinearity was not severe enough to affect our analysis’s overall findings. Page 10/20 Table 3 Table 3 Beta Regression for the IDER Score in a Sample of 255 Depressed and Non-depressed Adults Table 3 Beta Regression for the IDER Score in a Sample of 255 Depressed and Non-depressed Adults IDER Predictors Estimates CI p (Intercept) 0.16 0.07–0.35 < 0.001 Age 0.99 0.98–1.00 0.011 Disconnection and Rejection 1.07 1.03–1.10 < 0.001 Impaired Autonomy 1.08 1.05–1.12 < 0.001 Impaired Limits 1.02 0.99–1.04 0.126 Other-Directedness 1.00 0.98–1.03 0.939 Over-Vigilance/Inhibition 1.00 0.98–1.03 0.962 Number of Stressful Events 1.00 0.99–1.01 0.785 Sex [Male] 1.07 0.88–1.31 0.477 Negative Attribution [No] 0.66 0.53–0.83 < 0.001 Unexpected Attribution [No] 1.04 0.85–1.28 0.675 Out of Control Attribution [No] 0.74 0.59–0.94 0.012 Childhood Adversity [No] 0.71 0.59–0.86 < 0.001 Physical Excercise [No] 1.03 0.86–1.23 0.773 Smoking Cigarettes [No] 0.75 0.60–0.95 0.015 Alcohol Use [No] 0.95 0.74–1.23 0.707 Psychoactive Substance Use [No] 1.43 0.86–2.38 0.165 Observations 342 R2 0.596 The predictors age, disconnection and rejection, impaired autonomy, negative attribution, out of control attribution, childhood adversity, and smoking cigarettes, are all statistically significant at the 0.05 level, meaning that it is unlikely to have arisen by chance. The negative coefficients for Negative Attribution, Out of Control Attribution, Childhood Adversity, and Smoking Cigarettes suggest that not having these predictors is associated with a decrease in the log odds of experiencing NA (These coefficients are in comparison to the “yes” reference group). In contrast, having these predictors is associated with an increase in the log odds of experiencing NA. Sample Characteristics Additionally, a negative coefficient for age suggests that as age increases, the odds of experiencing NA decreases. The results also show that disconnection and rejection and impaired autonomy are significantly associated with the odds of experiencing NA. Each one unit increase in disconnection and rejection increases the log odds of experiencing NA by 0.06 units, while each one unit increase in impaired autonomy increases the log odds of experiencing NA by 0.08 units (in a scale from 0 to 1). Table 3 displays the precision results for the mean model, which includes the pseudo R-squared of 0.60 that demonstrates a strong fit for the model as noted by Giselmar and collegues [23]. However, it is important to keep in mind that the pseudo R-squared should not be directly compared to the R-squared of linear regression models. This is because the former only measures how well the model fits the data, and does not estimate the accounted variability. Regression tree results Regression tree results Tree-based methods are a type of predictive modeling that seeks to identify similar groups of people based on their outcomes using a series of yes/no questions. These questions are plotted in a graphical format as a tree or bush, with the root representing the initial question and subsequent branches representing subsequent questions. To make predictions for a new person, the tree is traversed from the root to the end of the last branch by answering the yes/no questions. Questions that appear close to the root or frequently throughout the tree are more important for grouping people with similar outcomes. Tree-based methods can produce very complex trees with small subgroups of people, which can be simplified by pruning or removing certain branches. The optimal amount of pruning is determined by testing different values of the cost complexity parameter (CP) on different parts of the data and choosing the value that results in the best performance on a separate set of data. By simplifying the tree, the model can be made more interpretable and easier to use in practice. The optimal CP parameter was found to be 0.001, as it produced the highest R-squared value and the lowest root mean square error (RMSE) value among all parameters tested. Using the value of CP the regression tree model in the testing data achieved an R-squared value of 0.7744057, which measures the proportion of variance in the IDER score that is explained by the predictors selected. A value of 0.7744057 indicates that the model is explaining a significant portion of Page 12/20 Page 12/20 the variance in the IDER score. Additionally, the RMSE of 4.01031 suggests that the model is providing reasonably accurate predictions for the IDER score. the variance in the IDER score. Additionally, the RMSE of 4.01031 suggests that the model is providing reasonably accurate predictions for the IDER score. Figure 1 displays the results of the regression tree analysis. The final decision tree model incorporates nine variables, including three cognitive domains: disconnection and rejection (dyrysq), impaired autonomy (padysq), and impaired limits (seiysq); two attributional styles: negative attribution (is_negative) and out of control attribution (is_no_control); two lifestyle variables: alcohol (is_alcohol) use and smoking cigarettes (is_smoke); and two demographic variables: age and childhood adversity (is_abuse). The root node of the tree includes 255 observations, with a mean IDER score of 21.88. Discussion This study aimed to examine the extent to which cognitive schemas, lifestyle factors, and demographic characteristics could accurately predict NA, as measured by the IDER questionnaire. It employed both Beta Regression and CART as quantitative methods to evaluate and compare the factors associated with NA in depressed and non-depressed participants. Our results indicate that both methods agree on the significant impact of cognitive schemas, attributional styles, lifestyle, and demographic variables in predicting NA. Furthermore, both models effectively adapt to the data, with the regression tree accounting for a considerable proportion of the variance in negative affect. These findings suggest that three maladaptive cognitive schemas, namely disconnection and rejection, impaired autonomy, and impaired limits, are strongly associated with higher levels of NA. Disconnection and rejection refers to a pattern of negative self-beliefs that lead an individual to feel disconnected and rejected by others. Impaired autonomy refers to a pattern of negative self-beliefs that lead an individual to feel like they lack control over their lives and are unable to make choices for themselves. Impaired limits refers to a pattern of negative beliefs and attitudes related to a lack of sense of self-control and an inability to set and maintain personal and interpersonal boundaries. These results are consistent with previous research that has shown that the five domains of maladaptive schemas, particularly impaired autonomy and disconnection and rejection, are closely related to the severity of depressive symptoms (Chen et al., 2019). This supports the idea that higher NA is related to more persistent and severe symptoms. Therefore, understanding the presence of cognitive schemas, such as these three maladaptive schemas, may be crucial in comprehending the impact of high levels of NA and the corresponding persistence and severity of depressive symptoms. Furthermore, this study emphasizes the significance of cognitive schemas over other factors in an individual’s life, as it demonstrates that the most critical factor in determining a higher score was a cognitive schema. While other variables were included in the model after the cognitive schema, this underscores the importance of therapies that aim to improve the flexibility of one’s interpretation and perception of life to achieve better outcomes and reduce negative feelings associated with NA. The results of this study suggest that negative attribution and the perception of uncontrollability during stressful situations contribute to higher levels of NA. Regression tree results Page 13/20 Page 13/20 In the end, both methods identify the same predictors as crucial in determining the IDER score’s values, but they complement each other by revealing further interactions. Additionally, the tree delves deeper and shows how alcohol consumption may also be relevant in determining the score. In the end, both methods identify the same predictors as crucial in determining the IDER score’s values, but they complement each other by revealing further interactions. Additionally, the tree delves deeper and shows how alcohol consumption may also be relevant in determining the score. Regression tree results The sisconnection and rejection variable (dyrysq) is the most critical factor in determining high or low IDER scores, with a cutoff value of 17. The tree also features three other main splits based on impaired autonomy (padysq), childhood adversity (is_abuse), and negative attribution (is_negative). These variables are crucial in determining high or low levels of NA. The decision tree indicates that a score of at least 35 points in the IDER instrument (split 63 in Fig. 1) is predicted by the interaction of a 16 or higher score in the disconnection and rejection cognitive domain (split 1), a score higher than 13 in the impaired autonomy schema domain (split 3), and not consuming alcohol (split 15) but smoking cigarettes (split 31). On the other hand, lower scores are determined by having low scores in the disconnection and rejection schema, lower scores in the impaired autonomy domain, not interpreting stressful situations as negative, and being 48 years of age or older. In summary, the decision tree depicts several interactions between cognitive, lifestyle, and demographic variables associated with higher levels of NA. Upon analyzing the results of the beta regression and regression tree, it is apparent that there is a correspondence between the two methods, and they provide complementary information. The beta regression demonstrates that cognitive schemas such as disconnection and rejection, and impaired autonomy significantly increase the likelihood of higher scores in the IDER test (p < 0.001). These two schemas’ effects are evident in the tree as they are the first two splits, indicating their higher importance in estimating higher levels of NA. The regression tree reveals an additional schema that is not significant in the beta regression, impaired autonomy. In combination with high levels of disconnection and rejection, impaired autonomy predicts higher scores in the IDER (split 14). Furthermore, both methods agree that childhood adversity’s presence or absence is critical in determining high or low levels of NA. The beta regression suggests that not being subjected to childhood adversity decreases the chances of high scores in the IDER, and the tree regression complements this by showing that low scores are possible in the absence of childhood adversity, but this is only achievable with the presence of interpreting stressful events as under control, not consuming alcohol, and lower scores in the disconnection and rejection schema. Discussion This finding is consistent with previous research that suggests that attributional style is more a product of current mood than a trait-like mode of thinking that increases susceptibility to clinical depression. This is supported by the stronger association between attributional style and current mood found in previous studies [24]. Furthermore, other studies have linked the interpretation of uncontrollability to the severity of depressive symptoms, indicating that the Page 14/20 Page 14/20 relationship between the perception of uncontrollability and negative affect may be influenced by the level of trait negative affectivity [25, 26]. Another significant finding of this study is the impact of childhood adversity on the likelihood of higher NA. The results showed that childhood adversity was a significant predictor in both models, and the decision tree analysis revealed that being a victim of childhood adversity in association with disconnection and rejection, as well as impaired autonomy, led to a score of approximately 26 in the IDER questionnaire, indicating moderate to high levels of trait NA. This finding aligns with previous research linking childhood adversity to the enduring experience of negative emotions and recurrent and persistent depressive episodes ([27]). It highlights the importance of early intervention and support for individuals who have experienced childhood adversity to decrease the negative impact on their emotional well-being, overall mental health and ideally, prevention of mental disorders like MDD. Finally, the results of the study demonstrate a significant relationship between smoking and alcohol use and NA. This finding is consistent with previous evidence showing strong associations between smoking and mental health disorders characterized by persistent NA [28]. Indeed, a large body of literature supports the positive association between smoking status and neuroticism [28], which is a disposition to experience negative affects such as anger and anxiety [29]. As for alcohol use, the tree-based model revealed that it was a relevant factor at two different splits to predict the IDER score. Previous research has shown that higher levels of NA are specifically linked to the initiation and relapse of alcohol and other substance use disorders [30–33]. Moreover, the use of two different quantitative methods to explore the relationship between NA and the introduced predictors is an innovative approach to understanding psychological constructs. This method offers evidence of replicability and internal validity, as both methods agree on most of the significant predictors. Conclusions In conclusion, our study supports and adds to the previous evidence linking NA to cognitive schemas, attributional style, childhood adversity, and lifestyle behaviors by demonstrating the interaction between these predictors. Our model provides a comprehensive picture of how the combination of these variables can accurately predict higher levels of NA thus contributing to a deeper understanding of the development and maintenance of MDD. The results of this study hold significant practical implications for clinical settings. The use of machine learning techniques, as demonstrated in this study, can offer valuable insights for predicting and making decisions by providing clinicians with pathways that lead to more severe levels of NA and possible poor outcomes for the mental disorders it has been related to. If clinicians are provided with these types of tools that contain heuristics specific to particular populations and problems, it could increase awareness of critical risk factors for the persistence and maintenance of depressive disorders. This research represents a significant advancement in the understanding of NA because it takes into account various aspects of an individual, such as their cognitive schemas and interpretations, lifestyle, and demographics, to show that the interaction of these three factors is crucial in determining high levels of NA and, consequently, more persistent symptoms and greater vulnerability to present depression. Discussion Moreover, the tree method provides clinicians with a way to identify which factors related to others are protective or riskier, and represents critical trends that contribute to higher NA scores. This can help clinicians make more informed decisions when designing treatment plans to improve outcomes for patients. There are several limitations to this study that should be considered. First, it is important to note that this study is a secondary data analysis and the data was not collected specifically to answer the research question being investigated. As a result, the external validity of the study may be limited, as the sample only includes a specific population of depressed individuals and their matched healthy controls. Another limitation is that the data was collected cross-sectionally, which means that causality cannot be established between the predictors and the outcome. Future studies on negative affect should consider longitudinal designs that take into account cultural diversity and include measurements of other psychosocial and medical factors. In order to improve the generalizability of the findings, future research should aim to use a more diverse sample that includes a wider range of individuals and populations. Additionally, longitudinal studies would provide more information on the development and persistence of negative affect over time, which could help to identify potential intervention points. Finally, a more Page 15/20 Page 15/20 comprehensive measurement of psychological and medical factors could help to identify other predictors of negative affect, and to better understand the complex interplay between these factors. comprehensive measurement of psychological and medical factors could help to identify other predictors of negative affect, and to better understand the complex interplay between these factors. Authors’ Contributions CC and YG: Wrote the first and last draft of the manuscript and all authors contributed to and have approved the final manuscript. CC: Designed the study and conducted the statistical analysis. YG: Supervised, reviewed and edited the manuscript. CAT: Conducted literature searches and provided summaries of previous research studies. EF and DMA: Reviewed and edited the manuscript. GO: Guided the application of statistical, mathematical, computational, and other formal techniques to analyze and synthesize the data. Acknowledgements We would like to express our gratitude to Minciencias (712.2015/908 /120471250970), to the Universidad of Los Andes, our assistants, and the psychiatry Institutions. Finally, this research would not have been possible without the voluntary cooperation of all the patients and people that participated in this study. Declarations Ethics Approval and Consent to Participate This study was approved by the Ethics Committee from Los Andes University, the Ethics Committee from Instituto Colombiano del Sistema Nervioso Clínica Montserrat, the Ethics Committee from Clínica la Inmaculada, and the CEI Campo Abierto Comité de Ética en Investigación. All methods were carried out adhering to the Declaration of Helsinki. Written informed consent was obtained from each participant included in the study. The participants voluntarily joined the study without receiving any financial compensation. Consent for Publication Not applicable. Availability of Data and Materials The datasets generated and analyzed in the current study are available from the corresponding author on reasonable request. Page 16/20 Page 16/20 Competing Interests The authors declare that they have no competing interests. This study was financed by a research grant from the Colombian Ministry of Science Minciencias (712.2015/908 /120471250970) . Neither the design, the collection, the analysis and interpretation of data nor the manuscript and submission of the article partook of the participation of the financing entity. Authors’ Contributions References 1. WHO. Depression [Internet]. World Health Organization. World Health Organization. ; 2021. Available from: https://www.who.int/news-room/fact-sheets/detail/depression. 1. WHO. Depression [Internet]. World Health Organization. World Health Organization. ; 2021. Available from: https://www.who.int/news-room/fact-sheets/detail/depression. 1. WHO. Depression [Internet]. World Health Organization. World Health Organization. ; 2021. Available from: https://www.who.int/news-room/fact-sheets/detail/depression. 2. Encuesta nacional de salud mental (tomo i). Ministerio de Salud y Protección Social; 2015. 2. Encuesta nacional de salud mental (tomo i). Ministerio de Salud y Protección Social; 2015. 2. Encuesta nacional de salud mental (tomo i). Ministerio de Salud y Protec 3. Clark LA, Watson D. Tripartite model of anxiety and depression: Psychometric evidence and taxonomic implications. J Abnorm Psychol. 1991;100(3):316–36. 3. Clark LA, Watson D. Tripartite model of anxiety and depression: Psychometric evidence and taxonomic implications. J Abnorm Psychol. 1991;100(3):316–36. 4. Jeronimus BF, Kotov R, Riese H, Ormel J. Neuroticism’s prospective association with mental disorders halves after adjustment for baseline symptoms and psychiatric history, but the adjusted association hardly decays with time: A meta-analysis on 59 longitudinal/prospective studies with 443 313 participants. Psychol Med. 2016;46(14). 4. Jeronimus BF, Kotov R, Riese H, Ormel J. Neuroticism’s prospective association with mental disorders halves after adjustment for baseline symptoms and psychiatric history, but the adjusted association hardly decays with time: A meta-analysis on 59 longitudinal/prospective studies with 443 313 participants. Psychol Med. 2016;46(14). 5. Watson D, Clark LA. Negative affectivity: The disposition to experience aversive emotional states. Psychol Bull. 1984;96(3):465–90. 6. Agudelo Vélez DM, Gómez Maquet Y, López PL. Propiedades psicométricas del inventario de depresión estado rasgo (IDER) con una muestra de población general colombiana. Av en Psicología Latinoam. 2014;32(1):71–84. Page 17/20 Page 17/20 7. Buela-Casal G, Agudelo Vélez DM. IDER inventario de depresión estado-rasgo. S.A.: TEA Ediciones; 2008. 8. Ramos J, Sánchez A, Doll A. Personalidad, afecto y estilo de afrontamiento: Interacciones en trastorno de personalidad grave. Behav Psychology/Psicología Conductual. 2021;29(3):699–719. 8. Ramos J, Sánchez A, Doll A. Personalidad, afecto y estilo de afrontamiento: Interacciones en trastorno de personalidad grave. Behav Psychology/Psicología Conductual. 2021;29(3):699–719. 9. Spielberger D, Carretero-Dios H, De los Santos Roig M, Gualberto B. Spanish experimental version of the state-trait depression questionnaire (ST-DEP): State sub-scale (s-DEP). Int J Clin Health Psychol. 2002 Jan;2. 10. Stanton K, Watson D. Positive and negative affective dysfunction in psychopathology. Social and Personality Psychology Compass [Internet]. 2014;8(9):555–67. Available from: https://compass.onlinelibrary.wiley.com/doi/abs/10.1111/spc3.12132. 10. Stanton K, Watson D. Positive and negative affective dysfunction in psychopathology. References Social and Personality Psychology Compass [Internet]. 2014;8(9):555–67. Available from: https://compass.onlinelibrary.wiley.com/doi/abs/10.1111/spc3.12132. 11. Cacioppo J, Berntson G. The affect system: Architecture and operating characteristics. Curr Dir Psychol Sci. 1999 Oct;8:133–7. 11. Cacioppo J, Berntson G. The affect system: Architecture and operating characteristics. Curr Dir Psychol Sci. 1999 Oct;8:133–7. 12. Young JE, Klosko JS, Weishaar ME. Schema therapy: A practitioner’s guide. New York: Guilford Publications; 2003. 13. Jiang T, Gradus JL, Rosellini AJ. Supervised machine learning: A brief primer. Behavior Therapy [Internet]. 2020;51(5):675–87. Available from: https://www.sciencedirect.com/science/article/pii/S0005789420300678. 13. Jiang T, Gradus JL, Rosellini AJ. Supervised machine learning: A brief primer. Behavior Therapy [Internet]. 2020;51(5):675–87. Available from: https://www.sciencedirect.com/science/article/pii/S0005789420300678. 14. Neumann M, Klatt T. Identifying predictors of inpatient verbal aggression in a forensic psychiatric setting using a tree-based modeling approach. J interpers Violence. 2021 Jun;37:1–26. 14. Neumann M, Klatt T. Identifying predictors of inpatient verbal aggression in a forensic psychiatric setting using a tree-based modeling approach. J interpers Violence. 2021 Jun;37:1–26. 15. Latigg M, Villamizar V, Gaviria A, Ángel J, Cañizares C, Ferro E, et al. BDNF methylation and stress response in a clinical population with major depressive disorder. Biol Psychiatry. 2017 May;81:96. 15. Latigg M, Villamizar V, Gaviria A, Ángel J, Cañizares C, Ferro E, et al. BDNF methylation and stress response in a clinical population with major depressive disorder. Biol Psychiatry. 2017 May;81:96. 16. American Psychiatric Association. Diagnostic and statistical manual of mental disorders: DSM-IV- TR. 4th ed., text rev. Washington, DC: Autor; 2000. 16. American Psychiatric Association. Diagnostic and statistical manual of mental disorders: DSM-IV- TR. 4th ed., text rev. Washington, DC: Autor; 2000. 17. Ferrando L, Bobes J, Giber J, Soto M, Soto O. M.i.n.i. Mini international neuropsychiatric interview. Versión en español 5.0.0. DSM-IV. 2000;2023(03). Available from: 17. Ferrando L, Bobes J, Giber J, Soto M, Soto O. M.i.n.i. Mini international neuropsychiatric interview. Versión en español 5.0.0. DSM-IV. 2000;2023(03). Available from: https://www.fundacionforo.com/pdfs/mini.pdf. 17. Ferrando L, Bobes J, Giber J, Soto M, Soto O. M.i.n.i. Mini international neuropsychiatric interview. Versión en español 5.0.0. DSM-IV. 2000;2023(03). Available from: https://www.fundacionforo.com/pdfs/mini.pdf. 18. Sandín B, Chorot P. Cuestionario de sucesos vitales (CSV): Estructura factorial, características psicométricas y datos normativos. Revista de Psicopatología y Psicología Clínica [Internet]. 2017;22(2):95–115. Available from: https://revistas.uned.es/index.php/RPPC/article/view/19729. 18. Sandín B, Chorot P. Cuestionario de sucesos vitales (CSV): Estructura factorial, características psicométricas y datos normativos. Revista de Psicopatología y Psicología Clínica [Internet]. 2017;22(2):95–115. Available from: https://revistas.uned.es/index.php/RPPC/article/view/19729. 19. References Londoño N, Schnitter M, Marín C, Calvete E, Ferrer A, Maestre K, et al. Young schema questionnaire- short form: Validación en colombia. Universitas Physiol. 2012 Mar;11:147–64. 19. Londoño N, Schnitter M, Marín C, Calvete E, Ferrer A, Maestre K, et al. Young schema questionnaire- short form: Validación en colombia. Universitas Physiol. 2012 Mar;11:147–64. 20. Danhauer SC, Legault C, Bandos H, Kidwell K, Costantino J, Vaughan L et al. Positive and negative affect, depression, and cognitive processes in the cognition in the study of tamoxifen and raloxifene (co-STAR) trial. Aging, Neuropsychology, and Cognition [Internet]. 2013;20(5):532–52. Available from: https://doi.org/10.1080/13825585.2012.747671. 20. Danhauer SC, Legault C, Bandos H, Kidwell K, Costantino J, Vaughan L et al. Positive and negative affect, depression, and cognitive processes in the cognition in the study of tamoxifen and raloxifene (co-STAR) trial. Aging, Neuropsychology, and Cognition [Internet]. 2013;20(5):532–52. Available from: https://doi.org/10.1080/13825585.2012.747671. 21. Smithson M, Verkuilen J. A better lemon squeezer? Maximum-likelihood regression with beta- distributed dependent variables. Psychol Methods. 2006 Apr;11:54–71. 21. Smithson M, Verkuilen J. A better lemon squeezer? Maximum-likelihood regression with beta- distributed dependent variables. Psychol Methods. 2006 Apr;11:54–71. Page 18/20 Page 18/20 22. James G, Witten D, Hastie T, Tibshirani R. An introduction to statistical learning: With applications in r [Internet]. Springer; 2013. Available from: https://faculty.marshall.usc.edu/gareth-james/ISL/. 23. Hemmert GAJ, Schons LM, Wieseke J, Schimmelpfennig H. Log-likelihood-based pseudo-R2 in logistic regression: Deriving sample-sensitive benchmarks. Sociological Methods & Research [Internet]. 2018;47(3):507–31. Available from: https://doi.org/10.1177/0049124116638107. 24. Ball HA, McGuffin P, Farmer AE. Attributional style and depression. Br J Psychiatry. 2008;192(4):275– 8. 24. Ball HA, McGuffin P, Farmer AE. Attributional style and depression. Br J Psychiatry. 2008;192(4):275– 8. 25. Usuga Jerez AJ, Lemos Ramírez NV, Pinzón Ardila JL, Pérez Rivero PF, Uribe Rodríguez AF. Sucesos vitales estresantes, ansiedad y depresión en estudiantes de una universidad privada de bucaramanga. Informes Psicológicos [Internet]. 2021;21(2):61–74. Available from: https://revistas.upb.edu.co/index.php/informespsicologicos/article/view/7204. 26. Camuñas N, Mavrou E, Miguel Tobal JJ. Ansiedad y tristeza-depresión: Una aproximación desde la teoría de la indefensión-desesperanza. Revista de Psicopatología y Psicología Clínica [Internet]. 2019;24(1). Available from: https://revistas.uned.es/index.php/RPPC/article/view/23003. 27. Nanni V, Uher R, Danese A. Childhood maltreatment predicts unfavorable course of illness and treatment outcome in depression: A meta-analysis. Am J Psychiatry. 2012;169(2):141–51. 27. Nanni V, Uher R, Danese A. Childhood maltreatment predicts unfavorable course of illness and treatment outcome in depression: A meta-analysis. Am J Psychiatry. 2012;169(2):141–51. 28. Kassel JD, Stroud LR, Paronis CA. References Smoking, stress, and negative affect: Correlation, causation, and context across stages of smoking. Psychol Bull. 2003;129(2):270–304. 28. Kassel JD, Stroud LR, Paronis CA. Smoking, stress, and negative affect: Correlation, causation, and context across stages of smoking. Psychol Bull. 2003;129(2):270–304. 29. Widiger TA, Oltmanns JR. Neuroticism is a fundamental domain of personality with enormous public health implications. World Psychiatry. 2017;16(2):144–5. 29. Widiger TA, Oltmanns JR. Neuroticism is a fundamental domain of personality with enormous public health implications. World Psychiatry. 2017;16(2):144–5. 30. Stasiewicz PR, Maisto SA. Two-factor avoidance theory: The role of negative affect in the maintenance of substance use and substance use disorder. Behavior Therapy [Internet]. 1993;24(3):337–56. Available from: https://www.sciencedirect.com/science/article/pii/S0005789405802102. 31. Cooney NL, Litt MD, Morse PA, Bauer LO, Gaupp L. Alcohol cue reactivity, negative-mood reactivity, and relapse in treated alcoholic men. J Abnorm Psychol. 1997;106(2):243–50. 31. Cooney NL, Litt MD, Morse PA, Bauer LO, Gaupp L. Alcohol cue reactivity, negative-mood reactivity, and relapse in treated alcoholic men. J Abnorm Psychol. 1997;106(2):243–50. 32. Shoal GD, Giancola PR. Cognition, negative affectivity and substance use in adolescent boys with and without a family history of a substance use disorder. Journal of Studies on Alcohol [Internet]. 2001;62(5):675–86. Available from: https://doi.org/10.15288/jsa.2001.62.675. 32. Shoal GD, Giancola PR. Cognition, negative affectivity and substance use in adolescent boys with and without a family history of a substance use disorder. Journal of Studies on Alcohol [Internet]. 2001;62(5):675–86. Available from: https://doi.org/10.15288/jsa.2001.62.675. 33. Measelle JR, Stice E, Springer DW. A prospective test of the negative affect model of substance abuse: Moderating effects of social support. Psychol Addict Behav. 2006;20(3):225–33. 33. Measelle JR, Stice E, Springer DW. A prospective test of the negative affect model of substance abuse: Moderating effects of social support. Psychol Addict Behav. 2006;20(3):225–33. Figures Page 19/20 Figure 1 Figure legend not available with this version. Figure legend not available with this version. Figure legend not available with this version. Page 20/20
https://openalex.org/W2964200492	https://europepmc.org/articles/pmc5531809?pdf=render	English	null	DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing	PloS one	2,017	cc-by	15,349	a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 RESEARCH ARTICLE Byunghan Lee1, Taesup Moon2, Sungroh Yoon1,3,4, Tsachy Weissman5 1 Electrical and Computer Engineering, Seoul National University, Seoul, Korea, 2 College of Information and Communication Engineering, Sungkyunkwan University, Suwon, Korea, 3 Interdisciplinary Program in Bioinformatics, Seoul National University, Seoul, Korea, 4 Neurology and Neurological Sciences, Stanford University, Stanford, California, United States of America, 5 Electrical Engineering, Stanford University, Stanford, California, United States of America * tsmoon@skku.edu (TM); sryoon@snu.ac.kr (SY) * tsmoon@skku.edu (TM); sryoon@snu.ac.kr (SY) a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 Editor: Junwen Wang, Mayo Clinic Arizona, UNITED STATES UNITED STATES Received: March 20, 2017 Accepted: June 30, 2017 Published: July 27, 2017 Copyright: © 2017 Lee et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. U S S Received: March 20, 2017 Accepted: June 30, 2017 Published: July 27, 2017 Copyright: © 2017 Lee et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Received: March 20, 2017 Accepted: June 30, 2017 Published: July 27, 2017 Copyright: © 2017 Lee et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. OPEN ACCESS Citation: Lee B, Moon T, Yoon S, Weissman T (2017) DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing. PLoS ONE 12(7): e0181463. https://doi.org/10.1371/ journal.pone.0181463 Abstract We consider the correction of errors from nucleotide sequences produced by next- generation targeted amplicon sequencing. The next-generation sequencing (NGS) plat- forms can provide a great deal of sequencing data thanks to their high throughput, but the associated error rates often tend to be high. Denoising in high-throughput sequencing has thus become a crucial process for boosting the reliability of downstream analyses. Our methodology, named DUDE-Seq, is derived from a general setting of reconstructing finite- valued source data corrupted by a discrete memoryless channel and effectively corrects substitution and homopolymer indel errors, the two major types of sequencing errors in most high-throughput targeted amplicon sequencing platforms. Our experimental studies with real and simulated datasets suggest that the proposed DUDE-Seq not only outperforms existing alternatives in terms of error-correction capability and time efficiency, but also boosts the reliability of downstream analyses. Further, the flexibility of DUDE-Seq enables its robust application to different sequencing platforms and analysis pipelines by simple updates of the noise model. DUDE-Seq is available at http://data.snu.ac.kr/pub/dude-seq. DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing Byunghan Lee1, Taesup Moon2, Sungroh Yoon1,3,4, Tsachy Weissman5 Citation: Lee B, Moon T, Yoon S, Weissman T (2017) DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing. PLoS ONE 12(7): e0181463. https://doi.org/10.1371/ journal.pone.0181463 * tsmoon@skku.edu (TM); sryoon@snu.ac.kr (SY) DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing this paper, is a cost-effective method that enables researchers to focus on investigating areas of interest that are likely to be involved in a particular phenotype. According to previous studies [8, 9], targeted sequencing often results in the complete coverage of exons of disease-related genes, while alternative methods result in approximately 90–95% coverage. Hence, in clinical settings, researchers tend to rely on targeted sequencing for diagnostic evaluations. Funding: This work was supported in part by the National Research Foundation of Korea (NRF) grant funded by the Korea government (Ministry of Science, ICT and Future Planning) [2014M3A9E2064434 and 2016M3A7B4911115], in part by a grant of the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare [HI14C3405030014], in part by the Basic Science Research Program through the National Research Foundation of Korea [NRF-2016R1C1B2012170], in part by the ICT R&D program of MSIP/IITP [2016-0-00563, Research on Adaptive Machine Learning Technology Development for Intelligent Autonomous Digital Companion], and in part by NIH Grant 5U01CA198943-03. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. To detect sequences based on fluorescent labels at the molecular level, NGS technologies normally rely on imaging systems requiring templates that are amplified by emulsion polymer- ase chain reaction (PCR) or solid-phase amplification [1]. These amplification and imaging processes can generate erroneous reads, the origin of which can be traced to the incorrect determination of homopolymer lengths, the erroneous insertion/deletion/substitution of nucleotide bases, and PCR chimeras [6]. Substitution errors dominate for many platforms, including Illumina, while homopolymer errors, manifested as insertions and deletions (indels), are also abundant for 454 pyrosequencing and Ion Torrent. Erroneous reads must be properly handled because they complicate downstream analyses (e.g., variant calling and genome assembly), often lowering the quality of the whole analysis pipeline [7] Soft clipping, in which 3’-ends of a read are trimmed based on the quality scores of individual bases, may be the simplest approach, but it results in a loss of information [10]. More sophisticated methods focus on detecting and correcting errors in sequence data [11–20]. Given the widespread use of Illumina sequencing platforms, most error-correction algorithms have targeted substitution errors [10]. Competing interests: The authors have declared that no competing interests exist. Introduction A new generation of high-throughput, low-cost sequencing technologies, referred to as next- generation sequencing (NGS) technologies [1], is reshaping biomedical research, including large-scale comparative and evolutionary studies [2–4]. Compared with automated Sanger sequencing, NGS platforms produce significantly shorter reads in large quantities, posing vari- ous new computational challenges [5]. Data Availability Statement: All relevant data are within the paper, its Supporting Information files, and its supporting website (http://data.snu.ac.kr/ pub/dude-seq). All the used datasets are also available on the Sequence Read Archive (SRA) under the accession number SRP000570 (SRS002051–SRS002053) at https://www.ncbi. nlm.nih.gov/sra/SRP000570 and the European Nucleotide Archive (ENA) under the accession number PRJEB6244 (ERS671332–ERS671344) at http://www.ebi.ac.uk/ena/data/view/PRJEB6244. There are several DNA sequencing methodologies that use NGS [6, 7] including whole genome sequencing (WGS), chromatin immunoprecipitation (ChIP) sequencing, and targeted sequencing. WGS is used to analyze the genome of an organism to capture all variants and identify potential causative variants; it is also used for de novo genome assembly. ChIP sequencing identifies genome-wide DNA binding sites for transcription factors and other pro- teins. Targeted sequencing (e.g., exome sequencing and amplicon sequencing), the focus of 1 / 25 PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 As summarized in recent reviews [10, 21], current error-correction methods for NGS data can be categorized as follows: k-mer (i.e., oligonucleotides of length k) frequency/spectrum- based, multiple sequence alignment (MSA)-based, and statistical error model-based methods. The idea behind k-mer-based methods [13, 20, 22–25] is to create a list of “trusted” k-mers from the input reads and correct untrusted k-mers based on a consensus represented by this spectrum. In addition to the length of the k-mer, coverage (k-mer occurrences) information is important to determine trusted k-mers. Under the assumption that errors are rare and random and that coverage is uniform, for sufficiently large k, it is reasonable to expect that most errors alter k-mers to inexistent ones in a genome. Thus, for high-coverage genome sequences obtained by NGS, we may identify suspicious k-mers and correct them based on a consensus. MSA-based methods [12, 16, 26] work by aligning related sequences according to their similar- ities and correcting aligned reads, usually based on a consensus in an alignment column, using various techniques. This alignment-based scheme is inherently well-suited for correcting indel errors. Early methods suffered from computational issues, but recent approaches utilize advanced indexing techniques to expedite the alignments. In statistical error model-based methods [27–29], a statistical model is developed to capture the sequencing process, including error generation. In this regard, an empirical confusion model is often created from datasets, exploiting the information obtained from, e.g., alignment results, Phred quality scores (a mea- sure of the quality of nucleobases generated by automated DNA sequencing) [30], or other parameters. While the above methods often exhibit good performance for various platforms, they also have several limitations. First, k-mer-based schemes tend to be ineligible when the coverage is expected to vary over the queried sequences, as in transcriptomics, metagenomics, heteroge- neous cell samples, or pre-amplified libraries [21]. Second, MSA-based methods, which do not suffer from the above issue related to non-uniform coverage, often require the application of heuristic and sophisticated consensus decision rules for the aligned columns, and such rules may be sensitive to specific applications or sequencing platforms. Third, statistical error model-based methods typically use computationally expensive schemes (e.g., expectation- maximization) owing to additional stochastic modeling assumptions for the underlying DNA PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 2 / 25 DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing sequences. Moreover, little attention is given to the validity and accuracy of such modeling assumptions, let alone to theoretical analysis of whether near optimum or sound error-correc- tion performance is attained. Finally, many existing schemes applying the three methods often return only representative (consensus) denoised sequences created by merging input sequences; hence, the number of sequences is often not preserved after denoising. In some applications, this may result in inconsistencies in downstream analyses. To address these limitations, many existing tools combine the three methods in a complementary manner to improve performance [10, 21]. In this paper, as an alternative, we applied an algorithm called Discrete Universal DEnoiser (DUDE) [31] for accurate DNA sequence denoising. DUDE was developed for a general set- ting of reconstructing sequences with finite-valued components (source symbols) corrupted by a noise mechanism that corrupts each source symbol independently and statistically identi- cally. In the DNA denoising literature, such a noise model is equivalent to the confusion matrix commonly used in statistical error-model-based methods. As demonstrated in the orig- inal paper [31], DUDE exhibits rigorous performance guarantee for the following setting; even when no stochastic modeling assumptions are made for the underlying clean source data, only with the assumption of known noise mechanism, DUDE is shown to universally attain the optimum denoising performance for any source data the data increase. We note that the above setting of DUDE naturally fits the setting for DNA sequence denoising, i.e., it is difficult to establish accurate stochastic models for clean DNA sequences, but it is simple and fairly realis- tic to assume noise models (i.e., confusion matrices) for sequencing devices based on reference sequences. The DUDE algorithm, which will be explained in details in the next section, possesses fla- vors that are somewhat connected to all three representative methods mentioned above, in a single scheme. Specifically, DUDE works with double-sided contexts of a fixed size that are analogous to k-mers. Moreover, like MSA, DUDE applies a denoising decision rule to each noisy symbol based on aggregated information over certain positions in the reads. However, unlike MSA, which makes a decision based on the information collected from the symbols in the same aligned column, DUDE makes a decision using the information collected from posi- tions with the same double-sided context. PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 Finally, the denoising decision rule of DUDE utilizes information from the assumed noise model, like in most statistical error model-based meth- ods, but does not assume any stochastic model on the underlying sequence, thus resulting in a computationally efficient method. The method of incorporating the noise model is also simple, making it easy to flexibly apply DUDE to different sequencing platforms by simply changing the confusion matrix model in the algorithm. With the above unique nature of the DUDE algorithm, we show in our experiments that it outperforms other state-of-the-art schemes, particularly for applications to targeted amplicon sequencing. Specifically, among the applicable areas of targeted amplicon sequencing (e.g., cancer gene, 16S rRNA, plant, and animal sequencing [32]), we used 16S rRNA benchmark datasets obtained with different library preparation methods and DNA polymerases to confirm the robustness of our algorithm across various sequencing preparation methods. Targeted amplicon sequencing datasets often have deeper sequencing coverage than those of WGS or ChIP datasets, which frequently makes conventional k-mer-based techniques often suffer from the amplification bias problem [33]. By contrast, for DUDE-Seq, as the sequencing coverage becomes deeper, context-counting vectors can accumulate more probable contexts, and the robustness of denoising typically improves. We apply two versions of DUDE separately for substitution and homopolymer errors, the two major types of sequencing error. For substitu- tion errors, our approach directly utilizes the original DUDE with appropriate adaptation to DNA sequences and is applicable to reads generated by any sequencing platform. For PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 3 / 25 DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing homopolymer errors, however, we do not apply the original DUDE, which was developed in a framework that does not cover errors of the homopolymer type. To correct homopolymer errors, we therefore adopt a variant of DUDE for general-output channels [34]. Our homopol- ymer-error correction is applicable to cases in which base-called sequences and the underlying flowgram intensities are available (e.g., pyrosequencing and Ion Torrent). For brevity, we refer to both of these DUDE-based approaches as DUDE-Seq, but the correction type will be easily distinguishable by the reader. Discrete Universal DEnoiser (DUDE) In this section, we formally introduce the DUDE algorithm along with its notation and its con- nection to DNA sequence denoising. Fig 1 shows the concrete setting of the discrete denoising problem. We denote the underlying source data as {xi} and assume each component takes values in some finite set X. The resulting noisy version of the source corrupted by a noise mechanism is denoted as {Zi}, and its components take values in, again, some finite set Z. As mentioned in the Introduction, DUDE assumes that the noise mechanism injects noises that are independent and statistically identical, and such a mechanism is often referred to as a Dis- crete Memoryless Channel (DMC) in information theory. The DMC is completely character- ized by the channel transition matrix, also known as the confusion matrix, P 2 RjXjjZj, of which the (x, z)-th element, P(x, z), stands for Pr(Zi = z\|xi = x), i.e., the conditional probability that the noisy symbol takes value z, given that the original source symbol is x. We denote ran- dom variables with uppercase letters and the individual samples of random variables or deter- ministic symbols with lowercase letters. Thus, the underlying source data, which are treated by DUDE as individual sequences (and not a stochastic process), are denoted by the lowercase {xi}, and the noise-corrupted sequences, i.e., sequences of random variables, are denoted by uppercase {Zi}. Furthermore, throughout this paper, we generally denote a sequence (n-tuple) as an = (a1,. . .,an), for example, where a j i refers to the subsequence (ai,. . .,aj). As shown in Fig 1, a discrete denoiser observes the entire noisy data Zn and reconstructs the original data with ^X n ¼ ð^X 1ðZnÞ; . . . ; ^X nðZnÞÞ. The goodness of the reconstruction by a discrete denoiser ^X n is measured by the average loss, L^X nðxn; ZnÞ ¼ 1 n X n i¼1 Lðxi; ^X iðZnÞÞ; ð1Þ ð1Þ where Lðxi; ^xiÞ is a single-letter loss function that measures the loss incurred by estimating xi with ^xi at location i. The loss function can be also represented with a loss matrix L 2 RjXjj ^X j. DUDE in [31] is a two-pass algorithm that has linear complexity with respect to the data size n. Discrete Universal DEnoiser (DUDE) Thus, the denoising rule Eq (2), re-expressed in Eq (3), finds a reconstruction symbol ^x that mini- mizes the expected loss with respect to the empirical estimate (obtained by utilizing the inverse of P) of the count vector of the underlying xi given the noisy context ziþk ik. At a high level, DUDE is not a simple majority voting rule based on m; instead, it incorporates the DMC model P (the confusion matrix) and loss function Λ to obtain a more accurate estimation of the clean source symbol. For more detailed and rigorous arguments on the intuitive descrip- tion of Eq (2), we refer readers to the original paper [31, Section IV-B]. Note that formula Eq (2) assumes X ¼ Z ¼ ^X and P is invertible for simplicity, but Weissman et al. [31] deal with more general cases as well. The form of Eq (2) also shows that DUDE is a sliding window denoiser with window size 2k + 1; i.e., DUDE returns the same denoised symbol at all locations with the same value ziþk ik. DUDE is guaranteed to attain the optimum performance by the sliding window denoisers with the same window size as the observation length n increases. For more details on the theoretical performance analyses, see Weissman et al. [31, Section V]. The original DUDE dealt exclusively with the case of jXj and jZj finite. Dembo and Weiss- man [34] DUDE to the case of discrete input and general output channels; the noisy outputs do not have to have their values in some finite set, but can have continuous values as well. As in [31], the memoryless noisy channel model, which is characterized in this case by the set of densities ffxgx2X, was assumed to be known. As shown in [34, Fig 1], the crux of the arguments is to apply a scalar quantizer Q() to each continuous-valued noisy output {Yi} and to derive a virtual DMC, Γ 2 RjXjjZj, between the discrete input {Xi} and the quantized (hence, discrete) output {Zi}. Such Γ can be readily obtained by the knowledge of ffxgx2X and evaluating the fol- lowing integral for each (x, z): Γ(x, z) = R y:Q(y) = z fx(y)dy. Discrete Universal DEnoiser (DUDE) During the first pass, given the realization of the noisy sequence zn, the algorithm col- lects the statistics vector mðzn; lk; rkÞ½a ¼ jfi : k þ 1 i n k; ziþk ik ¼ lkarkgj; Fig 1. The general setting of discrete denoising. Fig 1. The general setting of discrete denoising. Fig 1. The general setting of discrete denoising. https://doi.org/10.1371/journal.pone.0181463.g001 PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 4 / 25 DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing for all a 2 Z, which is the count of the occurrence of the symbol a 2 Z along the noisy sequence zn that has the double-sided context ðlk; rkÞ 2 Z2k. Note that m is similar to the counts across the aligned columns for the simple majority voting in MSA-based denoising methods. However, in DUDE, the count is collected regardless of whether the positions in the reads are aligned or not, but considering whether the position has the same context. Additionally, the context length k is analogous to the k-mer length. Once the m vector is collected, for the sec- ond pass, DUDE then applies the rule ^XiðznÞ ¼ arg min ^x2X mTðzn; zi1 ik; ziþk iþ1ÞΠ1½l^x pzi ð2Þ ð2Þ for each k + 1 i n −k, where πzi is the zi-th column of the channel matrix P, and l^x is the ^x-th column of the loss matrix Λ. Furthermore, stands for the element-wise product opera- tor for two vectors. The intuitive explanation of Eq (2) is as follows: when we rearrange the right-hand side of Eq (2), we obtain for each k + 1 i n −k, where πzi is the zi-th column of the channel matrix P, and l^x is the ^x-th column of the loss matrix Λ. Furthermore, stands for the element-wise product opera- tor for two vectors. The intuitive explanation of Eq (2) is as follows: when we rearrange the right-hand side of Eq (2), we obtain ð2Þ ¼ arg min ^x2X lT ^x fpzi ΠTmTðzn; zi1 ik; ziþk iþ1Þg; ð3Þ ð2Þ ð3Þ and we can show that πa P−T mT(zn, lk, rk) approximates the empirical count vector of the underlying clean symbol at the middle location that resulted in the noisy context lkark. PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 Substitution errors As described in the previous section, the setting of the original DUDE algorithm naturally aligns with the setting of substitution-error correction in DNA sequence denoising. We can set X ¼ Z ¼ fA; C; G; Tg, and the loss function as the Hamming loss, namely, Lðx; ^xÞ ¼ 0, if x ¼ ^x, and Lðx; ^xÞ ¼ 1, otherwise. Then, the two-pass sliding window procedure of DUDE for collecting the statistics vector m and the actual denoising can be directly applied as shown in the toy example in Fig 2. Before we formally describe our DUDE-Seq for substitution-error correction, however, we need to address some subtle points. First, the original DUDE in Eq (2) assumes that the DMC matrix P is known beforehand, but in real DNA sequence denoising, we need to estimate P for each sequencing device. As described in the Experimental Results section in detail, we performed this estimation following the typical process for obtaining the empirical confusion matrix, i.e., we aligned the predefined reference sequence and its noise-corrupted sequence and then determined the ratio of substi- tution errors and obtain the estimated P. Second, the original DUDE assumes that the noise mechanism is memoryless, i.e., the error rate does not depend on the location of a base within the sequence. In contrast, for real sequencing devices, the actual error rate, namely, the condi- tional probability Pr(Zi = z\|Xi = x) may not always be the same for all location index values i. For example, for Illumina sequencers, the error rate tends to increase towards the ends of reads, as pointed out in [21]. In our DUDE-Seq, however, we still treat the substitution error mechanism as a DMC and therefore use the single estimated P obtained as above, which is essentially the same as that obtained using the average error rate matrix. Our experimental results show that such an approach still yields very competitive denoising results. Thirdly, the optimality of the original DUDE relies on the stationarity of the underlying clean sequence, thus requiring a very large observation sequence length n to obtain a reliable statistics vector m. In contrast, most sequencing devices generate multiple short reads of lengths 100 * 200. Fig 2. A sliding window procedure of the DUDE-Seq with the context size k = 3. During the first pass, DUDE-Seq updates the m(zn, l3, r3) for the encountered double-sided contexts (l3, r3). Discrete Universal DEnoiser (DUDE) Once the virtual DMC is obtained, the rest of the algorithm in [34] proceeds similarly as the original DUDE; specifically, it obtains the statistics vector m for the quantized noisy outputs {Zi} during the first pass, and then applies a sliding window denoising rule similar to Eq (2), which depends on the statistics vec- tor m, the virtual DMC Γ, ffxgx2X, and the noisy sequence Yn, during the second pass. A con- crete denoising rule can be found in [34, Eqs (16), (19) and (20)]. In [34], a formal analysis of the generalized DUDE shows that it attains the optimum denoising performance among slid- ing window denoisers with the same window size, that base their denoising decisions on the PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 5 / 25 DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing original continuous-valued outputs Yn. We refer readers to the paper for more details. In the next section, we show how we adopt this generalized DUDE in our DUDE-Seq to correct homopolymer errors in DNA sequencing. Substitution errors Then, for the second pass, DUDE-Seq uses the obtained m(zn, l3, r3) and Eq (2) for the denoising. https://doi.org/10.1371/journal.pone.0181463.g002 Fig 2. A sliding window procedure of the DUDE-Seq with the context size k = 3. During the first pass, DUDE-Seq updates the m(zn, l3, r3) for the encountered double-sided contexts (l3, r3). Then, for the second pass, DUDE-Seq uses the obtained m(zn, l3, r3) and Eq (2) for the denoising. https://doi.org/10.1371/journal.pone.0181463.g002 Fig 2. A sliding window procedure of the DUDE-Seq with the context size k = 3. During the first pass, DUDE-Seq updates the m(zn, l3, r3) for the encountered double-sided contexts (l3, r3). Then, for the second pass, DUDE-Seq uses the obtained m(zn, l3, r3) and Eq (2) for the denoising. 6 / 25 PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing Hence, in DUDE-Seq, we combined all statistics vectors collected from multiple short reads to generate a single statistics vector m to use in Eq (2). Addressing the above three points, a formal summary of DUDE-Seq for the substitution errors is given in Algorithm 1. Note that the pseudocode in Algorithm 1 skips those bases whose Phred quality score s are higher than a user-specified threshold and invokes DUDE-Seq only for the bases with low quality scores (lines 10–14). This is in accord with the common practice in sequence preprocessing and is not a specific property of the DUDE-Seq algorithm. Furthermore, for simplicity, we denoted zn as the entire noisy DNA sequence, and mTðzn; zi1 ik; ziþk iþ1Þ represents the aggregated statistics vector obtained as described above. mTðzn; zi1 ik; ziþk iþ1Þ represents the aggregated statistics vector obtained as described above. Algorithm 1 The DUDE-Seq for substitution errors Require: Observation zn, Estimated DMC matrix P 2 R44, Hamming loss Λ 2 R44, Context size k, Phred quality score Qn Ensure: The denoised sequence ^X n 1: Define mðzn; lk; rkÞ 2 R4 for all (lk, rk)2{A,C,G,T}2k. 2: Initializem(zn, lk, rk)[a] = 0 for all (lk, rk)2{A,C,G,T}2k and for all a 2 {A,C,G,T} 3: For i k + 1,. . ., n −k do ⊳First pass 4: mðzn; zi1 ik; ziþk iþ1Þ½zi ¼ mðzn; zi1 ik; ziþk iþ1Þ½zi þ 1 ⊳Update the count statisticsvector 5: end for 6: for i 1,. . Remarks 1. Incorporating flanking sequences in DUDE-Seq is quite straightforward; we can simply use the one-sided contexts l2k or r2k once DUDE-Seq reaches the flanking regions. In our exper- iments, however, we did not perform such modification (lines 7–8 of Algorithm 1) since we normally used small k values (around k = 5). As demonstrated in our experimental results, the effect of such small flanking regions is not significant on the final denoising results, and we can achieve satisfactory results without considering flanking regions. However, in gen- eral, should longer values of k be needed, we can easily modify the algorithm to incorporate one-sided contexts in the flanking regions, and such modification will clearly improve the final denoising result. 2. DUDE-Seq does not need to consider reverse complements of the input sequences to collect m’s, since forward and reverse reads are handled separately in our experiments. Reverse complements are typically considered when we need to handle double-stranded sequences without knowing whether each read corresponds to the forward or reverse strand. Substitution errors ., n do ⊳Second pass 7: if i k or i n −k + 1 then 8: ^X i ¼ zi 9: else 10: if Qi > threshold then ⊳Quality score 11: ^X i ¼ zi 12: else 13: ^X iðznÞ ¼ argmin ^x2fA;C;G;Tg mTðzn; zi1 ik; ziþk iþ1ÞΠ1½l^x pzi ⊳Apply the denoising rule 14: end if 15: end if 16: end for Require: Observation zn, Estimated DMC matrix P 2 R44, Hamming loss Λ 2 R44, Context size k, Phred quality score Qn 1: Define mðzn; lk; rkÞ 2 R4 for all (lk, rk)2{A,C,G,T}2k. k k k k 4: mðzn; zi1 ik; ziþk iþ1Þ½zi ¼ mðzn; zi1 ik; ziþk iþ1Þ½zi þ 1 ⊳Update the count statisticsvector 5: end for ⊳Quality score Homopolymer errors Homopolymer errors, particularly in pyrosequencing, occur while handling the observed flow- gram, and a careful understanding of the error injection procedure is necessary to correct 7 / 25 PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing Fig 3. Conditional intensity distributions for N = 0, 1, 2, 3. https://doi.org/10.1371/journal.pone.0181463.g003 Fig 3. Conditional intensity distributions for N = 0, 1, 2, 3. https://doi.org/10.1371/journal.pone.0181463.g003 Fig 3. Conditional intensity distributions for N = 0, 1, 2, 3. https://doi.org/10.1371/journal.pone.0181463.g003 Fig 3. Conditional intensity distributions for N = 0, 1, 2, 3. https://doi.org/10.1371/journal.pone.0181463.g003 Fig 3. Conditional intensity distributions for N = 0, 1, 2, 3. https://doi.org/10.1371/journal.pone.0181463.g003 these errors. As described in [35], in pyrosequencing, the light intensities, i.e., flowgram, that correspond to a fixed order of four DNA bases {T, A, C, G} are sequentially observed. The intensity value increases when the number of consecutive nucleotides (i.e., homopolymers) for each DNA base increases, and the standard base-calling procedure rounds the continuous-val- ued intensities to the closest integers. For example, when the observed light intensities for the two frames of DNA bases are [0.03 1.03 0.09 0.12; 1.89 0.09 0.09 1.01], the corresponding rounded integers are [0.00 1.00 0.00 0.00; 2.00 0.00 0.00 1.00]. Hence, the resulting sequence is ATTG. The insertion and deletion errors are inferred because the observed light intensities do not perfectly match the actual homopolymer lengths; thus, the rounding procedure may result in the insertion or deletion of DNA symbols. In fact, the distribution of the intensities f, given the actual homopolymer length N, {P(f\|N)}, can be obtained for each sequencing device, and Fig 3 shows typical distributions given various lengths. Exploiting the fact that the order of DNA bases is always fixed at {T, A, C, G}, we can apply the setting of the generalized DUDE in [34] to correct homopolymer errors as follows. Because we know the exact DNA base that corresponds with each intensity value, the goal is the correct estimatimation of homopolymer lengths from the observed intensity values. Hence, we can interpret the intensity distributions {P(f\|N)} as the memoryless noisy channel models with a continuous-output, where the channel input is the homopolymer length N. We set the upper bound of N to 9 according to the convention commonly used for handling flowgram distribu- tions in the targeted amplicon sequencing literature [35–37]. PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 Setup We used both real and simulated NGS datasets and compared the performance of DUDE-Seq with that of several state-of-the-art error correction methods. The list of alternative tools used for comparison and the rationale behind our choice s are described in the next subsection. When the flowgram intensities of base-calling were available, we corrected both homopolymer and substitution errors; otherwise, we only corrected substitution errors. The specifications of the machine we used for the analysis are as follows: Ubuntu 12.04.3 LTS, 2 × Intel Xeon X5650 CPUs, 64 GB main memory, and 2 TB HDD. We used both real and simulated NGS datasets and compared the performance of DUDE-Seq with that of several state-of-the-art error correction methods. The list of alternative tools used for comparison and the rationale behind our choice s are described in the next subsection. When the flowgram intensities of base-calling were available, we corrected both homopolymer and substitution errors; otherwise, we only corrected substitution errors. The specifications of the machine we used for the analysis are as follows: Ubuntu 12.04.3 LTS, 2 × Intel Xeon X5650 CPUs, 64 GB main memory, and 2 TB HDD. Algorithm 2 The DUDE-Seq for homopolymer errors Algorithm 2 The DUDE Seq for homopolymer errors Require: Flowgram data fn, Flowgram densities fPðf jNÞg 9 N¼0, Hamming loss L 2 R1010, Context size k Ensure: The denoised sequence ^D 1: Let QR(f) be the rounding quantizer in Eq (4) of the main text 2: Let Base(i) 2 {T, A, C, G} be the DNA base corresponding to fi 3: Define mðf n; lk; rkÞ 2 R10 for all (lk, rk) 2 {0, 1,.. .,9}2k. 4: Initializem(fn, lk, rk)[a] = 0 for all (lk, rk) 2 {0, 1,. . .,9}2k and for all a 2 {0, 1,. . .,9} 5: Let ^D ¼ , I = 0 6: for i 0,. . .,9 do 7: for j 0,.. .,9 do 8: Compute Γ(i, j) following Eq (5) of the main text ⊳Computing the vir- tual DMC Γ 9: end for 10: end for 11: for i 1,.. .,n do Obtain zi = QR(fi) ⊳Note zi 2 {0,.. .,9} 12: end for 13: for i k + 1,. . .,n −k do ⊳First pass 14: mðf n; zi1 ik; ziþk iþ1Þ½zi ¼ mðf n; zi1 ik; ziþk iþ1Þ½zi þ 1 15: end for 16: for i 1,.. Homopolymer errors When the usual rounding func- tion QRðf Þ ¼ argmin i2f0;...;9g ji f j ð4Þ ð4Þ 8 / 25 PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing is used as a scalar quantizer, as mentioned above, and the virtual DMC G 2 R1010 can be obtained by calculating the integral Gði; jÞ ¼ R jþ0:5 j0:5 Pðf jiÞdf ð5Þ ð5Þ for each 0 i 9, 1 j 9 and Gði; 0Þ ¼ R 0:5 0 Pðf jiÞdf . With this virtual DMC model, we apply a scheme inspired by the generalized DUDE to cor- rectly estimate the homopolymer lengths, which results in correcting the insertion and dele- tion errors. That is, we set X ¼ Z ¼ f0; 1; . . . ; 9g, and again use the Hamming loss 1010 L 2 R1010. With this setting, we apply QR(f) to each fi to obtain the quantized discrete output zi, and obtain the count statistics vector m from zn during the first pass. Then, for the second pass, instead of applying the more involved denoising rule in [34, we employ the same rule as Eq (2) with Γ in place of P to obtain the denoised sequence of integers ^X n based on the quan- tized noisy sequence Zn. Although it is its implementation is easier and it has a faster running time than that of the generalized DUDE. Once we obtain ^X n, from the knowledge of the DNA base for each i, we can reconstruct the homopolymer error-corrected DNA sequence ^D (the length of which may not necessarily be equal to n). Algorithm 2 summarizes the pseudo-code of DUDE-Seq for homopolymer-error correction. PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 Setup .,n do ⊳Second pass g q p y Require: Flowgram data fn, Flowgram densities fPðf jNÞg 9 N¼0, Hamming loss L 2 R1010, Context size k L 2 R1010, Context size k Ensure: The denoised sequence ^D Ensure: The denoised sequence ^D 1: Let QR(f) be the rounding quantizer in Eq (4) of the main text 1: Let QR(f) be the rounding quantizer in Eq (4) of the main text 2: Let Base(i) 2 {T, A, C, G} be the DNA base corresponding to fi 2: Let Base(i) 2 {T, A, C, G} be the DNA base corresponding to fi 3: Define mðf n; lk; rkÞ 2 R10 for all (lk, rk) 2 {0, 1,.. .,9}2k. 4: Initializem(fn, lk, rk)[a] = 0 for all (lk, rk) 2 {0, 1,. . .,9}2k and for all a 2 {0, 1,. . .,9} 5: Let ^D ¼ , I = 0 6: for i 0,. . .,9 do 7: for j 0,.. .,9 do 8: Compute Γ(i, j) following Eq (5) of the main text ⊳Computing the vir- tual DMC Γ 9: end for 10: end for 11: for i 1,.. .,n do Obtain zi = QR(fi) ⊳Note zi 2 {0,.. .,9} 12: end for 2: Let Base(i) 2 {T, A, C, G} be the DNA base corresponding to fi 3: Define mðf n; lk; rkÞ 2 R10 for all (lk, rk) 2 {0, 1,.. .,9}2k. 4: Initializem(fn, lk, rk)[a] = 0 for all (lk, rk) 2 {0, 1,. . .,9}2k and for all a 2 {0, 1,. . .,9} 5: Let ^D ¼ , I = 0 6: for i 0,. . .,9 do 7: for j 0,.. .,9 do 8: Compute Γ(i, j) following Eq (5) of the main text ⊳Computing the vir- 2: Let Base(i) 2 {T, A, C, G} be the DNA base corresponding to fi 3: Define mðf n; lk; rkÞ 2 R10 for all (lk, rk) 2 {0, 1,.. .,9}2k. 4: Initializem(fn, lk, rk)[a] = 0 for all (lk, rk) 2 {0, 1,. . .,9}2k and for all a 2 {0, 1,. . .,9} 5: Let ^D ¼ , I = 0 3: Define mðf n; lk; rkÞ 2 R10 for all (lk, rk) 2 {0, 1,.. .,9}2k. 3: Define mðf n; lk; rkÞ 2 R10 for all (lk, rk) 2 {0, 1,.. .,9}2k. 4: Initializem(fn, lk, rk)[a] = 0 for all (lk, rk) 2 {0, 1,. Setup . .,9}2k and for all a 2 {0, 1,. . .,9} 6: for i 0,. . .,9 do 7: for j 0,.. .,9 do 7: for j 0,.. .,9 do j , , 8: Compute Γ(i, j) following Eq (5) of the main text ⊳Computing the vir- tual DMC Γ ⊳Note zi 2 {0,.. .,9} ⊳First pass ⊳Second pass PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 9 / 25 DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing 17: if i k or i n −k + 1 then ^X iðf nÞ ¼ zi 18: else 19: ^X iðf nÞ ¼ argmin ^x2X mTðf n; zi1 ik; ziþk iþ1ÞG1½l^x gzi ⊳Note ^X iðznÞ 2 f0; . . . ; 9g 20: end if 21: if ^X iðf nÞ 1 then 22: for j 1; . . . ; ^X iðf nÞ do ^DIþj ¼ BaseðiÞ ⊳Reconstructingthe DNA sequence 23: end for 24: end if 25: I I þ ^X iðf nÞ 26: end for DUDE-Seq has a single hyperparameter k, the context size, that needs to be determined. Similar to the popular k-mer-based schemes, there is no analytical method for selecting the best k for finite data size n, except for the asymptotic order result of kjXj 2k ¼ oðn= lognÞ in [31], but a heuristic rule of thumb is to try values between 2 and 8. Furthermore, as shown in Eq (2), the two adjustable matrices, Λ and P, are required for DUDE-Seq. The loss Λ used for both types of errors is the Hamming loss. According to Marinier et al. [38], adjusting the sequence length by one can correct most homopolymer errors, which justifies our use of Ham- ming loss in DUDE-Seq. In our experiments, the use of other types of loss functions did not result in any noticeable performance differences. The DMC matrix P for substitution errors is empirically determined by aligning each sampled read to its reference sequence, as in [35]. Fig 4 shows the non-negligible variation in the empirically obtained P’s across the sequencing platforms, where each row corresponds to the true signal x and each column corresponds to the observed noisy signal z. In this setting, each cell represents the conditional probability P(z\|x). In our experiments, dataset P1–P8 used P for GS FLX, Q19–Q31 used P for Illumina, and S5, A5 used P for Simulation data. Evaluation metric As a performance measure, we define the per-base error rate of a tool after denoising as As a performance measure, we define the per-base error rate of a tool after denoising as etool ¼ # mismatched bases # aligned bases ; ð6Þ ð6Þ in which ‘# aligned bases’ represents the number of mapped bases (i.e., matches and mis- matches) after mapping each read to its reference sequence, and ‘# mismatched bases’ repre- sents the number of the erroneous bases (i.e., insertions, deletions, and substitutions) among the aligned bases. We also employ an alternative definition that adjusts the error rate by incorporating the degree of alignment. To this end, we define the relative gain of the number of aligned bases after denoising by a tool over raw data as gðatoolÞ ¼ # aligned bases after denoising # aligned bases in raw # aligned bases in raw : ð7Þ ð7Þ Based on this, the adjusted error rate ^etool of a denoising tool is defined as follows: ^etool ¼ ð1 þ gðatoolÞÞ etool gðatoolÞ eraw; ð8Þ ð8Þ ^etool ¼ ð1 þ gðatoolÞÞ etool gðatoolÞ eraw; where etool and eraw represent the (unadjusted) error rates of the denoised data and the raw data, respectively. In other words, Eq (8) is a weighted average of etool and eraw, in which the weights are determined by the relative number of aligned bases of a tool compared to the raw sequence. We believe ^etool is a fairer measure as it penalizes the error rate of a denoiser when there is a small number of aligned bases. The relative gain of the adjusted error rate over raw data is then defined as gð^etoolÞ ¼ eraw ^etool eraw ; ð9Þ ð9Þ which we use to evaluate the denoiser performance. which we use to evaluate the denoiser performance. While evaluating a clustering result, we employ a measure of concordance (MoC) [42] which is a popular similarity measure for pairs of clusterings. For two pairs of clusterings P and Q with I and J clusters, respectively, the MoC is defined as MoCðP; QÞ ¼ 1 ﬃﬃﬃﬃIJ p 1 X I i¼1 X J j¼1 f 2 ij piqj 1 ! Evaluation metric ð10Þ ð10Þ where fij is the number of the common objects between cluster Pi and Qj when pi and qj are the numbers of the objects in cluster Pi and Qj, respectively. A MoC of one or zero represents per- fect or no concordance, respectively, between the two clusters. Setup The details of each dataset are explained in the follow- ing sections. In order to evaluate the results, we used Burrows-Wheeler Aligner (BWA) [39] and SAM- tools [40]. We aligned all reads to their reference genome using BWA with the following parameters: [minimum seed length: 19, matching score: 1, mismatch penalty: 4, gap open pen- alty: 6, gap extension penalty: 1]. After the mapped regions were determined using BWA in SAM format, we chose uniquely mapped pairs using SAMtools. The Compact Idiosyncratic Gapped Alignment Report (CIGAR) string and MD tag (string for mismatching positions) for each of the resultant pairs in the SAM file were reconstructed to their pairwise alignments using sam2pairwise [41]. Fig 4. Adjustable DMC matrix Π of DUDE-Seq. Empirically obtained Π’s for different sequencing platforms (colors are on a log scale). Fig 4. Adjustable DMC matrix Π of DUDE-Seq. Empirically obtained Π’s for different sequencing platforms (colors are on a log scale). PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 10 / 25 DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing Software chosen for comparison It is impossible to compare the performance of DUDE-Seq with that of all other schemes. Hence, we selected representative baselines using the following reasoning. 1. We included tools that can represent different principles outlined in the Introduction, namely, k-mer-based (Trowel, Reptile, BLESS, and fermi), MSA-based (Coral), and statisti- cal error model-based (AmpliconNoise) methods. 2. We considered the recommendations of [21] to choose baseline tools that are competitive for different scenarios, i.e., for 454 pyrosequencing data (AmpliconNoise), non-uniform PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 11 / 25 DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing coverage data, such as metagenomics data (Trowel, fermi, Reptile), data dominated by sub- stitution errors, such as Illumina data (Trowel, fermi, Reptile), and data with a high preva- lence of indel errors (Coral). coverage data, such as metagenomics data (Trowel, fermi, Reptile), data dominated by sub- stitution errors, such as Illumina data (Trowel, fermi, Reptile), and data with a high preva- lence of indel errors (Coral). 3. For multiple k-mer-based tools, we chose those that use different main approaches/data structures: BLESS (k-mer spectrum-based/hash table and bloom filter), fermi (k-mer spec- trum and frequency-based/hash table and suffix array), Trowel (k-mer spectrum-based/ hash table), and Reptile (k-mer frequency and Hamming graph-based/replicated sorted k-mer list). 4. The selected tools were developed quite recently; Trowel and BLESS (2014), fermi (2012), Coral and AmpliconNoise (2011), and Reptile (2010). 5. We mainly chose tools that return read-by-read denoising results to make fair error-rate comparisons with DUDE-seq. We excluded tools that return a substantially reduced num- ber of reads after error correction (caused by filtering or forming consensus clusters). Examples of excluded tools are Acacia, ALLPATHS-LG, and SOAPdenovo. 6. We also excluded some recently developed tools that require additional mandatory infor- mation (e.g., the size of the genome of the reference organism) beyond the common setting of DNA sequence denoising in order to make fair error-rate comparisons. Examples of excluded tools are Fiona, Blue, and Lighter. Incorporating those tools that require addi- tional information into the DUDE-Seq framework and comparisons with the excluded tools would be another future directions. DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing Fig 5. Hyperparameter k of DUDE-Seq. Effects of varying context size k [k1 is for Algorithm 1 (substitution-error correction) and k2 is for Algorithm 2 (homopolymer-error correction); data: [35]]. Fig 5. Hyperparameter k of DUDE-Seq. Effects of varying context size k [k1 is for Algorithm 1 (substitution-error correction) and k2 is for Algorithm 2 (homopolymer-error correction); data: [35]]. https://doi.org/10.1371/journal.pone.0181463.g005 while P2 had 23 clones that were mixed in equal proportions. In P3, P4, and P5 and P6, P7, and P8, there are 87 mock communities mixed in even and uneven proportions, respectively. In all datasets, both homopolymer and substitution errors exist, and the flowgram intensity values as well as the distributions are available [35]. Therefore, DUDE-Seq tries to correct both types of errors using the empirically obtained P and the flowgram intensity distributions {P(f\|N)}. f\| We first show the effect of k on the performance of DUDE-Seq in Fig 5. The vertical axis shows the ratio between the number of OTUs assigned after denoising with DUDE-Seq and the ground truth number of OTUs for the P1, P2, and P8 dataset. The horizontal axis shows the k values used for correcting the substitution errors (i.e., for Algorithm 1), and color-coded curves were generated for different k values used for homopolymer-error correction (i.e., for Algorithm 2). As shown in the figure, correcting homopolymer errors (i.e., with k = 2 for Algo- rithm 2) always enhanceed the results in terms of the number of OTUs in comparison to cor- recting substitution errors alone (i.e., Algorithm 1 alone). We observe that k = 5 for Algorithm 1 and k = 2 for Algorithm 2 produce the best results in terms of the number of OTUs. Larger k value work better for substitution errors owing to the smaller alphabet size of the data, i.e., 4, compared to that of homopolymer errors, i.e., 10. Motivated by this result, we fixed the context sizes of substitution error correction and homopolymer error correction to k = 5 and k = 2, respectively, for all subsequent experiments. In Fig 6(a), we report a more direct analysis of error correction performance. We compared the performance of DUDE-Seq with that of Coral [16], which is an MSA-based state-of-the-art scheme. Real data: 454 pyrosequencing Pyrosequenced 16S rRNA genes are commonly used to characterize microbial communities because the method yields relatively longer reads than those of other NGS technologies [43]. Although 454 pyrosequencing is gradually being phased out, we test ed DUDE-Seq with 454 pyrosequencing data for the following reasons: (1) the DUDE-Seq methodology for correcting homopolymeric errors in 454 sequencing data is equally applicable to other sequencing tech- nologies that produce homopolymeric errors, such as Ion Torrent; (2) using pyrosequencing data allows us to exploit existing (experimentally obtained) estimates of the channel transition matrix Γ (e.g., [35]), which is required for denoising noisy flowgrams by DUDE-Seq (see Algo- rithm 2); (3) in the metagenomics literature, widely used standard benchmarks consist of data- sets generated by pyrosequencing. In metagenome analysis [44], grouping reads and assigning them to operational taxonomic units (OTUs) (i.e., binning) are essential processes, given that the majority of microbial species have not been taxonomically classified. By OTU binning, we can computationally identify closely related genetic groups of reads at a desired level of sequence differences. However, owing to erroneous reads, nonexistent OTUs may be obtained, resulting in the common prob- lem of overestimating ground truth OTUs. Such overestimation is a bottleneck in the overall microbiome analysis; hence, removing errors in reads before they are assigned to OTUs is a critical issue [35]. With this motivation, in some of our experiments below, we used the differ- ence between the number of assigned OTUs and the ground truth number of OTUs as a proxy for denoising performance; the number of OTUs was determined using UCLUST [45] at iden- tity threshold of 0.97 which is for species assignment. We tested the performance of DUDE-Seq with the eight datasets used in [35], which are mixtures of 94 environmental clones library from eutrophic lake (Priest Pot) using primers 787f and 1492r. Dataset P1 had 90 clones that are mixed in two orders of magnitude difference PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 12 / 25 PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing Fig 6. Comparison of reads correction performance on eight real 454 pyrosequencing datasets (labeled P1–P8; [35]). [parameters: k = 5 (Algorithm 1) and k = 2 (Algorithm 2) for DUDE-Seq; (sPyroNoise, cPyroNoise, sSeqNoise, cSeqNoise) = (60, 0.01, 25, 0.08) for AmpliconNoise; (k, mr, mm, g) = (21, 2, 2, 3) for Coral]: (a) Per-base error rates [1 and 2 represents substitution error-correction (Algorithm 1) and homopolymer error-correction (Algorithm 2), respectively.] (b) Measure of concordance (MoC), a similarity measure for pairs of clusterings (c) Running time (the type and quantity of processors used for each case are shown in legend). Fig 6. Comparison of reads correction performance on eight real 454 pyrosequencing datasets (labeled P1–P8; [35]). [parameters: k = 5 (Algorithm 1) and k = 2 (Algorithm 2) for DUDE-Seq; (sPyroNoise, cPyroNoise, sSeqNoise, cSeqNoise) = (60, 0.01, 25, 0.08) for AmpliconNoise; (k, mr, mm, g) = (21, 2, 2, 3) for Coral]: (a) Per-base error rates [1 and 2 represents substitution error-correction (Algorithm 1) and homopolymer error-correction (Algorithm 2), respectively.] (b) Measure of concordance (MoC), a similarity measure for pairs of clusterings (c) Running time (the type and quantity of processors used for each case are shown in legend). https://doi.org/10.1371/journal.pone.0181463.g006 In Fig 6(b), we compare the error correction performance of three schemes, Amplicon- Noise, Coral, and DUDE-Seq, in terms of the MoC. AmpliconNoise assumes a certain statisti- cal model on the DNA sequence and runs an expectation-maximization algorithm for denoising. Here, the two clusterings in the comparison are the golden OTU clusterings and the clusterings returned by denoisers. We observe that for all eight datasets, the number of OTUs generated by DUDE-Seq is consistently closer to the ground truth, providing higher MoC values than those of the other two schemes. Furthermore, Fig 6(c) compares the running time of the three schemes for the eight data- sets. We can clearly see that DUDE-Seq is substantially faster than the other two. Particularly, we stress that the running time of DUDE-Seq, even when implemented and executed with a single CPU, is two orders of magnitude faster than that of parallelized AmpliconNoise, run on four powerful GPUs. We believe that this substantial boost over state-of-the-art schemes with respect to running time is a compelling reason for the adoption of DUDE-Seq in microbial community analysis. PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 It aligns multiple reads by exploiting the k-mer neighborhood of each base read and produces read-by-read correction results for pyrosequencing datasets, similar to DUDE-Seq. Furthermore, as a baseline, we also present ed the error rates for the original, uncorrected sequences (labeled ‘Raw’). We did not include the results of AmpliconNoise [35], a state-of- the-art scheme for 454 pyrosequencing data, in the performance comparison because it does not provide read-by-read correction results, making a fair comparison of the per-base error correction performance with DUDE-Seq difficult. We observeed that DUDE-Seq(1+2), which corrects both substitution errors and homopolymer errors, always outperforms Coral, and the relative error reductions of DUDE-Seq(1+2) with respect to ‘Raw,’ without any denoising, was up to 23.8%. Furthermore, the homopolymer error correction further drives down the error rates obtained by substitution-error correction alone; hence, DUDE-Seq(1+2) always outper- forms DUDE-Seq(1). PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 13 / 25 DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing Table 1. Details of the Illumina datasets [32] used for our experiments shown in Fig 7. dataset ID region sequencer Taq organism forward & reverse primer Q19 V4 MiSeq2 Q5 AT 515 & 805RA Q20 V4 MiSeq2 Q5 BT 515 & 805RA Q21 V4 MiSeq2 Q5 BV 515 & 805RA Q22 V4 MiSeq2 Q5 CS 515 & 805RA Q23 V4 MiSeq2 HF AT 515 & 805RA Q24 V4 MiSeq2 HF BT 515 & 805RA Q25 V4 MiSeq2 HF BV 515 & 805RA Q26 V4 MiSeq2 HF CS 515 & 805RA Q27 V3/V4 MiSeq1 Q5 AT 314f & 806rcb Q28 V3/V4 MiSeq1 Q5 BT 314f & 806rcb Q29 V3/V4 MiSeq1 Q5 BV 314f & 806rcb Q30 V3/V4 MiSeq1 Q5 CS 314f & 806rcb Q31 V3/V4 MiSeq1 HF AT 314f & 806rcb Taqs: HiFI Kapa (HF), Q5 neb (Q5); Organisms: Anaerocellum thermophilum Z-1320 DSM 6725 (AT), Bacteroides thetaiotaomicron VPI-5482 (BT), Bacteroides vulgatus ATCC 8482 (BV), Caldicellulosiruptor saccharolyticus DSM 8903 (CS), Herpetosiphon aurantiacus ATCC 23779 (HA), Rhodopirellula baltica SH 1 (RBS), Leptothrix cholodnii SP-6 (LC) Table 1. Details of the Illumina datasets [32] used for our experiments shown in Fig 7. Caldicellulosiruptor saccharolyticus DSM 8903) targeting two hypervariable regions, V3 and V4, using different configurations (see the caption for Table 1 and Fig 7 for details). To exam- ine how the number of reads in a dataset affects denoising performance, we derived 10 subsets from the original datasets by randomly subsampling 10,000 to 100,000 reads in increments of 10,000 reads. In addition to Coral, we compared the performance of DUDE-Seq with that of BLESS [48], fermi [49], and Trowel [25], which are representative k-mer-based state-of-the-art tools. BLESS corrects “weak” k-mers that exist between consecutive “solid” k-mers, assuming that a weak k-mer has only one error. Fermi corrects sequencing errors in underrepresented k- mers using a heuristic cost function based on quality scores and does not rely on a k-mer occurrence threshold. Trowel does not use a coverage threshold for its k-mer spectrum and iteratively boosts the quality values of bases after making corrections with k-mers that have high quality values. Fig 7 shows the per-base error rates, defined in Eq (6), for the tools under comparison using the first eight datasets (Q19–Q26) and their subsets created as described above (thus, a total of 80 datasets per tool). Real data: Illumina sequencing Illumina platforms, such as GAIIx, MiSeq, and HiSeq, are currently ubiquitous platforms in genome analysis. These platforms intrinsically generate paired-end reads (forward and reverse reads), due to the relatively short reads compared to those obtained by automated Sanger sequencing [46]. Merging the forward and reverse reads from paired-end sequencing yeilds elongated reads (e.g., 2 × 300 bp for MiSeq) that improve the performance of downstream pipelines [47]. Illumina platforms primarily inject substitution errors. A realistic error model is not the DMC, though, as the error rates of the Illumina tend to increase from the beginning to the end of reads. Thus, the assumptions under which the DUDE was originally developed do not exactly apply to the error model of Illumina. In our experiments with DUDE-Seq, however, we still used the empirically obtained DMC model P in Fig 4, which was computed by averaging all error rates throughout different Illumina platforms. In our experiments, we used 13 real Illumina datasets (named Q19–Q31) reported previ- ously [32], including sequencing results from four organisms (Anaerocellum thermophilum Z- 1320 DSM 6725, Bacteroides thetaiotaomicron VPI-5482, Bacteroides vulgatus ATCC 8482, and PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 14 / 25 Taqs: HiFI Kapa (HF), Q5 neb (Q5); Organisms: Anaerocellum thermophilum Z-1320 DSM 6725 (AT), Bacteroides thetaiotaomicron VPI-5482 (BT), Bacteroides vulgatus ATCC 8482 (BV), Caldicellulosiruptor saccharolyticus DSM 8903 (CS), Herpetosiphon aurantiacus ATCC 23779 (HA), Rhodopirellula baltica SH 1 (RBS), Leptothrix cholodnii SP-6 (LC) PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 https://doi.org/10.1371/journal.pone.0181463.t001 nisms: Anaerocellum thermophilum Z-1320 DSM 6725 (AT), Bacteroides thetaiotaomicron VPI-5482 (BT), DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing Fig 7. Comparison of reads correction performance on real Illumina datasets (labeled Q19–Q26; see Table 1 for more details). [parameters: (k, mr, mm, g) = (21, 1, 1, 1000) for Coral; k = 21 for Trowel; (k, O, C, s) = (21, 3, 0.3, 5) for fermi; k = 5 for DUDE-Seq; no BLESS result shown since it did not work on these data] [Organisms: Anaerocellum thermophilum Z-1320 DSM 6725 (Q19 and Q23), Bacteroides thetaiotaomicron VPI-5482 (Q20 and Q24), Bacteroides vulgatus ATCC 8482 (Q21 and Q25), Caldicellulosiruptor saccharolyticus DSM 8903 (Q22 and Q26)] [Q19–Q22: Miseq (Library: nested single index, Taq: Q5 neb, Primer: 515 & 805RA)] [Q23–Q26: Miseq (Library: NexteraXT, Taq: Q5 neb, Primer: 341f & 806rcb)]. https://doi.org/10.1371/journal.pone.0181463.g007 Fig 7. Comparison of reads correction performance on real Illumina datasets (labeled Q19–Q26; see Table 1 for more details). [parameters: (k, mr, mm, g) = (21, 1, 1, 1000) for Coral; k = 21 for Trowel; (k, O, C, s) = (21, 3, 0.3, 5) for fermi; k = 5 for DUDE-Seq; no BLESS result shown since it did not work on these data] [Organisms: Anaerocellum thermophilum Z-1320 DSM 6725 (Q19 and Q23), Bacteroides thetaiotaomicron VPI-5482 (Q20 and Q24), Bacteroides vulgatus ATCC 8482 (Q21 and Q25), Caldicellulosiruptor saccharolyticus DSM 8903 (Q22 and Q26)] [Q19–Q22: Miseq (Library: nested single index, Taq: Q5 neb, Primer: 515 & 805RA)] [Q23–Q26: Miseq (Library: NexteraXT, Taq: Q5 neb, Primer: 341f & 806rcb)]. https://doi.org/10.1371/journal.pone.0181463.g007 sequence, but it can have a low error rate calculated as in Eq (6). In our experiments with the datasets Q27-Q31, we detected a large variance in the number of aligned bases across different denoising tools; thus, it was difficult to make a fair comparison among the performance of dif- ferent tools with Eq (6). We note that in the experiments presented in Figs 6(a) and 7, such a large variance was not detected. To alleviate this issue, we employ the alternative definition of the per-base error rate of a tool in Eq (8). Fig 8 shows the results obtained for 100,000-read subsets of each of the Q19–Q31 datasets, i.e., all datasets, for DUDE-Seq and the four alternative denoisers. Because the datasets Q27– Q31 had two subsets of 100,000 reads, we used a total of 18 datasets to draw Fig 8, one each from Q19–Q26 and two each from Q27–Q31. BLESS did not run successfully on these datasets, and hence its results are not shown. First, we can confirm that DUDE-Seq is effective in reducing substitu- tion errors for data obtained using the Illumina platform in all tested cases of targeted ampli- con sequencing, with relative error rate reductions of 6.40–49.92%, compared to the ‘Raw’ sequences. Furthermore, among the tools included in the comparison, DUDE-Seq produced the best results for the largest number of datasets. For Q24 and Q25, fermi was most effective, but was outperformed by DUDE-Seq in many other cases. Coral was able to denoise to some extent but was inferior to DUDE-Seq and fermi. Trowel gave unsatisfactory results in this experiment. Before presenting our next results, we note that while the error rate defined in Eq (6) is widely used for DNA sequence denoising research as a performance measure, it occasionally misleading and cannot be used to fairly evaluate the performance of denoisers. This is because only errors at aligned bases are counted in the error rate calculation; hence, a poor denoiser may significantly reduce the number of aligned bases, potentially further corrupting the noisy PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 15 / 25 DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing Experiments on simulated data We performed more detailed experiments using Illumina simulators in order to further high- light the strong denoising performance of DUDE-Seq, including the effects on downstream analyses. Fig 9(a) shows the results obtained using the Grinder simulator [50] and a comparison with Coral. Trowel and Reptile require quality scores as input, which are provided by the GemSIM simulator, but not by the Grinder simulator; hence, we could not include Trowel and Reptile in Fig 9(a). We generated nine synthetic datasets of forward reads that had error rates at the end of the sequence varying from 0.2% to 1.0%, as denoted on the horizontal axis. For all cases, the error rate at the beginning of the sequence was 0.1%. We again used the average DMC model for the entire sequence for DUDE-Seq. Note that the error rates for the ‘Raw’ data, i.e., the red bars, match the average of the error rates at the beginning and the end of the sequence. From the figure, consistent with the real datasets analyzed in Section, we clearly see that DUDE-Seq significantly outperforms Coral for all tested error rates. To evaluate the performance of DUDE-Seq for paired-end reads, we generated datasets, shown in Table 2, with the GemSIM sequencing data simulator [51]. As shown in the table, we used 23 public reference sequences [35] to generate the dataset A5 and a single reference sequence for S5. We used the error model v5 that has error rate s of 0.28% for forward reads and 0.34% for reverse reads. In Fig 9(b), in addition to DUDE-Seq, Coral, fermi, and Trowel, we included the results obtained using Reptile [20], another k-mer spectrum-based method that outputs read-by-read denoising results. We again observe from the figure that DUDE-Seq outperforms the alternatives by significant margins. In Table 3, we show that the error-corrected reads produced by DUDE-Seq can also improve the performance of downstream pipelines, such as paired-end merging. We applied four different paired-end merging schemes, CASPER [52], COPE [53], FLASH [47], and PAN- DAseq [54], for the two datasets A5 and S5 in Table 2. The metrics are defined as usual. A true positive (TP) is defined as a merge with correct mismatching resolution in the overlap region, and a false positive (FP) is defined as a merge with incorrect mismatching resolution in the overlap region. DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing As mentioned previously, BLESS could not run successfully on Q19–Q26; hence, there are only 10 points for BLESS in the plots. Fig 8(a), 8(b) and 8(c) presents the distribution of gð^etoolÞ, g(atool), and running times for each tool, respec- tively. For each distribution, the average value is marked with a solid circle. As shown in Fig 8 (b), we clearly see that Coral and Trowel show a large variance in the number of aligned bases. Fig 8. Performance comparison. (a) Relative gain of adjusted error rates over ‘Raw’ data Eq (9). (b) Relative gain of aligned bases Eq (7). (c) Running time on real Illumina datasets (labeled Q19–Q31; see the caption for Fig 7). [parameters: kmerlength = 21 for BLESS; (k, mr, mm, g) = (21, 1, 1, 1000) for Coral; k = 21 for Trowel; (k, O, C, s) = (21, 3, 0.3, 5) for fermi; k = 5 for DUDE-Seq] [BLESS did not work on Q19–Q26]. https://doi.org/10.1371/journal.pone.0181463.g008 Fig 8. Performance comparison. (a) Relative gain of adjusted error rates over ‘Raw’ data Eq (9). (b) Relative gain of aligned bases Eq (7). (c) Running time on real Illumina datasets (labeled Q19–Q31; see the caption for Fig 7). [parameters: kmerlength = 21 for BLESS; (k, mr, mm, g) = (21, 1, 1, 1000) for Coral; k = 21 for Trowel; (k, O, C, s) = (21, 3, 0.3, 5) for fermi; k = 5 for DUDE-Seq] [BLESS did not work on Q19–Q26]. https://doi.org/10.1371/journal.pone.0181463.g008 PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 16 / 25 DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing For example, Coral only aligns 30% of bases compared to the raw sequence after denoising for some datasets. With the effect of this variance in aligned bases adjusted, Fig 8(a) shows that DUDE-Seq produces the highest average gð^etoolÞ, i.e., 19.79%, among all the compared tools. Furthermore, the variability of g(atool) was the smallest for DUDE-Seq, as shown in Fig 8(b), suggesting its robustness. Finally, in Fig 8(c), we observe that the running times were signifi- cantly shorter for DUDE-Seq and Trowel than for Coral and fermi. Overall, we can conclude that DUDE-Seq is the most robust tool, with a fast running time and the highest average accu- racy after denoising. In summary, we observe from Figs 7 and 8 that DUDE-Seq robustly outperforms the com- peting schemes for most of the datasets tested. DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing We specifically emphasize that DUDE-Seq shows a strong performance, even though the DMC assumption does not hold for the sequencer. We believe that the better performance of DUDE-Seq relative to other state-of-the- art algorithms (based on MSA or k-mer spectrums) on real Illumina datasets strongly demon- strates the competitiveness of DUDE-Seq as a general DNA sequence denoiser for targeted amplicon sequencing. PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 The accuracy and F1 score results show that correcting errors with DUDE-Seq consistently yields better paired-end merg- ing performance, not only compared to the case with no denoising, but also compared to the d-end reads merging performance statistics [parameters: k = 5 for DUDE-Seq; (k, mr, mm, g) = (21, 1, 1, 1000) or fermi] Table 3. Paired-end reads merging performance statistics [parameters: k = 5 for DUDE-Seq; (k, mr, mm, g) = (21, 3, 0.3, 5) for fermi]. Table 3. Paired-end reads merging performance statistics [parameters: k = 5 for DUDE-Seq; (k, mr, mm, g) = (21, 1, 1, 1000) for Coral; (k, O, C, s) = (21, 3, 0.3, 5) for fermi]. Experiments on simulated data Furthermore, a false negative (FN) is a merge that escapes the detection, and a true negative (TN) is defined as a correct prediction for reads that do not truly overlap. The accuracy and F1 score are computed based on the above metrics [55]. For each dataset, we compared the results for four cases: performing paired-end merging without any denoising, PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 17 / 25 DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing Fig 9. Reads correction performance on simulated dataset. [parameters: k = 5 for DUDE-Seq; k = 10 fo Trowel; (k, mr, mm, g) = (21, 1, 1, 1000) for Coral; optimal values set by tool seq-analy for Reptile; (k, O, C s) = (21, 3, 0.3, 5) for fermi]: (a) Varying error rates using the Grinder simulator [50]. (b) Varying reads composition using the GemSIM simulator [51] (values on top of each bar represent the error rates). https://doi org/10 1371/journal pone 0181463 g009 DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequenc Fig 9. Reads correction performance on simulated dataset. [parameters: k = 5 for DUDE-Seq; k = 10 for Trowel; (k, mr, mm, g) = (21, 1, 1, 1000) for Coral; optimal values set by tool seq-analy for Reptile; (k, O, C, s) = (21, 3, 0.3, 5) for fermi]: (a) Varying error rates using the Grinder simulator [50]. (b) Varying reads composition using the GemSIM simulator [51] (values on top of each bar represent the error rates). https://doi.org/10.1371/journal.pone.0181463.g009 Fig 9. Reads correction performance on simulated dataset. [parameters: k = 5 for DUDE-Seq; k = 10 for Trowel; (k, mr, mm, g) = (21, 1, 1, 1000) for Coral; optimal values set by tool seq-analy for Reptile; (k, O, C, s) = (21, 3, 0.3, 5) for fermi]: (a) Varying error rates using the Grinder simulator [50]. (b) Varying reads composition using the GemSIM simulator [51] (values on top of each bar represent the error rates). https://doi.org/10.1371/journal.pone.0181463.g009 PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 18 / 25 DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing after correcting errors with Coral, after correcting errors with fermi, and after correcting errors with DUDE-Seq. Reptile and Trowel were not included in this experiment because they were generally outperformed by Coral and fermi, as shown in Fig 9(b). The accuracy and F1 score results show that correcting errors with DUDE-Seq consistently yields better paired-end merg- ing performance, not only compared to the case with no denoising, but also compared to the Table 2. Details of the public data [52] used for our experiments on simulated data shown in Table 3. dataset ID # total reads # refs fragment length read length overlap length simulator (error model) or sequencer used S5 1,000,000 [1] 160 100 40 GemSIM (v5#) A5 1,000,000 [23] 160–190 100 10–40 GemSIM (v5 #) # Error model v5 (forward rate 0.28%, reverse 0.34%) https://doi.org/10.1371/journal.pone.0181463.t002 Table 3. Paired-end reads merging performance statistics [parameters: k = 5 for DUDE-Seq; (k, mr, mm, g) = (21, 1, 1, 1000) for Coral; (k, O, C, s) = (21, 3, 0.3, 5) for fermi]. Table 2. Details of the public data [52] used for our experiments on simulated data shown in Table 3. dataset ID # total reads # refs fragment length read length overlap length simulator (error model) or sequencer used S5 1,000,000 [1] 160 100 40 GemSIM (v5#) A5 1,000,000 [23] 160–190 100 10–40 GemSIM (v5 #) # Error model v5 (forward rate 0.28%, reverse 0.34%) https://doi org/10 1371/journal pone 0181463 t002 Table 2. Details of the public data [52] used for our experiments on simulated data shown in Table 3. https://doi.org/10.1371/journal.pone.0181463.t002 after correcting errors with Coral, after correcting errors with fermi, and after correcting errors with DUDE-Seq. Reptile and Trowel were not included in this experiment because they were generally outperformed by Coral and fermi, as shown in Fig 9(b). The accuracy and F1 score results show that correcting errors with DUDE-Seq consistently yields better paired-end merg- ing performance, not only compared to the case with no denoising, but also compared to the after correcting errors with Coral, after correcting errors with fermi, and after correcting errors with DUDE-Seq. Reptile and Trowel were not included in this experiment because they were generally outperformed by Coral and fermi, as shown in Fig 9(b). https://doi.org/10.1371/journal.pone.0181463.t003 Discussion Our experimental results show that DUDE-Seq can robustly outperform k-mer-based, MSA- based, and statistical error model-based schemes for both real-world datasets, such as 454 pyr- osequencing and Illumina data, and simulated datasets, particularly for targeted amplicon sequencing. This performance advantage in denoising further allowed us to obtain improved results in downstream analysis tasks, such as OTU binning and paired-end merging. Further- more, the time demand of DUDE-Seq-based OTU binning is order(s) of magnitude lower than that of the current state-of-the-art schemes. We also demonstrated the robustness and flexibility of DUDE-Seq by showing that a simple change in P matrix is enough to apply the exact same DUDE-Seq to data obtained using different sequencing platforms. In particular, we experimentally showed that even when the memoryless channel assumption does not hold, as in Illumina data, DUDE-Seq still solidly outperforms state-of-the-art schemes. The sliding window nature of DUDE-Seq resemble s the popular k-mer-based schemes in the literature. However, while all existing k-mer-based schemes rely on heuristic threshold selection for determining errors in the reads, regardless of the error model of the sequencing platform, DUDE-Seq applies an analytic denoising rule that explicitly takes the error model P into account. Therefore, even for identical noisy reads zn, DUDE-Seq may result in different denoised sequences, depending on the P’s of different sequencing platforms, whereas the k- mer-based scheme will always result in the exact same denoised sequence. The performance gains reported in this paper compared to state-of-the-art baselines, including those for k-mer- based schemes, substantiate the competitiveness of our method for targeted amplicon sequencing. Another advantage of DUDE-Seq is its read-by-read error-correction capability. This fea- ture is important for a number of bioinformatics tasks, including de novo sequencing, metage- nomics, resequencing, targeted resequencing, and transcriptome sequencing, which typically require the extraction of subtle information from small variants in each read. In addition to the types of tasks presented in this paper (i.e., per-based error correction, OTU binning, and paired-end merging), we plan to apply DUDE-Seq to additional tasks, as mentioned above. Additional venues for further investigation include the procedure for estimating the noise mechanism represented by P, which is currently empirically determined by aligning each read to the reference sequence and is therefore sensitive to read mapping and alignment. For more robust estimation, we may employ an expectation-maximization-based algorithm, as was recently proposed for estimating substitution emissions for the data obtained using nanopore technology [56]. tool dataset # merges TP FP FN accuracy F1 CASPER S5 1,000,000 997,303 2,697 0 0.997 0.999 COPE 974,219 961,366 12,853 25,781 0.961 0.980 FLASH 999,921 977,431 22,490 79 0.977 0.989 PANDAseq 999,947 976,807 23,140 53 0.977 0.988 CASPER S5 w/ Coral 1,000,000 997,510 2,490 0 0.998 0.999 COPE 975,803 963,717 12,086 24,197 0.964 0.982 FLASH 999,942 978,835 21,107 58 0.979 0.989 PANDAseq 999,949 978,270 21,679 51 0.978 0.989 CASPER S5 w/ fermi 1,000,000 997,356 2,644 0 0.997 0.999 COPE 994,025 969,451 24,574 5,975 0.969 0.984 FLASH 999,933 972,025 27,908 67 0.972 0.986 PANDAseq 999,952 971,567 28,385 48 0.972 0.986 CASPER S5 w/ DUDE-Seq 1,000,000 999,320 680 0 0.999 1.000 COPE 987,238 983,639 3,599 12,762 0.984 0.992 FLASH 999,958 992,915 7,043 42 0.993 0.996 PANDAseq 999,949 991,146 8,803 51 0.991 0.996 CASPER A5 999,973 997,202 2,771 27 0.997 0.999 COPE 924,634 915,981 8,653 75,366 0.916 0.956 FLASH 999,578 977,355 22,223 422 0.977 0.989 PANDAseq 999,122 978,720 20,402 878 0.979 0.989 CASPER A5 w/ Coral 999,974 995,899 4,075 26 0.996 0.998 COPE 927,757 918,733 9,024 72,243 0.919 0.958 FLASH 999,742 978,814 20,928 258 0.979 0.989 PANDAseq 999,351 979,899 19,452 649 0.980 0.990 CASPER A5 w/ fermi 999,969 997,288 2,681 31 0.997 0.999 COPE 939,986 923,252 16,734 60,014 0.923 0.960 FLASH 999,732 974,903 24,829 268 0.975 0.987 PANDAseq 999,328 975,320 24,008 672 0.975 0.988 CASPER A5 w/ DUDE-Seq 999,971 998,078 1,893 29 0.998 0.999 COPE 943,531 939,555 3,976 56,469 0.940 0.969 FLASH 999,638 989,860 9,778 362 0.990 0.995 PANDAseq 999,354 989,250 10,104 646 0.989 0.995 htt //d i /10 1371/j l 0181463 t003 mance statistics [parameters: k = 5 for DUDE-Seq; (k, mr, mm, g) = (21, 1, 1, 1000) for Coral; (k, O, C, s) = Table 3. Paired-end reads merging performance statistics [parameters: k = 5 for DUDE-Seq; (k, mr, mm, g) = (21, 1, 1, 1000) for Coral; (k, O, C, s) = (21, 3, 0.3, 5) for fermi]. PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 19 / 25 DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing cases with Coral and fermi error correct ion. This result highlights the potential application of DUDE-Seq for boosting the performance of downstream DNA sequence analyses. PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 Supporting information S1 File. Fig A, DUDE-Seq web interface. This is a screenshot of the website accompanying the proposed DUDE-Seq method (http://data.snu.ac.kr/pub/dude-seq). For users who prefer a graphical user interface, this website provides a web-based execution environments for DUDE-Seq. Through this screen, a user can specify the parameters for each of the two error types (in the figure, DUDE-Seq (1) stands for for the substitution error correction described in Algorithm 1 and DUDE-Seq (2) stands for the homopolymer error correction shown in Algo- rithm 2), and upload the input file of her choice. The DUDE-Seq process starts automatically by clicking the “SUBMIT” button. For advanced users who prefer batch processing, the source code of DUDE-Seq is also available at http://github.com/datasnu/dude-seq. All the used data- sets are also available on the Sequence Read Archive (SRA) under the accession number SRP000570 (SRS002051–SRS002053) at https://www.ncbi.nlm.nih.gov/sra/SRP000570 and the European Nucleotide Archive (ENA) under the accession number PRJEB6244 (ERS671332– ERS671344) at http://www.ebi.ac.uk/ena/data/view/PRJEB6244. Fig B, Website output: sequence complexity. The DUDE-Seq website provides analysis results from applying the DUST algorithm [60] and block-entropy to the outputs from denoising by DUDE-Seq. The DUST algorithm masks low-complexity regions that have highly biased distribution of nucleo- tides based on counting 3-mer frequencies in 64-base windows. The DUST score is computed based on how often different trinucleotides occur as follows: score ¼ X k i¼1 niðni 1Þðw 2Þs 2ðl 1Þl where k = 43 is the trinucleotide size, w = 64 is the window size, ni is the number of the words i in a window, l is the number of the possible words in a window, and s is the scaling factor. The score is scaled from 0 to 100 and a high score implies a low complexity metagenome. The block-entropy is calculated using Shannon’s diversity index [61]. The block-entropy evaluates the entropy of the trinucleotides in a sequence as follows: entropy ¼ X k i¼1 ð ni l Þlogkðni l Þ where k = 43 is the trinucleotide size, ni is the number of the words i in a window, and l is the number of the possible words in a window. The entropy is also scaled from 0 to 100 and a low entropy implies a low complexity metagenome. Fig C, Website output: tag sequence proba- bility. Discussion Considering uncertainties in P may also be helpful; hence, it may be useful to investigate the relevance of the framework in [57]. Additionally, it will likely be fruitful to uti- lize the information in Phred quality scores to make decisions about noisy bases and to fine- tune the objective loss function in our approach. Using a lossy compressed version of the qual- ity scores is one possible direction for boosting the inferential performance of some down- stream applications, as shown in [58]. Furthermore, particularly for the homopolymer error correction, there are several hyperparameters whose choices can be experimented with in the future to potentially achieve substantial performance boosts. Examples include the choice of alphabet size (in lieu of the current value of 10), the choice of the loss function that may be pro- portional to the difference between the true and estimated value of N (in lieu of the current Hamming loss), and the choice of quantization (in lieu of Eq (4)). Moreover, we may apply the full generalized DUDE in [34] for homopolymer error correction to determine if better PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 20 / 25 DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing performance can be achieved at the cost of increased complexity. Applying DUDE-Seq to other types of sequencing technology with homopolymer errors (e.g., Ion Torrent) would also be possible as long as we can acquire flow (e.g., ionogram) density distributions to estimate Γ. Currently, there exists no public data repository that includes such information for Ion Tor- rent, and thus existing Ion Torrent denoisers often ignore homopolymer errors or rely on sim- plistic noise modeling and iterative updates that unrealistically limit the maximum length of homopolymer errors that can be handled, let alone computational efficiency [36]. Finally, we plan to test DUDE-Seq on several other sequencing platforms, such as PacBio and Oxford Nanopore, which tend to result in longer and more noisy sequences, to further substantiate the robustness and effectiveness of our algorithm. Applying the recently developed deep neural networks -based Neural DUDE algorithm [59] to DNA sequence denoising beyond targeted amplicon sequencing could be another fruitful direction. PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 Supporting information Another output from the DUDE-Seq website is the tag sequence probability of reads where k = 43 is the trinucleotide size, ni is the number of the words i in a window, and l is the number of the possible words in a window. The entropy is also scaled from 0 to 100 and a low entropy implies a low complexity metagenome. Fig C, Website output: tag sequence proba- bility. Another output from the DUDE-Seq website is the tag sequence probability of reads PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 21 / 25 DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing [62]. This is to reveal the existence of artifacts at the ends, i.e., adapter or barcode sequences at the 5’- or 3’-end. Fig D, Website output: sequence duplication. The accompanying website also carries out sequence duplication analysis based on the denoised outputs from DUDE-Seq, in order to reveal artificial duplicates. As shown in the figure, five types of duplication statistics [63] are provided: exact duplicates, 5’ duplicates, 3’ duplicates, exact duplicates with the reverse complement of another sequence, and 5’ or 3’ duplicates with the reverse complement of another sequence. (PDF) Author Contributions Conceptualization: Taesup Moon, Sungroh Yoon. Data curation: Byunghan Lee. Formal analysis: Byunghan Lee, Taesup Moon, Sungroh Yoon, Tsachy Weissman. Funding acquisition: Taesup Moon, Sungroh Yoon. Investigation: Byunghan Lee, Taesup Moon, Sungroh Yoon, Tsachy Weissman. Methodology: Byunghan Lee, Taesup Moon, Sungroh Yoon, Tsachy Weissman. Investigation: Byunghan Lee, Taesup Moon, Sungroh Yoon, Tsachy Weissman. Methodology: Byunghan Lee, Taesup Moon, Sungroh Yoon, Tsachy Weissman. Software: Byunghan Lee. Supervision: Sungroh Yoon. Supervision: Sungroh Yoon. Writing – original draft: Byunghan Lee, Taesup Moon, Sungroh Yoon, Tsachy Weissman. Writing – original draft: Byunghan Lee, Taesup Moon, Sungroh Yoon, Tsachy Weissman. Writing – review & editing: Byunghan Lee, Taesup Moon, Sungroh Yoon, Tsachy Weissman. Writing – review & editing: Byunghan Lee, Taesup Moon, Sungroh Yoon, Tsachy Weissman. Acknowledgments This work was supported in part by the National Research Foundation of Korea (NRF) grant funded by the Korea government (Ministry of Science, ICT and Future Planning) [2014M3A9E2064434 and 2016M3A7B4911115], in part by a grant of the Korea Health Tech- nology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare [HI14C3405030014], in part by the Basic Science Research Program through the National Research Foundation of Korea [NRF- 2016R1C1B2012170], in part by the ICT R&D program of MSIP/IITP [2016-0-00563, Research on Adaptive Machine Learning Technology Development for Intelligent Autonomous Digital Companion], and in part by NIH Grant 5U01CA198943-03. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. References 1. Metzker ML. Sequencing technologies—the next generation. Nature Reviews Genetics. 2010; 11(1): 31–46. https://doi.org/10.1038/nrg2626 PMID: 19997069 1. Metzker ML. Sequencing technologies—the next generation. Nature Reviews Genetics. 2010; 11(1): 31–46. https://doi.org/10.1038/nrg2626 PMID: 19997069 2. Astbury WT. Molecular biology or ultrastructural biology? 1961;. PMID: 13684868 3. Bateson W. Materials for the Study of Variation, Treated with Especial Regard to Discontinuity in the Origin of Species. Macmillan; 1894. 3. Bateson W. Materials for the Study of Variation, Treated with Especial Regard to Discontinuity in the Origin of Species. Macmillan; 1894. 4. Riesenfeld CS, Schloss PD, Handelsman J. Metagenomics: genomic analysis of microbial communi- ties. Annu Rev Genet. 2004; 38:525–552. https://doi.org/10.1146/annurev.genet.38.072902.091216 PMID: 15568985 4. Riesenfeld CS, Schloss PD, Handelsman J. Metagenomics: genomic analysis of microbial communi- ties. Annu Rev Genet. 2004; 38:525–552. https://doi.org/10.1146/annurev.genet.38.072902.091216 PMID: 15568985 PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 22 / 25 DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing 5. Pop M, Salzberg SL. Bioinformatics challenges of new sequencing technology. Trends in Genetics. 2008; 24(3):142–149. 5. Pop M, Salzberg SL. Bioinformatics challenges of new sequencing technology. Trends in Genetics. 2008; 24(3):142–149. 6. Shendure J, Ji H. Next-generation DNA sequencing. Nature biotechnology. 2008; 26(10):1135–1145. https://doi.org/10.1038/nbt1486 PMID: 18846087 7. Goodwin S, McPherson JD, McCombie WR. Coming of age: ten years of next-generation sequencing technologies. Nature Reviews Genetics. 2016; 17(6):333–351. https://doi.org/10.1038/nrg.2016.49 PMID: 27184599 8. Bamshad MJ, Ng SB, Bigham AW, Tabor HK, Emond MJ, Nickerson DA, et al. Exome sequencing as a tool for Mendelian disease gene discovery. Nature Reviews Genetics. 2011; 12(11):745–755. https:// doi.org/10.1038/nrg3031 PMID: 21946919 9. Jamuar SS, Tan EC. Clinical application of next-generation sequencing for Mendelian diseases. Human genomics. 2015; 9(1):1. https://doi.org/10.1186/s40246-015-0031-5 10. Yang X, Chockalingam SP, Aluru S. A survey of error-correction methods for next-generation sequenc- ing. Briefings in bioinformatics. 2013; 14(1):56–66. https://doi.org/10.1093/bib/bbs015 PMID: 22492192 11. Ilie L, Fazayeli F, Ilie S. HiTEC: accurate error correction in high-throughput sequencing data. Bioinfor- matics. 2011; 27(3):295–302. https://doi.org/10.1093/bioinformatics/btq653 PMID: 21115437 12. Kao WC, Chan AH, Song YS. ECHO: a reference-free short-read error correction algorithm. Genome research. 2011; 21(7):1181–1192. https://doi.org/10.1101/gr.111351.110 PMID: 21482625 13. Kelley DR, Schatz MC, Salzberg SL, et al. Quake: quality-aware detection and correction of sequencing errors. Genome Biol. 2010; 11(11):R116. https://doi.org/10.1186/gb-2010-11-11-r116 PMID: 21114842 14. Qu W, Hashimoto Si, Morishita S. Efficient frequency-based de novo short-read clustering for error trim- ming in next-generation sequencing. Genome research. 2009; 19(7):1309–1315. https://doi.org/10. 1101/gr.089151.108 PMID: 19439514 15. Salmela L. PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 References https://doi.org/10.1093/ bioinformatics/btu440 PMID: 25161220 30. Ewing B, Hillier L, Wendl MC, Green P. Base-calling of automated sequencer traces using Phred. I. Accuracy assessment. Genome research. 1998; 8(3):175–185. https://doi.org/10.1101/gr.8.3.175 PMID: 9521921 31. Weissman T, Ordentlich E, Seroussi G, Verdu´ S, Weinberger MJ. Universal discrete denoising: Known channel. IEEE Transactions on Information Theory. 2005; 51(1):5–28. https://doi.org/10.1109/TIT. 2004.839518 32. Schirmer M, Ijaz UZ, D’Amore R, Hall N, Sloan WT, Quince C. Insight into biases and sequencing errors for amplicon sequencing with the Illumina MiSeq platform. Nucleic acids research. 2015; p. gku1341. 33. Yan B, Hu Y, Ng C, Ban KH, Tan TW, Huan PT, et al. Coverage analysis in a targeted amplicon-based next-generation sequencing panel for myeloid neoplasms. Journal of clinical pathology. 2016; p. jclin- path–2015. https://doi.org/10.1136/jclinpath-2015-203580 34. Dembo A, Weissman T. Universal denoising for the finite-input general-output channel. Information Theory, IEEE Transactions on. 2005; 51(4):1507–1517. https://doi.org/10.1109/TIT.2005.844104 35. Quince C, Lanzen A, Davenport RJ, Turnbaugh PJ. Removing noise from pyrosequenced amplicons. BMC bioinformatics. 2011; 12(1):38. https://doi.org/10.1186/1471-2105-12-38 PMID: 21276213 36. Bragg LM, Stone G, Butler MK, Hugenholtz P, Tyson GW. Shining a light on dark sequencing: charac- terising errors in Ion Torrent PGM data. PLoS Comput Biol. 2013; 9(4):e1003031. https://doi.org/10. 1371/journal.pcbi.1003031 PMID: 23592973 37. Fichot EB, Norman RS. Microbial phylogenetic profiling with the Pacific Biosciences sequencing plat- form. Microbiome. 2013; 1(1):10. https://doi.org/10.1186/2049-2618-1-10 PMID: 24450498 38. Marinier E, Brown DG, McConkey BJ. Pollux: platform independent error correction of single and mixed genomes. BMC bioinformatics. 2015; 16(1):10. https://doi.org/10.1186/s12859-014-0435-6 PMID: 25592313 39. Li H, Durbin R. Fast and accurate short read alignment with Burrows–Wheeler transform. Bioinformat- ics. 2009; 25(14):1754–1760. https://doi.org/10.1093/bioinformatics/btp324 PMID: 19451168 40. Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, et al. The sequence alignment/map format and SAMtools. Bioinformatics. 2009; 25(16):2078–2079. https://doi.org/10.1093/bioinformatics/btp352 PMID: 19505943 41. LaFave MC, Burgess SM. sam2pairwise version 1.0.0; 2014. Available from: https://doi.org/10.5281/ zenodo.11377. 42. Pfitzner D, Leibbrandt R, Powers D. Characterization and evaluation of similarity measures for pairs of clusterings. Knowledge and Information Systems. 2009; 19(3):361–394. https://doi.org/10.1007/ s10115-008-0150-6 43. Reeder J, Knight R. Rapid denoising of pyrosequencing amplicon data: exploiting the rank-abundance distribution. Nature methods. 2010; 7(9):668. https://doi.org/10.1038/nmeth0910-668b PMID: 20805793 44. Schloss PD, Handelsman J. Introducing DOTUR, a computer program for defining operational taxo- nomic units and estimating species richness. Applied and environmental microbiology. 2005; 71(3): 1501–1506. https://doi.org/10.1128/AEM.71.3.1501-1506.2005 PMID: 15746353 45. Edgar RC. Search and clustering orders of magnitude faster than BLAST. Bioinformatics. 2010; 26(19):2460–2461. References Correction of sequencing errors in a mixed set of reads. Bioinformatics. 2010; 26(10): 1284–1290. https://doi.org/10.1093/bioinformatics/btq151 PMID: 20378555 16. Salmela L, Schro¨der J. Correcting errors in short reads by multiple alignments. Bioinformatics. 2011; 27(11):1455–1461. https://doi.org/10.1093/bioinformatics/btr170 PMID: 21471014 17. Schro¨der J, Schro¨der H, Puglisi SJ, Sinha R, Schmidt B. SHREC: a short-read error correction method. Bioinformatics. 2009; 25(17):2157–2163. https://doi.org/10.1093/bioinformatics/btp379 PMID: 19542152 18. Wijaya E, Frith MC, Suzuki Y, Horton P. Recount: expectation maximization based error correction tool for next generation sequencing data. In: Genome Inform. vol. 23. World Scientific; 2009. p. 189–201. 19. Yang X, Aluru S, Dorman KS. Repeat-aware modeling and correction of short read errors. BMC bioin- formatics. 2011; 12(Suppl 1):S52. https://doi.org/10.1186/1471-2105-12-S1-S52 PMID: 21342585 20. Yang X, Dorman KS, Aluru S. Reptile: representative tiling for short read error correction. Bioinformat- ics. 2010; 26(20):2526–2533. https://doi.org/10.1093/bioinformatics/btq468 PMID: 20834037 21. Laehnemann D, Borkhardt A, McHardy AC. Denoising DNA deep sequencing data-high-throughput sequencing errors and their correction. Briefings in Bioinformatics. 2016; 17(1):154–179. https://doi.org/ 10.1093/bib/bbv029 PMID: 26026159 22. Medvedev P, Scott E, Kakaradov B, Pevzner P. Error correction of high-throughput sequencing data- sets with non-uniform coverage. Bioinformatics. 2011; 27(13):i137–i141. https://doi.org/10.1093/ bioinformatics/btr208 PMID: 21685062 23. Nikolenko SI, Korobeynikov AI, Alekseyev MA. BayesHammer: Bayesian clustering for error correction in single-cell sequencing. BMC genomics. 2013; 14(Suppl 1):S7. https://doi.org/10.1186/1471-2164- 14-S1-S7 PMID: 23368723 24. Greenfield P, Duesing K, Papanicolaou A, Bauer DC. Blue: correcting sequencing errors using consen- sus and context. Bioinformatics. 2014; 30(19):2723–2732. https://doi.org/10.1093/bioinformatics/ btu368 PMID: 24919879 25. Lim EC, Mu¨ller J, Hagmann J, Henz SR, Kim ST, Weigel D. Trowel: a fast and accurate error correction module for Illumina sequencing reads. Bioinformatics. 2014; p. btu513. 26. Bragg L, Stone G, Imelfort M, Hugenholtz P, Tyson GW. Fast, accurate error-correction of amplicon pyr- osequences using Acacia. Nature Methods. 2012;9(5):425–426. 27. Meacham F, Boffelli D, Dhahbi J, Martin DI, Singer M, Pachter L. Identification and correction of system- atic error in high-throughput sequence data. BMC bioinformatics. 2011; 12(1):451. https://doi.org/10. 1186/1471-2105-12-451 PMID: 22099972 28. Yin X, Song Z, Dorman K, Ramamoorthy A. PREMIER—PRobabilistic error-correction using Markov inference in errored reads. In: 2013 IEEE International Symposium on Information Theory Proceedings (ISIT); 2013. p. 1626–1630. 23 / 25 PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing 29. Schulz MH, Weese D, Holtgrewe M, Dimitrova V, Niu S, Reinert K, et al. Fiona: a parallel and automatic strategy for read error correction. Bioinformatics. 2014; 30(17):i356–i363. References https://doi.org/10.1093/bioinformatics/btq461 PMID: 20709691 46. Bartram AK, Lynch MD, Stearns JC, Moreno-Hagelsieb G, Neufeld JD. Generation of multimillion- sequence 16S rRNA gene libraries from complex microbial communities by assembling paired-end Illu- mina reads. Applied and environmental microbiology. 2011; 77(11):3846–3852. https://doi.org/10.1128/ AEM.02772-10 PMID: 21460107 47. Magoč T, Salzberg SL. FLASH: fast length adjustment of short reads to improve genome assemblies. Bioinformatics. 2011; 27(21):2957–2963. https://doi.org/10.1093/bioinformatics/btr507 PMID: 21903629 48. Heo Y, Wu XL, Chen D, Ma J, Hwu WM. BLESS: bloom filter-based error correction solution for high- throughput sequencing reads. Bioinformatics. 2014; p. btu030. 49. Li H. Exploring single-sample SNP and INDEL calling with whole-genome de novo assembly. Bioinfor- matics. 2012; 28(14):1838–1844. https://doi.org/10.1093/bioinformatics/bts280 PMID: 22569178 50. Angly FE, Willner D, Rohwer F, Hugenholtz P, Tyson GW. Grinder: a versatile amplicon and shotgun sequence simulator. Nucleic acids research. 2012; 40(12):e94–e94. https://doi.org/10.1093/nar/gks251 PMID: 22434876 24 / 25 PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 DUDE-Seq: Fast, flexible, and robust denoising for targeted amplicon sequencing 51. McElroy KE, Luciani F, Thomas T. GemSIM: general, error-model based simulator of next-generation sequencing data. BMC genomics. 2012; 13(1):74. https://doi.org/10.1186/1471-2164-13-74 PMID: 22336055 52. Kwon S, Lee B, Yoon S. CASPER: context-aware scheme for paired-end reads from high-throughput amplicon sequencing. BMC bioinformatics. 2014; 15(Suppl 9):S10. https://doi.org/10.1186/1471-2105- 15-S9-S10 PMID: 25252785 53. Liu B, Yuan J, Yiu SM, Li Z, Xie Y, Chen Y, et al. COPE: an accurate k-mer-based pair-end reads con- nection tool to facilitate genome assembly. Bioinformatics. 2012; 28(22):2870–2874. https://doi.org/10. 1093/bioinformatics/bts563 PMID: 23044551 54. Masella AP, Bartram AK, Truszkowski JM, Brown DG, Neufeld JD. PANDAseq: paired-end assembler for illumina sequences. BMC bioinformatics. 2012; 13(1):31. https://doi.org/10.1186/1471-2105-13-31 PMID: 22333067 55. Witten IH, Frank E. Data Mining: Practical machine learning tools and techniques. Morgan Kaufmann; 2005. 56. Jain M, Fiddes IT, Miga KH, Olsen HE, Paten B, Akeson M. Improved data analysis for the MinION nanopore sequencer. Nature methods. 2015;. https://doi.org/10.1038/nmeth.3290 57. Gemelos GM, Sigurjonsson S, Weissman T. Algorithms for discrete denoising under channel uncer- tainty. Signal Processing, IEEE Transactions on. 2006; 54(6):2263–2276. https://doi.org/10.1109/TSP. 2006.874295 58. Ochoa I, Hernaez M, Goldfeder R, Ashley E, Weissman T. Effect of lossy compression of quality scores on variant calling. Bioinformatics, under review. 2016;. 59. Moon T, Min S, Lee B, Yoon S. Neural universal discrete denoiser. In: Proceedings of Neural Informa- tion Processing Systems (NIPS); 2016. 60. Morgulis A, Gertz EM, Scha¨ffer AA, Agarwala R. A fast and symmetric DUST implementation to mask low-complexity DNA sequences. PLOS ONE \| https://doi.org/10.1371/journal.pone.0181463 July 27, 2017 References Journal of Computational Biology. 2006; 13(5):1028–1040. https://doi. org/10.1089/cmb.2006.13.1028 PMID: 16796549 61. Shannon CE. A mathematical theory of communication. ACM SIGMOBILE Mobile Computing and Communications Review. 2001; 5(1):3–55. https://doi.org/10.1145/584091.584093 62. Schmieder R, Lim YW, Rohwer F, Edwards R. TagCleaner: Identification and removal of tag sequences from genomic and metagenomic datasets. BMC bioinformatics. 2010; 11(1):1. https://doi.org/10.1186/ 1471-2105-11-341 63. Schmieder R, Edwards R. Quality control and preprocessing of metagenomic datasets. Bioinformatics. 2011; 27(6):863–864. https://doi.org/10.1093/bioinformatics/btr026 PMID: 21278185 25 / 25
https://openalex.org/W2727248232	https://touroscholar.touro.edu/cgi/viewcontent.cgi?article=1067&context=shs_pubs	English	null	Efficacy of a Stress Management Module in Managing Stress and Clean Time in Dual Diagnosis (Mental Illness and Substance Misuse) Clients	InTech eBooks	2,017	cc-by	6,037	Efficacy of a Stress Management Module in Managing Stress and Efficacy of a Stress Management Module in Managing Stress and Clean Time in Dual Diagnosis (Mental Illness and Substance Clean Time in Dual Diagnosis (Mental Illness and Substance Misuse) Clients Misuse) Clients Pat J. Precin Touro College, patricia.precin@touro.edu Follow this and additional works at: https://touroscholar.touro.edu/shs_pubs Part of the Occupational Therapy Commons Part of the Occupational Therapy Commons Touro Scholar Touro Scholar School of Health Sciences Publications and Research School of Health Sciences 3,050+ OPEN ACCESS BOOKS 102,000+ INTERNATIONAL AUTHORS AND EDITORS 100+ MILLION DOWNLOADS BOOKS DELIVERED TO 151 COUNTRIES AUTHORS AMONG TOP 1% MOST CITED SCIENTIST 12.2% AUTHORS AND EDITORS FROM TOP 500 UNIVERSITIES Selection of our books indexed in the Book Citation Index in Web of Science™ Core Collection (BKCI) Chapter from the book Occupational Therapy - Occupation Focused Holistic Practice in Rehabilitation Downloaded from: http://www.intechopen.com/books/occupational-therapy- occupation-focused-holistic-practice-in-rehabilitation PUBLISHED BY World's largest Science, Technology & Medicine Open Access book publisher PUBLISHED BY Recommended Citation Recommended Citation Recommended Citation Recommended Citation Precin, P. J. (2017). Efficacy of a stress management module in managing stress and clean time in dual diagnosis (mental illness and substance misuse) clients. In M. Huri (Ed.), Occupational therapy - Occupation focused holistic practice in rehabilitation (pp. 67-79). Rijeka, Croatia: IN TECH. This Book Chapter is brought to you for free and open access by the School of Health Sciences at Touro Scholar. It has been accepted for inclusion in School of Health Sciences Publications and Research by an authorized administrator of Touro Scholar. For more information, please contact touro.scholar@touro.edu. World's largest Science, Technology & Medicine Open Access book publisher Selection of our books indexed in the Book Citation Index in Web of Science™ Core Collection (BKCI) Chapter from the book Occupational Therapy - Occupation Focused Holistic Practice in Rehabilitation Downloaded from: http://www.intechopen.com/books/occupational-therapy- occupation-focused-holistic-practice-in-rehabilitation Chapter from the book Occupational Therapy - Occupation Focused Holistic Practice in Rehabilitation Interested in publishing with InTechOpen? Contact us at book.department@intechopen.com Chapter 4 Chapter 4 Efficacy of a Stress Management Module in Managing Stress and Clean Time in Dual Diagnosis (Mental Illness and Substance Misuse) Clients Patricia Precin Additional information is available at the end of the chapter http://dx.doi.org/10.5772/intechopen.68314 Additional information is available at the end of the chapter Additional information is available at the end of the chapter http://dx.doi.org/10.5772/intechopen.68314 Abstract A 1‐year pilot quasi‐experimental efficacy study of the Stress Management for Recovery Module (SM) was performed with 37 dual diagnosis (DD) clients from a DD outpatient clinic in the United States. It was hypothesized that clients who received the SM would show more improvement in their ability to manage stress and clean time than controls and when compared to themselves before and after the SM intervention. Outcome data showed that clients who received the SM learned new material and used it to make changes in their lives. Results from paired sample t tests demonstrated that clients who received the SM showed a significant improvement in their number of clean days during intervention as compared to before (p = 0.008). Clients showed a significant improvement in their knowledge of stress after the intervention as compared to before (pre‐ versus post‐test) (p = 0.033), but there was no significant difference when compared to the con- trol group. These results indicate that this SM is an effective method for improving stress management skills and clean time in DD clients at this clinic and a need for future ran- domized and controlled experimentation. Keywords: living skills, occupational therapy, addiction © 2017 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2017 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. 1. Introduction and literature review Over the last 30 years, clinical researchers have been establishing best practices for dual diagnosis (DD) clients (clients diagnosed with a chronic major mental illness and substance abuse or dependence). Treatment techniques often involved motivational enhancement, peer support, harm minimization, and relapse prevention. Group treatment usually focused on psychoeducation on drug use and mental health, reasons for drug use, reasons to change, Occupational Therapy - Occupation Focused Holistic Practice in Rehabilitation 68 harm reduction strategies, planning for the future, assertiveness training to cope with high risk situations [1–3], leisure activity development [4], and skills training (stress management, time management, and social skills including assertiveness training) with an emphasis on prob- lem solving [5, 6], refusal of drugs, and coping with cravings throughout topics. Treatment facilities included both inpatient and outpatient programs. Most of these researchers used an intervention group that ran anywhere from 2 to 18 months [7] with a closed cohort group, or an open group of ongoing duration where clients were discharged after meeting the group’s objectives. Timko, Dixon, and Moss [8] reported 70% of the 298 nationwide psychiatric resi- dential Veterans Affairs treatment facilities in their study offered some form of stress manage- ment treatment for their DD clients and 90% out of 114 nationwide substance abuse residential Veterans Affairs treatment facilities they studied offered some form of stress management treatment for their DD clients. Even though the efficacy studies above all included some form of stress management to help clients develop alternative ways to manage their problems, there have been few studies published to date that examine the effects of a treatment group focused exclusively on stress management training for DD clients. Yet, many clinician/authors have stated that the inclu- sion of stress management training in a DD rehabilitation program is imperative. Hodgson et al. [4] stated that DD clients need to develop alternative coping behaviors to substance misuse. Lindsay [9] believed that one of the major roles of a therapist working in alcohol treatment facilities is to help clients identify daily problems and learn to cope with them in new ways that do not include substance misuse. Patrick [10] stated that persons with schizophrenia and substance misuse are particularly susceptible to stress, both perceived and anticipated, and that stress management helps prevent relapse in these clients. 1. Introduction and literature review Goldman and Barr [11] offered an explanation for increased anxiety and depression upon drug ces- sation, the rapid decrease in abnormally high (from substance misuse) levels of dopamine. Gutman [12] recommended stress management to deal with these intense initial emotions but also ongoing emotions, since addiction effects neurological pathways throughout the lifespan even after drug cessation. Buijsse et al. [13] suggest stress management training to teach techniques that can be used to decrease the effects of environmental stress on people who misuse substances. The Living Skills Recovery Curriculum (LSRC) [14] is a treatment intervention that helps DD clients acquire basic living skills. It contains four different modules: Activities of Daily living for Abstinence, Social Skills for Sobriety, Time Management for 12‐Step Treatment, and Stress Management for Recovery. Each skill is taught in relation to how it aids in relapse prevention and recovery for each client’s personal lifestyle and pattern of addiction. The purpose of this study was to examine the efficacy of the Stress Management Recovery training module on reducing substance misuse and increasing the ability to manage stress in DD clients with the hope that it can be utilized in other settings by occupational therapists. A reduction in substance misuse is defined as an increase in the number of days sober, or a decrease in the length of drug relapses, or a decrease in the number of drug relapses. The abil- ity to manage stress is defined as the use of healthy coping skills to manage daily stressors. The hypothesis is that clients who received the LSRC’s SM would show more improvement in Efficacy of a Stress Management Module in Managing Stress and Clean Time in Dual Diagnosis... http://dx.doi.org/10.5772/intechopen.68314 69 69 their ability to manage stress and clean time than controls and when compared to themselves before and after SM intervention. their ability to manage stress and clean time than controls and when compared to themselves before and after SM intervention. Table 1. Demographics of experimental and control groups. cs of experimental and control groups. 2.1. Sample The subjects in this 1‐year quasi‐experimental efficacy study (both experimental and control) were adults (over 18 years of age) from a DD outpatient clinic in a metropolitan hospital in the United States where the average length of stay was 5 years. The clinic was in operation Monday through Friday from 9:00 am to 2:00 pm. Both experimental and control group cli- ents received treatment through the DD clinic as clinically necessitated. Treatment for both groups included the possibility of substance misuse groups, vocational groups, task groups, music therapy, nursing intervention, and psychiatric services. All received once‐a‐week case management services and random drug screens. All gave written consent to be in the study. The total number of subjects in the experimental group (those that received the LSRC’s SM) was 21. For their demographics, see Tables 1 and 2. Some of the clients’ clean time data were not available before the SM began because these clients started the SM when they started the program. For some other clients, clean time measures could not be obtained 4 months after the SM because they graduated from the program. Therefore, when statistical analyses on clean time were performed, the number of clients (N) in the groups varied. Occasionally, some of the clients preferred not to take the pre‐ or post‐test; so the N was adjusted accord- ingly and reported separately for each outcome. The total number of subjects in the control group was 16. Their demographics are also reported in Tables 1 and 2. Experimental Control Demographic # % # % Median age 18 37 16 36 Male 12 67 11 66 Female 6 33 5 34 African‐American 12 67 11 68.75 Caucasian 3 12 2 12.5 Hispanic 2 11 2 12.5 Other 1 9 1 6.25 Mean # psychiatric hospitalizations 18 6 16 6 Mean # detoxifications 18 1 16 1 Mean # of rehabilitations 18 0.5 16 1 Experimental Control Demographic # % # % Median age 18 37 16 36 Male 12 67 11 66 Female 6 33 5 34 African‐American 12 67 11 68.75 Caucasian 3 12 2 12.5 Hispanic 2 11 2 12.5 Other 1 9 1 6.25 Mean # psychiatric hospitalizations 18 6 16 6 Mean # detoxifications 18 1 16 1 Mean # of rehabilitations 18 0.5 16 1 Table 1. Demographics of experimental and control groups. 2.1. Sample Occupational Therapy - Occupation Focused Holistic Practice in Rehabilitation 0 70 Experimental Control Diagnoses # % # % Schizophrenia with Polysubstance 5 28 8 50 Crack 3 17 4 25 Alcohol & crack 3 17 Alcohol & cocaine 1 5.5 Alcohol & marijuana 1 5.5 1 6.25 Cocaine 1 5.5 Alcohol 1 5.5 2 12.5 Marijuana 1 5.5 Marijuana & crack 1 5.5 Other 1 5.5 1 6.25 Total 18 100 16 100 Table 2. Diagnostic Statistical Manual diagnoses. Table 2. Diagnostic Statistical Manual diagnoses. 2.2. Treatment The SM of the LSRC included 16 topic areas dealing with stress management for recov- ery. The topics provided a structured skeleton useful to elicit personal information from clients on their strengths and problem areas in coping with stressful recovery situations and identifying stressful situations and their personal signs of stress. Topics also provided stress management techniques that had to do with recovery, such as developing alternative coping strategies that did not involve drugs, managing raw emotions (anger management), identifying triggers and warning signs, relaxation skills, stretching exercises, biofeedback, nutrition, music, poetry, crafts, and how to work through relapses. The SM utilized a cogni- tive behavioral approach to recovery and living skills acquisition. Paradigms of treatment included peer support (universality from group intervention), harm minimization, and relapse prevention. Goal setting and problem solving skills were emphasized throughout all topics. 2.3.2. Post‐test A post‐test was administered to both the experimental group and control group the day after the SM ended in order to determine how much of the SM material was learned and/or relearned, stored, and recalled after 4 months of SM intervention. The post‐test was the same as the pre‐test, and clients were able to complete it in 5–10 min. 2.3.3. Outcome measures 2.3.3.1. Attendance Attendance was a measure of the number of sessions attended per client in the experimental group. 2.3.3.2. Objectives Each session of the SM had approximately 4–6 objectives to be learned by each client as listed in the LSRC Group Leader Plans [14]. Scores were percentages of total possible objectives that a client met on the days he or she attended the SM. This was a measure of the amount of mate- rial learned each session and only gathered for the experimental group. 2.3. Measures 2.3. Measures 2.3.1. Pre‐test 2.3.1. Pre‐test A pre‐test was administered to both the experimental group and control group the day before the SM began in order to determine the clients’ knowledge of stress management prior to Efficacy of a Stress Management Module in Managing Stress and Clean Time in Dual Diagnosis... http://dx.doi.org/10.5772/intechopen.68314 7 Efficacy of a Stress Management Module in Managing Stress and Clean Time in Dual Diagnosis... http://dx.doi.org/10.5772/intechopen.68314 71 71 intervention. The pre‐test used was a paper and pencil open‐ended questionnaire with six questions on stress management that was replicated from Precin’s Living Skills Recovery Workbook [14] for this population. No reliability or validity studies have yet been published using this questionnaire. Clients completed the questionnaire in 5–10 min. 2.3.3.7. Clean time Clean time was collected three different ways to increase the accuracy of measuring substance use. The number of clean days (#CD), the number of relapses (#R), and the average length of relapse (ALR) were counted 4 months before, during, and 4 months after intervention. To con- trol for the influence of other aspects of treatment taking place in the dual diagnosis clinic, the same clean time measures (#CD, #R, ALR) were gathered at enrollment. Clean time measures were obtained from the clients’ charts through the substance abuse counselor’s documenta- tion of drug screen results. 2.4. Procedures Treatment began after the facility’s Internal Review Board approved this study under an exempt status because no risks were involved and no invasive procedures were used. The LSRC’s SM was run in the DD clinic by an occupational therapist for 4 months three times consecutively in 1 year. Clients attended twice‐a‐week. Each time, the module was run with seven clients in the group, so that at the end of a year, a total of 21 clients received the SM and constituted the experimental group. The control group consisted of clients in the DD clinic not currently assigned to the module who gave consent to be in the study through their case managers. There were three control groups of six, five, and five clients each with a total of 16 clients. Each time, outcome measure- ments were taken from the SM experimental group, and the same outcome measures were gathered from each control group. Data were gathered, recorded, and analyzed by the author. 2.3.3.6. Staff observations (SO) At the end of the 4‐month intervention period, staff members not involved in the LSRC reported whether they thought their clients in the experimental group’s skills in stress man- agement improved, stayed the same, or got worse during the 4‐month intervention period. 2.3.3.3. Goals Goals were the number of SM‐related goals that the clients in the experimental group achieved during the 4‐month treatment period. This is a measure of how well material generalized to the outside. 2.3.3.4. Members report that they learned new material (MRLNM) At the end of the intervention, each client in the experimental group completed a satisfaction questionnaire [14] in which he or she stated whether or not they learned new material. 2.3.3.5. Members report that they made changes in their lives (MRMCL) At the end of the intervention, each client in the experimental group completed a satisfaction questionnaire [14] in which he or she stated whether or not they made changes in their lives due to the SM intervention. Occupational Therapy - Occupation Focused Holistic Practice in Rehabilitation 2 Occupational Therapy - Occupation Focused Holistic Practice in Rehabilitation 72 72 2.3.3.6. Staff observations (SO) 2.5. Statistical and data analysis Statistical data were analyzed using the Statistical Package for the Social Sciences (SPSS) pro- gram. Percentages were calculated by the author. In order to examine the effectiveness of the LSRC SM, a within subjects, paired t‐test was used to compare the difference in post‐test scores from the pre‐test scores. In addition, an independent t‐test was used to compare the partici- pants of the SM to controls to see whether the change in score was due to the intervention or to participation in the program. The data distribution was evaluated using Leven’s test for Equality of Variance. For non‐normal data, the Mann‐Whitney U test (a nonparametric statistic) was employed. Findings with a p value < or = to 0.05 were considered statistically sig- nificant. To further examine the effectiveness of the LSRC, percentages of the number of objec- tives met, attendance, MRLNM, MRMCL, and SO, along with the number of goals met were Efficacy of a Stress Management Module in Managing Stress and Clean Time in Dual Diagnosis... http://dx.doi.org/10.5772/intechopen.68314 73 73 calculated for clients in the experimental group. Within subjects analyses, using paired t tests were performed to investigate whether the DD members significantly increased their clean time during and after receiving the SM as compared to their previous amount of clean time before intervention began. Correlations using a Pearson‐product moment correlation coeffi- cient were used to answer the following investigative questions. Was newly learned material lost over time? Was attendance a factor in clients’ progress? Did staff’s observations correlate with members’ self‐reports of progress and/or objective findings? Did members’ self‐report of progress correlate with other objective findings? Did the number of goals met on the outside correlate with the amount of material each patient learned throughout the session? Do patients who report having learned new material also tend to report that they made changes in their lives due to the SM? Pearson‐product moment correlation coefficients were also used to see whether there were any correlations between demographics and the ability to utilize the SM. 3. Results 3.1. Change in pre‐ and post‐test values Table 3. Average change from pre‐ to post‐test values between treatment and control groups over time. 3.1. Change in pre‐ and post‐test values The change in pre‐ and post‐test values between experimental and control groups over time is presented in Table 3. For 18 members in the SM experimental group, the mean pre‐test in SM was 0.30 (SD = 0.16). This increased to a 0.46 (SD = 0.26) post‐SM score. This increase in value was statistically significant (p = 0.033) within the experimental group as per a paired samples t test. For the 16 individuals in the control group, the mean pre‐test in SM was 0.15 (SD = 0.19). This increased to a 0.22 (SD = 0.21) post‐SM score. This increase in value was not statistically significant (p = 0.331) within the control group as per a paired samples t test. The rate of change between pre‐ and post‐test scores was compared in the experimental group with the control group. The average change from pre‐ to post‐test for the experimental group was 0.16, whereas the average change for the control group was 0.07. The resulting p value of 0.92 (t[19] = −0.10) generated from an independent t test reflecting the magnitude of change per groups (experimental verses control) was not sta- tistically significant. The 95% confidence interval for the mean difference between the two was −1.91 to 1.73. Treatment Control p‐Value Pre‐test Post‐test Pre‐test Post‐test N M (SD) N M (SD) N M (SD) N M (SD) 18 0.30 (16) 18 0.46 (0.26) 16 0.15 (0.19) 16 0.22 (0.21) 0.92 Note: p‐Value is derived from unpaired t test on the mean change from pre‐test to post‐test and reflects a paired analysis reflecting the magnitude of change per groups. Table 3. Average change from pre‐ to post‐test values between treatment and control groups over time. Occupational Therapy - Occupation Focused Holistic Practice in Rehabilitation 74 3.2. Effectiveness of SM on the experimental group The 21 clients in the SM experimental group achieved an average of 77% of the total number of objectives possible in SM on the days they attended. The average number of goals related to stress management achieved by each client during the SM was four. The average atten- dance throughout the 4 months was 63%. Ninety‐one percent of the clients reported that they learned new material, and 86% reported that they made changes in their lives as a result of the SM training. The staff observed that 73% of the clients showed improvement in their ability to manage stress during the intervention. Table 4. Clean time comparisons before, during, and after treatment through paired t tests, N = 18. Notes: P1 = the magnitude of difference (p) between 4MB and 4MD, P2 = the magnitude of difference (p) between the magnitude of difference between 4MD and 4MA, P3 = the magnitude of difference (p) between 4MB an 4MA, 4MB = 4 months before treatment began, 4MD = 4 months during treatment, 4MA = 4 months after treatment, * = significant at the p < 0.05 level, = significant at the p < 0.01 level. 3.3. Substance use 3.3.1. Clean days 3.3.2. Relapses For the results of the #R, see Table 4. For 18 individuals in the SM experimental group, the mean #R 4 months before intervention was 0.944 (SD = 1.4). This decreased to 0.389 (SD = 0.698) during intervention. This decrease in value was marginally significant (t[17] = 1.97, p = 0.066) within the experimental group as per paired samples t test. The 95% confidence interval for the mean difference between the two was −0.04 to 1.15. The mean #R 4 months after intervention was 0.500 (SD = 0.857). The difference between the average #R 4 months before intervention and 4 months after intervention was not significant (t[17] = −1.72, p = 0.104, 95% CI −0.99 to 0.10), nor was the difference between the average #R during intervention as com- pared to 4 months after intervention (t[17] = 0.42, p = 0.682, 95% CI −0.452 to 0.674). 3.3.3. Average length of relapses For the results of the ALR, see Table 4. For 18 individuals in the SM experimental group, the ALR 4 months before intervention was 30 days (SD = 43.9). This decreased to 8.2 (SD = 19.3) days during intervention. This decrease in value was statistically significant (t[17] = 2.65, p = 0.017) within the experimental group as per paired samples t test. The 95% confidence interval for the mean difference between the two was 4.45 to 39.1. The ALR 4 months after intervention was 16.1 days (SD = 37.4). The difference between the ALR 4 months before intervention as com- pared to after intervention was not significant (t[17] = −1.28, p = 0.217, 95% CI −36.87 to 8.98), nor was the difference between the ALR during intervention and 4 months after intervention (t[17] = 0.97, p = 0.343, 95% CI −9.12 to 24.78). 3.3.1. Clean days For the results of the #CD, see Table 4. For 18 individuals in the SM, the mean #CD 4 months before treatment was 84.4 (SD = 44.1). This increased to 108.9 (SD = 23.6) during intervention. This increase in value was statistically significant (t[17] = −3.01, p = 0.008) within the experi- mental group as per paired samples t test. The 95% confidence interval for the mean dif- ference between the two was −41.58 to −7.30. There was a slight drop in the #CD 4 months after intervention (M = 102.2, SD = 38.1). This drop was not significant when compared to the #CD 4 months before treatment (t[17] = 1.69, p = 0.11, 95% CI −4.46 to 40.02) or the #CD during intervention (t[17] = −0.76, p = 0.46, 95% CI −25.07 to 11.74). Measure Clean time 4MB 4MD 4MA P1 P2 P3 M (SD) M (SD) M (SD) Clean days 84.40 (44.10) 108.90 (23.60) 0.008 108.90 (23.60) 102.20 (38.10) 0.460 84.40 (44.10) 102.20 (38.10) 0.110 # Relapses 0.94 (1.40) 0.389 (0.698) 0.067 0.389 (0.698) 0.50 (0.86) 0.682 0.94 (1.40) 0.50 (0.86) 0.104 ALR 30 (43.9) 8.2 (19.3) 0.017* 8.2 (19.3) 16.1 (37.4) 0.343 30 (43.9) 16.1 (37.4) 0.217 Notes: P1 = the magnitude of difference (p) between 4MB and 4MD, P2 = the magnitude of difference (p) between the magnitude of difference between 4MD and 4MA, P3 = the magnitude of difference (p) between 4MB an 4MA, 4MB = 4 months before treatment began, 4MD = 4 months during treatment, 4MA = 4 months after treatment, * = significant at the p < 0.05 level, ** = significant at the p < 0.01 level. Notes: P1 = the magnitude of difference (p) between 4MB and 4MD, P2 = the magnitude of difference (p) between the magnitude of difference between 4MD and 4MA, P3 = the magnitude of difference (p) between 4MB an 4MA, 4MB = 4 months before treatment began, 4MD = 4 months during treatment, 4MA = 4 months after treatment, * = significant at the p < 0.05 level, = significant at the p < 0.01 level. Efficacy of a Stress Management Module in Managing Stress and Clean Time in Dual Diagnosis... http://dx.doi.org/10.5772/intechopen.68314 75 75 3.4. Correlations Correlation matrix of LSRC outcomes using Pearson‐product moment coefficient. 3.4. Correlations For the results of correlations between SM outcomes in the experimental group see Table 5. In the SM experimental group, the following positive correlations were significant at the p < 0.05 level: attendance and number of objectives met, attendance and number of goals met, attendance and MRLNM, attendance and MRMCL, number of objectives met and MRLNM, number of objectives met and SO, MRLNM and MRMCL, MRLNM and SO, MRMCL and SO. All other correlations in the stress management experimental group were not significant at the p < 0.05 level. There were no significant correlations between the demographics of the experimental group and any of the outcome measures. Attendance Objectives Goals MRLNM MRMCL Attendance Objectives r = 0.553 N = 21 p = 0.009 Goals r = 0.723 r = 0.292 N = 22 N = 18 p = 0.000 p = 0.240 Occupational Therapy - Occupation Focused Holistic Practice in Rehabilitation 76 76 Attendance Objectives Goals MRLNM MRMCL MRLNM r = 0.710 r = 0.618 r = 0.330 N = 21 N = 18 N = 21 p = 0.000 p = 0.006 p = 0.144 MRMCL r = 0.645 r = 0.414 r = 0.393 r = 0.795 N = 21 N = 18 N = 21 N = 22 p = 0.002 p = 0.09 p = 0.078 p = 0.000 SO r = 0.274 r = 0.535 r = 0.226 r = 0.513* r = 0.647 N = 22 N = 18 N = 22 N = 21 N = 21 p = 0.218 p = 0.022 p = 0.311 p = 0.017 p = 0.002 Notes: Correlation is significant at the 0.01 level (2‐tailed). Correlation is significant at the 0.05 level (two‐tailed). MRLNM = members report that they learned new material as a result of SM, MRNCL = members report that they made changes in their lives as a result of SM, and SO = staff observations. Table 5. Correlation matrix of LSRC outcomes using Pearson‐product moment coefficient. Notes: Correlation is significant at the 0.01 level (2‐tailed). Correlation is significant at the 0.05 level (two‐tailed). MRLNM = members report that they learned new material as a result of SM, MRNCL = members report that they made changes in their lives as a result of SM, and SO = staff observations. Table 5. Correlation matrix of LSRC outcomes using Pearson‐product moment coefficient. ble 5. Table 5. Correlation matrix of LSRC outcomes using Pearson‐product moment coefficient. 4. Discussion The experimental group significantly increased their knowledge of stress management infor- mation after completing the SM, demonstrating the ability to learn, store, and recall new information, whereas the control group did not show a significant increase. However, the increase noted in the experimental group when compared to the control group was no longer significant. Since there were improvements in so many other aspects of stress management in the experimental group, it could be that stress management is best learned and utilized through hands on experience and talking about feelings instead of obtaining knowledge about the subject. It is one thing to “know about” stress reduction, but a different experience to “feel it in one’s bones.” The fact that the SM does both could account for the difference in outcomes. For instance, the objectives for each session of the SM incorporate knowledge about stress management with hands on experience. The objective for day 26, “clients will use biofeedback as a stress management technique,” presents the knowledge of how to count breaths and gives the client the physical experience that respiration rate can be controlled and decreasing respiration rate can be calming. Clients did very well in achieving the objectives for each session (an average of 77% of the objectives was met in the SM). The fact that clients achieved on the average of four stress management goals that they set during the SM demon- strates that the material learned and experienced generalized to their lives outside the clinic. The other outcome measures also support the efficacy of the LSRC’s SM. Clients in the exper- imental group (91%) reported that they learned new material and 86% reported that they made changes in their lives as a result of the SM. This is consistent with staff members report- ing that 73% of their clients in SM were better able to manage stress. Clients who participated Efficacy of a Stress Management Module in Managing Stress and Clean Time in Dual Diagnosis... http://dx.doi.org/10.5772/intechopen.68314 7 Efficacy of a Stress Management Module in Managing Stress and Clean Time in Dual Diagnosis... http://dx.doi.org/10.5772/intechopen.68314 77 77 in the SM significantly improved their #CD and their ALR. Four months after SM ended, they experienced a slight decrease in #CD and ALR. 4.1. Limitations This study was a quasi‐experiment. The author used a sample of convenience that followed the selection/referral procedure found in most clinics. Although the demographic profile of these clients closely approximated those in the literature for dual diagnosis outpatient clin- ics, caution should be used if generalizing the results to other settings because of a possible sample bias. Validity and reliability of the pre‐/post‐test had not been established. 4. Discussion Although this decrease was statistically insig- nificant when compared to their clean time during SM, it was also statistically insignificant when compared to before SM, indicating that clients were almost back to where they started before intervention. In order for stress management for recovery intervention to be effective, it should be longer than 4 months. Clients showed no significant change in the #R before, during, or after SM. This could be because the average #R was and remained one, so there was not a lot of improvement to be made. Correlations indicated that newly learned material was not lost over time (percentage of objectives met x post‐test scores). This may be due to repetition, review, and multiple modes of training (visual, auditory, bodily sensations, and eliciting prior experiences), which have been built into the curriculum. Attendance was an important factor in clients’ progress. The more clients came to SM, the more objectives and goals they met, the more they reported that they learned new material, and the more staff observed improvements in their ability to manage stress. Even though the attendance rate of 63% achieved by the experimental group is standard for what is reported in psychiatric clin- ics, it may be beneficial, given the significance of attendance, to generate ways to improve it. Staff’s observations correlated positively with members’ self‐report of progress and the number of objectives met. Staff’s sensitivity to improvement is necessary to encourage and provide positive feedback to clients and provide continuity across intervention modalities throughout the clinic, just as their sensitivity to ongoing needs of the clients can be helpful in referring future clients to the LSRC. Members’ self‐report of progress correlated with one of two objective findings. There was a positive correlation between member’s reporting that they learned new material and achieving objectives, but not with the number of goals met outside the clinic. Perhaps the clients in answering this question did not consider goal achievement an indicator of change. If so, this would be an important connection to help the clients make in order for their self‐esteem to fully benefit from their progress. The number of goals met on the outside did positively correlate with the amount of material each patient learned throughout the session. Patients who report having learned new material also tend to report that they made changes in their lives due to the SM. 4.2. Conclusion The LSRC’s SM is an effective method for improving stress management skills and clean time in DD clients in this DD clinic. Clients significantly increased their number of clean days, decreased their average length of relapse, and were able to learn, store, and recall informa- tion on stress management. They achieved over three‐fourths of the daily objectives, reported Occupational Therapy - Occupation Focused Holistic Practice in Rehabilitation 8 78 learning new material, and were able to make changes in their lives by generalizing what they learned/experienced to their environment outside the clinic. These results lend support for future randomized controlled experiments to investigate the efficacy of this SM with DD clients and also for the use of this module by occupational therapists working with the DD population. Acknowledgements This research was supported by a grant from the Metropolitan New York District of the New York State Occupational Therapy Association. Author details Patricia Precin Address all correspondence to: patricia.precin@touro.edu Occupational Therapy at Touro College, Licensed Psychoanalyst in Private Practice, Faculty of the National Psychological Association for Psychoanalysis and Stony Brook University, New York, NY, USA References [1] Chaney E, O’Leary M, Marlatt G. Skill training with alcoholics. Journal of Consulting Clinical Psychology. 1976;46:1092‐1104. DOI: 10.1037/0022‐006X.46.5.1092 [2] Jones S, Kanfer R, Lanyon R. Skills training with alcoholics: A clinical extension. Add ictive Behavior. 1982;7:285‐290. DOI: 10.1016/0306‐4603(82)90057‐0 [3] Marlatt G, Gordon J, editors. Relapse Prevention: Maintenance Strategies in the Treat ment of Addictive Behavior. New York: Guilford; 1985 [4] Hodgson S, Lloyd C, Schmid T. The leisure participation of clients with a dual diagnosis. British Journal of Occupational Therapy. 2001;64(10):487‐492. DOI: 10.1177/03080226010 6401003 [5] Abrams D. Psychosocial assessment of alcohol and stress interactions: Bridging the gap between laboratory and treatment outcome research. In Pohorecky L, Brick J, editors. Stress and Alcohol Use. New York: Elsevier Biomedical; 1983. pp. 61‐86. [6] Monti P, Abrams D, Kadden R, Cooney N.. Treating Alcohol Dependence: A Coping Skills Guide. New York: Guilford; 1990 79 [7] Jerrell JM, Ridgely MS. Comparative effectiveness of three approaches to serving people with severe mental illness and substance abuse disorders. The Journal of Nervous and Mental Disease. 1995;183(9):566‐576. DOI: 10.1097/00005053‐199509000‐00002 [8] Timko C, Dixon K, Moos RH. Treatment for dual diagnosis patients in the psychiatric and substance abuse systems. Mental Health Services Research. 2005;7(4):229‐242. DOI: 10.1007/s11020‐005‐7455‐9 [9] Lindsay W. The role of the occupational therapist: Treatment of alcoholism. American Journal of Occupational Therapy. 1983;37:36‐43. DOI: 10.5014/ajot.37.1.36 [10] Patrick DD. Dual diagnosis: Substance‐related and psychiatric disorders. Nursing Clinics of North America. 2003;38(1):67‐73. DOI: 10.1016/S0029‐6465(02)00066‐X [11] Goldman D, Barr CS. Clinical implications of basic research. Restoring the addicted brain. New England Journal of Medicine. 2002;347:843‐845. [12] Gutman, S. Why addiction has a chronic, relapsing course. The neurobiology of addic- tion: implications for occupational therapy practice. Occupational Therapy in Mental Health. 2006;22:1‐29. DOI: 10.1300/J004v22n02_01 [13] Buijsse N, Woody C, Davis SF. Occupational therapy in the treatment of addictive behaviors. British Journal of Therapy and Rehabilitation. 1999;6:300‐307. DOI: 10.12968/ bjtr.1999.6.6.13973 [14] Precin P. Living Skills Recovery Workbook. Brattleboro, VM: Echo Point Books and Media, LLC; 2015
https://openalex.org/W3046072710	https://iris.uniroma1.it/bitstream/11573/1433470/1/Corcos_Contrasting_2020.pdf	English	null	Contrasting effects of exotic plant invasions and managed honeybees on plant–flower visitor interactions	Diversity and distributions	2,020	cc-by	9,309	B I O D I V E R S I T Y R E S E A R C H B I O D I V E R S I T Y R E S E A R C H Daria Corcos1 \| Andree Cappellari1,2 \| Maurizio Mei2 \| Dino Paniccia3 \| Pierfilippo Cerretti2 \| Lorenzo Marini1 1Department of Agronomy, Food, Natural Resources, Animals and Environment (DAFNAE), University of Padua, Legnaro, Padua, Italy 2Department of Biology and Biotechnology “Charles Darwin”, Sapienza University of Rome, Rome, Italy 3Via Colle 13, Frosinone, Italy Correspondence Daria Corcos, Department of Agronomy, Food, Natural Resources, Animals and Environment (DAFNAE), University of Padua, Legnaro, Padua, Italy. Email: daria.corcos@unipd.it Funding information STARS Consolidator Grant, Grant/Award Number: STARS-CoG–2017 Editor: Andrew Barnes Diversity and Distributions. 2020;00:1–12. Received: 17 January 2020 \| Revised: 23 June 2020 \| Accepted: 30 June 2020 Received: 17 January 2020 \| Revised: 23 June 2020 \| Accepted: 30 June 2020 DOI: 10.1111/ddi.13132 B I O D I V E R S I T Y R E S E A R C H Contrasting effects of exotic plant invasions and managed honeybees on plant–flower visitor interactions Daria Corcos1 \| Andree Cappellari1,2 \| Maurizio Mei2 \| Dino Paniccia3 \| Pierfilippo Cerretti2 \| Lorenzo Marini1 1 \| INTRODUCTION Plant diversity is expected to play a key role in influencing plant–pol- linator interactions as it is related to the availability and heteroge- neity of floral resources. For instance, pollinator specialization has been showed to increase with increasing richness of flowering plant species (Ebeling, Klein, & Tscharntke, 2011). p ( g ) Here, we explored how a highly invasive plant species (Buddleja davidii Franch.) impacted plant–flower visitor interactions over its whole elevational range distribution, and whether this effect was modified by managed honeybees and available alternative flower resources. We compared invaded and equivalent non-invaded ripar- ian habitats along a gradient in honeybee abundance and species richness of flower resources by sampling Hymenoptera, Diptera and Lepidoptera pollinators. First, we tested for a potential interaction between B. davidii and honeybee in shaping the stability of plant– flower visitor networks. We expected that, in the presence of the highly rewarding invasive plant species, honeybees would focus on B. davidii, thus decreasing network stability. In non-invaded sites in- stead, honeybees would forage on many plant species, consequently resulting in a higher network stability via increased network con- nectance. Second, we tested how B. davidii, honeybee abundance and species richness of flowering plants influenced resource overlap of wild flower visitors with honeybees and their species-level spe- cialization, and whether these effects changed among insect orders. We expected that, in sites invaded by the exotic plant species, a large proportion of flower visitors would concentrate their foraging on the novel resource, consequently increasing their specialization. However, due to their aggressive foraging behaviour, honeybees are expected to influence how other species visit flowers (Hung et al., 2019; Santos et al., 2012), usually diverging wild flower visi- tors on less abundant resources (Goulson, 2003; Hung et al., 2019; Thomson, 2016) and thus increasing their apparent specialization. Among biotic drivers, introduced and managed pollinator species can also alter plant–pollinator interactions (Geslin et al., 2017; Magrach, González-Varo, Boiffier, Vilà, & Bartomeus, 2017; Montero-Castaño & Vilà, 2017; Norfolk, Gilbert, & Eichhorn, 2018; Valido, Rodríguez-Rodríguez, & Jordano, 2019). The honey- bee (Apis mellifera Linnaeus), in particular, is the most widespread pollinator species in natural and agricultural ecosystems (Hung, Kingston, Albrecht, Holway, & Kohn, 2018; Montero-Castaño & Vilà, 2017; Rader et al., 2009). Honeybees usually focus their foraging on the plants providing the most abundant resources (Magrach et al., 2017; Montero-Castaño & Vilà, 2017). 1 \| INTRODUCTION switch to less rewarding resources (Hung, Kingston, Lee, Holway, & Kohn, 2019). Mountain ecosystems are considered biodiversity hotspots char- acterized by a high number of rare and endemic species. Even though mountains are usually less prone to invasions than other ecosystems, there is increasing attention to the expansion of ex- otic plants in response to climate and land use change (Alexander et al., 2011; Dainese et al., 2017). Although many exotic plants have become naturalized without imposing strong impacts on native plant communities (Timóteo et al., 2018), invasive species with attractive flowers rich in nectar and pollen can compete with native plants for pollinators, ultimately altering the structure of mutualistic networks (Benadi, Hovestadt, Poethke, & Blüthgen, 2014; Grass, Berens, Peter, & Farwig, 2013; Hansen et al., 2018; Maruyama et al., 2016; Russo, Nichol, & Shea, 2016; Stout & Tiedeken, 2017). At the same time, invasions of mass-flowering plants can have positive, long-term effects on pollinator popula- tion dynamics by complementing nectar and pollen resources pro- vided by native plants (Albrecht, Ramis, & Traveset, 2016; Davis, Kelly, Maggs, & Stout, 2018; Giovanetti, Ramos, & Máguas, 2018; Russo et al., 2016; Stout & Tiedeken, 2017). Despite this vari- ability, most of the available research has focused on analysing changes in plant–pollinator interactions in invaded vs. uninvaded plant communities without considering potential interactions be- tween exotic plant invasion and other abiotic and biotic drivers of environmental change (González-Varo et al., 2013, but see Grass et al., 2013). In addition to the anthropogenic drivers described above, natural processes can also influence plant–pollinator interactions. In moun- tains, the degree of specialization of plant–pollinator networks usu- ally decreases with elevation (Hoiss, Krauss, & Steffan-Dewenter, 2015) due to a more unpredictable and unfavourable environment that narrows resource accessibility at extreme cold temperatures (Adedoja, Kehinde, & Samways, 2018; Johnson & Steiner, 2000). This reduction in network specialization is often linked to the re- placement of specialized pollinator guilds (e.g., many wild bees) with groups with wider niche breadth, such as flies and wasps (Benadi et al., 2014; Chacoff et al., 2012; Ramos-Jiliberto et al., 2010). However, elevation and plant diversity are often related in moun- tains, with native plants showing a hump-shaped relationship with elevation (Guo et al., 2013) and exotic plants typically decreasing with increasing elevation (Alexander et al., 2011; Marini et al., 2013). Abstract Aim: To explore how a highly invasive plant species (Buddleja davidii Franch.), man- aged honeybees and flower diversity affected plant–flower visitor interactions over the whole elevational range distribution of the exotic plant. Location: Italian Alps. Methods: We selected nine pairs of sites (one invaded and one non-invaded by B. davidii) across gradients in honeybee abundance and diversity of flower resources. We observed plant–flower visitor interactions every three weeks, for a total of five surveys covering the full flowering season of B. davidii (June–August). We tested how B. davidii, honeybee abundance and flowering plant diversity affected network robustness, overlap in flower resource use of wild flower visitors with honeybees and flower visitor specialization. We also tested for an interaction between B. davidii presence and honeybee abundance, and tested whether the effects of the two vari- ables changed among insect orders. Results: Buddleja davidii and honeybees had contrasting effects on network ro- bustness and on several species-level metrics. Network robustness increased with increasing honeybee abundance and flower diversity. Increasing honeybee abundance generally increased specialization of lepidopterans and dipterans that tended to switch to less visited plant species, possibly in order to avoid competition. Specialization of flower visitors declined in sites invaded by B. davidii, indicating that the invasive plant attracted pollinators, which in turn also visited co-occurring spe- cies in the neighbourhood. Main conclusions: Although increasing honeybee abundance was associated with higher network stability, it also modified plant–flower visitor interactions by forcing species to shift their diet irrespective of floral diversity. The effect was particularly strong for non-bee flower visitors. The consequences of these changes in plant– flower visitor interactions for the reproductive success of flowering plants are still largely unknown. Diversity and Distributions. 2020;00:1–12. \| 1 wileyonlinelibrary.com/journal/ddi This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2020 The Authors. Diversity and Distributions published by John Wiley & Sons Ltd Corcos and Cappellari contributed equally to the study. \| 1 wileyonlinelibrary.com/journal/ddi 2 \| 2 \| CORCOS et al. Alien species, Apis mellifera, bipartite networks, competition, elevational gradient, network stability, specialization 2.1 \| Study system Fieldwork was carried out in the mountain areas of Veneto and Trentino-Alto Adige regions (Northern Italy; Figure S1). Climate varies with elevation: it is continental in the lowlands, with warm summers and mild winters, whereas higher elevations are character- ized by alpine climate, with cool summers, cold winters and frequent snowfalls. Precipitation is abundant year round, especially at higher elevations (c. 1,200 mm). The area is invaded by the perennial shrub B. davidii, which can be found up to 1,200 m a.s.l., especially in dis- turbed areas such as railways and river banks. It is native of China and Japan and was introduced in Europe in the late 1800s. Self- incompatibility is high (>95%), and the species requires insect cross- pollination for reproduction (Ebeling, Schreiter, Hensen, Durka, & Auge, 2012). The species is highly attractive to pollinators due to its conspicuous inflorescences rich in nectar. Because of their flo- ral morphology (narrow and long corolla tube), species of the genus Buddleja are mostly pollinated by butterflies, honeybees and other large bees (Gong et al., 2015). The species exhibits a long flowering period that usually starts in late spring and lasts until late summer. 2.2 \| Sampling design We selected nine pairs of 40 × 20 m sites across the whole eleva- tional range distribution of B. davidii (c. 100–1,200 m a.s.l.; Table S1). All the sampled sites were open riparian habitats located along val- ley bottoms. For each pair, we chose one site invaded and one non- invaded by B. davidii. Within each pair, the surrounding landscape of each site was similar in terms of habitat composition (Table S2). In invaded sites, the abundance of B. davidii ranged from 2.3% to 38% of all flowering species (mean = 15.9, SD = 9.7%), whereas it was completely absent in control sites. We tested for differences in plant species composition (all flowering species excluding B. davidii) between invaded and non-invaded sites using the Multiple Response Permutation Procedure (MRPP) test. For MRPP, we used 1,000 per- mutations and the Jaccard distance matrix. Site pair was chosen as grouping factor. Invaded vs. non-invaded sites did not differ in flowering plant composition (MRPP statistics: A = −0.015, signifi- cance of delta = 0.989). Visualization of the similarity in plant com- munity composition among sites was obtained by running a NMDS (non-metric multidimensional scaling) analysis on the matrix of cover of flowering plants in each site (Figure S2). The MRPP and NMDS analyses were performed using the functions mrpp and metaMDS of the vegan library (Oksanen et al., 2019) in R version 3.2.3 (R Core Development Team, 2015). With the exception of one pair, the geo- graphical distance between sites within pairs was less than 2 km. This design assured that the two sites within each pair could poten- tially receive the same visits from managed honeybees, being the distance below the foraging range of this species (Steffan-Dewenter & Kuhn, 2003). Keeping a constant honeybee pressure between 2.3 \| Surveys of flowering plants and flower visitors During spring and summer 2018, we observed plant–flower visitor networks in the 18 study sites. Sites were visited approximately every three weeks, for a total of five surveys covering the full flow- ering season of B. davidii. The first survey was conducted in June, while the last survey occurred at the end of August, as the flower- ing started to cease. Plants were identified in the field when pos- sible, or collected and prepared in a herbarium and identified later. Insect sampling occurred between 09.00 hr and 17.00 hr in dry and sunny conditions with low wind and temperature above 18°C. At each round, each flowering species was observed for a 5-min period, during which all insects touching the reproductive parts of the flow- ers were counted or collected using a butterfly net (Ø 35 cm). Every 15 min, we paused for 10 min to reduce the level of disturbance and let the flower visitor community recover. Our survey included the most abundant and diverse groups of flower visitors, that is all the bees and sphecids (Hymenoptera), hoverflies, conopids and tachinid flies (Diptera) and butterflies (Lepidoptera). Coleopterans, less than 1% of all observed insects, were not collected and identified. Insects were identified in the field when possible or placed in vials filled with 70% ethanol and sorted and identified in the laboratory to species or morphospecies level. As females of Bombus terrestris (Linnaeus) and Bombus lucorum (Linnaeus) are often difficult to distinguish, we pooled both species (males and females) as B. terrestris/lucorum. 2 \| METHODS sites within each pair was the only way to account for differences in attractiveness of sites, as a direct manipulation of honeybee hives would have been impossible due to the unpredictable foraging be- haviour of honeybees. The density of beehives in the sampling areas is quite high with an average of 5.0 beehives km−2 (data provided by the National Data Bank of the Zootechnical Registry established by the Ministry of Health at the National Service Centre of the “G. Caporale” Institute of Teramo). We did not opt for a plant removal design (e.g. Lopezaraiza-Mikel, Hayes, Whalley, & Memmott, 2007), due to the large size of B. davidii populations involved and the size of B. davidii plants (tall shrubs up to 5 m). 1 \| INTRODUCTION On the one hand, being an abundant and generalist species, the honey- bee is responsible for most interactions in many plant–pollinator networks, potentially increasing network robustness against pol- linator extinctions (Burgos et al., 2007; Geslin et al., 2017). On the other hand, being sometimes a less efficient pollinator, it can also impair pollination service reducing the reproductive success of those plant species highly visited by honeybees (Valido et al., 2019). Moreover, the honeybee is expected to modify the inter- actions between wild pollinators and plants, increasing potential competition in terms of resource overlap, and so forcing species to \| 3 3 CORCOS et al. 2.4.3 \| Effect of B. davidii and honeybee abundance on network stability A linear mixed-effects model, with the site pair as random factor to account for spatial dependence in the design, was used to test how network stability (i.e., robustness index) responded to the presence of B. davidii (invaded vs. control sites), honeybee abundance and flowering plant species richness. We also included in the model the interaction between B. davidii and honeybee abundance. Both abun- dance of honeybees and species richness of flowering plants were ln-transformed to improve linearity and model residual distribution. The model was fitted using the function lme in the nlme package (Pinheiro, Bates, DebRoy, Sarkar, & Team, 2017). Interactions were removed using a backward stepwise model selection procedure (p- value > .05). Model assumptions were checked using residual diag- nostic plots. Effects were represented using partial residual plots using the visreg package (Breheny & Burchett, 2017). 4 \| CORCOS et al. species-level specialization indices excluding the honeybee from the network. The three specialization metrics were strongly cor- related with those calculated with the honeybee in the network, so we present and discuss only the values calculated including the honeybee within the network. However, we run the analyses with both methods and found similar results (Table S3). can seriously limit the interpretation of raw network measures and their use for multiple network comparison (Chagnon, 2015). In order to account for network size effect when comparing different networks, the robustness index was standardized using null mod- els as ΔRobustness = observed robustness–robustnessnull, where robustnessnull represents the mean robustness value from 1,000 randomized networks obtained using the Patefield, the Vazquez and the swap.web algorithms (Dormann, Fründ, & Gruber, 2008; Patefield, 2012; Schleuning et al., 2012; Vázquez et al., 2007). The three algorithms respectively constraint marginal totals of rows and columns, connectance and both marginal totals and con- nectance. As standardized robustness values obtained with the different algorithms were highly correlated with each other and with non-standardized ones (Figure S3), we decided to present and discuss only the values obtained with the Vazquez algorithm. Both raw and standardized robustness values were calculated using the bipartite package (Dormann et al., 2008). 2.4.2 \| Resource overlap and flower visitor species specialization For each wild flower visitor species within each network, the re- source overlap with honeybees was calculated using the Morisita's index (Morisita, 1959) in the spaa package (Zhang, 2016). The metric ranges from 0 (no overlap) to 1 (complete overlap). As species-level specialization can be defined in many differ- ent ways, we calculated three metrics that convey complementary information (Maglianesi, Blüthgen, Böhning-Gaese, & Schleuning, 2014): (1) normalized degree, (2) dʹ and (3) PDI (Paired Difference Index). The normalized degree is one of the simplest specializa- tion indices, based on the presence/absence of flower visitors on plants, and calculated as the standardized ratio between the num- ber of plant species on which a flower visitor species was observed and the total richness of flowering plant species in the network. It ranges from 0 (highly specialized) to 1 (highly generalized). The dʹ index (standardized Kullback–Leibler distance) additionally takes into account the abundance of flower visitors observed on a plant species, and how much the resource is shared by different flower visitor species (Benadi et al., 2014; Blüthgen et al., 2006). In PDI, the strongest link between a flower visitor species and its most vis- ited plant is contrasted with those over all remaining plant species within the network (Poisot, Canard, Mouquet, & Hochberg, 2012). Both dʹ and PDI range from 0 (no selectiveness) to 1 (extreme se- lectiveness). Normalized degree was transformed as the inverse of the index (1–normalized degree) in order to make it comparable with the other two specialization indices. Defining specialization for rare species may be critical because of the risk of overesti- mating their specialization (Dorado, Vázquez, Stevani, & Chacoff, 2011; Dormann, 2011). To calculate the species-level specializa- tion indices, we thus removed all singletons and doubletons from the dataset before using the specieslevel function in the bipartite package (Dormann et al., 2008). Moreover, we computed the three 2.4.1 \| Network stability The network stability was quantified using the robustness index (Burgos et al., 2007; Memmott, Waser, & Price, 2004). Robustness was calculated by removing flower visitor species from the net- work according to their abundance, from the rarest to the most abundant. Robustness ranges from 0 (highly unstable network) to 1 (highly stable network). It is well known that number of species (i.e. network size) and sampling effort can affect the number of observed interactions in real ecological networks, and thus several measures of network structure (Delmas et al., 2019). This issue 2.4.5 \| Potential collinearity between predictors 1,200 m a.s.l.) that implied a narrow temperature gradient between our low and high elevation sites. In temperate mountains such as the European Alps, pollinator communities do not usually exhibit strong species turnover between lowlands and mid-altitudes (c. 1,500 m a.s.l.; Lefebvre, Villemant, Fontaine, & Daugeron, 2018), meaning that foraging preferences and species composition are quite stable along our elevational gradient. The selection of the sites guaranteed statistical independence between honeybee abundance and elevation (Pearson's correla- tion = −0.207, p-value = .409) and between honeybee abundance and flowering plant species richness (Pearson's correlation = 0.368, p-value = .133), while the species richness of flowering plants sig- nificantly increased with elevation (Pearson's correlation = 0.549, p-value = .018). We also measured multicollinearity in our models by computing the variance inflation factors (VIF). The VIF values were all below 1.5, indicating no multicollinearity in our models. 2.4.4 \| Effect of B. davidii and honeybee abundance on resource overlap and flower visitor specialization (a) FI G U R E 1 Partial residual plots showing the effects of (a) honeybee abundance (ln-transformed) and (b) species richness of flowering plants (ln- transformed) on standardized network robustness. The robustness index was standardized as ΔRobustness = observed robustness–robustnessnull, where robustnessnull represents the mean robustness value from 1,000 randomized networks obtained using the Vazquez algorithm (Vázquez et al., 2007) (b) Estimate SE df t p- value Intercept −0.152 0.167 8 −0.913 .388 Species richness of flowering plants 0.185 0.049 6 3.810 .009 Honeybee 0.031 0.009 6 3.186 .019 B. davidii (invaded vs. uninvaded) 0.016 0.019 6 0.858 .424 B. davidii (invaded vs. uninvaded) × Honeybee – – – – – Note: Honeybee abundance and flowering plant species richness were ln-transformed to improve linearity and model residual distribution. Interactions are presented only when significant (p-value < .05). Note: Honeybee abundance and flowering plant species richness were ln-transformed to improve linearity and model residual distribution. Interactions are presented only when significant (p-value < .05). 2.4.4 \| Effect of B. davidii and honeybee abundance on resource overlap and flower visitor specialization Linear mixed-effects models were used to test whether resource overlap and species-level specialization changed in relation to the presence of B. davidii (invaded vs. control sites), honeybee abun- dance and species richness of flowering plants, and whether the effects of B. davidii presence and honeybee abundance changed among insect orders (Hymenoptera, Diptera and Lepidoptera). We also tested the interaction between the presence of B. davidii and the abundance of honeybees. The response variables were the mean values across species of resource overlap and species specialization indices in each site (n = 18), separate for each flower visitor guild (i.e. wild Hymenoptera, Diptera and Lepidoptera). In the case of resource overlap, one pair was removed from the dataset due to honeybee absence in one of the site. We included sampling site nested within the pair as random factor to account for spatial dependence in the design. Resource overlap, normalized degree, honeybee abundance and flowering plant species richness were ln-transformed in order to improve linearity and model residual distribution. Non-significant in- teractions were removed using a backward stepwise model selection procedure (p-value > .05). Model assumptions were checked using residual diagnostic plots. All models were fitted using the function lme in the nlme package (Pinheiro et al., 2017). Effects were repre- sented using partial residual plots using the visreg package (Breheny & Burchett, 2017). \| 5 CORCOS et al. FI G U R E 1 Partial residual plots showing the effects of (a) honeybee abundance (ln-transformed) and (b) species richness of flowering plants (ln- transformed) on standardized network robustness. The robustness index was standardized as ΔRobustness = observed robustness–robustnessnull, where robustnessnull represents the mean robustness value from 1,000 randomized networks obtained using the Vazquez algorithm (Vázquez et al., 2007) (a) (b) Estimate SE df t p- value Intercept −0.152 0.167 8 −0.913 .388 Species richness of flowering plants 0.185 0.049 6 3.810 .009 Honeybee 0.031 0.009 6 3.186 .019 B. davidii (invaded vs. uninvaded) 0.016 0.019 6 0.858 .424 B. davidii (invaded vs. uninvaded) × Honeybee – – – – – Note: Honeybee abundance and flowering plant species richness were ln-transformed to improve linearity and model residual distribution. Interactions are presented only when significant (p-value < .05). TA B LE 1 Results from the linear mixed-effects model testing the effects of explanatory variables on network robustness standardized using the Vazquez algorithm (a) (b) CORCOS et al. 4 \| DISCUSSION The effect of invasive plants on pollinators has been intensively stud- ied, but often in isolation from other biotic and abiotic drivers. Here, for the first time, we tested for the interactive effect of a highly in- vasive plant species (B. davidii) and a competitive managed pollina- tor species (the honeybee) on plant–flower visitor interactions. We found no interaction between B. davidii and honeybee abundance on both network- and species-level metrics. However, the two ex- planatory variables showed contrasting effects on network stabil- ity and flower visitor specialization. While the presence of B. davidii decreased flower visitor specialization, honeybee abundance gener- ally increased both network robustness against flower visitor extinc- tions and species specialization, but the latter with some differences among insect orders. Network robustness standardized using the Vazquez algorithm was positively influenced by both the abundance of honeybees (Figure 1a) and the species richness of flowering plants (Figure 1b), while no effect of B. davidii presence was observed (Table 1). The effects were consistent between models using robustness standard- ized with the Patefield and the swap.web algorithms, while effects were slightly different for the model using the raw robustness values (Table S8). For the latter, the effect of species richness of flowering plants disappeared, while the effect of honeybees was consistent. 2.4.6 \| Elevation versus number of flowering plants Overall, we recorded 9,563 interactions between 370 flower visi- tor species and 150 flowering plant species (25 plant species were not visited by any insect; Tables S4 and S5), for a total of 1,737 unique plant–flower visitor interactions. Beside B. davidii, we found 10 more exotic plant species (Table S5). As often found in real mutu- alistic networks (Devoto, Bailey, Craze, & Memmott, 2012), we only observed a subset of the expected interactions based on rarefac- tion curves, but with no evident systematic differences in sampling completeness between sites invaded by B. davidii and control sites (Figure S4). As elevation was correlated with flowering plant species richness and exhibited a narrow range of variation, we only used species rich- ness of flowering plants as explanatory variable in all the models de- scribed above. However, in preliminary analyses, we ran all models using elevation instead of number of flowering plants as explana- tory variable, and found no effect of elevation on network stabil- ity, resource overlap and species specialization. This result may be explained by the relatively low upper limit of B. davidii distribution (c. 6 \| CORCOS et al. The flower visitor community was composed of 68.1% hy- menopterans (172 species, 6,514 individuals), 27.0% dipterans (152 species, 2,585 individuals) and 4.9% lepidopterans (47 species, 464 individuals; Tables S6 and S7). The honeybee was the most abundant flower visitor (Figure S5), accounting for 43% of the total observed individuals, and the third most generalist (collected on 67 plant spe- cies), after two hoverfly species, Eristalis tenax (Linnaeus) (collected on 73 plant species) and Sphaerophoria scripta (Linnaeus) (collected on 70 plant species). The honeybee was found in all sites, except for one, independently from site elevation. Overall, a flower visitor species was observed, on average, on 4.6 plant species (min = 1; max = 73), which in turn received, on average, visits from 11.6 flower visitor species (min = 1; max = 83). only positively influenced by the number of flowering plant species (Table 3; Figure 5). 4.1 \| Network stability increased with honeybee abundance and richness of flowering plant species The level of resource overlap with honeybees was different among insect orders, with lepidopterans having the highest over- lap and dipterans sharing less resources with honeybees (Table 2; Figure 2). No effects of B. davidii, honeybee abundance and flower- ing plant species richness on resource overlap were observed. Super-generalist species such as the honeybee are able to exploit many plant species and thus to potentially increase network robust- ness against flower visitor extinctions. Although a previous study found that the removal of honeybees from networks did not af- fect overall stability (Santos et al., 2012), here on the contrary we showed that increasing honeybee abundance increased network stability. Highly stable networks, that is those with a high honeybee abundance, are more robust to the loss of rare flower visitor spe- cies, as only a limited number of plant species will not be visited any- more when rare species become extinct. However, as honeybees are sometimes less efficient pollinator than wild species, this increased robustness might not necessarily translate into higher reproductive success of those plants that are visited by honeybees (Magrach et al., 2017; Valido et al., 2019). The positive effect of species richness of flowering plants on network stability was probably related to the The normalized degree specialization index was influenced by the presence of B. davidii, insect order and number of flowering plants (Table 3). It was lower in sites invaded by the exotic plant compared with control sites (Figure 3a). Lepidopterans were the most special- ized flower visitors, whereas dipterans were the less specialized (Figure 3b). The number of flowering plants had a positive effect on species specialization in terms of normalized degree (Figure 3c). The dʹ specialization index was only influenced by honeybee abundance, with the effect changing among insect orders (Table 3); that is, dʹ increased with increasing honeybee abundance for dipterans and lepidopterans, whereas for hymenopterans, it was high irrespective of honeybee abundance (Figure 4). The PDI specialization index was Estimate SE df t p-value Intercept 1.675 2.086 30 0.803 .428 Species richness of flowering plants −0.844 0.498 5 −1.696 .151 Honeybee −0.069 0.114 5 −0.602 .574 B. davidii (invaded vs. 4.2 \| Resource overlap between wild flower visitors and honeybees depended on insect order The interactions between honeybees and wild flower visitors may be explored using the resource overlap index, where high levels of overlap indicate potential competition between wild and managed species (Franco, Aguiar, Ferreira, & De Oliveira-Rebouças, 2009; Paini, 2004). Resource overlap between honeybees and other pol- linators has been recorded before (Goulson, 2003), and an increase in honeybee abundance is expected to intensify competition for floral resources between wild and managed bees (Thomson, 2016). Here, we found that resource overlap was higher for lepidopter- ans than for hymenopterans and dipterans. Despite B. davidii being highly attractive for both butterflies and honeybees, the observed effect did not change in invaded vs. uninvaded sites, meaning that the high level of overlap between lepidopterans and honeybees is probably due to other reasons. One hypothesis might be related to the morphology and length of the mouthparts in the three insect orders: while lepidopterans have medium–long proboscises that allow to efficiently exploit flowers with different shapes and corolla depth, most hymenopteran (except for bumblebees) and dipteran species are short-tongued, and usually prefer open flowers (Kevan & Baker, 1983). Therefore, the honeybee, as a medium-tongued bee (c. 6 mm), mostly overlaps with lepidopterans. Consistent with previous studies (Gillespie & Elle, 2018; Steffan-Dewenter & Tscharntke, 2000), we found that honeybees did not appear to The attractive effect of exotic plants was previously found to be particularly evident for highly generalist pollinator species, such as the honeybee (Hung et al., 2019; Magrach et al., 2017). Because of its foraging behaviour, this species may force wild pollinators to shift on less abundant resources (Goulson, 2003; Hung et al., 2019; Thomson, 2016). Depending on how flexible other species are in modifying their diet breadth, the effect of honeybee on wild polli- nators may be different. Here, we found that an increase in honey- bee abundance did not modify the number of visited plant species or how flower visitors were partitioned among flower resources, but affected how the latter were shared by flower visitor species (dʹ). As honeybee abundance increased, lepidopterans and dipterans tended to switch to less visited plant species, possibly in order to avoid competition (Hung et al., 2019). On the other hand, the effect of honeybee abundance on wild hymenopterans was less evident. FI G U R E 2 Partial residual plot showing the effect of insect order on resource overlap between managed honeybees and wild flower visitors (ln-transformed) share flower resources with both hymenopterans and dipterans. The different mouthpart morphologies of the three orders thus result in a partitioning of flower resources that is independent by honeybee abundance and B. davidii presence, as well as the species richness of flowering species. On the other hand, the low level of resource overlap between honeybees and wild hymenopterans and dipterans might also be an effect of competitive exclusion, leading less-competitive wild flower visitors to shift to alternative flower resources. 4.3 \| Flower visitor specialization decreased with B. davidii presence and increased with increasing honeybee abundance As invasive plants are often rich in pollen and nectar, they can poten- tially compete with native plants for pollinators (Dietzsch, Stanley, & Stout, 2011; Flanagan, Mitchell, & Karron, 2010), causing a reduction in the range of consumed resources and thus potentially resulting in increased specialization of pollinator species (Maruyama et al., 2016). Conversely, we found that, in invaded sites, flower visitors used a broader range of plants, reflected by a decrease in their specializa- tion. A possible explanation is that the invasive plant attracts flower visitors, which in turn visit also co-occurring species in the neigh- bourhood (Molina-Montenegro, Badano, & Cavieres, 2008). For example, some bumblebee species perceive adjacent plants (closer than 6 m) as a single resource, even when they belong to different species (Klinkhamer, de Jong, & Linnebank, 2001). Additionally, when foraging on the preferred species, other flowering species are regu- larly visited in order to compare their resource levels (Albrecht et al., 2016). B. davidii, thanks to its scented and showy inflorescences, is highly attractive and might act as a facilitator for co-occurring na- tive plants. This “magnet species” effect was also recorded in areas invaded by other exotic plant species (Albrecht et al., 2016; Molina- Montenegro et al., 2008). FI G U R E 2 Partial residual plot showing the effect of insect order on resource overlap between managed honeybees and wild flower visitors (ln-transformed) FI G U R E 2 Partial residual plot showing the effect of insect order on resource overlap between managed honeybees and wild flower visitors (ln-transformed) FI G U R E 2 Partial residual plot showing the effect of insect order on resource overlap between managed honeybees and wild flower visitors (ln-transformed) positive correlation between the number of flowering species and both pollinator richness and abundance; that is, sites rich in plant species hosted more species and individuals that in turn promoted network robustness (Thébault & Fontaine, 2010). 4.1 \| Network stability increased with honeybee abundance and richness of flowering plant species uninvaded) −0.250 0.225 5 −1.110 .318 Order: Hymenoptera 0.486 0.223 30 2.183 .037 Order: Lepidoptera 0.977 0.223 30 4.388 <.001 Order: Hymenoptera × Honeybee – – – – – Order: Lepidoptera × Honeybee – – – – – Note: Resource overlap values were ln-transformed to improve linearity and model residual distribution. Interactions are presented only when significant (p-value < .05). TA B LE 2 Results from the linear mixed-effects model testing the effects of explanatory variables on resource overlap of wild flower visitors with honeybees, and how it changes among insect orders (i.e., Hymenoptera, Diptera and Lepidoptera) Estimate SE df t p-value Intercept 1.675 2.086 30 0.803 .428 Species richness of flowering plants −0.844 0.498 5 −1.696 .151 Honeybee −0.069 0.114 5 −0.602 .574 B. davidii (invaded vs. uninvaded) −0.250 0.225 5 −1.110 .318 Order: Hymenoptera 0.486 0.223 30 2.183 .037 Order: Lepidoptera 0.977 0.223 30 4.388 <.001 Order: Hymenoptera × Honeybee – – – – – Order: Lepidoptera × Honeybee – – – – – Note: Resource overlap values were ln-transformed to improve linearity and model residual distribution. Interactions are presented only when significant (p-value < .05). \| 7 \| 7 7 CORCOS et al. TA B LE 3 Results from the linear mixed-effects models testing the effects of explanatory variables on three species-level specialization indices: (a) 1–normalized degree, (b) dʹ (standardized Kullback–Leibler distance) and (c) PDI (Paired Difference Index), differently for wild insect orders (i.e., Hymenoptera, Diptera and Lepidoptera) 4.2 \| Resource overlap between wild flower visitors and honeybees depended on insect order It did not lead to a switch on less visited species, but it decreased the evenness of wild hymenopteran visits on floral resources, suggesting that some resources became more preferred. 8 \| Increasing values of the index correspond to an increase in specialization FI G U R E 5 Partial residual plot showing the effect of the number of flowering plant species (ln-transformed) on PDI (Paired Difference Index). Increasing values of the index correspond to an increase in specialization CORCOS et al. \| 9 FI G U R E 5 Partial residual plot showing the effect of the number of flowering plant species (ln-transformed) on PDI (Paired Difference Index). Increasing values of the index correspond to an increase in specialization FI G U R E 4 Partial residual plot showing the interaction between insect order and honeybee abundance (ln-transformed) on dʹ (standardized Kullback–Leibler distance). Increasing values of the index correspond to an increase in specialization FI G U R E 4 Partial residual plot showing the interaction between insect order and honeybee abundance (ln-transformed) on dʹ (standardized Kullback–Leibler distance). Increasing values of the index correspond to an increase in specialization FI G U R E 5 Partial residual plot showing the effect of the number of flowering plant species (ln-transformed) on PDI (Paired Difference Index). Increasing values of the index correspond to an increase in specialization FI G U R E 5 Partial residual plot showing the effect of the number of flowering plant species (ln-transformed) on PDI (Paired Difference Index). Increasing values of the index correspond to an increase in specialization Despite the common knowledge of lepidopterans having inter- mediate level of specialization compared with hymenopterans (more specialized) and dipterans (less specialized) (Benadi et al., 2014; Chacoff et al., 2012), in our study lepidopterans displayed the high- est degree of specialization in term number of visited plant species. This is probably due the high attractiveness of some plant species to butterflies, including B. davidii (Gong et al., 2015). Indeed, in our study over the 76% of lepidopterans was collected in sites invaded by the exotic species, and the 36% was directly foraging on B. davi- dii. Among the remaining 24% lepidopterans collected in uninvaded sites, over the 33% was collected on Campanula persicifolia L. in a single sampling site. The observed high selectivity of lepidopterans is thus likely to depend on the identity of the plant species rather than the ecology of the order. 8 \| compared with wild bees (Magrach et al., 2017; Valido et al., 2019), its increased abundance might have positive effects on network sta- bility by increasing network robustness against pollinator extinction. However, honeybees also impacted wild flower visitors, causing a shift in their diets and an increase in their specialization. In contrast, the presence of B. davidii facilitated the visits on neighbouring plants with inferior rewards, as this invasive plant is an excellent source of nectar that can be exploited by many insect species. The conse- quences of these changes in mutualistic interactions for the repro- ductive success of flowering plants are still largely unknown, calling for more research on the long-term effects of honeybee on plant community dynamics. In contrast to previous studies (Ebeling et al., 2011; Fründ, Linsenmair, & Blüthgen, 2010), we found that the number of flower- ing plants positively influenced flower visitor specialization in terms of normalized degree and PDI. When the plant community was rich in flower resources, flower visitors reduced their diet breath (i.e. number of resources they visited) and thus increased their special- ization, probably because species were more selective choosing more nutrient-rich species compared with situations where only few resources were available. However, dʹ did not change with increasing flowering plant richness, suggesting that the reduction in diet breath of flower visitors might not result in a decreased resource overlap with other flower visitor species. 5 \| CONCLUSIONS Primary data are provided in the main text or in the Appendix S1. Primary data are provided in the main text or in the Appendix S1. Contrary to our expectations, we found no interactive effect of B. davidii and honeybee abundance on plant–flower visitor interactions. However, we showed that the invasive plant and managed honey- bees can jointly affect flower visitor communities. Although the honeybee is sometimes considered to be a less efficient pollinator PEER REVIEW The peer review history for this article is available at https://publo ns.com/publon/10.1111/ddi.13132. ACKNOWLEDGEMENTS The research was supported by the University of Padua STARS Consolidator Grant (STARS-CoG–2017) to LM. We would like to thank Filippo Prosser (Museum of Rovereto, Italy) for helping in the identification of plants and Paolo Paolucci (University of Padua) for the identification of butterflies. We also thank Francesca Zampieri and Valerio Caruso for support in the fieldwork. 8 \| CORCOS et al. 8 Estimate SE df t p-value (a) 1–normalized degree Intercept −0.755 0.148 30 −5.103 <.001 Species richness of flowering plants 0.124 0.044 6 2.819 .030 Honeybee 0.010 0.009 6 1.205 .274 B. davidii (invaded vs. uninvaded) 0.051 0.021 6 2.423 .052 Order: Hymenoptera 0.075 0.023 30 3.245 .003 Order: Lepidoptera 0.129 0.025 30 5.164 <.001 Order: Hymenoptera × Honeybee – – – – – Order: Lepidoptera × Honeybee – – – – – (b) dʹ Intercept −0.163 0.317 28 −0.514 .611 Species richness of flowering plants 0.106 0.091 6 1.160 .290 Honeybee 0.046 0.025 6 1.822 .118 B. davidii (invaded vs. uninvaded) 0.046 0.044 6 1.064 .328 Order: Hymenoptera 0.383 0.158 28 2.423 .022 Order: Lepidoptera −0.373 0.173 28 −2.152 .040 Order: Hymenoptera × Honeybee −0.074 0.031 28 −2.375 .025 Order: Lepidoptera × Honeybee 0.048 0.034 28 1.394 .174 (c) PDI Intercept 0.768 0.055 30 13.974 <.001 Species richness of flowering plants 0.044 0.016 6 2.692 .036 Honeybee 0.003 0.003 6 0.827 .440 B. davidii (uninvaded vs. invaded) 0.010 0.008 6 1.303 .240 Order: Hymenoptera 0.009 0.009 30 1.006 .323 Order: Lepidoptera 0.018 0.009 30 1.944 .061 Order: Hymenoptera × Honeybee – – – – – Order: Lepidoptera × Honeybee – – – – – Note: Honeybee abundance, flowering plant species richness and normalized degree were ln- transformed to improve linearity and model residual distribution. Interactions are presented only Note: Honeybee abundance, flowering plant species richness and normalized degree were ln- transformed to improve linearity and model residual distribution. Interactions are presented only when significant (p-value < .05). (b) FI G U R E 3 Partial residual plots showing the effect of (a) presence of B. davidii, (b) insect order and (c) number of flowering plant species (ln-transformed) on normalized degree (ln-transformed). Increasing values of the index correspond to an increase in specialization (a) (b) (c) (a) (c) (a) (b) (c) FI G U R E 3 Partial residual plots showing the effect of (a) presence of B. davidii, (b) insect order and (c) number of flowering plant species (ln-transformed) on normalized degree (ln-transformed). Increasing values of the index correspond to an increase in specialization \| 9 CORCOS et al. FI G U R E 4 Partial residual plot showing the interaction between insect order and honeybee abundance (ln-transformed) on dʹ (standardized Kullback–Leibler distance). REFERENCES Adedoja, O. A., Kehinde, T., & Samways, M. J. (2018). Insect-flower inter- action networks vary among endemic pollinator taxa over an eleva- tion gradient. PLoS ONE, 13(11), 1–17. https://doi.org/10.1371/journ al.pone.0207453 Ebeling, S., Schreiter, S., Hensen, I., Durka, W., & Auge, H. (2012). Outcrossing breeding system does not compromise invasiveness in Buddleja davidii. Flora, 207(12), 843–848. https://doi.org/10.1016/j. flora.2012.09.010 Albrecht, M., Ramis, M. R., & Traveset, A. (2016). Pollinator-mediated im- pacts of alien invasive plants on the pollination of native plants: the role of spatial scale and distinct behaviour among pollinator guilds. Biological Invasions, 18(7), 1801–1812. https://doi.org/10.1007/ s10530-016-1121-6 Flanagan, R. J., Mitchell, R. J., & Karron, J. D. (2010). Increased relative abundance of an invasive competitor for pollination, Lythrum sali- caria, reduces seed number in Mimulus ringens. Oecologia, 164(2), 445–454. https://doi.org/10.1007/s00442-010-1693-2 Alexander, J. M., Kueffer, C., Daehler, C. C., Edwards, P. J., Pauchard, A., Seipel, T., … Walsh, N. (2011). Assembly of nonnative floras along elevational gradients explained by directional ecological filtering. Proceedings of the National Academy of Sciences of the United States of America, 108(2), 656–661. https://doi.org/10.1073/pnas.10131 36108 Franco, E. L., Aguiar, C. M. L., Ferreira, V. S., & De Oliveira-Rebouças, P. L. (2009). Plant use and niche overlap between the introduced honey bee (Apis mellifera) and the native bumblebee (Bombus atratus) (Hymenoptera: Apidae) in an Area of tropical mountain vegetation in northeastern Brazil. Sociobiology, 53(1), 141–150. Benadi, G., Hovestadt, T., Poethke, H.-J., & Blüthgen, N. (2014). Specialization and phenological synchrony of plant–pollinator inter- actions along an altitudinal gradient. Journal of Animal Ecology, 83(3), 639–650. https://doi.org/10.1111/1365-2656.12158 Fründ, J., Linsenmair, K. E., & Blüthgen, N. (2010). Pollinator diversity and specialization in relation to flower diversity. Oikos, 119(10), 1581– 1590. https://doi.org/10.1111/j.1600-0706.2010.18450.x Blüthgen, N., Menzel, F., & Blüthgen, N. (2006). Measuring specializa- tion in species interaction networks. BMC Ecology, 6, 9. https://doi. org/10.1186/1472-6785-6-9 Geslin, B., Gauzens, B., Baude, M., Dajoz, I., Fontaine, C., Henry, M., … Vereecken, N. J. (2017). Massively introduced managed species and their consequences for plant–pollinator interactions. Advances in Ecological Research, 57, 147–199. https://doi.org/10.1016/ bs.aecr.2016.10.007 Breheny, P., & Burchett, W. (2017). Visualization of regression mod- els using visreg. R Journal, 9(2), 56–71. https://doi.org/10.32614/ rj-2017-046 Gillespie, S., & Elle, E. (2018). Non-native plants affect generalist pol- linator diet overlap and foraging behavior indirectly, via impacts on native plant abundance. Biological Invasions, 20(11), 3179–3191. https://doi.org/10.1007/s10530-018-1767-3 Burgos, E., Ceva, H., Perazzo, R. P. J., Devoto, M., Medan, D., Zimmermann, M., & María Delbue, A. REFERENCES (2007). Why nestedness in mutualistic net- works? Journal of Theoretical Biology, 249(2), 307–313. https://doi. org/10.1016/j.jtbi.2007.07.030 Giovanetti, M., Ramos, M., & Máguas, C. (2018). Why so many flowers? A preliminary assessment of mixed pollination strategy enhancing sex- ual reproduction of the invasive Acacia longifolia in Portugal. Web Ecology, 18(1), 47–54. https://doi.org/10.5194/we-18-47-2018 Chacoff, N. P., Vázquez, D. P., Lomáscolo, S. B., Stevani, E. L., Dorado, J., & Padrón, B. (2012). Evaluating sampling completeness in a desert plant–pollinator network. Journal of Animal Ecology, 81(1), 190–200. https://doi.org/10.1111/j.1365-2656.2011.01883.x Gong, W. C., Chen, G., Vereecken, N. J., Dunn, B. L., Ma, Y. P., & Sun, W. B. (2015). Floral scent composition predicts bee pollination sys- tem in five butterfly bush (Buddleja, Scrophulariaceae) species. Plant Biology, 17(1), 245–255. https://doi.org/10.1111/plb.12176 Chagnon, P. L. (2015). Characterizing topology of ecological net- works along gradients: The limits of metrics’ standardization. Ecological Complexity, 22, 36–39. https://doi.org/10.1016/j. ecocom.2015.01.004 González-Varo, J. P., Biesmeijer, J. C., Bommarco, R., Potts, S. G., Schweiger, O., Smith, H. G., … Vilà, M. (2013). Combined effects of global change pressures on animal-mediated pollination. Trends in Ecology & Evolution, 28(9), 524–530. https://doi.org/10.1016/j. tree.2013.05.008 Dainese, M., Aikio, S., Hulme, P. E., Bertolli, A., Prosser, F., & Marini, L. (2017). Human disturbance and upward expansion of plants in a warming climate. Nature Climate Change, 7(8), 577–580. https://doi. org/10.1038/nclimate3337 Davis, E. S., Kelly, R., Maggs, C. A., & Stout, J. C. (2018). Contrasting im- pacts of highly invasive plant species on flower-visiting insect com- munities. Biodiversity and Conservation, 27(8), 2069–2085. https:// doi.org/10.1007/s10531-018-1525-y Goulson, D. (2003). Effects of introduced bees on native ecosystems. Annual Review of Ecology, Evolution, and Systematics, 34(1), 1–26. https://doi.org/10.1146/annurev.ecolsys.34.011802.132355 Grass, I., Berens, D. G., Peter, F., & Farwig, N. (2013). Additive effects of exotic plant abundance and land-use intensity on plant–pollinator interactions. Oecologia, 173(3), 913–923. https://doi.org/10.1007/ s00442-013-2688-6 Delmas, E., Besson, M., Brice, M. H., Burkle, L. A., Dalla Riva, G. V., Fortin, M. J., … Poisot, T. (2019). Analysing ecological networks of species in- teractions. Biological Reviews, 94(1), 16–36. https://doi.org/10.1111/ brv.12433 Guo, Q., Kelt, D. A., Sun, Z., Liu, H., Hu, L., Ren, H., & Wen, J. (2013). Global variation in elevational diversity patterns. Scientific Reports, 3(1), 3007. https://doi.org/10.1038/srep03007 Devoto, M., Bailey, S., Craze, P., & Memmott, J. (2012). Understanding and planning ecological restoration of plant-pol- linator networks. Ecology Letters, 15(4), 319–328. https://doi. org/10.1111/j.1461-0248.2012.01740.x Hansen, S., Roets, F., Seymour, C. L., Thébault, E., van Veen, F. J. Ebeling, A., Klein, A.-M., & Tscharntke, T. (2011). Plant–flower visitor interaction webs: Temporal stability and pollinator specialization increases along an experimental plant diversity gradient. Basic and Applied Ecology, 12(4), 300–309. https://doi.org/10.1016/j. baae.2011.04.005 ORCID Daria Corcos https://orcid.org/0000-0002-6771-1769 Andree Cappellari https://orcid.org/0000-0002-6726-1323 Pierfilippo Cerretti https://orcid.org/0000-0002-9204-3352 Lorenzo Marini https://orcid.org/0000-0001-7429-7685 CORCOS et al. 10 Journal of the Royal Statistical Society. Series C (Applied Statistics), 30(1), 91–97. Journal of the Royal Statistical Society. Series C (Applied Statistics), 30(1), 91–97. Hung, K.-L. J., Kingston, J. M., Lee, A., Holway, D. A., & Kohn, J. R. (2019). Non-native honey bees disproportionately dominate the most abun- dant floral resources in a biodiversity hotspot. Proceedings of the Royal Society B: Biological Sciences, 286(1897), 20182901. https://doi. org/10.1098/rspb.2018.2901 Pinheiro, J., Bates, D., DebRoy, S., Sarkar, D., &R Core Team (2017). nlme: Linear and Nonlinear Mixed Effects Models. R Package Version 3.1-131. Retrieved from https://cran.r-project.org/package=nlme Johnson, S. D., & Steiner, K. E. (2000). Generalization versus specializa- tion in plant pollination systems. Trends in Ecology & Evolution, 15(4), 140–143. https://doi.org/10.1016/S0169-5347(99)01811-X Poisot, T., Canard, E., Mouquet, N., & Hochberg, M. E. (2012). A comparative study of ecological specialization estimators. Methods in Ecology and Evolution, 3(3), 537–544. https://doi.org/10.1111/j.2041-210X.2011.00174.x R Development Core Team (2015). R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. Retrieved from http://www.r-project.org/ Kevan, P. G., & Baker, H. G. (1983). Insects as flower visitors and pol- linators. Annual Review of Entomology, 28(1), 407–453. https://doi. org/10.1146/annurev.en.28.010183.002203 Rader, R., Howlett, B. G., Cunningham, S. A., Westcott, D. A., Newstrom- Lloyd, L. E., Walker, M. K., … Edwards, W. (2009). Alternative pollina- tor taxa are equally efficient but not as effective as the honeybee in a mass flowering crop. Journal of Applied Ecology, 46(5), 1080–1087. https://doi.org/10.1111/j.1365-2664.2009.01700.x Klinkhamer, P. G. L., de Jong, T. J., & Linnebank, L. A. (2001). Small-scale spatial patterns determine ecological relationships: an experimental example using nectar production rates. Ecology Letters, 4(6), 559– 567. https://doi.org/10.1046/j.1461-0248.2001.00267.x Lefebvre, V., Villemant, C., Fontaine, C., & Daugeron, C. (2018). Altitudinal, temporal and trophic partitioning of flower-visitors in Alpine communities. Scientific Reports, 8(1), 4706. https://doi. org/10.1038/s41598-018-23210-y Ramos-Jiliberto, R., Domínguez, D., Espinoza, C., López, G., Valdovinos, F. S., Bustamante, R. O., & Medel, R. (2010). Topological change of Andean plant–pollinator networks along an altitudinal gradient. Ecological Complexity, 7(1), 86–90. https://doi.org/10.1016/j.ecocom.2009.06.001 Lopezaraiza-Mikel, M. E., Hayes, R. B., Whalley, M. R., & Memmott, J. (2007). The impact of an alien plant on a native plant–pollinator net- work: An experimental approach. Ecology Letters, 10(7), 539–550. https://doi.org/10.1111/j.1461-0248.2007.01055.x Russo, L., Nichol, C., & Shea, K. (2016). Pollinator floral provisioning by a plant invader: Quantifying beneficial effects of detrimental species. Diversity and Distributions, 22(2), 189–198. https://doi.org/10.1111/ddi.12397 Maglianesi, M. A., Blüthgen, N., Böhning-Gaese, K., & Schleuning, M. (2014). Morphological traits determine specialization and resource use in plant–hummingbird networks in the neotropics. Ecology, 95(12), 3325–3334. https://doi.org/10.1890/13-2261.1 Santos, G. M., Aguiar, C. M. L., Genini, J., Martins, C. F., Zanella, F. C. V., & Mello, M. A. R. (2012). Invasive Africanized honeybees change the structure of native pollination networks in Brazil. Biological Invasions, 14(11), 2369–2378. https://doi.org/10.1007/s10530-012-0235-8 Magrach, A., González-Varo, J. P., Boiffier, M., Vilà, M., & Bartomeus, I. (2017). Honeybee spillover reshuffles pollinator diets and affects plant reproductive success. Nature Ecology and Evolution, 1(9), 1299– 1307. https://doi.org/10.1038/s41559-017-0249-9 Schleuning, M., Fründ, J., Klein, A.-M., Abrahamczyk, S., Alarcón, R., Albrecht, M., … Blüthgen, N. (2012). Specialization of mutual- istic interaction networks decreases toward tropical latitudes. Current Biology, 22(20), 1925–1931. https://doi.org/10.1016/j. cub.2012.08.015 Marini, L., Bertolli, A., Bona, E., Federici, G., Martini, F., Prosser, F., & Bommarco, R. (2013). Beta-diversity patterns elucidate mechanisms of alien plant invasion in mountains. Global Ecology and Biogeography, 22(4), 450–460. https://doi.org/10.1111/geb.12006 Steffan-Dewenter, I., & Kuhn, A. (2003). Honeybee foraging in differ- entially structured landscapes. Proceedings of the Royal Society of London. Series B: Biological Sciences, 270(1515), 569–575. https://doi. org/10.1098/rspb.2002.2292 Maruyama, P. K., Vizentin-Bugoni, J., Sonne, J., Martín González, A. M., Schleuning, M., Araujo, A. C., … Dalsgaard, B. (2016). The integration of alien plants in mutualistic plant-hummingbird networks across the Americas: The importance of species traits and insularity. Diversity and Distributions, 22(6), 672–681. https://doi.org/10.1111/ddi.12434 Steffan-Dewenter, I., & Tscharntke, T. (2000). Resource overlap and possible competition between honey bees and wild bees in Central Europe. Oecologia, 122(2), 288–296. Stout, J. C., & Tiedeken, E. J. (2017). Direct interactions between invasive plants and native pollinators: evidence, impacts and approaches. Functional Ecology, 31(1), 38–46. https://doi. org/10.1111/1365-2435.12751 Memmott, J., Waser, N. M., & Price, M. V. (2004). Tolerance of pollination networks to species extinctions. Proceedings of the Royal Society of London. Series B: Biological Sciences, 271(1557), 2605–2611. https:// doi.org/10.1098/rspb.2004.2909 Thébault, E., & Fontaine, C. (2010). Stability of ecological communities and the architecture of mutualistic and trophic networks. Science, 329(5993), 853–856. https://doi.org/10.1126/science.1188321 Molina-Montenegro, M. A., Badano, E. I., & Cavieres, L. A. (2008). Positive interactions among plant species for pollinator service: assessing the ‘magnet species’ concept with invasive species. Oikos, 117(12), 1833– 1839. https://doi.org/10.1111/j.0030-1299.2008.16896.x Thomson, D. M. (2016). Local bumble bee decline linked to recovery of honey bees, drought effects on floral resources. Ecology Letters, 19(10), 1247–1255. https://doi.org/10.1111/ele.12659 Montero-Castaño, A., & Vilà, M. (2017). REFERENCES F., & Pryke, J. S. (2018). Alien plants have greater impact than habitat frag- mentation on native insect flower visitation networks. Diversity and Distributions, 24(1), 58–68. https://doi.org/10.1111/ddi.12656 Dietzsch, A. C., Stanley, D. A., & Stout, J. C. (2011). Relative abun- dance of an invasive alien plant affects native pollination pro- cesses. Oecologia, 167(2), 469–479. https://doi.org/10.1007/s0044 2-011-1987-z butions, 24(1), 58–68. https://doi.org/10.1111/ddi.12656 Hoiss, B., Krauss, J., & Steffan-Dewenter, I. (2015). Interactive effects of elevation, species richness and extreme climatic events on plant–pol- linator networks. Global Change Biology, 21(11), 4086–4097. https:// doi.org/10.1111/gcb.12968 Dorado, J., Vázquez, D. P., Stevani, E. L., & Chacoff, N. P. (2011). Rareness and specialization in plant–pollinator networks. Ecology, 92(1), 19–25. https://doi.org/10.1890/10-0794.1 Hung, K.-L. J., Kingston, J. M., Albrecht, M., Holway, D. A., & Kohn, J. R. (2018). The worldwide importance of honey bees as pollinators in natural habitats. Proceedings of the Royal Society B: Biological Sciences, 285(1870), 20172140. https://doi.org/10.1098/rspb.2017.2140 Dormann, C. F. (2011). How to be a specialist? Quantifying specialisation in pollination networks. Network Biology, 1(1), 1–20. Dormann, C. F., Fründ, J., & Gruber, B. (2008). Introducing the bipartite Package: Analysing Ecological Networks. R News, 8, 8–11. Package: Analysing Ecological Networks. R News, 8, 8–11. CORCOS et al. 11 SUPPORTING INFORMATION SUPPORTING INFORMATION Influence of the honeybee and trait similarity on the effect of a non-native plant on pollination and network rewiring. Functional Ecology, 31(1), 142–152. https://doi. org/10.1111/1365-2435.12712 Timóteo, S., O'Connor, C. J., López-Núñez, F. A., Costa, J. M., Gouveia, A. C., & Heleno, R. H. (2018). Pollination networks from natural and anthropogenic-novel communities show high structural similar- ity. Oecologia, 188(4), 1155–1165. https://doi.org/10.1007/s0044 2-018-4281-5 Morisita, M. (1959). Measuring of the dispersion of individuals and anal- ysis of the distributional patterns. Memoires of the Faculty of Science, Kyushu University. Series E. Biology, 2, 215–235. Valido, A., Rodríguez-Rodríguez, M. C., & Jordano, P. (2019). Honeybees disrupt the structure and functionality of plant–pollinator networks. Scientific Reports, 9(1), 4711. https://doi.org/10.1038/s41598-019- 41271-5 Norfolk, O., Gilbert, F., & Eichhorn, M. P. (2018). Alien honeybees in- crease pollination risks for range-restricted plants. Diversity and Distributions, 24(5), 705–713. https://doi.org/10.1111/ddi.12715 Vázquez, D. P., Melián, C. J., Williams, N. M., Blüthgen, N., Krasnov, B. R., & Poulin, R. (2007). Species abundance and asymmetric interaction strength in ecological networks. Oikos, 116(7), 1120–1127. https:// doi.org/10.1111/j.2007.0030-1299.15828.x Oksanen, J., Blanchet, F. G., Friendly, M., Kindt, R., Legendre, P., Mcglinn, D., … Wagner, H. (2019). Package “vegan.” Community Ecology Package. Paini, D. R. (2004). Impact of the introduced honey bee (Apis mellifera) (Hymenoptera: Apidae) on native bees: A review. Austral Ecology, 29(4), 399–407. https://doi.org/10.1111/j.1442-9993.2004.01376.x Zhang, J. (2016). Package “spaa”: SPecies Association Analysis. R Package Version 0.2.2. Retrieved from https://CRAN.R-Project.Org/Packa ge=spaa Patefield, W. (2012). Algorithm AS 159: An Efficient Method of Generating Random R × C tables with given row and column totals. CORCOS et al. 12 BIOSKETCH This work is the result of a collaboration between researchers from the University of Padua and the University of Rome. The group in Padua performed research on applied ecology and global change, while the team in Rome focused on insect system- atics, phylogeny and evolution. Additional supporting information may be found online in the Supporting Information section. How to cite this article: Corcos D, Cappellari A, Mei M, Paniccia D, Cerretti P, Marini L. Contrasting effects of exotic plant invasions and managed honeybees on plant–flower visitor interactions. Divers Distrib. 2020;00:1–12. https://doi. org/10.1111/ddi.13132 Author contributions: L.M. and D.C. conceived the ideas and de- signed the sampling; D.C. and A.C. collected the field data; A.C., M.M., D.P. and P.C. identified insects; D.C., A.C. and L.M. ana- lysed the data; D.C. and A.C. led the writing of the manuscript. All authors contributed critically to the drafts and gave final ap- proval for publication.
https://openalex.org/W2546861366	https://europepmc.org/articles/pmc5090330?pdf=render	English	null	Colour formation on the wings of the butterfly Hypolimnas salmacis by scale stacking	Scientific reports	2,016	cc-by	8,874	Radwanul Hasan Siddique1,2,†, Silvia Vignolini2, Carolin Bartels3, Irene Wacker3 & Hendrik Hölscher1 The butterfly genus Hypolimnas features iridescent blue colouration in some areas of its dorsal wings. Here, we analyse the mechanisms responsible for such colouration on the dorsal wings of Hypolimnas salmacis and experimentally demonstrate that the lower thin lamina in the white cover scales causes the blue iridescence. This outcome contradicts other studies reporting that the radiant blue in Hypolimnas butterflies is caused by complex ridge-lamellar architectures in the upper lamina of the cover scales. Our comprehensive optical study supported by numerical calculation however shows that scale stacking primarily induces the observed colour appearance of Hypolimnas salmacis. Many butterflies (Order: Lepidoptera) possess very fascinating colouration, and most species of butterfly can be identified solely by the colour pattern on their wings1–3. Such a diversity has attracted a wealth of research to determine the mechanisms responsible for such colours4–6. From a functional perspective, wing coloura- tion can be important in a multitude of ways, ranging from mating3,7, camouflage8,9 and warning purposes10,11. Moreover, structurally unique, visually chromatic and complex colour mechanisms found in numerous butter- fly wings inspired various advanced technical applications. For example, a hierarchical multilayer air-cuticle pattern inspired by Morpho butterflies was mimicked for selective gas sensing12 and colourful hydropho- bic coatings13. Colour mixing due to multilayer microcavities inspired from Papilio blumei was replicated for polarization-sensitive optical signatures14. The reverse diffracting grating effect of Pierella luna was copied arti- ficially and might be useful for bio-sensing and anti-counterfeiting15. Hence, analysing butterfly colour patterns does not only enrich our understanding of inter/intra-specific communication and the evolution of exaggerated signalling of butterflies, it can also be the source of new photonic devices.l g gl p Hypolimnas salmacis, also known as the “Blue Diadem”, is a butterfly in the family Nymphalidae from the Afro-tropic eco-zone16 (Fig. 1a). It is one of 23 butterflies found in the genus Hypolimnas16 that is well-known for its sexual dimorphism. In other words, the two sexes of the same species are distinct in colour, shape, size, and structure. On the dorsal wings, almost all Hypolimnas species, including Hypolimnas salmacis, carry UV-white, blue and brown/black patches17. As the respective colours, including their brightness and saturation, are crucial for the animals’ selective mate, species or rival recognition18,19, it is of utmost importance to understand the basis of their colouration. www.nature.com/scientificreports www.nature.com/scientificreports www.nature.com/scientificreports received: 09 June 2016 accepted: 12 October 2016 Published: 02 November 2016 Colour formation on the wings of the butterfly Hypolimnas salmacis by scale stacking OPEN received: 09 June 2016 accepted: 12 October 2016 Published: 02 November 2016 Radwanul Hasan Siddique1,2,†, Silvia Vignolini2, Carolin Bartels3, Irene Wacker3 & Hendrik Hölscher1 Radwanul Hasan Siddique1,2,†, Silvia Vignolini2, Carolin Bartels3, Irene Wacker3 & Hendrik Hölscher1 Scientific Reports \| 6:36204 \| DOI: 10.1038/srep36204 Results Optical appearance of scales on the wing. The dorsal side of the Hypolimnas salmacis wings features dark brown or black, white, and blue regions. On the ventral side, white regions mimic the dorsal side while everywhere else the wings are covered by light brown scales. The basal region is almost black near the thorax for both forewing and hindwing. The outer margin of the whole wing is mostly brown and covered with blue and white patches in the middle. A stereo-microscopy image of the Hypolimnas salmacis wing is shown in Fig. 1b. Interestingly, the blue colouration is visible only in diffused illumination, suggesting that the origin of colouration is purely physical. Moreover, tilting the wing shifts the blue colour to purple-violet (Fig. 1c) that demonstrates the iridescent property of the Hypolimnas salmacis wing. As in most other Lepidoptera species, both white and black/ brown areas are uniformly covered with cover and ground scales4. y g As clearly visible in Fig. 1b, some of the cover scales in the blue areas are not completely blue. While examin- ing thoroughly scales in blue areas, we observed that white cover scales stacked on brown ground scales appear blue. And, the same cover scales look white when lying on top of white ground scales. To further confirm our observation, we picked a single white cover and a single brown ground scale from the blue region. We manually overlapped the white cover and brown ground scales whereupon the blue colouration appears as a result of the stacking (see Fig. 1d). The cover scale appeared white when the bottom brown scale was absent. Hence, the for- mation of the blue colour on the Hypolimnas salmacis wings occurs due to scale stacking when a white scale is on top of a brown one. Electron microscopy of scales. Creating blue colour out of white and brown does not accord with con- ventional colour blending or filtering mechanism. To understand this effect, we first analysed the scales by elec- tron microscopy (Fig. 2a–d). The blue areas consist of stacks of white and brown scales (Fig. 2a,b) as already observed from the optical analysis. The SEM images of the white and brown scales show the typical oval shape with a width of 100 μm and length of 200 μm. Both of them consist of grating-like ridges with a typical distance of (2 ± 0.2) μm. Radwanul Hasan Siddique1,2,†, Silvia Vignolini2, Carolin Bartels3, Irene Wacker3 & Hendrik Hölscher1 Studies on Hypolimnas bolina suggested that structurally coloured ornaments serve as sexual signals20 and that they carry potentially useful information on phenotypic quality21. After investigating the reflec- tance of these butterfly wings, Kemp et al.22 reported that the angle-dependent reflectance arises as a consequence of the optical mechanisms including constructive interference and diffraction at the complex ridge-lamellar for- mation of the upper lamina. In a more recent study on another genus, Hypolimnas alimena, the dull colouration mechanism of the scales is explained via simpler micro-architecture of the upper lamina as well19. Here, we ana- lyse the colour formation on the wings of Hypolimnas salmacis and show that its blue iridescence is rather caused by the lower thin lamina of the cover scale. We demonstrate explicitly that the blue colouration arises primarily from the stacking of brown and white scales on the wing. 1Institute of Microstructure Technology, Karlsruhe Institute of Technology (KIT), Hermann-von-Helmholtz-Platz 1, 76344 Eggenstein-Leopoldshafen, Germany. 2Department of Chemistry, University of Cambridge, Lensfield Road, Cambridge CB2 1EW, UK. 3Centre for Advanced Materials (CAM), University of Heidelberg, 69120 Heidelberg, Germany. †Present address: Medical Engineering, California Institute of Technology (Caltech), MC 136-93, Pasadena, CA, USA. Correspondence and requests for materials should be addressed to R.H.S. (email: rhs@caltech.edu) or H.H. (email: hendrik.hoelscher@kit.edu) Scientific Reports \| 6:36204 \| DOI: 10.1038/srep36204 www.nature.com/scientificreports/ www.nature.com/scientificreports/ Figure 1. Scales of the butterfly Hypolimnas salmacis. (a) Photo of a Hypolimnas salmacis butterfly. Its dorsal wings feature three regions: blue, white and dark brown/black. (b) Optical image of the forewing showing all three regions without and with diffused light from the top. Blue coloured scales only appear under diffused light condition. A zoom shows the scales stacking on the wing. (c) By tilting of the wing, the colour of blue regions shifts to purple-violet. This effect demonstrates the iridescence of Hypolimnas salmacis butterfly wings. (d) Overlapping of single white and brown scales causes the blue appearance of the white scales and reveals the colouration mechanism by scale stacking. Figure 1. Scales of the butterfly Hypolimnas salmacis. (a) Photo of a Hypolimnas salmacis butterfly. Its dorsal wings feature three regions: blue, white and dark brown/black. (b) Optical image of the forewing showing all three regions without and with diffused light from the top. Blue coloured scales only appear under diffused light condition. A zoom shows the scales stacking on the wing. Radwanul Hasan Siddique1,2,†, Silvia Vignolini2, Carolin Bartels3, Irene Wacker3 & Hendrik Hölscher1 (c) By tilting of the wing, the colour of blue regions shifts to purple-violet. This effect demonstrates the iridescence of Hypolimnas salmacis butterfly wings. (d) Overlapping of single white and brown scales causes the blue appearance of the white scales and reveals the colouration mechanism by scale stacking. Results The brown scales feature thin membranes between the ridges (Fig. 2d) while the white ones do not (Fig. 2c). Cross-ribs across the ridges create open areas with a size of (2 ± 0.2) μm × (1 ± 0.1) μm. Such windows created by cross-ribs are commonly termed alveoli and found in Papilionids6. The sides of the ridges are covered with microribs. The cross-sectional SEM image of the blue region shown in Fig. 2e reveals longitudinal ridges of cover and ground scales both consisting of very small microribs. The ridges are standing on a single thin film (lower lamina) with a thickness of around (190 ± 20) nm. Thin membranes can be observed in the windows of the brown scales between the ridges (Fig. 2d,e). Optical spectroscopy of scales on the wing. Total diffusive reflection spectra were measured in differ- ent regions of the Hypolimnas salmacis wings with an integrating sphere. Figure 3 shows the obtained spectra of blue, white and brown regions marked in the inset. Blue areas feature a weak reflection of only 20% reflectance Scientific Reports \| 6:36204 \| DOI: 10.1038/srep36204 2 www.nature.com/scientificreports/ Figure 2. Micro- and nanostructure of the Hypolimnas salmacis scales. (a,b) SEM images cover scales on brown ground scales. (c) Detail of a single white scale. Longitudinal grating li scale are connected with cross-ribs. The sides of the ridges are covered with small microribs. brown scale. In terms of dimension, it mimics almost exactly the white scale but the windows cross-ribs, are closed by thin pigmented membranes. (e) SEM image of a cross section of the f a white and brown scale. Both scales show similar features with upper lamina of ridges and ti well as lower lamina of thin films. Thin membranes between the ridges are visible only in the Figure 2. Micro- and nanostructure of the Hypolimnas salmacis scales. (a,b) SEM images of stacks of white cover scales on brown ground scales. (c) Detail of a single white scale. Longitudinal grating like ridges along the scale are connected with cross-ribs. The sides of the ridges are covered with small microribs. (d) Detail of single brown scale. In terms of dimension, it mimics almost exactly the white scale but the windows, created by the cross-ribs, are closed by thin pigmented membranes. (e) SEM image of a cross section of the forewing including a white and brown scale. Results However, some small mismatch can be noticed at long wavelengths, which might be due to a partial coherence effect during the light propagation which is not considered in the equation above23. In a next step we placed a white scale on top of a brown scale and the measured reflection spectrum (solid blue line, Fig. 4d) shows a reflection peak at 420 nm in the same blue spectral area as observed in the reflection spectrum of the isolated white scale. These observations suggest that a white scale on an absorbing background exhibits blue colouration. To check the consistency of our measurements and assumptions we also calculated the non-coherent scattering of a white scale on a brown scale assuming non-coherent scattering during the light propagation in the stack. The resulting spectrum (black dashed line) is shown in Fig. 4c with a concise schematic noting the terms as inset. The calculated spectrum fits well with the measured stacked spectrum. However, some small mismatch can be noticed at long wavelengths, which might be due to a partial coherence effect during the light propagation which is not considered in the equation above23.l In a next step we placed a white scale on top of a brown scale and the measured reflection spectrum (solid blue line, Fig. 4d) shows a reflection peak at 420 nm in the same blue spectral area as observed in the reflection spectrum of the isolated white scale. These observations suggest that a white scale on an absorbing background exhibits blue colouration. To check the consistency of our measurements and assumptions we also calculated the non-coherent scattering of a white scale on a brown scale = + × × . R R T R T (2) w,b white white brown white = + × × . R R T R T w,b white white brown white (2) The calculated spectrum (black dashed line) of the stacked system in this case matches the measured spectrum completely as depicted in Fig. 4d. Hence, the overall result explains the scale stacking phenomena satisfactorily. The calculated spectrum (black dashed line) of the stacked system in this case matches the measured spectrum completely as depicted in Fig. 4d. Hence, the overall result explains the scale stacking phenomena satisfactorily. Translucency of white scales. The above presented micro-spectroscopy of a white scale resulted in a trans- mittance of about 90%. Results Both scales show similar features with upper lamina of ridges and tiny microribs as well as lower lamina of thin films. Thin membranes between the ridges are visible only in the brown scale. Figure 2. Micro- and nanostructure of the Hypolimnas salmacis scales. (a,b) SEM images of stacks of white cover scales on brown ground scales. (c) Detail of a single white scale. Longitudinal grating like ridges along the scale are connected with cross-ribs. The sides of the ridges are covered with small microribs. (d) Detail of single brown scale. In terms of dimension, it mimics almost exactly the white scale but the windows, created by the cross-ribs, are closed by thin pigmented membranes. (e) SEM image of a cross section of the forewing including a white and brown scale. Both scales show similar features with upper lamina of ridges and tiny microribs as well as lower lamina of thin films. Thin membranes between the ridges are visible only in the brown scale. with a broad peak at ≈430 nm. In the reflection spectrum of the white area, there is no particular peak in the visible regime, resulting in the overall white appearance. Nonetheless, a considerably broad reflection peak can be noticed in the UV at ≈375 nm. A similar UV reflection was reported in the white patch of H. bolina22. Although the overall structure of the brown scales of H. salmacis is comparable to that of the white scales, the brown area has no reflection peak neither in UV nor in the visible spectral regime. This is presumably due to high melanin pigmentation in the membranes and ridges of the brown scales leading to high absorption. Optical spectroscopy of single scales. In order to understand the observed scale stacking effects in more details, we performed also micro-spectrometry on individual scales and stacks of scales. All the reflection Scientific Reports \| 6:36204 \| DOI: 10.1038/srep36204 3 www.nature.com/scientificreports/ Figure 3. Reflectance spectra of Hypolimnas salmacis recorded on different wing areas. Total reflectance spectra of blue, white and brown regions measured as indicated in the inset. Blue areas show a considerably low intense reflection with a broad peak at ≈430 nm. No distinct peak is observed in the visible regime (380– 760 nm) of the reflection spectra of the white area. In the UV, however, there is a broad peak at ≈375 nm. Scientific Reports \| 6:36204 \| DOI: 10.1038/srep36204 Results Brown scales barely reflect in the UV and the visible regime but the reflection increases towards the infrared. Figure 3. Reflectance spectra of Hypolimnas salmacis recorded on different wing areas. Total reflectance spectra of blue, white and brown regions measured as indicated in the inset. Blue areas show a considerably low intense reflection with a broad peak at ≈430 nm. No distinct peak is observed in the visible regime (380– 760 nm) of the reflection spectra of the white area. In the UV, however, there is a broad peak at ≈375 nm. Brown scales barely reflect in the UV and the visible regime but the reflection increases towards the infrared. measurements are performed on black background to avoid stray reflection/scattering. The resulting spectra are plotted in Fig. 4. The white scale is almost transparent in the visible regime with a high transmittance of ≈90% (dashed line, Fig. 4a. Thus, the white scale can be considered as nearly pigment-free. However, an isolated white scale has a reflection peak at 420 nm, i.e., in the blue spectral area (solid line, Fig. 4a). We carried out similar experiments on a single isolated brown scale (Fig. 4b). The resultant spectra indicate very low reflecting and transmitting properties of the brown scale and therefore point out its strong absorbing behaviour.il To explain how the appearance of white and blue is created by scale stacking, we first measured the reflection spectrum of white scales lying on the wing membrane of Hypolimnas salmacis (Fig. 4c). In this arrangement, the white appearance indeed coincides with a broad spectrum (grey line) over all visible wavelengths. The spectrum of the wing membrane itself shows also a broadband reflection and whitish colouration (not shown). Defining the individual reflectance of a white scale, a brown scale and a wing membrane as Rwhite, Rbrown and Rmembrane, respec- tively and the transmittance of white scale as Twhite, the reflectance of a white scale stacking on the wing membrane can be calculated from = + × × R R T R T (1) w,m white white membrane white (1) assuming non-coherent scattering during the light propagation in the stack. The resulting spectrum (black dashed line) is shown in Fig. 4c with a concise schematic noting the terms as inset. The calculated spectrum fits well with the measured stacked spectrum. Results However, this experiment does not provide the scattering properties, i.e. whether white scales are transparent or translucent. In order to examine this property, we passed laser light with two different wavelengths (blue with 445 nm and red with 635 nm) through the white region of a butterfly wing and captured the resultant diffraction patterns on a screen (Fig. 5). A transparent medium does not allow diffuse light scatter- ing and diffracts only the zeroth order. In our case, however, two diffracted orders including the zeroth order are observed on the screen indicating large diffuse light scattering at the micro-ridge patterns (2 ± 0.1 μm) of the white scales24. The diffraction pattern of a single white scale shown in Fig. 5c was obtained in transmission mode using a K-space imaging system with Bertrand lens25. This conoscopic imaging allows to record the directionality of the scattered beam, i.e. to record the Fourier plane. The zeroth order corresponds to a direct transmission of the white light source through the scale. For higher orders, colours with increasing wavelengths (from blue to red) are diffracted at increasing angles. However, the higher order diffraction signals are relatively broader in angle due to Scientific Reports \| 6:36204 \| DOI: 10.1038/srep36204 4 www.nature.com/scientificreports/ Figure 4. Single scale spectroscopy of Hypolimnas salmacis. (a) Spectra of a single white scale in reflection (solid line) and transmission (dashed line). (b) Spectra of a single brown scale in reflection (solid line) and transmission (dashed line). (c) Reflection spectrum measured on stack of a white scale on a wing membrane (solid line). Individually measured reflectance and transmittance properties are used to calculate the white scale-wing membrane stack reflection (dashed line) and compared with the experimental stack measurement (solid line). A schematics of a single white scale spectrometry on a wing membrane with individual reflectance and transmittance terms is provided in the inset. (d) Reflection spectrum measured on stack of a white and brown scale (solid line) which demonstrates the blue colouration due to the peak at the 420 nm. Individually measured reflectance and transmittance properties are also used to calculate (dashed line) and confirm the experimental stack measurement (solid line). The insets show a schematic of the measurement. Figure 4. Single scale spectroscopy of Hypolimnas salmacis. (a) Spectra of a single white scale in reflection (solid line) and transmission (dashed line). www.nature.com/scientificreports/ (a) Thin film reflection is calculated at normal incidence considering a thin chitin film surrounded by air. In order to account for the local variations observed in the lower lamina of the white scales, the thickness of the film is modeled with a Gaussian distribution as shown in the inset with a mean thickness (d) of 190 nm and a variance (σd) of 31 nm. The simulated mean reflectance (dash-dotted line) of the distribution shows good agreement with the experimental reflection spectrum (solid line). (b) The experimental specular reflection spectra (solid line) at oblique incident angles are compared with the developed model in unpolarised light condition (average of s- and p-polarisation) for a mean thickness of 190 nm and a variance of 31 nm. Calculated reflection spectra of the bio-inspired thin film (dash-dotted line, σd = 31 nm) encounters a red-shift of 10 nm in the peak wavelength of reflection at oblique incident angles with respect to a simple flat thin film (dashed line, σd = 0 nm). a typical refractive index of chitin, i.e. nc = 1.5626. The reflectance of a thin film surrounded by air at any incident angle of θ for a given polarisation can be calculated from27 λ δ δ = − − − − R d r i r i ( , ) (1 exp( 2 )) 1 exp( 2 ) , (3) 2 2 (3) where δ = (2πncd cos θ)/λ is the phase delay introduced by the film thickness of d, and r is the reflection coef- ficient at the air-chitin boundary governed by Fresnel’s equation for a given polarisation, i.e., r = (cos θ − nc cos(sin−1(sin θ/nc)))/(cos θ + nc cos (sin−1(sin θ/nc))) for s-polarisation or r = (cos(sin−1(sin θ/nc)) − nc cos θ)/ (cos(sin−1(sin θ/nc)) + nc cos θ) for p-polarisation. Additionally, we have to consider that the thickness of the lamina is not perfectly constant. We observed in the electron microscopy images that it varies locally from 170 to 210 nm within the scale. In order to include these local variations of the thin film, we modeled the film thickness by a Gaussian distribution σ π σ =    − −    f d d d ( ) 1 2 exp ( ) 2 , (4) d 2 d 2 (4) with mean thickness d and variance σd. www.nature.com/scientificreports/ www.nature.com/scientificreports/ Figure 6. Simulation of the lower thin lamina of Hypolimnas salmacis white scales. (a) Thin film reflection is calculated at normal incidence considering a thin chitin film surrounded by air. In order to account for the local variations observed in the lower lamina of the white scales, the thickness of the film is modeled with a Gaussian distribution as shown in the inset with a mean thickness (d) of 190 nm and a variance (σd) of 31 nm. The simulated mean reflectance (dash-dotted line) of the distribution shows good agreement with the experimental reflection spectrum (solid line). (b) The experimental specular reflection spectra (solid line) at oblique incident angles are compared with the developed model in unpolarised light condition (average of s- and p-polarisation) for a mean thickness of 190 nm and a variance of 31 nm. Calculated reflection spectra of the bio-inspired thin film (dash-dotted line, σd = 31 nm) encounters a red-shift of 10 nm in the peak wavelength of reflection at oblique incident angles with respect to a simple flat thin film (dashed line, σd = 0 nm). igure 6. Simulation of the lower thin lamina of Hypolimnas salmacis white scales. (a) Thin film reflection is Figure 6. Simulation of the lower thin lamina of Hypolimnas salmacis white scales. (a) Thin film reflection is calculated at normal incidence considering a thin chitin film surrounded by air. In order to account for the local variations observed in the lower lamina of the white scales, the thickness of the film is modeled with a Gaussian distribution as shown in the inset with a mean thickness (d) of 190 nm and a variance (σd) of 31 nm. The simulated mean reflectance (dash-dotted line) of the distribution shows good agreement with the experimental reflection spectrum (solid line). (b) The experimental specular reflection spectra (solid line) at oblique incident angles are compared with the developed model in unpolarised light condition (average of s- and p-polarisation) for a mean thickness of 190 nm and a variance of 31 nm. Calculated reflection spectra of the bio-inspired thin film (dash-dotted line, σd = 31 nm) encounters a red-shift of 10 nm in the peak wavelength of reflection at oblique incident angles with respect to a simple flat thin film (dashed line, σd = 0 nm). Figure 6. Simulation of the lower thin lamina of Hypolimnas salmacis white scales. www.nature.com/scientificreports/ In this way, the local thickness variation of the lamina can be easily com- bined with Eq. (3) and the averaged reflectance of the wavy membrane for a given polarisation can be calculated from with mean thickness d and variance σd. In this way, the local thickness variation of the lamina can be easily com- bined with Eq. (3) and the averaged reflectance of the wavy membrane for a given polarisation can be calculated from ∫ λ λ = . ∞ R R d f d dd ( ) ( , ) ( ) (5) 0 (5) With this equation, first we calculated the thin film reflection for a mean thickness of 190 nm and varied the variance σd at normal angle of incidence (Fig. 6a). For σd = 0 nm, the calculated spectrum is simply the thin film reflectance for a 190 nm slab surrounded by air (dashed line). The effect of local variation on the reflection properties by the roughness factor σd is indicated by dotted arrows. The corresponding reflection spectrum is shown in Fig. 6a and compared with the single white scale reflectance obtained from the micro-spectrometry. The simulated thin film reflectance with a variance σd = 31 nm agrees well with the experimental data. Furthermore, we calculated the reflection spectra in unpolarised light condition by averaging out the s- and p-polarisation for various oblique angles of incidence and compared with the experimental ones in Fig. 6b. The roughness factor of σd = 31 nm in the model also corresponds well with the experimental data at off-normal light condition. A blue-shift is noticed in the peak wavelength of reflection spectra towards UV spectral region at large angles of incidence. However, the inclusion of roughness factor of σd = 31 nm in the thin film model red-shifts the peak wavelength of reflection of a flat thin film (σd = 0 nm) by 10 nm (Fig. 6b). Overall, the theoretical thin film model verifies that the blue colouration results from the thin film nature of the lower lamina. Results (b) Spectra of a single brown scale in reflection (solid line) and transmission (dashed line). (c) Reflection spectrum measured on stack of a white scale on a wing membrane (solid line). Individually measured reflectance and transmittance properties are used to calculate the white scale-wing membrane stack reflection (dashed line) and compared with the experimental stack measurement (solid line). A schematics of a single white scale spectrometry on a wing membrane with individual reflectance and transmittance terms is provided in the inset. (d) Reflection spectrum measured on stack of a white and brown scale (solid line) which demonstrates the blue colouration due to the peak at the 420 nm. Individually measured reflectance and transmittance properties are also used to calculate (dashed line) and confirm the experimental stack measurement (solid line). The insets show a schematic of the measurement. Figure 5. Light diffraction of the white scales of Hypolimnas salmacis. (a) A blue (445 nm) and (b) a red laser (635 nm) shine through the white region of a Hypolimnas salmacis wing. The resultant higher diffraction orders on the screen demonstrate the transmission grating like behaviour of the white scales. (c) K-space imaging of a single white scale in transmission mode showing high order diffraction. Figure 5. Light diffraction of the white scales of Hypolimnas salmacis. (a) A blue (445 nm) and (b) a red laser (635 nm) shine through the white region of a Hypolimnas salmacis wing. The resultant higher diffraction orders on the screen demonstrate the transmission grating like behaviour of the white scales. (c) K-space imaging of a single white scale in transmission mode showing high order diffraction. the inherent disorder of the natural grating. This outcome clarifies that white scales act as a translucent medium in the visible regime. Due to the grating like ridge architecture, it scatters all wavelengths non-coherently if it lies on top of a white or random surface which diffuses the light. Simulation of the thin film interference of the white scales. The experimental results presented above already demonstrated that the effect of the blue colouration is caused by scale stacking which amplifies the broad blue peak of the white scales. As the upper lamina works as a diffuser, we speculate that the lower feature-less thin lamina is the origin of this blue peak. To prove that, we simulated the reflection of a thin film with Scientific Reports \| 6:36204 \| DOI: 10.1038/srep36204 5 Discussion The theoretical modelling of a thin film including surface variation also con- firms this hypothesis (Fig. 6b).h i The local variation of the laminar thickness can be modelled by a Gaussian distribution and the variance (standard deviation) can act as a roughness factor. Such surface roughness smoothens the reflection spectrum of a single thin flat film (Fig. 6a). This explains the dull appearance of the blue colour. Although an isolated white scale has a peak blue reflectance of around 10%, blue areas on the wing reach up to 20% due to multiple scattering of overlapping scales23. Moreover, the peak wavelength of reflection encounters a red-shift of 10 nm because of the surface roughness (Fig. 6b). This can be explained by the incoherent effect caused by the local surface varia- tion that disrupts partially the coherent scattering of a thin 1D photonic crystal i.e. the lamina. Hence, the local thickness variation of the lower lamella weakens the iridescent property of the scales towards UV spectral region.l p p y p g Micro-ridges in butterfly scales usually act as good scatters for longer wavelengths36. Again, windows created by cross-ribs are often reported to be responsible for the diffusion of the incoming or outgoing light6. This can explain the translucency of the white scales when white light passes through the upper lamina. Due to the high translucency of the white scales, the weak blue reflection from the lower thin lamina is eliminated by the under- lying incoherent diffuse scattering of the wing membrane. The broadband absorbing pigment melanin in black scales effectively reduces stray-light and back-scattering, at the same time uncovers the blue appearance of the white scales. The overall mechanism of structural colouration by scale stacking in Hypolimnas salmacis wing is sketched in Fig. 7. The pigment melanin is also found in other blue, e.g. Morpho butterflies and mainly prevents the decrease in saturation of the colour37. The thin lower lamina in Morpho butterflies works as a thin reflector too, but the intense blue reflection of the scales is dominated by the upper lamina multi-layered lamellar archi- tecture38. In contrast, the upper lamina of Hypolimnas salmacis white scales works as a broadband light scatterer.hl pp yp g The typical colours of butterfly wing scales may be solely produced from either micro- and nanoarchitectures, or pigments, but most frequently from a combination of both37,39. Discussion Very little research on the butterfly Hypolimnas salmacis has been published so far although some other Hypolimnas species were studied extensively19,22,28. We observed structural UV reflection from the white patches owing to the multilayer microrib architecture. UV reflection from ridge-microrib structures has been already observed in nymphalids, pierids and other butterfly families29,30. However, to create blue reflection from a similar 6 Scientific Reports \| 6:36204 \| DOI: 10.1038/srep36204 www.nature.com/scientificreports/ Figure 7. The mechanism of structural colouration by scale stacking in Hypolimnas salmacis butterfly wing. Schematics of scale stacking colouration mechanism of Hypolimnas salmacis butterfly wing demonstrating the role of a diffusive white wing membrane and an absorbing brown scale on colour appearan of a translucent white scale. Figure 7. The mechanism of structural colouration by scale stacking in Hypolimnas salmacis butterfly wing. Schematics of scale stacking colouration mechanism of Hypolimnas salmacis butterfly wing demonstrating the role of a diffusive white wing membrane and an absorbing brown scale on colour appearan of a translucent white scale. structure, longitudinal and transverse dimensions of the microribs need to be sufficiently large. This statement is supported by the Morpho butterfly structural pattern as a very good natural example of such architecture, as well as by several experimental biomimetic structures12,31,32. It was also reported that the origin of the intense and directional blue reflection of Hypolimnas bolina is similar kind of complex structures22. However, we demonstrated that the colouration of the blue areas of the Hypolimnas salmacis wing is caused by scale stacking where the origin of the blue colour is the thin lower lamina of the white scales. Our experimental analysis as well as theoretical analytical modelling confirm that the phase delay introduced by the thin film lower lamina causes light interference in blue spectrum. Indeed, the lower lamina works as a one dimensional photonic crystal which is commonly found in butterfly wing scales33,34. However, due to the low index contrast between chitin (n = 1.56) and air (n = 1), the thin lamina can not form a complete photonic bandgap. It rather creates a pseudo-bandgap which causes the iridescence of the Hypolimnas salmacis35. The pseudo-photonic bandgap in the visible spectrum created by photonic crystals is responsible for iridescence in many other butterflies and insects5,34. Hence, the proposition of thin film interference explains the violet-purple appearance of the wing at oblique angles as shown in Fig. 1c. Scientific Reports \| 6:36204 \| DOI: 10.1038/srep36204 Methods Sample preparation and imaging. Dried samples of Hypolimnas salmacis were kindly supplied by the Stadtpark Mannheim GmbH, Germany and a dried sample was bought from Bug Under Glass©, USA. Scales were carefully removed from the wings with tweezers for subsequent imaging by optical microscopy. A stereo-mi- croscope (SteREO Discovery.V8, Carl Zeiss Microscopy GmbH, Germany) was used in reflection mode to image the scales under epi-illumination condition (Fig. 1)l p g For the imaging by scanning electron microscopy (SEM), the examined regions of the butterfly wings were coated with a 15 nm thin gold layer (K575X sputter coater, Quorum Technologies Ltd.). Surface patterns were subsequently imaged by SEM (SUPRA® 60 VP, Carl Zeiss Microscopy GmbH, Germany) operated at 5 kV (see Fig. 2a–d). For cross-section imaging, small pieces of wings were embedded in epoxide resin. First, the pieces were dipped in a combination of 70% acetone and 30% epoxide mix (42.4 g Glycidether, 29.6 g DDSA, 18.4 g MNA, all chemicals from SERVA Electrophoresis GmbH, Germany). Subsequently, they were exposed to vacuum for a few minutes to remove air bubbles. Afterwards, the mixture was shaken for an hour. This step was repeated first with a combination of 30% acetone and 70% epoxide mix, then twice with 100% epoxide mix to make sure the viscous resin penetrated into all the tiny cavities of the butterfly scales. The resin-infiltrated wing pieces were placed into a silicone mold and covered with epoxide mix plus accelerator (10 g epoxide mix and 0.265 g BDMA) and baked at 65 °C for 2 days. The polymerized block was removed from the mold and trimmed to expose the wing piece. Thin (≈70 nm) sections from the block-face were prepared using an ultramicrotome (Leica Microsystems, Germany) and transferred to small pieces of silicon wafer for SEM imaging. The cross section shown in Fig. 2e) was imaged at 1.5 kV in a SEM (Ultra, Carl Zeiss Microscopy GmbH, Germany). Optical spectroscopy. The macroscopic reflection spectra of an intact Hypolimnas salmacis wing shown in Fig. 3 were recorded with a UV-Vis spectrometer (Lambda 1050, PerkinElmer Inc., USA). The total diffuse reflec- tion was measured with an InGaAs 150 mm integrating sphere averaging over a 2 mm2 area on different regions (blue, white, brown) of an intact wing. A customized Zeiss Axio microscope was used for the micro-spectroscopic analysis of individual scales (Fig. Methods 4) with a spot size of ≈25 μm in bright field (BF) reflection mode with a halogen lamp in Koehler illumina- tion. Unpolarised light from the halogen lamp was illuminated via a 10X objective (EC Epiplan-APOCHROMAT, Zeiss) with a numerical aperture of NA = 0.3. The reflected light was collected with a spectrometer (AvaSpec-HS2048, Avantes, UK) through a 200 μm core optical fiber (Avantes, UK) mounted in confocal configuration.hl i g The angular-resolved specular reflection was measured using home-built optical goniometric setup (Fig. 6b). A light source from a stabilised Tungsten light source (SLS201, Thorlabs, USA) is collimated with a pinhole and a long working distance objective lens to form a 50 μm wide parallel incident beam that illuminates a single scale at a fixed angle. The specularly reflected light is detected at different angles with an aperture of 2° and coupled into an optical fiber connected to the spectrometer (AvaSpec-ULS2048x64-USB2 Avantes, USA). All the spectra reported are referenced to a lambda/20 UV fused silica mirror (Thorlabs, USA). Discussion This is in fact also true for other animals in nature40–42. Often the cover scales of butterfly wings consist of complex architectures with ridges, lamella, micror- ibs, crossribs with or without pigments and contribute strongly to the wing colouration. Otherwise, the cover scales are reported to be transparent and scatter the underlying ground scale reflection26. Only a few examples of scale stacking were reported so far and mainly described as an enhancer of scattering properties in terms of perception and brightness23,43. Spectral alteration by scale stacking was only recently observed in small regions of European Nymphaline butterflies44. A similar colouration mechanism is also found in the feathers of the bird Steller’s Jay where the colour difference of white and blue feathers arises primarily due to the inherent melanin pigment content difference below spongy nanostructures45,46. p gf p gy In summary, we experimentally demonstrated that the blue wing regions of the Hypolimnas salmacis originate from scale stacking of white translucent scales on top of brown absorbing scales. Our single-scale micro-spectrometry and thin film simulation showed that the blue colouration is caused by the lower thin lamina of the white scales. Our detailed study also revealed the cause of the weak iridescence of Hypolimnas salmacis and might lead to better understanding of mating preferences and signal variation in other Hypolimnas butterflies3,19. Scientific Reports \| 6:36204 \| DOI: 10.1038/srep36204 7 www.nature.com/scientificreports/ Moreover, such a mechanism of colour interplay with different thin plates might be useful for technical appli- cations in the field of nano-optics and photonics. For instance, by tuning the specular and/or diffuse reflection factor of a top optical layer, and the absorption properties of a bottom film, different optical signatures might be encoded. Referencesi J f p gy , ( ) 3. Kemp, D. J. & Rutowski, R. L. The Role of Coloration in Mate Choice and Sexual Interactions in Butterflies vol. 43 (Elsevier Inc., 2011), 1 edn. J f p gy , ( ) 3. Kemp, D. J. & Rutowski, R. L. The Role of Coloration in Mate Choice and Sexual Interactions in Butterflies vol. 43 (Elsevier Inc., 2011) 1 edn. 4. Ghiradella, H. Structure of iridescent lepidopteran scales: variations on several themes. Annals of the Entomological Socie America 77, 637–645 (1984). 5. Vukusic, P. & Sambles, J. R. Photonic structures in biology. Nature 424, 852–855 (2003). gy 6. Ingram, A. L. & Parker, A. R. A review of the diversity and evolution of photonic structures in butterflies, incorporating the work of John Huxley (The Natural History Museum, London from 1961 to 1990). Philos. Trans. R. Soc. Lond. B. Biol. Sci. 363, 2465–2480 (2008).l gy 6. Ingram, A. L. & Parker, A. R. A review of the diversity and evolution of photonic structures in butterflies, incorporating the work of John Huxley (The Natural History Museum, London from 1961 to 1990). Philos. Trans. R. Soc. Lond. B. Biol. Sci. 363, 2465–2480 (2008).l 7. Kronforst, M. R. et al. Linkage of butterfly mate preference and wing color preference cue at the genomic location of wing Proceedings of the National Academy of Sciences 103, 6575–6580 (2006).l g f y f 8. Wilts, B. D., Pirih, P., Arikawa, K. & Stavenga, D. G. Shiny wing scales cause spec(tac)ular camouflage of the angled sunbeam butterfly, curetis acuta. Biological Journal of the Linnean Society 109, 279–289 (2013).hl g f y f 8. Wilts, B. D., Pirih, P., Arikawa, K. & Stavenga, D. G. Shiny wing scales cause spec(tac)ular camouflage of the angled sunbeam butterfly, curetis acuta. Biological Journal of the Linnean Society 109, 279–289 (2013).hl l y g f y 9. Siddique, R. H., Gomard, G. & Hölscher, H. The role of random nanostructures for the omnidirectional anti-reflection properties of the glasswing butterfly. Nature Communications 6, 6909 (2015).l l y g f y 9. Siddique, R. H., Gomard, G. & Hölscher, H. The role of random nanostructures for the omnidirectional anti-reflection propert the glasswing butterfly. Nature Communications 6, 6909 (2015).l g gl y ( ) 0. Pinheiro, C. E. G. Palatablility and escaping ability in neotropical butterflies: tests with wild kingbirds (tyrannus melancholicus tyrannidae). Scientific Reports \| 6:36204 \| DOI: 10.1038/srep36204 1. Brakefield, P. M. & French, V. Butterfly wings: the evolution of development of colour patterns. BioEssays 21, 391–401 (1999).l ki , ,l y g p p y , ( ) 2. Prum, R. O., Quinn, T. & Torres, R. H. Anatomically diverse butterfly scales all produce structural colours by coherent scattering Journal of Experimental Biology 209, 748–765 (2006).hl Referencesi D., Matsushita, A., Arikawa, K. & Stavenga, D. G. Spectrally tuned structural and pigmentary coloration of bird butterfly wing scales. Journal of The Royal Society Interface 12, 20150717 (2015).i l y g fh y y f 0. Stavenga, D. G., Tinbergen, J., Leertouwer, H. L. & Wilts, B. D. Kingfisher feathers-colouration by pigments, spongy nanostructure and thin films. Journal of Experimental Biology 214, 3960–3967 (2011).ll nd thin films. Journal of Experimental Biology 214, 3960–3967 (20 i f p gy 41. Saba, M., Wilts, B. D., Hielscher, J. & Schröder-Turk, G. E. Absence of circular polarisation in reflections of butterfly wing scales chiral gyroid structure. Materials Today: Proceedings 1, 193–208 (2014). 42. Hsiung, B.-K., Blackledge, T. A. & Shawkey, M. D. Structural color and its interaction with other color-producing elem perspectives from spiders. In SPIE Optical Engineering+ Applications 91870B–91870B (International Society for Optics Photonics, 2014).f ) 43. Yoshioka, S. & Kinoshita, S. Wavelength–selective and anisotropic light–diffusing scale on the wing of the morpho butterfly. Proceedings of the Royal Society of London B: Biological Sciences 271, 581–587 (2004).li 43. Yoshioka, S. & Kinoshita, S. Wavelength–selective and anisotropic light–diffusing scale on the wing of the morpho butt Proceedings of the Royal Society of London B: Biological Sciences 271, 581–587 (2004).li 4. Stavenga, D. G., Leertouwer, H. L. & Wilts, B. D. Coloration principles of nymphaline butterflies - thin films, melanin, ommochromes and wing scale stacking. J. Exp. Biol. 217, 2171–2180 (2014).i and wing scale stacking. J. Exp. Biol. 217, 2171–2180 (2014). g g p 5. Shawkey, M. D. & Hill, G. E. Significance of a basal melanin layer to production of non-iridescent structural plumage color: evidence from an amelanotic steller’s jay (cyanocitta stelleri). Journal of Experimental Biology 209, 1245–1250 (2006). from an amelanotic stellers jay (cyanocitta stelleri). Journal of Experimental Biology 209, 1245 1250 (2006). 46. Igic, B., D’Alba, L. & Shawkey, M. D. Manakins can produce iridescent and bright feather colours without melanosomes. Journal of Experimental Biology 219 1851 1859 (2016) 46. Igic, B., D’Alba, L. & Shawkey, M. D. Manakins can produce iridescent and bright feather colours without melanosomes. Journ Experimental Biology 219, 1851–1859 (2016). Acknowledgements g We thank Andrea Gerstner (Stadtpark Mannheim GmbH) for supplying the Hypolimnas salmacis butterfly samples, and Paul Abaffy (Karlsruhe Institute of Technology) for performing the SEM. Furthermore, we acknowledge fruitful discussions with all members of the Biomimetics group at KIT and Bio-inspired photonics group at University of Cambridge. R.H.S. acknowledges the funding by the Karlsruhe House of Young Scientists for a research stay at Cambridge. This work was partly carried out with the support of the Karlsruhe School of Optics and Photonics (KSOP, www.ksop.idschools.kit.edu) and the Karlsruhe Nano Micro Facility (KNMF, www. kit.edu/knmf), a Helmholtz Research Infrastructure at Karlsruhe Institute of Technology (KIT, www.kit.edu). Author Contributions R.H.S. and H.H. conceived the idea. C.B. and I.W. worked on the cross section preparation and imaging. R.H.S., C.B. and I.W. conducted the microscopy of the butterfly scales. R.H.S. and S.V. performed and analysed the optical spectroscopy of the butterfly scales. R.H.S. and H.H. conducted and discussed the optical simulation. All authors commented on the manuscript. Referencesi Further studies on the courtship behaviour of african mimetic butterflies. Animal Behaviour 6, 224–230 (1958).ll l 29. Ghiradella, H., Aneshansley, D., Eisner, T., Silberglied, R. E. & Hinton, H. E. Ultraviolet reflection of a male butterfly: interference color caused by thin-layer elaboration of wing scales. Science 178, 1214–1217 (1972).fl y y g 0. Silberglied, R. E. & Taylor, O. R. Ultraviolet Differences between the Sulphur Butterflies, Colias eurytheme and C. philodice, and a Possible Isolating Mechanism. Nature 241, 406–408 (1973).h f Possible Isolating Mechanism. Nature 241, 406–408 (1973).h g 31. Siddique, R. H., Diewald, S., Leuthold, J. & Hölscher, H. Theo 31. Siddique, R. H., Diewald, S., Leuthold, J. & Hölscher, H. Theoretical and exp for the iridescence of Morpho butterflies. Opt. Express 21, 14351 (2013). 31. Siddique, R. H., Diewald, S., Leuthold, J. & Hölscher, H. Theoretical and experimental analysis of the structural pattern responsible for the iridescence of Morpho butterflies. Opt. Express 21, 14351 (2013). pl p p 2. Siddique, R. H., Ruben, H., Faisal, A., Lemmer, U. & Hölscher, H. Fabrication of hierarchical photonic nanostructures inspired by Morpho butterflies utilizing laser interference lithography Optical Material Express 5 2608 2612 (2015) l 32. Siddique, R. H., Ruben, H., Faisal, A., Lemmer, U. & Hölscher, H. Fabrication of hierarchical photonic nanostructures inspired by Morpho butterflies utilizing laser interference lithography. Optical Material Express 5, 2608–2612 (2015).lh l 33. Michielsen, K. & Stavenga, D. G. Gyroid cuticular structures in butterfly wing scales: biological photonic crystals. Journal Royal Society Interface 5, 85–94 (2008).l y y f 34. Zaccaria, R. P. Butterfly wing color: A photonic crystal demonstration. Optics and Lasers in Engineering 76, 70–73 (2016).l l 35. Joannopoulos, J. D., Johnson, S. G., Winn, J. N. & Meade, R. D. Photonic crystals: molding the flow of light (Princeton university press l 36. Zhao, Q. et al. Art of blackness in butterfly wings as natural solar collector. Soft Matter 7, 11433 (2011).l lt 37. Kinoshita, S., Yoshioka, S. & Kawagoe, K. Mechanisms of structural colour in the Morpho butterfly: cooperation of regularity irregularity in an iridescent scale. Proc. Biol. Sci. 269, 1417–1421 (2002).li g y 8. Giraldo, M. & Stavenga, D. Brilliant iridescence of morpho butterfly wing scales is due to both a thin film lower lamina and a multilayered upper lamina. Journal of Comparative Physiology A 202, 381–388 (2016). y pp f p y gy 39. Wilts, B. Referencesi g y g j 18. Kemp, D. J. Female butterflies prefer males bearing bright iridescent ornamentation. Proc. Biol. Sci. 274, l 19. Kemp, D. J., Jones, D., Macedonia, J. M. & Krockenberger, A. K. Female mating preferences and male signal variation in iride Hypolimnas butterflies. Anim. Behav. 87, 221–229 (2013). ypl ( ) 0. Huheey, J. E. Warning coloration and mimicry. In Bell, W. J. & Cardé, R. T. (eds.) Chemical Ecology of Insects 257–297 (Springer US, 1984) h l fi ld l f h l b fl h l b l (l ) ( h l d ) ypl 20. Huheey, J. E. Warning coloration and mimicry. In Bell, W. J. & Cardé, R. T. (eds.) Chemical Ecology o y J g y J ( ) gy f ( p g ) 21. Kemp, D. J. & Jones, R. E. Phenotypic plasticity in field populations of the tropical butterfly hypolimnas bolina (l.) (nymphalidae). Biological Journal of the Linnean Society 72, 33–45 (2001).l g f y 2. Kemp, D. J. & Macedonia, J. M. Structural ultraviolet ornamentation in the butterfly Hypolimnas bolina L.(Nymphalidae): visual morphological and ecological properties. Aust. J. Zool. 54, 235 (2006).l 22. Kemp, D. J. & Macedonia, J. M. Structural ultraviolet orn p g g p p 3. Stavenga, D. G., Giraldo, M. A. & Hoenders, B. J. Reflectance and transmittance of light scattering scales stacked on the wings o pierid butterflies. Opt. Express 14, 4880–4890 (2006).f l p p 24. Ko, Y. H. & Yu, J. S. Highly transparent sapphire micro-grating structures with large diffuse light scattering. Opt. Express 19, 15574–15583 (2011).h ( ) 25. Vignolini, S., Moyroud, E., Glover, B. J. & Steiner, U. Analysing photonic structures in plants. Journal of The Royal Society Interface 10, 20130394 (2013).ifl ( ) 6. Vukusic, P., Sambles, J. R., Lawrence, C. R. & Wootton, R. J. Quantified interference and diffraction in single Morpho butterfly scales Proc. R. Soc. B Biol. Sci. 266, 1403–1411 (1999).i , ( ) 27. Heavens, O. S. Optical properties of thin solid films (Courier Corporation, 1991).l ( ) O. S. Optical properties of thin solid films (Courier Corporation, 199 27. Heavens, O. S. Optical properties of thin solid films (Courier Cor p p p f fi p 28. Stride, G. O. Further studies on the courtship behaviour of african mimetic butterfli p p p f fi p 28. Stride, G. O. Referencesi Biological Journal of the Linnean Society 59, 351–365 (1996).tt y g f y 11. Mallet, J. Shift happens! shifting balance and the evolution of diversity in warning colour and mimicry. Ecological Entomology 35, 90–104 (2010). y g f y 11. Mallet, J. Shift happens! shifting balance and the evolution of diversity in warning colour and mimicry. Ecological Entomology 35, 90–104 (2010). 12. Potyrailo, R. A. et al. Towards outperforming conventional sensor arrays with fabricated individual photonic vapour sensors inspired by Morpho butterflies. Nature Communications 6, 7959 (2015). 12. Potyrailo, R. A. et al. Towards outperforming conventional sensor arrays with fabricated individual photonic vapour sensors inspired by Morpho butterflies. Nature Communications 6, 7959 (2015). l 13. Siddique, R. H. et al. Utilizing laser interference lithography to fabricate hierarchical optical active nanostructures inspired by the blue Morpho butterfly. In Liang, R. & Shaw, J. A. (eds) The Nature of Light: Light in Nature V, edited by Rongguang Liang, Joseph A. Shaw, Proc. of SPIE vol. 9187, 91870E (2014).l l 13. Siddique, R. H. et al. Utilizing laser interference lithography to fabricate hierarchical optical active nanostructures inspired by the blue Morpho butterfly. In Liang, R. & Shaw, J. A. (eds) The Nature of Light: Light in Nature V, edited by Rongguang Liang, Joseph A. Shaw, Proc. of SPIE vol. 9187, 91870E (2014).l , f , ( ) 14. Kolle, M. et al. Mimicking the colourful wing scale structure of the Papilio blumei butterfly. Nature nanotechnology 5, 511 (2010). Scientific Reports \| 6:36204 \| DOI: 10.1038/srep36204 8 www.nature.com/scientificreports/ 5. England, G. et al. Bioinspired micrograting arrays mimicking the reverse color diffraction elements evolved by the butterfly Pierella luna. Proceedings of the National Academy of Sciences 111, 15630–15634 (2014).l g f y f 16. Larsen, T. B. Butterflies of West Africa (Apollo Books, 2005). g f y f 16. Larsen, T. B. Butterflies of West Africa (Apollo Books, 2005). fl f f p 17. Seitz, A. Die Gross-Schmetterlinge der Erde: eine systematische Be eitz, A. Die Gross-Schmetterlinge der Erde: eine systematische Bearbeitung der bis jetzt bekannten Grossschmetterlinge (Kernen, 1909). emp, D. J. Female butterflies prefer males bearing bright iridescent ornamentation. Proc. Biol. Sci. 274, 1043–1047 (2007). 7. Se t , . ie Gross Schmetterlinge der rde: eine systematische earbeitung der bis jetzt bekannten Grossschmetterlinge (Ke e , 18. Kemp, D. J. Female butterflies prefer males bearing bright iridescent ornamentation. Proc. Biol. Sci. 274, 1043–1047 (2007). Scientific Reports \| 6:36204 \| DOI: 10.1038/srep36204 Additional Informationi Competing financial interests: The authors declare no competing financial interests. How to cite this article: Siddique, R. H. et al. Colour formation on the wings of the butterfly Hypolimnas salmacis by scale stacking. Sci. Rep. 6, 36204; doi: 10.1038/srep36204 (2016). Scientific Reports \| 6:36204 \| DOI: 10.1038/srep36204 9 www.nature.com/scientificreports/ Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/ © The Author(s) 2016 Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/ © The Author(s) 2016 Scientific Reports \| 6:36204 \| DOI: 10.1038/srep36204 10
https://openalex.org/W4360892455	https://www.dora.lib4ri.ch/psi/islandora/object/psi%3A51895/datastream/PDF/Ajeesh-2023-Interplay_of_structure_and_magnetism-%28published_version%29.pdf	English	null	Interplay of structure and magnetism in <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"><mml:mrow><mml:msub><mml:mi>LuFe</mml:mi><mml:mn>4</mml:mn></mml:msub><mml:msub><mml:mi>Ge</mml:mi><mml:mn>2</mml:mn></mml:msub></mml:mrow></mml:math> tuned by hydrostatic pressure	Physical review. B./Physical review. B	2,023	cc-by	9,811	Interplay of structure and magnetism in LuFe4Ge2 tuned by hydrostatic pressure M. O. Ajeesh ,1,* P. Materne ,2 R. D. dos Reis ,3 K. Weber,1 S. Dengre,4 R. Sarkar,4 R. Khasanov,5 I. Kraft,1 A W. Bi,2,6 J. Zhao,2 E. E. Alp,2 S. Medvedev,1 V. Ksenofontov,7 H. Rosner,1 H.-H. Klauss,4 C. Geibel,1 and M. Nicklas 1,† 1Max Planck Institute for Chemical Physics of Solids, Nöthnitzer Straße 40, 01187 Dresden, Germany 2Advanced Photon Source, Argonne National Laboratory, Lemont, Illinois 60439, USA 3Brazilian Synchrotron Light Laboratory (LNLS), Brazilian Center for Research in Energy and Materials (CNPEM), Campinas, 13083-970 Sao Paulo, Brazil 4Institut für Festkörper-und Materialphysik, Technische Universität Dresden, 01069 Dresden, Germany 5Laboratory for Muon Spin Spectroscopy, Paul Scherrer Institute, 5232 Villigen, Switzerland 6Department of Physics, University of Alabama at Birmingham, Birmingham, Alabama 35294, USA 7Institut für Anorganische und Analytische Chemie, Johannes Gutenberg-Universität, 55099 Mainz, Germany M. O. Ajeesh ,1,* P. Materne ,2 R. D. dos Reis ,3 K. Weber,1 S. Dengre,4 R. Sarkar,4 R. Khasanov,5 I. Kraft,1 W. Bi,2,6 J. Zhao,2 E. E. Alp,2 S. Medvedev,1 V. Ksenofontov,7 H. Rosner,1 H.-H. Klauss,4 C. Geibel,1 and M. Nicklas 1,† 1Max Planck Institute for Chemical Physics of Solids, Nöthnitzer Straße 40, 01187 Dresden, Germany 2Advanced Photon Source, Argonne National Laboratory, Lemont, Illinois 60439, USA 3Brazilian Synchrotron Light Laboratory (LNLS), Brazilian Center for Research in Energy and Materials (CNPEM), Campinas, 13083-970 Sao Paulo, Brazil 4Institut für Festkörper-und Materialphysik, Technische Universität Dresden, 01069 Dresden, Germany 5Laboratory for Muon Spin Spectroscopy, Paul Scherrer Institute, 5232 Villigen, Switzerland 6Department of Physics, University of Alabama at Birmingham, Birmingham, Alabama 35294, USA 7Institut für Anorganische und Analytische Chemie, Johannes Gutenberg-Universität, 55099 Mainz, Germany (Received 12 December 2022; accepted 28 February 2023; published 16 March 2023) LuFe4Ge2 crystallizes in the ZrFe4Si2-type structure, hosting chains of Fe tetrahedra giving rise to geometric frustration and low dimensionality. The compound orders antiferromagnetically at around 36 K accompanied by a simultaneous structural transition from a tetragonal phase to an orthorhombic phase. The hydrostatic pressure dependence of the magnetic and structural transitions is investigated using electrical transport, AC magnetic susceptibility, AC calorimetry, M¨ossbauer, muon-spin relaxation (μSR), and x-ray-diffraction measurements. External pressure suppresses the ﬁrst-order transition to the antiferromagnetic phase (AFM1) around 1.8 GPa. The structural transition is largely unaffected by pressure and remains between 30 to 35 K for pressures up to 2 GPa. A second antiferromagnetic phase (AFM2) is observed at higher pressures. Interplay of structure and magnetism in LuFe4Ge2 tuned by hydrostatic pressure The transition from the paramagnetic to the AFM2 phase is of second-order nature and appears to be connected to the structural transition. The magnetic volume fraction obtained from μSR and M¨ossbauer measurements reveal that the entire sample undergoes magnetic ordering in both magnetic phases. In addition, similar low-temperature muon-precession frequencies in AFM1 and AFM2 phases point at similar ordered moments and magnetic structures in both phases. Our results further indicate enhanced magnetic ﬂuctuations in the pressure-induced AFM2 phase. The experimental observations together with density functional theory calculations suggest that the magnetic- and structural-order parameters in LuFe4Ge2 are linked by magnetic frustration, causing the simultaneous magnetostructural transition. DOI: 10.1103/PhysRevB.107.125136 Present address: Los Alamos National Laboratory, Los Alamos, New Mexico 87545, USA; ajeesh@lanl.gov †Michael.Nicklas@cpfs.mpg.de PHYSICAL REVIEW B 107, 125136 (2023) PHYSICAL REVIEW B 107, 125136 (2023) Editors’ Suggestion Published by the American Physical Society I. INTRODUCTION frustration are of particular interest due to enhanced quantum ﬂuctuations. Therefore, identifying new systems and detailed investigations by tuning their ground-state properties are im- portant for improving our understanding of unconventional phenomena. Compounds with competing ground states have attracted tremendous attention in condensed-matter research. This is because novel properties, such as quantum criticality and un- conventional superconductivity are often observed in regions of competing energy scales in a variety of material classes [1–6]. In such compounds the interplay of magnetic, elec- tronic, and structural degrees of freedom dictates the emerging phenomena. A prime example is the unconventional supercon- ductivity observed in iron pnictides where the role of magnetic and nematic ﬂuctuations are crucially debated [7–11]. In this regard, compounds with low-dimensionality and magnetic Intermetallic AFe4X2 (A = rare-earth, X = Si, Ge) com- pounds are ideal candidates for studying unconventional phases and the effect of magnetic frustration on their proper- ties. These compounds crystallize in the ZrFe4Si2-type struc- ture (space-group P42/mnm) consisting of a slightly distorted tetrahedral arrangement of Fe atoms, a geometry well-known for exhibiting magnetic frustration [12]. Moreover, the Fe tetrahedra are edge shared to form chains along the crystallo- graphic c axis, resulting in a quasi-one-dimensional structure (see Fig. 1). These quasi-one-dimensional chains of geomet- rically frustrated Fe tetrahedra form a very different type of magnetic lattice compared to the 122 Fe pnictides where the magnetic frustration is caused by competing exchange inter- actions. In that way, the intermetallic AFe4X2 materials offer a new perspective on the entanglement of crystal structure and magnetism. Present address: Los Alamos National Laboratory, Los Alamos, New Mexico 87545, USA; ajeesh@lanl.gov †Michael.Nicklas@cpfs.mpg.de Published by the American Physical Society under the terms of the Creative Commons Attribution 4.0 International license. Further distribution of this work must maintain attribution to the author(s) and the published article’s title, journal citation, and DOI. Open access publication funded by the Max Planck Society. Previous investigations on the isostructural compound ZrFe4Si2 using chemical substitution and external pressure 125136-1 Published by the American Physical Society 2469-9950/2023/107(12)/125136(9) PHYSICAL REVIEW B 107, 125136 (2023) M. O. AJEESH et al. 0 20 40 60 80 100 120 140 0 20 40 60 80 Cp(Jmol-1K-1) Cp(Jmol-1K-1) T (K) (c) LuFe4Ge2 34 36 38 40 0 40 80 heating cooling T (K) 0 500 1000 0 0.2 0.4 0.6 T 2 (K2) Cp/T(Jmol-1K-2) 0 4 8 12 dρ/dT (arb. I. INTRODUCTION (b) Tempera- ture dependence of the electrical resistivity ρ(T ) (left axis) and the temperature derivative dρ(T )/dT (right axis). (c) Heat-capacity Cp of LuFe4Ge2 as a function of temperature. The upper inset displays the thermal hysteresis in Cp(T ) obtained from the analysis of heating and cooling parts of the thermal-relaxation cycle. The lower inset shows Cp/T vs T 2 where the straight line is a ﬁt to the data for 2 K ⩽T ⩽20 K using Cp(T ) = γ T + βT 3. In this article, we report a detailed investigation on the pressure evolution of the magnetic and structural transitions in LuFe4Ge2, carried out using electrical transport, mag- netic susceptibility, AC calorimetry, M¨ossbauer spectroscopy, muon-spin relaxation, and x-ray diffraction experiments un- der external pressure. The nature of the phase transitions and the details of the various magnetic phases are discussed. Furthermore, our experimental ﬁndings together with theo- retical calculations based on density functional theory (DFT) elucidate the role of magnetic frustration in the interplay of magnetic and structural degrees of freedom in LuFe4Ge2. ment μeff(= 3KBC/NAμ0μ2 B) of 2.2(2)μB/Fe and a Weiss temperature θW = −95(1) K. These CW parameters indicate relatively large Fe moments in the paramagnetic phase (PM) with dominant antiferromagnetic interaction among them. At low temperatures, a sharp drop in susceptibility is observed at TN = 36 K corresponding to the antiferromagnetic tran- sition. It should be noted that the observed TN differs from the earlier reported value of 32 K [14]. This may be due to a difference in sample quality where a large amount of impurity phases could have affected the precise determination of the ordering temperature using neutron-diffraction studies. We note that the powder x-ray diffraction (PXRD) pattern of the sample used in this paper does not show any additional peaks corresponding to a Fe3Ge impurity phase within the I. INTRODUCTION unit) (b) 0 50 100 150 200 250 300 0 0.2 0.4 0.6 0.8 1 ρ (mΩcm) T (K) LuFe4Ge2 0.05 0.1 0.15 0.2 LuFe4Ge2 χ (cm3/mol) (a) 0 4 8 12 χ-1 (mol/cm3) FIG. 2. (a) Temperature dependence of the DC magnetic suscep- tibility χ of LuFe4Ge2. The inverse susceptibility χ −1(T ) is plotted on the right axis along with a Curie-Weiss ﬁt to the data (red curve) in the temperature interval between 100 and 300 K. (b) Tempera- ture dependence of the electrical resistivity ρ(T ) (left axis) and the temperature derivative dρ(T )/dT (right axis). (c) Heat-capacity Cp of LuFe4Ge2 as a function of temperature. The upper inset displays the thermal hysteresis in Cp(T ) obtained from the analysis of heating and cooling parts of the thermal-relaxation cycle. The lower inset shows Cp/T vs T 2 where the straight line is a ﬁt to the data for 2 K ⩽T ⩽20 K using Cp(T ) = γ T + βT 3. FIG. 1. (a) Crystal structure of LuFe4Ge2 viewed along the c axis. In the low-temperature orthorhombic phase (Pnnm), the Fe sites split in to two sites marked as Fe1 and Fe2 [14]. (b) The chainlike arrangement of edge-shared Fe tetrahedra viewed along the b axis. 0 20 40 60 80 100 120 140 0 20 40 60 80 Cp(Jmol-1K-1) Cp(Jmol-1K-1) T (K) (c) LuFe4Ge2 34 36 38 40 0 40 80 heating cooling T (K) 0 500 1000 0 0.2 0.4 0.6 T 2 (K2) Cp/T(Jmol-1K-2) 0 4 8 12 dρ/dT (arb. unit) (b) 0 50 100 150 200 250 300 0 0.2 0.4 0.6 0.8 1 ρ (mΩcm) T (K) LuFe4Ge2 0.05 0.1 0.15 0.2 LuFe4Ge2 χ (cm3/mol) (a) 0 4 8 12 χ-1 (mol/cm3) 0 4 8 12 dρ/dT (arb. unit) (b) 0 50 100 150 200 250 300 0 0.2 0.4 0.6 0.8 1 ρ (mΩcm) T (K) LuFe4Ge2 0.05 0.1 0.15 0.2 LuFe4Ge2 χ (cm3/mol) (a) 0 4 8 12 χ-1 (mol/cm3) 4 8 χ-1 (mol/cm3) 2 dρ/dT (arb. unit) FIG. 1. (a) Crystal structure of LuFe4Ge2 viewed along the c axis. In the low-temperature orthorhombic phase (Pnnm), the Fe sites split in to two sites marked as Fe1 and Fe2 [14]. (b) The chainlike arrangement of edge-shared Fe tetrahedra viewed along the b axis. I. INTRODUCTION 0 20 40 60 80 100 120 140 0 20 40 60 80 Cp(Jmol-1K-1) Cp(Jmol-1K-1) T (K) (c) LuFe4Ge2 34 36 38 40 0 40 80 heating cooling T (K) 0 500 1000 0 0.2 0.4 0.6 T 2 (K2) Cp/T(Jmol-1K-2) ( ) revealed that this compound is close to a lattice-volume-tuned quantum critical point [13]. Furthermore, signiﬁcantly large electronic heat capacity observed at low temperatures has been ascribed to the effect of magnetic frustration. The combi- nation of magnetic frustration and low dimensionality in these compounds draws the attention for a detailed investigation on other materials in the family. Earlier studies on LuFe4Ge2 showed an antiferromagnetic (AFM) transition at 32 K with ﬁrst-order character accom- panied by a structural transition from tetragonal P42/mnm to orthorhombic Pnnm [14]. The results pointed at a canted arrangement of Fe moments on the ab plane yielding a com- mensurate antiferromagnetic phase with propagation vector q = 0. Moreover, the size of the ordered Fe moment of 0.44μB appeared to be highly reduced, which is attributed to the presence of magnetic frustration. However, a detailed study on the physical properties and the interplay of structure and magnetism is lacking. FIG. 2. (a) Temperature dependence of the DC magnetic suscep- FIG. 2. (a) Temperature dependence of the DC magnetic suscep- tibility χ of LuFe4Ge2. The inverse susceptibility χ −1(T ) is plotted on the right axis along with a Curie-Weiss ﬁt to the data (red curve) in the temperature interval between 100 and 300 K. (b) Tempera- ture dependence of the electrical resistivity ρ(T ) (left axis) and the temperature derivative dρ(T )/dT (right axis). (c) Heat-capacity Cp of LuFe4Ge2 as a function of temperature. The upper inset displays the thermal hysteresis in Cp(T ) obtained from the analysis of heating and cooling parts of the thermal-relaxation cycle. The lower inset shows Cp/T vs T 2 where the straight line is a ﬁt to the data for 2 K ⩽T ⩽20 K using Cp(T ) = γ T + βT 3. FIG. 2. (a) Temperature dependence of the DC magnetic suscep- tibility χ of LuFe4Ge2. The inverse susceptibility χ −1(T ) is plotted on the right axis along with a Curie-Weiss ﬁt to the data (red curve) in the temperature interval between 100 and 300 K. A. Ambient pressure characterization The temperature dependence of the DC magnetic sus- ceptibility χ = M/H of LuFe4Ge2 at ambient pressure is shown in Fig. 2(a). Here, the ferromagnetic contribution in the magnetization from a small amount of Fe3Ge phase (<4%) in the sample is estimated by measuring magnetization at different ﬁelds, which is then subtracted from the data mea- sured at 1 T to obtain the intrinsic magnetic susceptibility of LuFe4Ge2. High-temperature χ(T ) follows a Curie-Weiss (CW) behavior χ(T ) = C/(T −θW) with an effective mo- 125136-2 PHYSICAL REVIEW B 107, 125136 (2023) INTERPLAY OF STRUCTURE AND MAGNETISM IN … FIG. 3. (a) Electrical resistivity of LuFe4Ge2 as a function of temperature for several applied pressures. The inset: Resistivity isotherms as a function of pressure. The arrows indicate the phase boundary to a pressure-induced phase. (b) Temperature derivative of electrical resistivity dρ/dT vs T for several pressures. Various anomalies are marked by symbols. (c) Temperature dependence of the real part of the AC magnetic susceptibility (χ′) of LuFe4Ge2 for selected applied pressures (data shifted vertically). The pressure dependence of the anomalies are traced by the dashed lines and additional low-temperature anomalies at higher pressures are marked by ∗and # symbols. resolution of the data (see Appendix A). This suggests that the impurity volume fraction is rather small. In addition, using scanning electron micrograph on samples from different parts of the polycrystalline ingot, we obtain an upper estimate of the impurity volume fraction of 4%. Figure 2(b) shows the temperature dependence of the elec- trical resistivity ρ(T ) of LuFe4Ge2. ρ(T ) reveals a metallic behavior upon cooling followed by a sudden drop in the resis- tivity at 36 K coinciding with the magnetostructural transition. The decrease in the resistivity at the transition temperature can be understood as the reduction in the scattering contribution due to the transition from the disordered paramagnetic to the antiferromagnetically ordered phase. Upon further cooling, the resistivity decreases with a higher slope, indicating a fur- ther reduction of the scattering in the antiferromagnetic phase. The resistivity ratio is determined as ρ300K /ρ1.8 K ≈11 for the investigated sample. g p Heat-capacity Cp of LuFe4Ge2 as a function of tem- perature is presented in Fig. 2(c). A very sharp peak in Cp(T ) is observed at T = 36 K, consistent with TN obtained from magnetic-susceptibility and resistivity data. A. Ambient pressure characterization The pressure dependence of the anomalies are traced by the dashed lines and additional low-temperature anomalies at higher pressures are marked by ∗and # symbols. B. Expetiments under hydrostatic pressure 1. Electrial rsistivity To investigate the evolution of the magnetostructural tran- sition upon application of hydrostatic pressure, we ﬁrst turn to electrical-resistivity measurements. Figure 3(a) presents the evolution of the temperature dependence of the electri- cal resistivity ρ(T ) under external pressure. Under pressure, initially, the anomaly in the resistivity shifts to lower tem- peratures with increasing pressure whereas the sharp nature of the anomaly changes to a gradual decrease. The anomaly becomes much weaker and moves to around 12 K at a pressure of 1.7 GPa. Above 1.7 GPa, the feature becomes too small and not traceable within the resolution of our data. However, the resistivity isotherms plotted as a function of pressure [see the inset of Fig. 3(a)] display clear jumps at the phase boundary. These data suggest that the ﬁrst-order-like antiferromagnetic transition (AFM1) is suppressed to zero temperature by the application of a pressure of pc ≈1.8 GPa. In addition, a weak shoulderlike anomaly in ρ(T ) develops around T = 35 K, which becomes more prominent at higher pressures. This anomaly is better visible in the temperature derivative of the electrical resistivity dρ/dT presented in Fig. 3(b). With increasing pressure, this feature slowly shifts to higher tem- peratures, reaching about 40 K at p = 2.52 GPa, suggesting a phase boundary from the PM phase to a pressure-induced phase. Furthermore, for p ⩾1.86 GPa, two additional features are clearly seen in dρ/dT at low temperatures around 20 K and 10 K (marked by ∗and # symbols, respectively). We note that the residual resistance above pc, in the pressure-induced phase, is considerably larger than that in the AFM1 phase at low pressures. A. Ambient pressure characterization The low- temperature part of Cp(T ) is ﬁtted with Cp(T ) = γ T + βT 3 in the interval between 2 and 20 K [see the lower inset of Fig. 2(c)]. The ﬁt yields an enhanced value for the Som- merfeld coefﬁcient γ = 94(1) mJ mol−1 K−2 which indicates strong electron correlation effects. Cp(T ) upon heating and cooling, obtained by analyzing the thermal-relaxation curves recorded in a standard relaxation-type measurement setup fol- lowing a method outlined by Lashley et al. [15], is plotted in the upper inset of Fig. 2(b). A thermal hysteresis observed between heating and cooling curves conﬁrms the ﬁrst-order nature of the magnetostructural transition. FIG. 3. (a) Electrical resistivity of LuFe4Ge2 as a function of temperature for several applied pressures. The inset: Resistivity isotherms as a function of pressure. The arrows indicate the phase boundary to a pressure-induced phase. (b) Temperature derivative of electrical resistivity dρ/dT vs T for several pressures. Various anomalies are marked by symbols. (c) Temperature dependence of the real part of the AC magnetic susceptibility (χ′) of LuFe4Ge2 for selected applied pressures (data shifted vertically). The pressure dependence of the anomalies are traced by the dashed lines and additional low-temperature anomalies at higher pressures are marked by ∗and # symbols. FIG. 3. (a) Electrical resistivity of LuFe4Ge2 as a function of temperature for several applied pressures. The inset: Resistivity isotherms as a function of pressure. The arrows indicate the phase boundary to a pressure-induced phase. (b) Temperature derivative of electrical resistivity dρ/dT vs T for several pressures. Various anomalies are marked by symbols. (c) Temperature dependence of the real part of the AC magnetic susceptibility (χ′) of LuFe4Ge2 for selected applied pressures (data shifted vertically). The pressure dependence of the anomalies are traced by the dashed lines and additional low-temperature anomalies at higher pressures are marked by ∗and # symbols. FIG. 3. (a) Electrical resistivity of LuFe4Ge2 as a function of temperature for several applied pressures. The inset: Resistivity isotherms as a function of pressure. The arrows indicate the phase boundary to a pressure-induced phase. (b) Temperature derivative of electrical resistivity dρ/dT vs T for several pressures. Various anomalies are marked by symbols. (c) Temperature dependence of the real part of the AC magnetic susceptibility (χ′) of LuFe4Ge2 for selected applied pressures (data shifted vertically). 2. Magnetic susceptibility The results from ρ(T ) measurements are corroborated by AC magnetic susceptibility, which further conﬁrms their mag- netic origin. The results of the AC magnetic-susceptibility measurements on LuFe4Ge2 for several applied pressures are presented in Fig. 3(c). The susceptibility data at each pressure were normalized by the jump height of the superconducting transition of the Pb manometer in order to compare the data taken at different pressures. At p = 0.1 GPa, the real part of ac susceptibility (χ′) shows a sudden decrease at the antifer- romagnetic transition. Upon increasing pressure the drop in χ′ shifts to lower temperatures whereas the sharpness of the jump is reduced. A broad humplike feature develops around 40 K and shows a slight shift to higher temperatures with a further increase in pressure. These observations are in ex- cellent agreement with the resistivity data. Above 1.8 GPa, 125136-3 M. O. AJEESH et al. PHYSICAL REVIEW B 107, 125136 (2023) FIG. 4. (a) Temperature dependence of the magnetic-volume fraction Vmag of LuFe4Ge2 obtained using μSR at different pres- sures. The solid lines are ﬁts to the data using a phenomenological equation provided in the text. (b) Temperature dependence of the muon-spin precession frequency under different applied pressures. (c) Vmag obtained from time-domain 57Fe M¨ossbauer spectroscopy under several pressures. (b) Energy-domain 57Fe M¨ossbauer spectra measured at p = 2.9 GPa, T = 3 K, and an external magnetic ﬁeld of 6 T. The solid lines are ﬁts to the spectra by exact line-shape site analysis using RECOIL software. with the weak p-independent anomalies in the resistivity and susceptibility data. A 100% magnetic volume fraction in both phases is also conﬁrmed by M¨ossbauer data [see Fig. 4(c)]. Furthermore, the M¨ossbauer spectrum obtained at p = 2.9 GPa, T = 3 K, and an external magnetic ﬁeld of 6 T [see Fig. 4(d)] suggests that the pressure-induced phase is antiferromagnetically ordered (AFM2). The ﬁt to the spectrum by exact line-shape site analysis using RECOIL software revealed contributions from Fe moments parallel and antiparallel to the external magnetic ﬁeld. The fraction of magnetic moments on the Fe parallel to the external ﬁeld is about 66% with a hyperﬁne magnetic ﬁeld of Bhf = 11.8 T. This value is higher than that of the 34% of magnetic moments on Fe antiparallel to the external ﬁeld with Bhf = 3.6 T. 2. Magnetic susceptibility This is presumably because of the rotation of some part of the moments to the direction of the applied magnetic ﬁeld. Such an observation is consistent with the metamagnetic transition seen in the magnetoresistance data (see Appendix B). In the magnetoresistance (MR) data, the spin reorientation appears to occur at external ﬁelds as low as 5 T at high pressures. FIG. 4. (a) Temperature dependence of the magnetic-volume fraction Vmag of LuFe4Ge2 obtained using μSR at different pres- sures. The solid lines are ﬁts to the data using a phenomenological equation provided in the text. (b) Temperature dependence of the muon-spin precession frequency under different applied pressures. (c) Vmag obtained from time-domain 57Fe M¨ossbauer spectroscopy under several pressures. (b) Energy-domain 57Fe M¨ossbauer spectra measured at p = 2.9 GPa, T = 3 K, and an external magnetic ﬁeld of 6 T. The solid lines are ﬁts to the spectra by exact line-shape site analysis using RECOIL software. g p Further information regarding the strength of the local magnetic-ﬁeld Bloc at the muon site in the magnetically ordered phases is obtained from zero-ﬁeld μSR measure- ments. The temperature dependence of the muon-precession frequency ωμ for three different pressures is displayed in Fig. 4(b). At ambient pressure, the spontaneous precession occurs at T ≈36 K with a slight enhancement in ωμ upon further decreasing temperature. This sharp steplike behavior of ωμ(T ) is consistent with the ﬁrst-order nature of the phase transition at ambient pressure. At p = 1.39 GPa, the preces- sion starts at T ≈34 K. The frequency gradually increases upon cooling followed by a sudden jump at T ≈26 K. These features can be well ascribed to the two consecutive phase transitions upon lowering temperature: The ﬁrst from the paramagnetic to the pressure-induced AFM2 phase at 34 K and the second from the AFM2 to the AFM1 phase at 26 K. The gradual increase in ωμ at the ﬁrst phase transition implies that this transition is of the second-order type. The sharp jump in ωμ around 26 K arises from the transition from the AFM2 to the AFM1 phase and the sharpness points at a ﬁrst-order- type phase transition. At p = 2.26 GPa, the precession starts around 35 K where the compound undergoes the transition from the PM to the AFM2 phase. Upon further cooling, ωμ increases gradually until the lowest temperature in our exper- iment. 3. M¨ossbauer and muon-spin spectroscopy In order to get a microscopic understanding of the nature of the different magnetic phases of LuFe4Ge2, we have carried out muon-spin relaxation (μSR) and 57Fe M¨ossbauer spec- troscopy measurements under several pressures below and above pc ≈1.8 GPa. μSR measurements in a weak trans- verse ﬁeld of BTF = 50 Oe were performed to determine the magnetic-volume fraction (Vmag) and the transition tem- peratures. In Fig. 4(a), Vmag of LuFe4Ge2 as a function of temperature for three different pressures is shown. Vmag(T ) shows a sharp steplike increase at the magnetic-ordering temperature for all pressures, a characteristic feature of long- range magnetic order. It is also important to note that the entire sample volume undergoes magnetic ordering at all pres- sures. The data conﬁrm that the pressure-induced phase is also long-range magnetically ordered. The transition temperature TN is extracted by ﬁtting the magnetic-volume fraction with the phenomenological equation Vmag = 1 1+e(T −TN )/w where w is the transition width. Here, the magnetic-ordering temperature seems to be nearly independent of pressure and coincide 2. Magnetic susceptibility We note that the ωμ(T ) curve for p = 2.26 GPa shows noticeable slope changes at temperatures around 20 and 10 K. These features occur at the same temperatures as the low-T anomalies in the dρ(T )/dT and χ′(T ) data discussed earlier and suggest the possibility of multiple-order parameters as- sociated with the phase transition from the paramagnetic to the high-pressure AFM2 phase. It is worth noting that, at the lowest temperature, the muon-precession frequencies for the three pressures have similar values. This indicates that, at low temperatures, both the AFM1 and the pressure-induced AFM2 phase have similar local magnetic ﬁelds at the muon sites. the nature of the χ′(T ) curve is signiﬁcantly different from that in the low-pressure region. In the high-pressure region (p > 1.8 GPa), an additional humplike feature and a sharp de- crease in the susceptibility are observed at about 20 and 10 K coinciding with the low-temperature anomalies observed in dρ/dT . The origin and the nature of these anomalies are not fully understood, however, their implication in other physical properties are discussed later. 4. X-ray diffraction PM (a) 14.0 14.2 15.5 16.0 20 K 2θ (degree) 35 K 30 K 45 K p = 2.0 GPa 14.0 14.2 15.5 16.0 2θ (degree) 20 K 35 K 30 K 45 K p = 1.5 GPa 14.0 14.2 15.5 16.0 (130) (310) (200) (310) 20 K 35 K 30 K Intensity (arb. units) 2θ (degree) p = 0.5 GPa 45 K (200) Intensity (arb. units) 25 30 35 40 45 2 3 4 5 0.29 GPa C (arb.unit) T (K) 0.76 GPa cooling warming (b) 0 1 2 3 4 -2 -1 0 1 2 p > 1.8 GPa MR (%) B (T) p < 1.8 GPa (c) T = 2 K FIG. 5. Representative peaks in the x-ray diffraction patterns of LuFe4Ge2 at different temperatures and applied pressures. The structural transition from the tetragonal to the orthorhombic phase is evidenced by the splitting of the diffraction peak around 2θ ≈15.5◦. from tetragonal to orthorhombic symmetry is characterized by the splitting of certain peaks in the diffraction pattern. For example, the [310] peak splits in the orthorhombic phase whereas the [200] peak does not. Therefore, the evolution of these peaks with varying conditions of temperature and pressure provides a straight forward determination of the structural transition in the phase diagram. Figure 5 presents the [200] and [310] diffraction peaks obtained at various tem- peratures and pressures. Remarkably, the splitting of the [310] peak occurs between 30 and 35 K in the entire pressure range. This conﬁrms that the structural transition temperature does not signiﬁcantly change with pressure and it remains between 30 and 35 K. The lack of a more precise determination of the structural transition temperature is due to the limited temper- ature sampling available in our experiment. FIG. 6. (a) T −p phase diagram of LuFe4Ge2. The transition temperatures obtained from ρ(T ) (circle), χ ′(T ) (square), μSR (star), and M¨ossbauer (sphere) data are indicated. The ∗and # sym- bols stand for the low-temperature anomalies observed in ρ(T ) and χ ′(T ) data. (b) Thermal hysteresis in C(T ), measured using an AC- calorimetry technique under pressure, conﬁrms the ﬁrst-order nature of the transition at TN1. The arrow points at a very weak feature at TN2. (c) MR(B) = [ρ(B) −ρ(B = 0)]/ρ(B = 0) of LuFe4Ge2 measured at T = 2 K. FIG. 6. (a) T −p phase diagram of LuFe4Ge2. 4. X-ray diffraction Remarkably, the splitting of the [310] peak occurs between 30 and 35 K in the entire pressure range. This conﬁrms that the structural transition temperature does not signiﬁcantly change with pressure and it remains between 30 and 35 K. The lack of a more precise determination of the structural transition temperature is due to the limited temper- 25 30 35 40 45 2 3 4 5 0.29 GPa C (arb.unit) T (K) 0.76 GPa cooling warming (b) 0 1 2 3 4 -2 -1 0 1 2 p > 1.8 GPa MR (%) B (T) p < 1.8 GPa (c) T = 2 K ∗∗∗∗∗ # # # # # 0.5 1 1.5 2 2.5 3 0 0 20 40 60 ' μSR Mössbauer orth. AFM2 LuFe4Ge2 T (K) p (GPa) TN1 orth. AFM1 TN2 tetr. PM (a) FIG. 6. (a) T −p phase diagram of LuFe4Ge2. The transition temperatures obtained from ρ(T ) (circle), χ ′(T ) (square), μSR (star), and M¨ossbauer (sphere) data are indicated. The ∗and # sym- bols stand for the low-temperature anomalies observed in ρ(T ) and χ ′(T ) data. (b) Thermal hysteresis in C(T ), measured using an AC- calorimetry technique under pressure, conﬁrms the ﬁrst-order nature of the transition at TN1. The arrow points at a very weak feature at TN2. (c) MR(B) = [ρ(B) −ρ(B = 0)]/ρ(B = 0) of LuFe4Ge2 measured at T = 2 K. 14.0 14.2 15.5 16.0 20 K 2θ (degree) 35 K 30 K 45 K p = 2.0 GPa 14.0 14.2 15.5 16.0 2θ (degree) 20 K 35 K 30 K 45 K p = 1.5 GPa 14.0 14.2 15.5 16.0 (130) (310) (200) (310) 20 K 35 K 30 K Intensity (arb. units) 2θ (degree) p = 0.5 GPa 45 K (200) FIG. 5. Representative peaks in the x-ray diffraction patterns of LuFe4Ge2 at different temperatures and applied pressures. The structural transition from the tetragonal to the orthorhombic phase is evidenced by the splitting of the diffraction peak around 2θ ≈15.5◦. ∗∗∗∗∗ # # # # # 0.5 1 1.5 2 2.5 3 0 0 20 40 60 ' μSR Mössbauer orth. AFM2 LuFe4Ge2 T (K) p (GPa) TN1 orth. AFM1 TN2 tetr. 4. X-ray diffraction The transition structural transition. Figure 6(b) depicts heat-capacity data (measured using an AC-calorimetry technique) taken upon heating and cooling cycles under selected pressures of 0.29 and 0.76 GPa. The peak in C(T ) associated with the transition from the AFM2 to the AFM1 phase shows a thermal hys- teresis, conﬁrming the ﬁrst-order character of the transition. We note that the thermal hysteresis was measured with a very slow temperature sweep rate in order to minimize the effect of the relatively slow thermalization of the pressure cell. A similar hysteresis is also observed in resistivity and AC magnetic-susceptibility data (not shown). The ﬁrst-order nature of the transition is also consistent with the sharp jump in ωμ at this phase boundary. The transition from the PM to the AFM2 phase appears as very weak features in C(T ), ρ(T ), and χ′(T ) without any observable thermal hysteresis. Moreover, ωμ shows a gradual increase at this phase bound- ary, conﬁrming that the transition from the PM to the AFM2 phase is of second order. 4. X-ray diffraction We now turn to the evolution of the structural transition in LuFe4Ge2 under external pressure. The structural transition 125136-4 INTERPLAY OF STRUCTURE AND MAGNETISM IN … PHYSICAL REVIEW B 107, 125136 (2023) 25 30 35 40 45 2 3 4 5 0.29 GPa C (arb.unit) T (K) 0.76 GPa cooling warming (b) 0 1 2 3 4 -2 -1 0 1 2 p > 1.8 GPa MR (%) B (T) p < 1.8 GPa (c) T = 2 K ∗∗∗∗∗ # # # # # 0.5 1 1.5 2 2.5 3 0 0 20 40 60 ' μSR Mössbauer orth. AFM2 LuFe4Ge2 T (K) p (GPa) TN1 orth. AFM1 TN2 tetr. PM (a) FIG. 6. (a) T −p phase diagram of LuFe4Ge2. The transition temperatures obtained from ρ(T ) (circle), χ ′(T ) (square), μSR (star), and M¨ossbauer (sphere) data are indicated. The ∗and # sym- bols stand for the low-temperature anomalies observed in ρ(T ) and χ ′(T ) data. (b) Thermal hysteresis in C(T ), measured using an AC- calorimetry technique under pressure, conﬁrms the ﬁrst-order nature of the transition at TN1. The arrow points at a very weak feature at TN2. (c) MR(B) = [ρ(B) −ρ(B = 0)]/ρ(B = 0) of LuFe4Ge2 measured at T = 2 K. 14.0 14.2 15.5 16.0 20 K 2θ (degree) 35 K 30 K 45 K p = 2.0 GPa 14.0 14.2 15.5 16.0 2θ (degree) 20 K 35 K 30 K 45 K p = 1.5 GPa 14.0 14.2 15.5 16.0 (130) (310) (200) (310) 20 K 35 K 30 K Intensity (arb. units) 2θ (degree) p = 0.5 GPa 45 K (200) FIG. 5. Representative peaks in the x-ray diffraction patterns of LuFe4Ge2 at different temperatures and applied pressures. The structural transition from the tetragonal to the orthorhombic phase is evidenced by the splitting of the diffraction peak around 2θ ≈15.5◦. from tetragonal to orthorhombic symmetry is characterized by the splitting of certain peaks in the diffraction pattern. For example, the [310] peak splits in the orthorhombic phase whereas the [200] peak does not. Therefore, the evolution of these peaks with varying conditions of temperature and pressure provides a straight forward determination of the structural transition in the phase diagram. Figure 5 presents the [200] and [310] diffraction peaks obtained at various tem- peratures and pressures. D. Electronic structure calculations 0 0.5 1 1.5 2 0 2 4 6 8 10 2(a-b)/(a+b)×102 m ( B) LDA GGA (c) gy ( ) Electronic-structure calculations for LuFe4Ge2 based on DFT have been performed to understand its magnetic and structural phase transitions. The magnetic properties are de- termined by the dominating contribution of the Fe-3d related rather narrow bands near the Fermi level EF, resulting in a pronounced peak in the density of states (DOS) near EF [see Fig. 7(a)]. With a value of about 3.5 states (per Fe and eV) at the Fermi level, the Stoner criterion is fulﬁlled, evidencing a magnetic instability and the tendency for spontaneous mag- netic ordering. The magnetic and structural transitions to a collinear AFM state leads to a strong reduction of the DOS at EF [compare Figs. 7(a) and 7(b)]. To a large part, this drop in the DOS(EF) is caused by the band splitting related to the localized Fe-3d moments and, thus, similar for different magnetic structures. FIG. 7. Total and partial density of states of LuFe4Ge2 for (a) the nonmagnetic tetragonal and (b) the antiferromagnetic orthorhombic cases. The Fermi level is at zero energy. Panel (c) shows the depen- dence of the lattice distortion (in percentages) on the Fe magnetic moment for different exchange correlation potentials. A full relaxation of the crystal structure (at the experi- mental low-temperature volume) yields that the orthorhombic symmetry is energetically more stable than the tetragonal. In- terestingly, the resulting equilibrium distortion 2(a −b)/(a + b) of the lattice is strongly dependent on the size of the magnetic moment, essentially independent from the choice of the exchange correlation potential [see Fig. 7(c)]. This strong and nonlinear dependence indicates that the magnetism and the crystal lattice are coupled in a nontrivial way. The energy dependence of the DOS for both the tetragonal and the orthorhombic experimental crystal structures without spin polarization reveals that the magnetic ordering can occur independent of the structural transition since for both symme- tries we obtain a similarly large DOS near the Fermi level. Therefore, the reason for the magnetic ordering occurring simultaneously with the structural transition might not be directly related to the electronic band structure but rather due to the change in the strength of the magnetic frustra- tion due to the lattice distortion. C. Phase diagram These observations can be taken as an indication for stronger magnetic ﬂuctuations in the AFM2 phase compared to the AFM1 phase. The neg- ative MR stems from the reduction in the resistivity as the external magnetic ﬁeld suppresses the magnetic ﬂuctuations. The small reduction in the χ′ at the magnetic ordering at the phase boundary to AFM2 phase compared to the large drop in χ′ at the ordering at the phase boundary to AFM1 phase is also in support of the existence of strong magnetic ﬂuctuations in AFM2 phase. Moreover, the gradual increase in the muon-precession frequency, which is a measure of the local magnetic ﬁeld, in the AFM2 phase upon deceasing temperature suggests a strong temperature dependence of the magnetic ﬂuctuations. motion of the conduction electrons in the transverse magnetic ﬁeld. However, in the AFM2 phase, MR(B) shows an opposite behavior displaying negative values. These observations can be taken as an indication for stronger magnetic ﬂuctuations in the AFM2 phase compared to the AFM1 phase. The neg- ative MR stems from the reduction in the resistivity as the external magnetic ﬁeld suppresses the magnetic ﬂuctuations. -6 -5 -4 -3 -2 -1 0 1 2 0 10 20 DOS (states/eV/f.u.) Energy (eV) orthorhombic AFM (b) 0 10 20 OS (states/eV/f.u.) total Lu Fe Ge tetragonal non magnetic (a) The small reduction in the χ′ at the magnetic ordering at the phase boundary to AFM2 phase compared to the large drop in χ′ at the ordering at the phase boundary to AFM1 phase is also in support of the existence of strong magnetic ﬂuctuations in AFM2 phase. Moreover, the gradual increase in the muon-precession frequency, which is a measure of the local magnetic ﬁeld, in the AFM2 phase upon deceasing temperature suggests a strong temperature dependence of the magnetic ﬂuctuations. C. Phase diagram The temperature–pressure phase diagram of LuFe4Ge2 is established using the results from the different high-pressure studies, see Fig. 6(a). The results from bulk measurements as well as from microscopic probes are in excellent agree- ment. TN1 corresponding to the transition from the PM to the AFM1 phase is continuously suppressed toward zero tem- perature upon increasing pressure at a critical pressure of pc ≈1.8 GPa. A second antiferromagnetic phase (AFM2) is conﬁrmed by M¨ossbauer and μSR measurements. The structural transition temperature, which coincides with the AFM1-ordering temperature at ambient pressures does not signiﬁcantly change with application of pressure and remains at about 35 K, i.e., the AFM1 transition decouples from the structural transition. Moreover, the onset of the mag- netic ordering from the PMto the AFM2 phase appears to be connected to the structural transition, i.e., its transition temperature TN2 is also almost independent of pressure. The magnetoresistance and the magnetic-susceptibility data provide insights into the relative strength of magnetic ﬂuctuations in the two antiferromagnetic phases of LuFe4Ge2. In the AFM1 phase, the MR = [ρ(B) −ρ(B = 0)]/ρ(B = 0) exhibits a monotonous increase with a quadratic dependence on the magnetic ﬁeld [see Fig. 6(c)]. This is the typical behav- ior expected for metallic systems and is due to the cyclotron The nature of the transitions between the various phases have been studied in detail. As described earlier, the mag- netostructural transition at ambient pressure is of ﬁrst-order type. The application of pressure decouples the TN1 and the 125136-5 PHYSICAL REVIEW B 107, 125136 (2023) M. O. AJEESH et al. 0 0.5 1 1.5 2 0 2 4 6 8 10 2(a-b)/(a+b)×102 m ( B) LDA GGA (c) -6 -5 -4 -3 -2 -1 0 1 2 0 10 20 DOS (states/eV/f.u.) Energy (eV) orthorhombic AFM (b) 0 10 20 OS (states/eV/f.u.) total Lu Fe Ge tetragonal non magnetic (a) FIG. 7. Total and partial density of states of LuFe4Ge2 for (a) the nonmagnetic tetragonal and (b) the antiferromagnetic orthorhombic cases. The Fermi level is at zero energy. Panel (c) shows the depen- dence of the lattice distortion (in percentages) on the Fe magnetic moment for different exchange correlation potentials. motion of the conduction electrons in the transverse magnetic ﬁeld. However, in the AFM2 phase, MR(B) shows an opposite behavior displaying negative values. III. CONCLUSION sample space was determined at low temperatures by the shift in the superconducting transition temperature of a piece of Pb. The electrical resistivity was measured using a standard four-terminal method where electrical contacts to the sample were made using 25-μm gold wires and silver paint. Resistiv- ity was measured using an LR700 resistance bridge (Linear Research). AC magnetic susceptibility was measured using home-made AC susceptometer that ﬁts inside the pressure cell. The signal was measured using an LR700 mutual- inductance bridge (Linear Research). A static ﬁeld BDC of 0.01 T and a modulation ﬁeld BAC of 1.3 mT at a frequency of 16 Hz were used for the measurements. AC calorimetry measurements were performed using a commercial heater and Cernox thermometer following the method described in Ref. [17]. To summarize, LuFe4Ge2 presents an interesting interplay of magnetic, structural, and electronic degrees of free- doms. At ambient pressure LuFe4Ge2 undergoes simultaneous antiferromagnetic and structural transitions at 36 K with ﬁrst- order character. The pressure dependence of the magnetic transition in LuFe4Ge2 has been investigated using electri- cal transport, AC magnetic susceptibility, AC calorimetry, M¨ossbauer, μSR, and PXRD measurements. External pres- sure suppresses the ﬁrst-order magnetic transition (AFM1) to zero temperature around 1.8 GPa. The structural transition is largely unaffected by pressure. A new antiferromagnetic phase (AFM2) is observed at higher pressures, conﬁrmed by M¨ossbauer and μSR measurements. The transition from the paramagnetic to the AFM2 phase is of second order and appears to be connected to the structural transition. μSR and M¨ossbauer data revealed 100% magnetic volume fraction in both the magnetic phases. In addition, similar values of muon- precession frequency at low temperatures in the AFM1 and AFM2 phases point at similar ordered moments and closely related magnetic structures in the two phases. Our results also indicate enhanced magnetic ﬂuctuations in the pressure- induced AFM2 phase. The experimental observations are supported by DFT band-structure calculations, suggesting a scenario where the magnetic- and structural-order parameters in LuFe4Ge2 are linked by magnetic frustration, causing the simultaneous magnetostructural transition. Our results reveal an interesting and unusual interplay of structure and mag- netism in LuFe4Ge2, which differ from the situation observed in the AFe2As2 pnictides. Therefore, LuFe4Ge2 is an attractive system where further in-depth studies could provide deeper insight into the interaction between frustrated magnetism and structural instability, a topic of general interest, which is also relevant for other classes of quantum materials. III. CONCLUSION Muon-spin relaxation measurements under pressure were performed at the μE1 beam line using the GPD spectrom- eter at the Paul Scherrer Institute (PSI), Switzerland (see Refs. [18,19] for details of the high-pressure μSR tech- nique at PSI). Synchrotron 57Fe Mössbauer spectroscopy measurements under pressure were conducted at the Nuclear Resonance beamline (ID18) of the European Synchrotron Ra- diation Facility (ESRF), Grenoble [20] and at the beamline 3ID-B of the Advanced Photon Source, Argonne National Laboratory, USA [21,22]. X-ray-diffraction experiments us- ing a diamond-anvil cell for generating pressure were performed at the XDS beamline at the Brazilian Synchrotron facility LNLS [23]. DFT calculations were performed using the plane-wave pseudopotential method implemented in the Vienna ab initio simulation package [24], applying the local density approxi- mation [25] and the general gradient approximation [26] for the exchange-correlation functional. We use a plane-wave en- ergy cutoff of 500 eV and a regular Monkhorst-Pack grid of 8 × 8 × 12 to perform the ionic relaxation and 10 × 10 × 14 to achieve the self-consistent calculations. To obtain the den- sity of states, the k mesh was increased to 22 × 22 × 24 using the tetrahedron method. The optimization of the struc- tures was carried out with a force convergence tolerance of 1 meV/Å per atom. The tetragonal (T ) and orthorhom- bic (O) structures have P42/mnm and Pnnm space-group symmetry, respectively. In order to study the FM and vari- ous AFM magnetic states and respective structural distortion of the orthorhombic structure, we carried out collinear and noncollinear calculations. To perform our studies, we have considered the structural parameters given in this paper at 10 K and ambient pressure and the structural data given in the previous work [14]. IV. METHODS Polycrystalline samples of LuFe4Ge2 were synthesized by a standard arc-melting technique on a copper hearth. Constituent elements (at least, 99.9% purity), taken in the sto- ichiometric ratio, were melted in an arc furnace under argon atmosphere, followed by several ﬂipping and remelting of the resulting ingot to ensure homogeneity. Then, the as-cast sam- ples were annealed at 1150 ◦C under a static argon atmosphere for a week. The phase purity of the annealed samples was checked by powder x-ray diffraction using Cu Kα radiation (see Appendix A) and a scanning electron micrograph, which revealed only a small amount (up to 4%) of eutectic phase Fe3Ge in our samples. The stoichiometry of the samples was veriﬁed using energy dispersive x-ray analysis. INTERPLAY OF STRUCTURE AND MAGNETISM IN … INTERPLAY OF STRUCTURE AND MAGNETISM IN … PHYSICAL REVIEW B 107, 125136 (2023) D. Electronic structure calculations High-ﬁeld magnetization measurements on LuFe4Ge2 showed that the magnetization remains small up to high ﬁelds and the extrapolated saturation ﬁeld is more than 150 T [16]. This corresponds to strong antiferromagnetic exchange interactions on the scale of more than 100 K. The strong exchange interaction along with a relatively low-ordering temperature points at the signiﬁcance of the magnetic frustration. It is likely that the magnetic frus- tration, at least, to some extent, is released by the distortion of the Fe tetrahedra during the structural transition from the tetragonal to the orthorhombic phase and, thereby, facilitat- ing the magnetic ordering. This scenario is supported by the calculations, which predict the structural transition even with- out involving magnetic polarization. Here, the large magnetic entropy connected with a ﬂuctuating frustrated paramagnetic system, which stabilizes the tetragonal phase upon increasing temperature, might explain why the tetragonal phase is stable down to such comparatively low temperature. Our results sug- gest a mechanism where the magnetic- and structural-order parameters in LuFe4Ge2 are linked by the magnetic frustration causing the simultaneous magnetostructural transition. The commensurate magnetic ground state proposed by Schobinger-Papamantellos et al. [14] is also reproduced by our calculations. Moreover, a collinear spin structure is found to be energetically close to the noncollinear ground state. This points at the possibility that the pressure-induced AFM2 phase could be a spin rearrangement from noncollinear to collinear structure. Such a subtle spin rearrangement is also consistent with the similar values of low-temperature muon-precession frequency found in AFM1 and AFM2 phases. 125136-6 APPENDIX A: POWDER X-RAY DIFFRACTION AT AMBIENT PRESSURE It is also interesting to note that the normalized resistivity ρ/ρ300 K at higher magnetic ﬁelds appears to have similar values for all applied pressures as displayed in Fig. 9(b). This suggests that the ﬁeld-polarized phase could be the same in the entire pressure range, once again indicating the close similarity between the AFM1 and the AFM2 phases. ACKNOWLEDGMENTS The Bragg positions are indicated by green vertical bars, and the bottom solid blue line indicates the difference between the experimental and the calculated intensities. 10 20 30 40 50 60 70 80 90 Intensity (arb. unit) Obs. Calc. Diff. Bragg position LuFe4Ge2 2θ 0 1 2 3 4 5 6 7 0.1 0.15 0.2 LuFe4Ge2 T = 2 K ρ/ρ300K B (T) (b) -30 -20 -10 0 0.01 0.50 1.67 0.98 1.86 1.48 2.00 1.59 2.35 LuFe4Ge2 T = 2 K MR (%) p (GPa) (a) 0 1 2 3 4 5 6 7 -2 0 2 4 MR (%) B (T) FIG. 9. (a) MR(B) of LuFe4Ge2 measured at T = 2 K for differ- ent pressures. The inset shows an enlarged view of the low-pressure curves. (b) Normalized resistivity ρ/ρ300 K as a function of ﬁeld at 2 K for several pressures. 0 1 2 3 4 5 6 7 0.1 0.15 0.2 LuFe4Ge2 T = 2 K ρ/ρ300K B (T) (b) -30 -20 -10 0 0.01 0.50 1.67 0.98 1.86 1.48 2.00 1.59 2.35 LuFe4Ge2 T = 2 K MR (%) p (GPa) (a) 0 1 2 3 4 5 6 7 -2 0 2 4 MR (%) B (T) Intensity (arb. unit) FIG. 8. Powder XRD pattern of LuFe4Ge2 obtained at ambient pressure and room temperature. The solid red line is a Rietveld reﬁned ﬁt of the experimental data. The Bragg positions are indicated by green vertical bars, and the bottom solid blue line indicates the difference between the experimental and the calculated intensities. ACKNOWLEDGMENTS DC magnetic susceptibility was measured using a su- perconducting quantum interference device magnetometer (magnetic properties measurement system, Quantum Design). The heat capacity at ambient pressure was recorded by a thermal-relaxation method using a physical property mea- surement system (Quantum Design). Electrical-transport, AC magnetic susceptibility, and AC-calorimetry measurements under hydrostatic pressure were performed using a double- layered piston-cylinder-type pressure cell with silicon oil as the pressure-transmitting medium. The pressure inside the This work was partly supported by Deutsche Forschungs- gemeinschaft (DFG) through Research Training Group GRK 1621. U. Nitzsche is acknowledged for technical support for DFT calculations. R.D.d.R. acknowledges ﬁnancial support from the Sao Paulo Research Foundation (FAPESP) (Grant No. 2018/00823-0) and from the Max Planck Society under the auspices of the Max Planck Partner Group R. D. dos Reis of the MPI for Chemical Physics of Solids, Dresden, Germany. 125136-7 PHYSICAL REVIEW B 107, 125136 (2023) M. O. AJEESH et al. 10 20 30 40 50 60 70 80 90 Intensity (arb. unit) Obs. Calc. Diff. Bragg position LuFe4Ge2 2θ FIG. 8. Powder XRD pattern of LuFe4Ge2 obtained at ambient pressure and room temperature. The solid red line is a Rietveld reﬁned ﬁt of the experimental data. The Bragg positions are indicated by green vertical bars, and the bottom solid blue line indicates the difference between the experimental and the calculated intensities. APPENDIX A: POWDER X-RAY DIFFRACTION AT AMBIENT PRESSURE The powder x ray diffraction pattern of LuFe Ge obtained 0 1 2 3 4 5 6 7 0.1 0.15 0.2 LuFe4Ge2 T = 2 K ρ/ρ300K B (T) (b) -30 -20 -10 0 0.01 0.50 1.67 0.98 1.86 1.48 2.00 1.59 2.35 LuFe4Ge2 T = 2 K MR (%) p (GPa) (a) 0 1 2 3 4 5 6 7 -2 0 2 4 MR (%) B (T) FIG. 9. (a) MR(B) of LuFe4Ge2 measured at T = 2 K for differ- ent pressures. The inset shows an enlarged view of the low-pressure curves. (b) Normalized resistivity ρ/ρ300 K as a function of ﬁeld at 10 20 30 40 50 60 70 80 90 Intensity (arb. unit) Obs. Calc. Diff. Bragg position LuFe4Ge2 2θ FIG. 8. Powder XRD pattern of LuFe4Ge2 obtained at ambient pressure and room temperature. The solid red line is a Rietveld reﬁned ﬁt of the experimental data. APPENDIX A: POWDER X-RAY DIFFRACTION AT AMBIENT PRESSURE FIG. 9. (a) MR(B) of LuFe4Ge2 measured at T = 2 K for differ- ent pressures. The inset shows an enlarged view of the low-pressure curves. (b) Normalized resistivity ρ/ρ300 K as a function of ﬁeld at 2 K for several pressures. The powder x-ray diffraction pattern of LuFe4Ge2 obtained at ambient pressure and room temperature is presented in Fig. 8. A Rietveld reﬁnement of the experimental data yielded structural parameters that are in excellent agreement with pre- viously reported values [14]. The lack of impurity peaks in the diffractogram within the resolution of the experimental data and the goodness of ﬁt of the Rietveld reﬁnement (χ2 = 1.54), reﬂects the improved quality of the polycrystalline material. observed at higher pressures with MR reaching −35% at 7 T for p = 2.35 GPa. We note that high-ﬁeld magnetization measurements on LuFe4Ge2 at ambient pressure revealed a weak metamagnetic transition at an applied magnetic ﬁeld of 47 T, yet without the tendency of saturation of the magnetization [16]. Such a metamagnetic transition could be related to a spin reorientation under the inﬂuence of applied magnetic ﬁeld. This is also corroborated by the M¨ossbauer measurements at high pressure and a magnetic ﬁeld of 6 T where some part of the Fe moments seem to reorient in the direction of the applied ﬁeld. It is also interesting to note that the normalized resistivity ρ/ρ300 K at higher magnetic ﬁelds appears to have similar values for all applied pressures as displayed in Fig. 9(b). This suggests that the ﬁeld-polarized phase could be the same in the entire pressure range, once again indicating the close similarity between the AFM1 and the AFM2 phases. observed at higher pressures with MR reaching −35% at 7 T for p = 2.35 GPa. We note that high-ﬁeld magnetization measurements on LuFe4Ge2 at ambient pressure revealed a weak metamagnetic transition at an applied magnetic ﬁeld of 47 T, yet without the tendency of saturation of the magnetization [16]. Such a metamagnetic transition could be related to a spin reorientation under the inﬂuence of applied magnetic ﬁeld. This is also corroborated by the M¨ossbauer measurements at high pressure and a magnetic ﬁeld of 6 T where some part of the Fe moments seem to reorient in the direction of the applied ﬁeld. [8] P. Dai, Antiferromagnetic order and spin dynamics in iron- based superconductors, Rev. Mod. Phys. 87, 855 (2015). APPENDIX B: FIELD-INDUCED METAMAGNETIC TRANSITION The transverse magnetoresistance MR(B) = [ρ(B) − ρ(0)]/ρ(0) of LuFe4Ge2 as a function of magnetic ﬁeld measured at different pressures at T = 2 K is shown in Fig. 9(a). MR(B) presents evidence for a ﬁeld-induced metamagnetic transition under pressure; a tiny kink in MR(B) at about 5 T at pressures starting from 1.5 GPa [see the inset of Fig. 9(a)]. This feature becomes much pronounced upon increasing pressure and continuously shifts to lower ﬁelds. Eventually, a strong steplike decrease in the MR is [1] N. D. Mathur, F. M. Grosche, S. R. Julian, I. R. Walker, D. M. Freye, R. K. W. Haselwimmer, and G. G. Lonzarich, Magneti- cally mediated superconductivity in heavy fermion compounds, Nature (London) 394, 39 (1998). [5] B. Keimer, S. Kivelson, M. Norman, S. Uchida, and J. Zaanen, From quantum matter to high-temperature superconductivity in copper oxides, Nature (London) 518, 179 (2015). [6] T. Gruner, D. Jang, Z. Huesges, R. Cardoso-Gil, G. H. Fecher, M. M. Koza, O. Stockert, A. P. Mackenzie, M. Brando, and C. Geibel, Charge density wave quantum critical point with strong enhancement of superconductivity, Nat. Phys. 13, 967 (2017). [2] P. Gegenwart, Q. Si, and F. Steglich, Quantum criticality in heavy-fermion metals, Nat. Phys. 4, 186 (2008). [3] T. Shibauchi, A. Carrington, and Y. Matsuda, A quantum crit- ical point lying beneath the superconducting dome in iron pnictides, Annu. Rev. Condens. Matter Phys. 5, 113 (2014). [7] A. Chubukov, Pairing mechanism in Fe-based superconductors, Annu. Rev. Condens. Matter Phys. 3, 57 (2012). [4] D. J. Scalapino, A common thread: The pairing interaction for unconventional superconductors, Rev. Mod. Phys. 84, 1383 (2012). Annu. Rev. Condens. Matter Phys. 3, 57 (2012). [8] P. Dai, Antiferromagnetic order and spin dynamics in iron- based superconductors, Rev. Mod. Phys. 87, 855 (2015). 125136-8 125136-8 PHYSICAL REVIEW B 107, 125136 (2023) INTERPLAY OF STRUCTURE AND MAGNETISM IN … [18] R. Khasanov, Z. Guguchia, A. Maisuradze, D. Andreica, M. Elender, A. Raselli, Z. Shermadini, T. Goko, F. Knecht, E. Morenzoni, and A. Amato, High pressure research using muons at the Paul Scherrer Institute, High Press. Res. 36, 140 (2016). [9] D. S. Inosov, Spin ﬂuctuations in iron pnictides and chalco- genides: From antiferromagnetism to superconductivity, C. R. Phys. 17, 60 (2016). [10] H.-H. Kuo, J.-H. Chu, J. C. Palmstrom, S. A. Kivelson, and I. R. APPENDIX B: FIELD-INDUCED METAMAGNETIC TRANSITION Fisher, Ubiquitous signatures of nematic quantum criticality in optimally doped Fe-based superconductors, Science 352, 958 (2016). [19] R. Khasanov, Perspective on muon-spin rotation/relaxation under hydrostatic pressure, J. Appl. Phys. 132, 190903 (2022). [11] Q. Wang, L. Fanfarillo, and A. E. Böhmer, Editorial: Nematic- ity in iron-based superconductors, Front. Phys. 10, 1038127 (2022). [20] V. Potapkin, A. I. Chumakov, G. V. Smirnov, J.-P. Celse, R. Rüffer, C. McCammon, and L. Dubrovinsky, The 57Fe syn- chrotron Mössbauer source at the ESRF, J. Synchrotron Radiat. 19, 559 (2012). [12] Y. P. Yarmoluk, L. A. Lysenko, and E. I. Gladyshevski, Zirco- nium iron silicide (ZrFe4Si2) structure as a new structural type of ternary transition metal silicides, Dopov. Nac. akad. nauk Ukr. RSR, Series A 37, 279 (1975). [21] W. Bi, J. Zhao, J.-F. Lin, Q. Jia, M. Y. Hu, C. Jin, R. Ferry, W. Yang, V. Struzhkin, and E. E. Alp, Nuclear resonant in- elastic X-ray scattering at high pressure and low temperature, J. Synchrotron Radiat. 22, 760 (2015). [13] M. O. Ajeesh, K. Weber, C. Geibel, and M. Nicklas, Putative quantum critical point in the itinerant magnet ZrFe4Si2 with a frustrated quasi-one-dimensional structure, Phys. Rev. B 102, 184403 (2020). [22] J. Y. Zhao, W. Bi, S. Sinogeikin, M. Y. Hu, E. E. Alp, X. C. Wang, C. Q. Jin, and J. F. Lin, A compact membrane-driven diamond anvil cell and cryostat system for nuclear resonant scattering at high pressure and low temperature, Rev. Sci. Instrum. 88, 125109 (2017). [14] P. Schobinger-Papamantellos, K. H. J. Buschow, and J. Rodríguez-Carvajal, Magnetoelastic phase transitions in the LuFe4Ge2 and YFe4Si2 compounds: A neutron diffraction study, J. Magn. Magn. Mater. 324, 3709 (2012). [23] F. A. Lima, M. E. Saleta, R. J. S. Pagliuca, M. A. Eleotério, R. D. Reis, J. Fonseca Júnior, B. Meyer, E. M. Bittar, N. M. Souza-Neto, and E. Granado, XDS: A ﬂexible beamline for X-ray diffraction and spectroscopy at the Brazilian synchrotron, J. Synchrotron Radiat. 23, 1538 (2016). [15] J. Lashley, M. Hundley, A. Migliori, J. Sarrao, P. Pagliuso, T. Darling, M. Jaime, J. Cooley, W. Hults, L. Morales, D. Thoma, J. Smith, J. Boerio-Goates, B. Woodﬁeld, G. Stewart, R. Fisher, and N. Phillips, Critical examination of heat capacity measurements made on a Quantum Design physical property measurement system, Cryogenics 43, 369 (2003). J. Synchrotron Radiat. 23, 1538 (2016). [24] G. Kresse and J. APPENDIX B: FIELD-INDUCED METAMAGNETIC TRANSITION Furthmüller, Efﬁcient iterative schemes for ab initio total-energy calculations using a plane-wave basis set, Phys. Rev. B 54, 11169 (1996). [16] K. Weber, Ph.D. thesis, Intermetallic 3d systems close to a magnetic instability: New unusual cases, Technical University of Dresden, Dresden, Germany, 2017. [25] J. P. Perdew and Y. Wang, Accurate and simple analytic repre- sentation of the electron-gas correlation energy, Phys. Rev. B 45, 13244 (1992). [17] E. Gati, G. Drachuck, L. Xiang, L.-L. Wang, S. L. Bud’ko, and P. C. Canﬁeld, Use of Cernox thermometers in AC speciﬁc heat measurements under pressure, Rev. Sci. Instrum. 90, 023911 (2019). [26] J. P. Perdew, K. Burke, and M. Ernzerhof, Generalized Gra- dient Approximation Made Simple, Phys. Rev. Lett. 77, 3865 (1996). 125136-9 125136-9 125136-9
https://openalex.org/W3157651569	https://www.frontiersin.org/articles/10.3389/fmars.2021.674014/pdf	English	null	Scale-Specific Patterns of the Production of the Charophyte Chara aspera in the Brackish Baltic Sea: Linking Individual and Community Production and Biomass Growth	Frontiers in marine science	2,021	cc-by	10,288	Edited by: Iñigo Muxika, Technological Center Expert in Marine and Food Innovation (AZTI), Spain Edited by: Iñigo Muxika, Technological Center Expert in Marine and Food Innovation (AZTI), Spain Reviewed by: Joao M. Neto, University of Coimbra, Portugal Irmgard Blindow, University of Greifswald, Germany Correspondence: Jonne Kotta jonne@sea.ee Reviewed by: Joao M. Neto, University of Coimbra, Portugal Irmgard Blindow, University of Greifswald, Germany Correspondence: Jonne Kotta jonne@sea.ee Specialty section: This article was submitted to Marine Ecosystem Ecology, a section of the journal Frontiers in Marine Science Keywords: benthic ecology, macroalgae, photosynthesis, primary production, environmental forcing, seasonal variability Received: 28 February 2021 Accepted: 13 April 2021 Published: 30 April 2021 Scale-Speciﬁc Patterns of the Production of the Charophyte Chara aspera in the Brackish Baltic Sea: Linking Individual and Community Production and Biomass Growth Jonne Kotta*, Kaire Torn, Tiina Paalme, Merli Rätsep, Kaire Kaljurand, Martin Teeveer and Ilmar Kotta Estonian Marine Institute, University of Tartu, Tallinn, Estonia An understanding of the nature of scale-dependence in macroalgal production and to quantify how primary production is incorporated into autotrophic biomass requires an assessment of primary production at multiple scales. In this study we experimentally estimated seasonal variability in individual photosynthesis, community production and growth in biomass of the charophyte Chara aspera in the Baltic Sea together with the key environmental variables known to regulate the production of macroalgae. The experiments suggested that the production and growth of C. aspera was deﬁned largely by light and temperature at all studied scales. However, the algal production at the same light levels was systematically lower at the community than an individual level, suggesting the importance of shelf-shading and below-ground processes in natural plant assemblages. Moreover, the observation scale also deﬁned response types between the environment, plant production and growth. This implies that the patterns of variability of the production and growth of macrophytes should always be interpreted in the context of scale and any multiscale model development should involve experimental validation at all important scales. ORIGINAL RESEARCH published: 30 April 2021 doi: 10.3389/fmars.2021.674014 Citation: In this respect, laboratory experiments have shown species-speciﬁc responses of macroalgae to environmental change in which macroalgal growth may increase, decrease or remain unchanged with increasing rates of plant photosynthesis (de los Santos et al., 2000; Israel and Hophy, 2002; Kim et al., 2016). Thus, understanding the nature of scale-dependence in macroalgal production can help to elucidate some of the possible controls by which primary production is incorporated into autotrophic biomass. Despite this importance, knowledge of such controls is limited as few studies have been undertaken to date that quantify the patterns of macroalgal production at diﬀerent scales with most being performed under laboratory conditions (e.g., de los Santos et al., 2000; Israel and Hophy, 2002; Binzer and Middelboe, 2005; Pärnoja et al., 2014; Kim et al., 2016). Due to artiﬁcially induced conditions, as well as the use of separate phytoelements (thallus pieces, leaves), however, laboratory incubations often fail to characterize primary production to that which occurs in nature (Binzer and Middelboe, 2005; Binzer et al., 2006). for realized photosynthesis along vertical light gradient within macroalgal assemblages (Binzer et al., 2006; Tait and Schiel, 2010). The eﬀect is stronger with increasing algal densities (Kotta et al., 2008; Paalme et al., 2013; Pärnoja et al., 2014) and in extreme cases, e.g., within giant kelp forests, only a fraction of light above algal canopy reaches the basal subcanopy species (Reed and Foster, 1984). This has large but mostly unknown aﬀects on photosynthesis and growth of macroalgal assemblages. Even though algal photosynthesis is saturated in the upper layers of algal canopies with increasing irradiance, community photosynthesis will remain unsaturated because photosynthetic tissue in the lower layers of algal canopy has an unused photosynthetic potential (Binzer and Middelboe, 2005; Tait and Schiel, 2010). Nutrient availability strongly aﬀects the production of macroalgae but the responses are mostly taxon-speciﬁc and highly inﬂuenced by the local light climate (Raven and Hurd, 2012). Importantly, macroalgae can store nutrients in their tissue in order to overcome temporal disparity between nutrient availability and optimal growth season. For example, the marine Laminariales grow fastest during summer when irradiance is the highest, but when nutrients are depleted above the thermocline. During the peak growth the algae use their nitrogen and phosphorus reserves that were stored during winter when the nutrient availability is greater but light levels are lower (Chapman and Craigie, 1977; Lüning, 1990). Citation: Primary production is important to all ecosystems as it supplies energy to other living organisms, thereby forming the foundation of food webs. Macroalgae are the dominant primary producers in coastal temperate marine ecosystems, representing some of the highest recorded per-area basis production values (Field et al., 1998; Gattuso et al., 2006). As macroalgae contribute signiﬁcantly to total global system carbon production (Hill et al., 2015) it is important to understand better how macroalgal primary production links to growth in biomass and how diﬀerent environmental factors aﬀect these patterns. Kotta J, Torn K, Paalme T, Rätsep M, Kaljurand K, Teeveer M and Kotta I (2021) Scale-Speciﬁc Patterns of the Production of the Charophyte Chara aspera in the Brackish Baltic Sea: Linking Individual and Community Production and Biomass Growth. Front Mar Sci 8:674014 Traditionally, the primary production of macroalgae has been quantiﬁed using detached pieces of algal thally or whole plants (Littler and Littler, 1980; Gómez et al., 1997) and more Front. Mar. Sci. 8:674014. doi: 10.3389/fmars.2021.674014 April 2021 \| Volume 8 \| Article 674014 1 Frontiers in Marine Science \| www.frontiersin.org Scale-Speciﬁc Algal Production Kotta et al. recently, the production of macroalgae has been measured directly in natural assemblages (Binzer and Middelboe, 2005; Tait and Schiel, 2010). While the ﬁrst approach enables quantiﬁcation of accurate response functions between diﬀerent environmental variables and primary production (e.g., irradiance–net photosynthesis curves), the latter technique allows the scaling up of primary production in natural conditions. The estimates of primary production are highly dependent on the scale of observation with photosynthetic production being higher at individual than at community levels (Binzer et al., 2006; Pärnoja et al., 2014). Likewise, scale aﬀects how macroalgae respond to changes in ambient conditions. The photosynthesis of individual plants often shows a typical light saturation curve (Lobban et al., 1985), whereas macroalgal photosynthesis in assemblages increases in a roughly linear fashion up to maximum levels of natural irradiance (Middleboe and Binzer, 2004). Changes in biomass integrate the net photosynthetic ﬁxation of inorganic carbon into autotrophic biomass as well as the processes of natural loss of biomass and thereby permits estimation of large-scale variation in realized production over time. Expectedly, when compared to the patterns of photosynthesis, another set of combinations of environmental factors can inﬂuence change in macroalgal biomass or the same relationships can have very diﬀerent functional shapes. Citation: In addition, the Charales can eﬀectively transport nutrients from rhizoids (in sediment with high N and P but no light) to shoots (in water with low N and P but high light irradiance) (Vermeer et al., 2003). However, when nutrients are plentiful in the water column and light levels are suﬃcient to support photosynthesis, direct nutrient uptake from the water through the outer cells of the branchlets and stems is expected (Raven, 2003). Temperature controls primary productivity of macroalgae to a lesser extent than light and nutrients unless the temperature is above or below the tolerances of the species. Within tolerance limits, macroalgae exhibit relative uniform responses to temperature change (Raven and Smith, 1978; Wiencke and tom Dieck, 1990). The species-speciﬁc temperature demands for survival and production relate to diﬀerent types of temperature stress the macroalgal species have locally adapted (Breeman and Pakker, 1994). Controlled experimental trials further demonstrate the high plasticity of response of macroalgae to speciﬁc limiting temperature conditions in which a long-term acclimation of macroalgae at extreme temperatures signiﬁcantly relaxes this temperature limitation (Nejrup et al., 2013). Thus, estimations of primary production in natural assemblages are increasingly relevant given the growing need for obtaining realistic response patterns to a complex of environmental factors. Primary production is controlled by both abiotic and biotic factors (Field et al., 1998; Hauxwell et al., 2003) but abiotic constraints are often dominant in natural environments (Hill et al., 2015). The availability of light and nutrients plays a crucial role in regulating primary production of macroalgae (Field et al., 1998; Hauxwell et al., 2003; Binzer et al., 2006). Light limitation may arise from the amount of irradiance arriving at the sea surface, the optical characteristics of water and self-shading within algal assemblages. The ﬁrst two processes vary over multiple temporal and spatial scales and deﬁne the availability of light ﬁeld above underwater canopies and the maximum photosynthetic rates of macroalgae (Kirk, 1994; Anthony et al., 2004). The third process, shelf-shading, is one of the most critical biological limitations because it sets an actual threshold Charophytes are found in diverse aquatic habitats, so this group consists of both early colonizers of newly-formed water bodies as well as species preferring more stable environmental conditions and habitats (Blindow and van de Weyer, 2016). Frontiers in Marine Science \| www.frontiersin.org Sampling of Environmental Variables and Communities of C. aspera The study was carried out at depths of 0.9–1.3 m. The study design combined a traditional sampling of water quality variables and benthic communities with the results from the latest oceanographic technology to achieve an accurate environmental assessment of the Rame Bay area. In addition to these observations, the production of C. aspera was measured experimentally at three organizational scales (for more details see below). C. aspera can be visually determined from other charophytes in the study area as it has clearly distinguishable globular white bulbils on the rhizoids. This feature enabled to run the production of C. aspera without a need for prior determination in the laboratory. To address the importance of scale on the factors regulating macroalgal production, we quantiﬁed experimentally the seasonal variability in individual photosynthesis, community production and biomass growth of the charophyte C. aspera in the brackish Baltic Sea. We then analyzed how seasonal change in the assemblage of environmental variables manifested in the production of macroalgae at these diﬀerent scales. We hypothesized that C. aspera with its well-known broad environmental tolerances can maintain high photosynthetic and growth potential across wide gradients in light, temperature, and water nutrients. We also expected that the relationships between the environment and photosynthesis when assessed on individual plants are strong and simple. At the community level, however, charophytes have an ability to adjust photosynthesis and growth to a diﬀerent set of environmental conditions. Moreover, below-ground processes may occasionally contribute to the oxygen ﬂuxes. Therefore relationships between light, nutrient concentration, temperature, community photosynthesis and growth are weaker and more complex compared to individual plants. y The oceanographic instruments were deployed 20 cm above the seabed to record short-term variability of temperature, light availability, and water velocity in the study site from March 2014 to March 2015. These results provided information on the environment around the experimental site during and in between samplings. Temperature was measured with a General Oceanics thermologger. Light availability was recorded by an Odyssey photosynthetically active radiation (PAR) logger. Temperature and PAR was logged at 60 min intervals. Water velocity was recorded by a JFE Advantech Logger version of the INFINITY 2-D electro-magnetic current meter AEMD- USB. During deployment the instrument recorded current speed once per minute. Water level was measured using a JFE Advantech Logger version of the INFINITY-WH AWH- USB wave height meter. During deployment the instrument recorded pressure once per second. Citation: The lush charophyte meadows play a underpinning role by providing food and/or refuge to zooplankton (Puche et al., 2020), benthic invertebrates (Kotta et al., 2004), ﬁsh and ﬁsh larvae (Hargeby et al., 2005), microbes (Kataržyt˙e et al., 2017), April 2021 \| Volume 8 \| Article 674014 2 Scale-Speciﬁc Algal Production Kotta et al. hibernate both as detached mats and bulbils in the study area (current study, Torn et al., 2010). and waterfowl (Schmieder et al., 2006). They also reduce water turbidity by stabilizing the sediment with rhizoids (Nõges et al., 2003), immobilizing nutrients (Kufel and Kufel, 2002), and inhibiting phytoplankton biomass (Hilt, 2015). Charophytes are most common in freshwater habitats but are also found in brackish water. The Baltic Sea is one of the best-studied brackish regions with respect to charophytes. In this marginal environment, the shallow low salinity embayments are often dominated by charophytes (Schubert and Blindow, 2003). The most common and widespread species in such habitats is Chara aspera Willd., a species characterized by a broad environmental niche space (Herkül et al., 2018). Sampling of Environmental Variables and Communities of C. aspera Nutrient concentrations in water were measured triplicate once a month. Immediately after collection, samples were preserved and brought to the laboratory in an insulated ice box. We used standard ISO procedures for measuring total N and total P (ISO 11905 and ISO 15681). We did not use neither ﬁltering nor fractioning for our samples. The analyses were made using a Skalar San + + autoanalyzer. Frontiers in Marine Science \| www.frontiersin.org Description of Study Area The study was carried out in Rame Bay from March 2014 to March 2015. Rame Bay is a shallow and semi-enclosed bay in the northeastern Baltic Sea. Depending on the prevailing wind direction, the bay receives eutrophicated waters from the Gulf of Riga or less eutrophied waters from the northern Baltic Proper (Suursaar et al., 1998). The maximum depth of the bay is 1.5 m. The bottom is composed of ﬁne sand and a thick layer of ﬁne silt (Torn and Martin, 2003). Salinity varies between 2 and 5.6 psu with lower values during heavy rainfall. Rapidly increasing water temperatures during spring, high temperatures in summer, and a long period of subzero temperatures in winter are typical. The ice-cover of shallow bays in the region lasts 4–5 months on average (Jevrejeva and Leppäranta, 2002). The charophyte community was sampled for coverage and biomass once in a month. Total coverage of all macrophytes and percentage cover of all visually distinguishable species were recorded by SCUBA diver during each sampling. Biomass samples were collected in three replicates by SCUBA diving using a 20 × 20 cm frame. Samples were stored deep frozen (–18◦C) until analysis. In the laboratory all species were identiﬁed to the species level. The dry weight of species was obtained after drying the individuals at 60◦C for 2 weeks (i.e., the time needed to achieve the same weight for two consecutive measurements). Growth in biomass was measured as the diﬀerence in total biomass of the plant community between two consecutive sampling months. The chlorophyll-a concentration was measured spectrophotometrically by Libra S32 after extraction of chlorophyll from algal tissue in ethanol. To determine the phosphorus content of the algae, the material was mineralized by heating with HNO3 and HClO4 (Miller, 1998). Most of Rame Bay is covered by a rich charophyte community. Five species of charophytes are found in the bay with C. aspera being the most widespread species. The seasonal dynamic of charophytes follows general growth patterns of the northern part of the temperate zone, which is characterized by the highest growth rates at the beginning of summer and degradation of the community in autumn (Torn et al., 2006, 2010). C. aspera April 2021 \| Volume 8 \| Article 674014 3 Scale-Speciﬁc Algal Production Kotta et al. technique (BRT). Description of Study Area The Dumas method was used to measure the nitrogen and carbon content of the algae (Buckee, 1994). The content of the elements in the resulting solution was determined by MP-AES spectrometry. The Dumas method was used to measure the nitrogen and carbon content of the algae (Buckee, 1994). Seasonal Changes in Abiotic Environmental Conditions Water temperature and light availability above the seabed followed the same seasonal pattern with higher values measured in July and lower values in winter (Figure 1). The thickest ice cover (26 cm) was recorded in March 2014. During the winter 2014/2015 ice formation alternated repeatedly with melting. Winter months were characterized by the highest water velocity and nitrogen concentration. The phosphorus content in the water showed no clear seasonal pattern with the highest values observed in spring Production in Algal Communities The production of C. aspera communities was measured in triplicate from transparent plexiglas chambers holding 29 l of water with a surface area 850 cm2. The chambers were placed randomly on the seaﬂoor over the homogeneous monospeciﬁc stand of C. aspera. A rubber seal assured an airtight seal between the chamber and the surrounding environment. Oxygen concentration in the chamber was measured every minute using a calibrated Optode-type oxygen sensor (Aanderaa Instruments) connected to a data logger (data recorder by JFE Advantech Co., Ltd.). This instrument also provides data on water temperature. Changes in dissolved oxygen averaged over minute intervals were used as a proxy of community (algal assemblage + sediment) net production. During deployment irradiance above the charophytes was measured every minute using a calibrated spherical quantum sensor connected to a data logger (ultra-miniature logger for light intensity by JFE Advantech Co., Ltd.). Each production experiment lasted 3 h on average. After deployment, macroalgae within incubation chambers were harvested and stored in a deep freezer at −20◦C. The subsequent sorting and determination of species were performed in the laboratory using a stereomicroscope. The dry weight of each macroalgal community was obtained after drying the individuals at 60◦C for 2 weeks. Combining this information with the oxygen ﬂux measurements described above enabled us to express net primary production in mg O2 g dry weight per macrophyte min−1. Photosynthetic Production of Individual Plants In each study visit the net photosynthetic rate of C. aspera was measured in triplicate from 600 ml glass bottles ﬁlled with seawater from the study site. Bottles with about 0.5 g dry weight of C. aspera and without algae (controls) were incubated in situ on special frames at a depth of 1 meter around midday. The hourly net production rates (mg O2 g dry weight −1 h−1) were calculated from the diﬀerences in oxygen concentrations, measured over the incubation period (1–3 h). Oxygen concentration and water temperature in incubation bottles was measured with OptiOx optical dissolved oxygen sensor connected to SevenGo DO meter (Mettler Toledo). The dry weight of C. aspera was determined after drying at 60◦C for 48 h. During the experiment the photosynthetically active radiation (PAR; µmol m−2 s−1) at the incubation depth was measured using Li 188B (Li-Cor Inc.) quantum meter. Description of Study Area BRT models are capable of handling diﬀerent types of predictor variables and their predictive performance is superior to most traditional modeling methods [see e.g., comparisons with GLM, GAM and multivariate adaptive regression splines, Elith et al. (2006) and Leathwick et al. (2006)]. Overﬁtting is often regarded as a problem in statistical modeling, but can be overcome by using independent data sets. The BRT modeling iteratively develops a large ensemble of small regression trees constructed from random subsets of the data. Each successive tree predicts the residuals from the previous tree to gradually boost the predictive performance of the overall model (Elith et al., 2008). Important parameters in building BRT models are the learning rate and tree complexity. The learning rate determines the contribution of each tree to the growing model and tree complexity deﬁnes the depth of interactions allowed in a model. A tree complexity of 1 assesses only main eﬀects; A tree complexity >1 includes interactions. Diﬀerent combinations of these parameters may yield variable predictive performance but generally a lower learning rate and inclusion of interactions gives better results (Elith et al., 2008). In the current study, the model learning rate was kept at 0.001 and tree complexity at 5. In order to avoid potential problems of overﬁtting, unimportant variables were removed using a simplify tool. In order to eliminate non-informative variables, the simplify tool progressively simpliﬁes the model, then re-ﬁts the model and sequentially repeats the process until a stopping criterion is reached. Such simpliﬁcation is most useful for small data sets where redundant predictors may degrade performance by increasing variance. Model performance was evaluated using the cross-validation statistics calculated during model ﬁtting (Hastie et al., 2009). The BRT modeling was done in R using the gbm package (Elith et al., 2008). Standard errors for the predictions and pointwise standard errors for the partial dependence curves, produced by R package “pdp” (Greenwell, 2017), were estimated using bootstrap (100 replications). Multicollinearity can be an issue with BRT modeling when assessing if and when environmental variables are of ecological interest. Thus, prior to modeling, the Pearson correlation analysis between all environmental variables was calculated in order to avoid including highly correlated variables into the model. The correlation analysis showed that most variables were only weakly intercorrelated (r < 0.5). The content of the elements in the resulting solution was determined by MP-AES spectrometry. Seasonal Changes in the Community of C. aspera Charophytes had high biomass in summer and autumn, whereas the algal cover was stably high throughout the year including winter (Figure 2). C. aspera overwintered as a sparse and ﬂat biomass residue. Small fresh shoots were observed only in April and the full community established by July. The rapid decay in biomass occurred during the formation of ice cover. Nutrient content (nitrogen and phosphorus) of C. aspera was high in decomposing algal material in winter and low in summer. The chlorophyll a content of C. aspera plants varied between months with no clear seasonal patterns (somewhat lower values in winter and higher values in summer). Statistical Analyses The contribution of diﬀerent environmental variables on the individual, community photosynthesis and growth in biomass of C. aspera was explored using the Boosted Regression Trees April 2021 \| Volume 8 \| Article 674014 Frontiers in Marine Science \| www.frontiersin.org 4 Scale-Speciﬁc Algal Production Kotta et al. FIGURE 1 \| Seasonal dynamics of abiotic environmental variables at near-bottom layer in the study area (means ± Standard Error). FIGURE 2 \| Seasonal dynamics in the biomass and cover of C. aspera community in the study area (means ± Standard Error). FIGURE 1 \| Seasonal dynamics of abiotic environmental variables at near-bottom layer in the study area (means ± Standard Error). GURE 1 \| Seasonal dynamics of abiotic environmental variables at near-bottom layer in the study area (means ± Standard Error). FIGURE 2 \| Seasonal dynamics in the biomass and cover of C. aspera community in the study area (means ± Standard Error). FIGURE 2 \| Seasonal dynamics in the biomass and cover of C. aspera community in the study area (means ± Standard Error). Seasonal Changes in the Individual and Community Production of C. aspera and December. August stood out among other summer months with the highest nutrient concentration and variability of water movement. Net production rates of C. aspera measured at an individual and community level followed similar seasonal patterns over the whole study period with maximum values observed in June and July and lowest values during winter months. The only exceptions were abnormally low community production values in May and August related to the presence of high amounts of organic debris on the sediment. Similarly, the biomass growth of C. aspera was high from June to September and low in other months (Figure 3). Frontiers in Marine Science \| www.frontiersin.org Relationships Between Environment, Individual, Community Production and Biomass Growth of C. aspera Boosted regression trees models on the production of C. aspera accounted for a signiﬁcant proportion of the variability at the studied scales with r2 values as follows: individual photosynthesis April 2021 \| Volume 8 \| Article 674014 Frontiers in Marine Science \| www.frontiersin.org 5 Scale-Speciﬁc Algal Production Kotta et al. 0.855, community production 0.854 and change in algal biomass 0.573 (Table 1). Temperature was the best overall predictor in the models of individual photosynthesis and change in algal biomass and the second-best predictor in the model of community production. However, the thresholds of algal production to increasing temperature varied among the studied scales. Light availability was another generic predictor of the production of C. aspera in all models. Like temperature, the complexity of (functional form) relationships between light and production did not change among models and the diﬀerences were due mostly to diﬀerent light thresholds in diﬀerent models. Among the studied environmental variables, nutrient concentration in water and chloroplast content in algal tissue were not signiﬁcant and the variables were not retained in the models. Photosynthesis in individual plants was related to temperature and light intensity with these variables accounting for nearly 100% of the predicted variability in the production of C. aspera at this scale. Increasing temperature increased the production of individual plants up to a threshold of 20◦C. Above this tipping point, variability in plant production was independent of the ambient water temperature. The production of individual plants had two clear phases: low production at light levels <500 µmol quanta m−2 s−1 and high production at light levels >600 µmol quanta m−2 s−1 with light saturation remaining at 550–600 µmol quanta m−2 s−1 (Figure 4). FIGURE 3 \| Seasonal dynamics in individual and community net production rates and biomass growth of C. aspera in the study area (means ± Standard Error). Extreme events (negative values) of community net production associated with the presence of high amounts of organic debris on the sediment are shown in red. A moving average smoother was used to visualize the trend in biomass gro th Like photosynthesis in individual plants, the production of benthic communities correlated positively with light intensity. However, the relationship was characterized by a smoother response and a higher light saturation point at 800 µmol quanta m−2 s−1. Importantly, community production decreased when the coverage of C. aspera attained >95%, indicating greater light limitation in denser algal communities. Relationships Between Environment, Individual, Community Production and Biomass Growth of C. aspera The relationship between temperature and community production exhibited two phases: low production at temperatures <13◦C and high production at temperatures >14◦C. BRT analysis excluded abnormally low community production values in May and August related to the presence of high amounts of organic debris on the sediment and high levels of belowground respiration. This suggests that when temperature exceeds 21◦C, the community net production is expected to decline sharply (Figure 5). Temperature, light climate and underwater currents were the key predictors of change in algal biomass accounting for >70% of the predicted variability in the biomass change. Importantly, high levels of growth were sustained at much lower light intensities, already above 100 µmol quanta m−2 s−1 compared to thresholds of 600 and 800 µmol quanta m−2 s−1 in the models of individual and community production. Similarly, high growth values were predicted only at velocities <1 cm s−1. Moreover, increasing TABLE 1 \| Summary statistics of the BRT models on relationships between environment, individual, community production and biomass growth of C. aspera. Variability explained Individual production Community production Community growth Total variation explained (r2) 0.855 0.854 0.573 Temperature 62 25 29 PAR 38 62 21 Coverage of Chara aspera 13 10 Water velocity 23 Water level 17 The contributions of environmental variables are relative contributions to the models’ total variable explained (%). TABLE 1 \| Summary statistics of the BRT models on relationships between environment, individual, community production and biomass growth of C. aspera. Summary statistics of the BRT models on relationships between FIGURE 3 \| Seasonal dynamics in individual and community net production rates and biomass growth of C. aspera in the study area (means ± Standard Error). Extreme events (negative values) of community net production associated with the presence of high amounts of organic debris on the sediment are shown in red. A moving average smoother was used to visualize the trend in biomass growth. FIGURE 3 \| Seasonal dynamics in individual and community net production rates and biomass growth of C. aspera in the study area (means ± Standard Error). Extreme events (negative values) of community net production associated with the presence of high amounts of organic debris on the sediment are shown in red. A moving average smoother was used to visualize the trend in biomass growth. FIGURE 3 \| Seasonal dynamics in individual and community net production rates and biomass growth of C. Relationships Between Environment, Individual, Community Production and Biomass Growth of C. aspera This FIGURE 5 \| Standardized functional-form relationships (± Standard Error) showing the effect of environmental variables on the community production of C. aspera, whilst all other variables are held at their means. The variables are ordered by their relative contribution in the BRT model (shown in%). Upward tickmarks on x-axis show the frequency distribution of data along this axis. See the section of methods for further information on environmental variables. FIGURE 4 \| Standardized functional-form relationships (± Standard Error) showing the effect of environmental variables on the net photosynthesis of individual plants of C. aspera, whilst all other variables are held at their means. The variables are ordered by their relative contribution in the BRT model (shown in%). Upward tickmarks on x-axis show the frequency distribution of data along this axis. See the section of methods for further information on environmental variables. water level was associated with reduced biomass of macroalgal communities (Figure 6). FIGURE 5 \| Standardized functional-form relationships (± Standard Error) showing the effect of environmental variables on the community production of C. aspera, whilst all other variables are held at their means. The variables are ordered by their relative contribution in the BRT model (shown in%). Upward tickmarks on x-axis show the frequency distribution of data along this axis. See the section of methods for further information on environmental variables. FIGURE 5 \| Standardized functional-form relationships (± Standard Error) showing the effect of environmental variables on the community production of C. aspera, whilst all other variables are held at their means. The variables are ordered by their relative contribution in the BRT model (shown in%). Upward tickmarks on x-axis show the frequency distribution of data along this axis. See the section of methods for further information on environmental variables. Relationships Between Environment, Individual, Community Production and Biomass Growth of C. aspera aspera in the study area (means ± Standard Error). Extreme events (negative values) of community net production associated with the presence of high amounts of organic debris on the sediment are shown in red. A moving average smoother was used to visualize the trend in biomass growth. April 2021 \| Volume 8 \| Article 674014 Frontiers in Marine Science \| www.frontiersin.org Frontiers in Marine Science \| www.frontiersin.org 6 Kotta et al. Scale-Speciﬁc Algal Production FIGURE 4 \| Standardized functional-form relationships (± Standard Error) showing the effect of environmental variables on the net photosynthesis of individual plants of C. aspera, whilst all other variables are held at their means. The variables are ordered by their relative contribution in the BRT model (shown in%). Upward tickmarks on x-axis show the frequency distribution of data along this axis. See the section of methods for further information on environmental variables. FIGURE 4 \| Standardized functional-form relationships (± Standard Error) showing the effect of environmental variables on the net photosynthesis of individual plants of C. aspera, whilst all other variables are held at their means. The variables are ordered by their relative contribution in the BRT model (shown in%). Upward tickmarks on x-axis show the frequency distribution of data along this axis. See the section of methods for further information on environmental variables. water level was associated with reduced biomass of macroalgal communities (Figure 6). DISCUSSION In this study we experimentally quantiﬁed seasonal variability in individual photosynthesis, community production and growth in biomass of the charophyte C. aspera in the Baltic Sea along with the key environmental variables that are known to determine the production of macroalgae. We found that temperature and light availability were the best overall predictors of the production FIGURE 5 \| Standardized functional-form relationships (± Standard Error) showing the effect of environmental variables on the community production of C. aspera, whilst all other variables are held at their means. The variables are ordered by their relative contribution in the BRT model (shown in%). Upward tickmarks on x-axis show the frequency distribution of data along this axis. See the section of methods for further information on environmental variables. of C. aspera, followed by macroalgal coverage, ﬂow velocity and water level. Importantly, the observed scale notably regulated the importance of diﬀerent predictors, as well as the relationships between the environment and plant responses in the models. Frontiers in Marine Science \| www.frontiersin.org DISCUSSION In this study we experimentally quantiﬁed seasonal variability in individual photosynthesis, community production and growth in biomass of the charophyte C. aspera in the Baltic Sea along with the key environmental variables that are known to determine the production of macroalgae. We found that temperature and light availability were the best overall predictors of the production In this study we experimentally quantiﬁed seasonal variability in individual photosynthesis, community production and growth in biomass of the charophyte C. aspera in the Baltic Sea along with the key environmental variables that are known to determine the production of macroalgae. We found that temperature and light availability were the best overall predictors of the production of C. aspera, followed by macroalgal coverage, ﬂow velocity and water level. Importantly, the observed scale notably regulated the importance of diﬀerent predictors, as well as the relationships between the environment and plant responses in the models. This April 2021 \| Volume 8 \| Article 674014 Frontiers in Marine Science \| www.frontiersin.org 7 Kotta et al. Scale-Speciﬁc Algal Production FIGURE 6 \| Standardized functional-form relationships (± Standard Error) showing the effect of environmental variables on growth in C. aspera biomass, whilst all other variables are held at their means. The variables are ordered by their relative contribution in the BRT model (shown in%). Upward tickmarks on x-axis show the frequency distribution of data along this axis. See the section of methods for further information on environmental variables. FIGURE 6 \| Standardized functional-form relationships (± Standard Error) showing the effect of environmental variables on growth in C. aspera biomass, whilst all other variables are held at their means. The variables are ordered by their relative contribution in the BRT model (shown in%). Upward tickmarks on x-axis show the frequency distribution of data along this axis. See the section of methods for further information on environmental variables. implies that although the physiological responses of C. aspera should always be interpreted in the context of scale, temperature and light availability predict a signiﬁcant proportion in the variability of charophyte production. saturation point of photosynthesis with increasing irradiance from 100 to 400 µmol quanta m−2 s−1 (Blindow et al., 2003). Similar responses of net photosynthesis along light gradients have been observed for C. aspera in other areas (Kovtun- Kante et al., 2014) and for other charophyte species, such as C. baltica and C. canescens (Küster et al., 2004). DISCUSSION Beside shelf- shading, sediment respiration is an additional explanation for the diﬀerences between individual and community photosynthesis. At high load of organic debris and/or elevated temperatures algal respiration and sediment decomposition may dominate over benthic production (Carvalho et al., 2005). Low light can limit the production potential of temperate macroalgae, particularly during low-light seasons or in turbid waters (Wallentinus, 1978; Littler, 1980); many studies have suggested a direct dependence of charophytes on light availability (Libbert and Walter, 1985; Blindow, 2000; Schneider et al., 2006; Schubert et al., 2018). However, due to high tolerances of C. aspera to low light environments (Blindow, 1992; Sand- Jensen et al., 2000), we hypothesized that the species could maintain positive photosynthesis and growth potential in a wide range of light conditions. In general, this held true for individual photosynthesis, community production and growth in biomass. However, individual photosynthesis was typically higher than community production under the same light conditions. This diﬀerence in photosynthesis can be partly attributed to lower light availability (light limitation) in a three-dimensional algal canopy than to individual algae exposed to ambient light levels. For the same reason, the photosynthesis-irradiance curve at community level was smoother than that of individual plants. Speciﬁcally, diﬀerent parts of the algal canopy are expected to receive diﬀerent amounts of light and therefore are expected to respond diﬀerentially. For example, the upper parts may often be oversaturated with light, whereas lower parts are undersaturated. This results in diﬀerent light adaptation (diﬀerent light saturation points) along the spatial light gradient and a smooth response of the algal community to light. This also applies for C. aspera, which is known to increase the light Diﬀerential responses of photosynthesis and growth to light irradiance is also related to the diﬀerent approaches used to calculate production at diﬀerent scales. Speciﬁcally, the photosynthesis of individual plants and communities is expressed as relative change per plant biomass, whereas biomass growth is reported on an absolute scale. In spring and early summer with abundant light, photosynthesis of individual plants and communities is enhanced. However, the absolute change in biomass is actually low because the plants themselves are small. Later, when the plants are taller, biomass increases are greater despite moderate light conditions and low photosynthesis of individual plants and communities. Frontiers in Marine Science \| www.frontiersin.org DISCUSSION This is happening because some inorganic carbon is ﬁxed into autotrophic biomass even at low photosynthesis levels between 100 and 500 µmol quanta m−2 s−1. In addition to light irradiance reaching the algal canopy, water turbidity aﬀects light availability and the production of charophytes in many habitats (Hellström, 1991; Schneider et al., 2006). Likewise, the sediments in Rame Bay are April 2021 \| Volume 8 \| Article 674014 8 Scale-Speciﬁc Algal Production Kotta et al. composed of ﬁne fractions (very ﬁne sand, silt, and clay) and even slow underwater currents (1 cm s−1) can bring silt and mud particles into the water column and thereby modulate the light environment as well as deposit sediment particles on plant thalli. Consequently, less light reaches plant photopigments and the rate of community photosynthesis and biomass growth decreases. This process is supported by experimental evidence in which long exposure to high levels of suspended sediment leads to cessation of growth and ultimately death of charophytes (Henricson et al., 2006). On the other hand, an experimental reduction of light availability to 25% of the natural irradiance does not induce a chronic compensatory reduction of the photosynthesis of charophyte communities (Kovtun-Kante et al., 2014). This suggests that charophytes recover easier from longer-term light reduction than sedimentation and may explain why water velocity (i.e., the intensity of resuspension) was the key predictor in the models of biomass growth. In addition to underwater currents, water level explained a signiﬁcant part of charophyte growth in biomass. The current charophyte meadows are located in very shallow water (1 m) and a decline in water level of only a few tens of centimeters is expected to improve light conditions in the algal canopy. control photosynthesis and growth of charophytes (Raven et al., 1979; Libbert and Walter, 1985; Torn et al., 2006). Due to such broad temperature tolerance, charophytes are expected to beneﬁt from contemporary climate change and broaden their distribution range in the Baltic Sea region (Torn et al., 2020). Finally, we expected that C. aspera could maintain high photosynthesis and growth potential across wide gradients in water nutrients. This hypothesis was fully supported as nutrient concentrations in water were not signiﬁcant in all models. Although charophytes are able to acquire nutrients from the water column (Raven, 2003), nutrient concentration in water has been shown to play a marginal eﬀect on the physiological state of charophytes (Blindow, 1988; Munsterhjelm, 2005). DISCUSSION Charophytes in our study area seem to rely on nutrients stored in their tissues as shown by the inverse relationship between macroalgal nutrient content and the photosynthesis of charophyte individuals (R2 = 0.15). Such strategy is widely used among long-living macroalgae in order to overcome temporal and/or spatial mismatch between nutrient availability and optimal growth season (Chapman and Craigie, 1977; Lüning, 1990; Raven, 2003). Prior to the study, we anticipated that models of photosynthesis in individual plants would involve fewer important environmental predictors than the models of photosynthesis and growth of charophyte communities. This prediction also held true. Although water temperature and light irradiance account for most of variability in the photosynthesis of individual plants, this share is less than in models of community photosynthesis and growth. The existence of a weak negative relationship between algal cover and community photosynthesis at the community level suggests that density dependence does play some role in photosynthesis regulation (Binzer and Middelboe, 2005; Tait and Schiel, 2010). Moreover, water properties (water level and ﬂow velocity) were important in regulating charophyte growth. However, as judged by the importance of diﬀerent predictors in the models, the eﬀects of macroalgal density and water properties were much weaker than light dependence of production and growth of C. aspera. It has been suggested that an increased production of photosynthetic pigments and chloroplasts supports more eﬃcient production of charophytes and helps plants to recover when light becomes limiting (Andrews et al., 1984; Küster et al., 2000; Marquardt and Schubert, 2009). Our study did not support this, as variability in photosynthesis and growth along light gradient was independent of the chloroplast content of charophyte tissue. The chloroplast content was measured during ﬁeldwork but due to lack of signiﬁcance, the variable was not retained in the models and therefore not reported. It has been also suggested that charophytes elongate quickly when light levels are low. This adaptation enables charophytes to reach the water surface where light is more plentiful (Andrews et al., 1984; Henricson et al., 2006). However, the shoot elongation in C. aspera is salinity-dependent and brackish water plants fail to elongate under impoverished light conditions (Blindow et al., 2003). Our study corroborated this as underwater light conditions were not inversely related to plant production throughout the light gradient. g p p g p The availability of diﬀerent forms of inorganic carbon may limit the photosynthesis of submerged vegetation including C. DISCUSSION aspera (van den Berg et al., 2002). Shallow water habitats of our study area are also characterized by a large amplitude of natural variability in pH and pCO2 with a daily range of pH recorded between 8 and 9 (Pajusalu et al., 2013). Among diﬀerent charophyte communities inhabiting our study area C. aspera is way less sensitive to inorganic carbon limitation compared to C. horrida and C. tomentosa (Pajusalu et al., 2015). Thus, as compared to light and temperature, the limitation of inorganic carbon is likely not a prominent stress to the studied community. Our study did not focus on interspeciﬁc interactions that potentially shape the photosynthesis and growth of C. aspera. A wealth of experimental studies has demonstrated that marine herbivores play key roles in regulating macroalgal communities (Hayward, 1988; Sala and Graham, 2002; Kotta and Witman, 2009). However, previous experiments suggest that herbivory or plant fragmentation induced by herbivores is not crucial in the studied charophyte communities (Kotta et al., 2004). Moreover, The availability of diﬀerent forms of inorganic carbon may limit the photosynthesis of submerged vegetation including C. aspera (van den Berg et al., 2002). Shallow water habitats of our study area are also characterized by a large amplitude of natural variability in pH and pCO2 with a daily range of pH recorded between 8 and 9 (Pajusalu et al., 2013). Among diﬀerent charophyte communities inhabiting our study area C. aspera is way less sensitive to inorganic carbon limitation compared to C. horrida and C. tomentosa (Pajusalu et al., 2015). Thus, as compared to light and temperature, the limitation of inorganic carbon is likely not a prominent stress to the studied community. We also hypothesized that C. aspera could maintain positive production and growth potential in a wide range of temperature conditions. This hypothesis held true for individual and community photosynthesis but not for growth in biomass. Interestingly, photosynthesis in individual plants increased with increasing temperature throughout the gradient, whereas this temperature dependence translated into relatively stable positive community photosynthesis with somewhat higher photosynthesis at temperatures >14◦C. Nevertheless, at temperatures >22◦C benthic production declined substantially due to the dominance of below-ground processes (e.g., sediment decomposition) and algal respiration over algal production (Carvalho et al., 2005; Torn et al., 2006). Thus, C. Frontiers in Marine Science \| www.frontiersin.org REFERENCES Buckee, G. K. (1994). Determination of total nitrogen in Barley, Malt and Beer by Kjeldahl procedures and the Dumas combustion method. Collaborative trial. Buckee, G. K. (1994). Determination of total nitrogen in Barley, Malt and Beer by Kjeldahl procedures and the Dumas combustion method. Collaborative trial. J. Inst. Brew. 100, 57–64. doi: 10.1002/jib.1994.100.2.57 Andrews, M., Box, R., McInroy, S., and Raven, J. A. (1984). Growth of Chara hispida. II. Shade adaptation. J. Ecol. 72, 885–895. doi: 10.2307/2259538 Carvalho, P., Thomaz, S. M., and Bini, L. M. (2005). Eﬀects of temperature on decomposition of a potential nuisance species: the submerged aquatic macrophyte Egeria najas Planchon (Hydrocharitaceae). Braz. J. Biol. 65, 51–60. doi: 10.1590/S1519-69842005000100008 Anthony, K., Ridd, P. V., Orpin, A. R., Larcombe, P., and Lough, J. (2004). Temporal variation of light availability in coastal benthic habitats: eﬀects of clouds, turbidity, and tides. Limnol. Oceanogr. 49, 2201–2211. doi: 10.4319/lo. 2004.49.6.2201 Chambers, P., and Prepas, E. E. (1990). Competition and coexistence in submerged aquatic plant communities: the eﬀects of species interactions versus abiotic factors. Freshw. Biol. 23, 541–550. doi: 10.1111/j.1365-2427.1990.tb00293.x Binzer, T., and Middelboe, A. L. (2005). From thallus to communities: scale eﬀects and photosynthetic performance in macroalgae communities. Mar. Ecol. Prog. Ser. 287, 65–75. doi: 10.3354/meps287065 Chapman, A. R. O., and Craigie, J. S. (1977). Seasonal growth in Laminaria longicruris: relations with dissolved inorganic nutrients and internal reserves of nitrogen. Mar. Biol. 40, 197–205. doi: 10.1007/bf00390875 Binzer, T., Sand-Jensen, K., and Middelboe, A. L. (2006). Community photosynthesis of aquatic macrophytes. Limnol. Oceanogr. 51, 2722–2733. doi: 10.4319/lo.2006.51.6.2722 de los Santos, C. B., Pérez-Lloréns, J. L., and Vergara, J. J. (2000). Photosynthesis and growth in macroalgae: linking functional-form and power-scaling approaches. Mar. Ecol. Prog. Ser. 377, 113–122. doi: 10.3354/meps07844 Blindow, I. (1988). Phosphorus toxicity in Chara. Aq. Bot. 32, 393–395. doi: 10. 1016/0304-3770(88)90110-6 Blindow, I. (1992). Decline of charophytes during eutrophication – comparison with angiosperms. Freshw. Biol. 28, 9–14. doi: 10.1111/j.1365-2427.1992. tb00557.x Elith, J., Graham, C. H., Anderson, R. P., Dudík, M., Ferrier, S., Guisan, A., et al. (2006). Novel methods improve prediction of species’ distributions from occurrence data. Ecography 29, 129–151. doi: 10.1111/j.2006.0906-7590. 04596.x Blindow, I. (2000). Distribution of charophytes along the Swedish coast in relation to salinity and eutrophication. Int. Rev. Hydrobiol. 85, 707–717. doi: 10.1002/ 1522-2632(200011)85:5/6<707::aid-iroh707>3.0.co;2-w Elith, J., Leathwick, J. R., and Hastie, T. (2008). A working guide to boosted regression trees. J. Anim. Ecol. 77, 802–813. doi: 10.1111/j.1365-2656.2008. FUNDING This study was funded by the Estonian Science Foundation Grant No. 9439 and the Interreg Central Baltic Programme Project MAREA. This study was funded by the Estonian Science Foundation Grant No. 9439 and the Interreg Central Baltic Programme Project MAREA. DISCUSSION aspera should sustain positive growth in its Baltic Sea habitats even under relatively extreme temperature conditions, which substantiates earlier studies showing that temperature does not signiﬁcantly April 2021 \| Volume 8 \| Article 674014 9 Scale-Speciﬁc Algal Production Kotta et al. scale and any multiscale model development should involve experimental validation at all important scales. C. aspera largely dominates the studied charophyte community and interspeciﬁc competition among algae and submerged vegetation is slight. It is plausible that spatial heterogeneity or diﬀerential utilization of abiotic resources are behind the pattern in distribution of species (Chambers and Prepas, 1990). DATA AVAILABILITY STATEMENT The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation. We thank Robert Szava-Kovats for proofreading the article. We thank Robert Szava-Kovats for proofreading the article. CONCLUSION A large set of experimental data is a prerequisite to understand and accurately model the patterns of production of any macroalgal assemblages. In this study we experimentally quantiﬁed seasonal variability in individual photosynthesis, community production and growth in biomass of the charophyte C. aspera in the Baltic Sea along with the key environmental variables known to determine the production of macroalgae. The experiments suggested that the production and growth of C. aspera was deﬁned largely by light and partly by temperature at all studied scales. However, algal production was systematically lower at the community than at an individual level under the same light conditions, suggesting the importance of shelf- shading and below-ground processes in natural plant assemblages (e.g., Carvalho et al., 2005; Binzer et al., 2006). Moreover, the observation scale also deﬁned response types between the environment, plant production and growth. This implies that the patterns of variability of the production and growth of macrophytes should always be interpreted in the context of AUTHOR CONTRIBUTIONS KT, MR, and JK built the databases. MR and JK analyzed the data. JK, KT, and TP wrote the manuscript. All authors conceived the study, discussed the results, edited the manuscript, and approved the submitted version. REFERENCES Regime shifts between macrophytes and phytoplankton – concepts beyond shallow lakes, unravelling stabilizing mechanisms and practical consequences. Limnetica 34, 467–480. doi: 10.23818/limn.34.35 Nejrup, L. B., Staehr, P. A., and Thomsen, M. S. (2013). Temperature- and light-dependent growth and metabolism of the invasive red algae Gracilaria vermiculophylla – a comparison with two native macroalgae. Eur. J. Phycol. 48, 295–308. doi: 10.1080/09670262.2013.830778 Israel, A., and Hophy, M. (2002). Growth, photosynthetic properties and Rubisco activities and amounts of marine macroalgae grown under current and elevated seawater CO2 concentrations. Glob. Change Biol. 8, 831–840. doi: 10.1046/j. 1365-2486.2002.00518.x Nõges, P., Tuvikene, L., Feldmann, T., Tõnno, I., Künnap, H., Luup, H., et al. (2003). The role of charophytes in increasing water transparency: a case study of two shallow lakes in Estonia. Hydrobiol. 506, 567–573. doi: 10.1023/B:HYDR. 0000008625.36438.75 Jevrejeva, S., and Leppäranta, M. (2002). Ice conditions along the Estonian coast in a statistical view. Nord. Hydrol. 33, 241–262. doi: 10.2166/NH.2002. 0025 Kataržyt˙e, M., Vaiˇci¯ut˙e, D., Buˇcas, M., Gyrait˙e, G., and Petkuvien˙e, J. (2017). Microorganisms associated with charophytes under diﬀerent salinity conditions. Oceanologia 59, 177–186. doi: 10.1016/j.oceano.2016.10.002 Paalme, T., Kotta, J., and Kersen, P. (2013). Does the growth rate of drifting Furcellaria lumbricalis and Coccotylus truncatus depend on their proportion and density. Proc. Estonian Acad. Sci. 62, 141–147. doi: 10.3176/proc.2013.2.08 Pajusalu, L., Martin, G., Põllumäe, A., and Paalme, T. (2013). Results of laboratory and ﬁeld experiments of the direct eﬀect of increasing CO2 on net primary production of macroalgal species in the brackish water ecosystems. Proc. Estonian Acad. Sci. 62, 148–154. doi: 10.3176/proc.2013.2.09 Kim, J.-H., Kang, E. J., Edwards, M. S., Lee, K., Jeong, H. J., and Kim, K. Y. (2016). Species-speciﬁc responses of temperate macroalgae with diﬀerent photosynthetic strategies to ocean acidiﬁcation: a mesocosm study. Algae 31, 243–256. doi: 10.4490/algae.2016.31.8.20 Pajusalu, L., Martin, G., Põllumäe, A., Torn, K., and Paalme, T. (2015). Direct eﬀects of increased CO2 concentrations in seawater on the net primary production of charophytes in a shallow, coastal, brackish-water ecosystem. Boreal Env. Res. 20, 413–422. Kirk, J. T. O. (1994). Light and Photosynthesis in Aquatic Ecosystems. Cambridge: Cambridge University Press. Kotta, J., Paalme, T., Kersen, P., Martin, G., Herkül, K., and Möller, T. (2008). Density dependent growth of the red algae Furcellaria lumbricalis and Coccotylus truncatus in the West-Estonian Archipelago Sea, northern Baltic Sea. Oceanologia 50, 577–585. Pärnoja, M., Kotta, J., Orav-Kotta, H., and Paalme, T. (2014). REFERENCES 01390.x Blindow, I., Dietrich, J., Möllmann, N., and Schubert, H. (2003). Growth, photosynthesis and fertility of Chara aspera under diﬀerent light and salinity conditions. Aquat. Bot. 76, 213–234. doi: 10.1016/S0304-3770(03)00053-6 Field, C. B., Behrenfeld, M. J., Randerson, J. T., and Falkowski, P. (1998). Primary production of the biosphere: integrating terrestrial and oceanic components. Science 281, 237–240. doi: 10.1126/science.281.5374.237 Blindow, I., and van de Weyer, K. (2016). “Ökologie der Characeen,” in Armleuchteralgen: Die Characeen Deutschlands, eds R. Becker, I. Blindow, A. Doege, T. Franke, T. Gregor, U. Hamann, et al. (Berlin: Springer-Verlag), 529–538. Gattuso, J. P., Gentili, B., Duarte, C. M., Kleypas, J. A., Middelburg, J. J., and Antione, D. (2006). Light availability in the coastal ocean: impact on the distribution of benthic photosynthetic organisms and their contribution to primary productivity. Biogeosciences 3, 489–513. doi: 10.5194/bg-3-489-2006 Breeman, A. M., and Pakker, H. (1994). Temperature ecotypes in seaweed – adaptive signiﬁcance and biogeographic implications. Bot. Mar. 37, 171–180. doi: 10.1515/botm.1994.37.3.171 Gómez, I., Weykam, G., Klöser, H., and Weincke, C. (1997). Photosynthetic light requirements, metabolic carbon balance and zonation of sublittoral macroalgae April 2021 \| Volume 8 \| Article 674014 Frontiers in Marine Science \| www.frontiersin.org 10 Scale-Speciﬁc Algal Production Kotta et al. from King George Island (Antarctica). Mar. Ecol. Prog. Ser. 148, 281–293. doi: 10.3354/meps148281 from King George Island (Antarctica). Mar. Ecol. Prog. Ser. 148, 281–293. doi: 10.3354/meps148281 Leathwick, J. R., Elith, J., Francis, M. P., Hastie, T., and Taylor, P. (2006). Variation in demersal ﬁsh species richness in the oceans surrounding New Zealand: an analysis using boosted regression trees. Mar. Ecol. Prog. Ser. 321, 267–281. doi: 10.3354/meps321267 Greenwell, B. M. (2017). pdp: an R package for constructing partial dependence plots. R. J. 9, 421–436. doi: 10.32614/rj-2017-016 Hargeby, A., Blom, H., Blindow, I., and Andersson, G. (2005). Increased growth and recruitment of piscivorous perch, Perca ﬂuviatilis, during a transient phase of expanding submerged vegetation in a shallow lake. Freshw. Biol. 50, 2053–2062. doi: 10.1111/j.1365-2427.2005.01446.x Libbert, E., and Walter, T. (1985). Photosynthetic production of a brackish water community of Chara tomentosa L. and its dependence on environmental conditions. Int. Rev. Ges. Hydrobiol. 70, 359–368. doi: 10.1002/iroh. 19850700305 Hastie, T., Tibshirani, R., and Friedman, J. (2009). The Elements of Statistical Learning: Data Mining, Inference, and Prediction. New York, NY: Springer. Littler, M. M. (1980). Morphological form and photosynthetic performances of marine macroalgae: tests of a functional/form hypothesis. Bot. Mar. 22, 161– 165. REFERENCES Hauxwell, J., Cebrián, J., and Valiela, I. (2003). Eelgrass Zostera marina loss in temperate estuaries: relationship to landderived nitrogen loads and eﬀect of light limitation imposed by algae. Mar. Ecol. Prog. Ser. 247, 59–73. doi: 10.3354/ meps247059 Littler, M. M., and Littler, D. S. (1980). The evolution of thallus form and survival strategies in benthic marine macroalgae: ﬁeld and laboratory tests of Littler, M. M., and Littler, D. S. (1980). The evolution of thallus form and survival strategies in benthic marine macroalgae: ﬁeld and laboratory tests of a functional form model. Am. Nat. 116, 25–44. doi: 10.1086/283610 , , , ( ) survival strategies in benthic marine macroalgae: ﬁeld and laboratory tests o a functional form model. Am. Nat. 116, 25–44. doi: 10.1086/283610 g g y a functional form model. Am. Nat. 116, 25–44. doi: 10.1086/283610 Hayward, P. J. (1988). Animals on Seaweeds. Richmond: Richmond Publishing Company. Lobban, C. S., Harrison, P. J., and Duncan, M. J. (1985). The Physiological Ecology of Seaweeds. New York: Cambridge University Press. Lüning, K. (1990). Seaweeds: Their Environment, Biogeography and Ecophysiology. New York: Wiley. Hellström, T. (1991). The eﬀect of resuspension on algal production in a shallow lake. Hydrobiology 213, 183–190. doi: 10.1007/BF00016421 Marquardt, R., and Schubert, H. (2009). Photosynthetic characterisation of Chara vulgaris in bioremediation ponds. Charophytes 2, 1–8. Henricson, C., Sandberg-Kilpi, E., and Munsterhjelm, R. (2006). Experimental studies on the impact of turbulence, turbidity and sedimentation on Chara tomentosa L. Cryptogamie Algol. 27, 419–434. Middleboe, A. L., and Binzer, T. (2004). Importance of canopy structure on photosynthesis in single- and multi-species assemblages of marine macroalgae. Oikos 107, 422–432. doi: 10.1111/j.0030-1299.2004.13345.x Herkül, K., Torn, K., and Möller, T. (2018). The environmental niche separation between charophytes and angiosperms in the northern Baltic Sea. Bot. Lett. 165, 115–127. doi: 10.1080/23818107.2017.1399824 Miller, R. O. (1998). “Nitric-perchloric acid wet digestion in an open vessel,” in Handbook of Reference Methods for Plant Analysis, ed. Y. P. Kalra (Boca Raton, FL: CRC Press), 57–61. Hill, R., Bellgrove, A., Macreadie, P. I., Petrou, K., Beardall, J., Steven, A., et al. (2015). Can macroalgae contribute to blue carbon? An Australian perspective. Limnol. Oceanogr. 60, 1689–1706. doi: 10.1002/lno.10128 Munsterhjelm, R. (2005). Natural Succession and Human-Induced Changes in the Soft-Bottom Macrovegetation of Shallow Brackish Bays on the Southern Coast of Finland. Helsinki: Walter and Andree de Nottbeck Foundation. PhD Thesis. Hilt, S. (2015). REFERENCES Comparisons of individual and community photosynthetic production indicate light limitation in the shallow water macroalgal communities of the Northern Baltic Sea. Mar. Ecol. 35, 19–27. doi: 10.1111/maec.12074 Kotta, J., Torn, K., Martin, G., Orav-Kotta, H., and Paalme, T. (2004). Seasonal variation of invertebrate grazing on Chara connivens and C. tomentosa in Kõiguste Bay, NE Baltic Sea. Helgoland Mar. Res. 58, 71–76. doi: 10.1007/ s10152-003-0170-2 Puche, E., Rojo, C., Ramos-Jiliberto, R., and Rodrigo, M. A. (2020). Structure and vulnerability of the multi-interaction network in macrophyte-dominated lakes. Oikos 129, 35–48. doi: 10.1111/oik.06694 Kotta, J., and Witman, J. (2009). “Regional-scale patterns,” in Marine Hard Bottom Communities, ed. M. Wahl (Berlin: Springer-Verlag), 89–99. doi: 10.1007/ b76710_6 Raven, J. A. (2003). Long-distance transport in non-vascular plants. Plant Cell Environ. 26, 73–85. doi: 10.1046/j.1365-3040.2003.00920.x Kovtun-Kante, A., Torn, K., and Kotta, J. (2014). In situ production of charophyte communities under reduced light conditions in a brackish-water ecosystem. Estonian J. Ecol. 63, 28–38. doi: 10.3176/eco.2014.1.03 Raven, J. A., and Hurd, C. L. (2012). Ecophysiology of photosynthesis in macroalgae. Photosynth. Res. 113, 105–125. doi: 10.1007/s11120-012-9768-z Raven, J. A., and Smith, F. A. (1978). Eﬀect of temperature on ion content, ion ﬂuxes and energy metabolism in Chara corallina. Plant Cell Environ. 1, 231–238. doi: 10.1111/j.1365-3040.1978.tb00766.x Kufel, L., and Kufel, I. (2002). Chara beds acting as nutrient sinks in shallow lakes - a review. Aquat. Bot. 72, 249–260. doi: 10.1016/S0304-3770(01)00204-2 Raven, J. A., Smith, F. A., and Glidewell, S. M. (1979). Photosynthetic capacities and biological strategies of giant-celled and small-celled macro-algae. New Phytol. 83, 299–309. doi: 10.1111/j.1469-8137.1979.tb07455.x Küster, A., Schaible, R., and Schubert, H. (2000). Light acclimation of the charophyte Lamprothamnium papulosum. Aquat. Bot. 68, 205–216. doi: 10. 1016/S0304-3770(00)00122-4 Küster, A., Schaible, R., and Schubert, H. (2004). Light acclimation of photosynthesis in three charophyte species. Aquat. Bot. 79, 111–124. doi: 10. 1016/j.aquabot.2004.01.010 Reed, D. C., and Foster, M. S. (1984). The eﬀects of canopy shadings on algal recruitment and growth in a giant kelp forest. Ecology 65, 937–948. doi: 10.2307/ 1938066 April 2021 \| Volume 8 \| Article 674014 11 Frontiers in Marine Science \| www.frontiersin.org Kotta et al. Scale-Speciﬁc Algal Production Sala, E., and Graham, M. H. (2002). Community-wide distribution of predator- prey interaction strength in kelp forests. Proc. Natl. Acad. Sci. U.S.A. 99, 3678–3683. doi: 10.1073/pnas.052028499 Torn, K., Martin, G., and Paalme, T. (2006). REFERENCES Seasonal changes in biomass, elongation growth and primary production rate of Chara tomentosa in the NE Baltic Sea. Ann. Bot. Fenn. 43, 276–283. Sand-Jensen, K., Riis, T., Vestergaard, O., and Larsen, S. E. (2000). Macrophyte decline in Danish lakes and streams over the past 100 years. J. Ecol. 88, 1030–1040. doi: 10.1046/j.1365-2745.2000.00519.x Torn, K., Peterson, A., and Herkül, K. (2020). Predicting the impact of climate change on the distribution of the key habitat-forming species in the NE Baltic Sea. J. Coast. Res. SI 95, 177–181. doi: 10.2112/SI95-035.1 Sea. J. Coast. Res. SI 95, 177–181. doi: 10.2112/SI95-035.1 Schmieder, K., Werner, S., and Bauer, H.-G. (2006). Submersed macrophytes as a food source for wintering waterbirds at Lake Constance. Aquat. Bot. 84, 245–250. doi: 10.1016/j.aquabot.2005.09.006 van den Berg, M. S., Coops, H., Simons, J., and Pilon, J. (2002). A comparative study of the use of inorganic carbon resources by Chara aspera and Potamogeton pectinatus. Aquat. Bot. 72, 219–233. doi: 10.1016/s0304-3770(01)00202-9 Schneider, S., Ziegler, C., and Melzer, A. (2006). Growth towards light as an adaptation to high light conditions in Chara branches. New Phytol. 172, 83–91. doi: 10.1111/j.1469-8137.2006.01812.x Vermeer, C. P., Escher, M., Portielje, R., and de Klein, J. J. M. (2003). Nitrogen uptake and translocation by Chara. Aquat. Bot. 76, 245–258. doi: 10.1016/ s0304-3770(03)00056-1 Schubert, H., and Blindow, I. (2003). Charophytes of the Baltic Sea. Köningstein: A. R. G. Ganter Verlag Kommanditgesellschaft. Wallentinus, I. (1978). Productivity studies on Baltic macroalgae. Bot. Mar. 21, 365–380. Schubert, H., Blindow, I., Bueno, N. C., Casanova, M. T., Pełechaty, M., and Pukacz, A. (2018). Ecology of charophytes – permanent pioneers and ecosystem engineers. Perspect. Phycol. 5, 61–74. doi: 10.1127/pip/2018/0080 Wiencke, C., and tom Dieck, I. (1990). Temperature requirements for growth and survival of macroalgae from Antarctica and southern Chile. Mar. Ecol. Prog. Ser. 59, 157–170. doi: 10.1007/BF02430357 doi: 10.3354/meps059157 ngineers. Perspect. Phycol. 5, 61–74. doi: 10.1127/pip/2018/0080 Suursaar, Ü, Otsmann, M., Astok, V., and Kullas, T. (1998). Modelling the water exchange in the Irben, Suur, Soela and Hari Straits in 1995-1996. EMI Rep. Ser. 9, 61–90. Conﬂict of Interest: The authors declare that the research was conducted in the absence of any commercial or ﬁnancial relationships that could be construed as a potential conﬂict of interest. Tait, L. W., and Schiel, D. R. (2010). Primary productivity of intertidal macroalgal assemblages: comparison of laboratory and in situ photorespirometry. Mar. Ecol. Prog. Ser. 416, 115–125. Frontiers in Marine Science \| www.frontiersin.org REFERENCES doi: 10.3354/meps08781 Copyright © 2021 Kotta, Torn, Paalme, Rätsep, Kaljurand, Teeveer and Kotta. This is an open-access article distributed under the terms of the Creative Commons Copyright © 2021 Kotta, Torn, Paalme, Rätsep, Kaljurand, Teeveer and Kotta. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. Torn, K., and Martin, G. (2003). Changes in the distribution of charophyte species in enclosed seabays of western Estonia. Proc. Estonian Acad. Sci. Biol. Ecol. 52, 134–140. Torn, K., Martin, G., Kotta, J., and Kupp, M. (2010). Eﬀects of diﬀerent types of mechanical disturbances on a charophyte dominated macrophyte community. Estuar. Coast. Shelf Sci. 87, 27–32. doi: 10.1016/j.ecss.2009.12.006 April 2021 \| Volume 8 \| Article 674014 Frontiers in Marine Science \| www.frontiersin.org 12
https://openalex.org/W4240438721	https://www.ispras.ru/upload/uf/a56/a56d0d9fa3fa84ea3ac144e01619db30.pdf	Russian	null	null	Trudy Instituta sistemnogo programmirovaniâ RAN/Trudy Instituta sistemnogo programmirovaniâ	2,018	cc-by	169,254	ISSN 2079-8156 (Print), ISSN 2220-6426 (Online) Труды Института системного программирования РАН Proceedings of the Institute for System Programming of the RAS Proceedings of ISP RAS are a double- blind peer-reviewed journal publishing scientific articles in the areas of system programming, software engineering, and computer science. The journal's goal is to develop a respected network of knowledge in the mentioned above areas by publishing high quality articles on open access. The journal is intended for researchers, students, and practitioners. It covers a wide variety of topics including (but not limited to):  Operating Systems.  Operating Systems.  Operating Systems.  Compiler Technology.  Databases and Information Systems.  Parallel and Distributed Systems.  Software Engineering.  Software Modeling and Design Tools.  Verification, Validation, and Testing.  Static and Dynamic Analysis.  Software Safety and Security.  Computer Algorithms.  Artificial Intelligence. The journal is published one volume per year, six issues in each volume. Open access to the journal content allows to provide public access to the research results and to support global exchange of knowledge. Proceedings of ISP RAS is abstracted and/or indexed in:  Compiler Technology. Databases and Information Systems.  Software Engineering.  Static and Dynamic Analysis.  Software Safety and Security.  Computer Algorithms. Труды ИСП РАН реферируются и/или индексируются в: УДК004.45 Editorial Board Заместитель главного редактора - Кузнецов Сергей Дмитриевич, д.т.н., профессор, ИСП РАН (Москва, Российская Федерация) (Phys.–Math.), Professor, National (Москва, Российская Федерация) Ластовецкий Алексей Леонидович, д.ф.-м.н., профессор, Университет Дублина (Дублин, Ирландия) Ломазова Ирина Александровна, д.ф.-м.н., профессор, Национальный исследовательский университет «Высшая школа экономики» (Москва, Российская Федерация) Новиков Борис Асенович, д.ф.-м.н., профессор, Санкт- Петербургский государственный университет (Санкт- Петербург, Россия) П А К ф ИСП Ластовецкий Алексей Леонидович, д.ф.-м. Research University Higher School of Economics (Moscow, Russian Federation) Ластовецкий Алексей Леонидович, д.ф.-м.н., профессор, Университет Дублина (Дублин, Ирландия) Boris A. Novikov, Dr. Sci. (Phys.–Math.), Professor, St. Ломазова Ирина Александровна, д.ф.-м.н., профессор, Национальный исследовательский университет «Высшая Petersburg University (St. Petersburg, Russia) школа экономики» (Москва, Российская Федерация) Alexander K. Petrenko, Dr. Sci. (Phys.–Math.), Institute for System Programming of the RAS (Moscow, Russian Federation) Новиков Борис Асенович, д.ф.-м.н., профессор, Санкт- Петербургский государственный университет (Санкт- Петербург, Россия) Alexandre F. Petrenko, PhD, Computer Research Institute of Montreal (Montreal, Canada) Assaf Schuster, Ph.D., Professor, Technion - Israel Institute of Technology (Haifa, Israel) Vitaly A. Semenov, Dr. Sci. (Phys.–Math.), Professor, Institute for System Programming of the RAS (Moscow, Russian Federation) Alexandre F. Petrenko, PhD, Computer Research Institute of Montreal (Montreal, Canada) Assaf Schuster, Ph.D., Professor, Technion - Israel Institute of Technology (Haifa, Israel) Петренко Александр Константинович, д.ф.-м.н., ИСП РАН (Москва, Российская Федерация) Петренко Александр Федорович, д.ф.-м.н., р др д р , д ф , Исследовательский институт Монреаля (Монреаль, Канада) Vitaly A. Semenov, Dr. Sci. (Phys.–Math.), Professor, Institute for System Programming of the RAS (Moscow, Russian Federation) Семенов Виталий Адольфович, д.ф.-м.н., профессор, ИСП РАН (Москва, Российская Федерация) Томилин Александр Николаевич, д.ф.-м.н., профессор, ИСП РАН (Москва, Российская Федерация) Черных Андрей, д.ф.-м.н., профессор, Научно- исследовательский центр CICESE (Энсенада, Нижняя Калифорния, Мексика) Victor Z. Shnitman, Dr. Sci. (Eng.), Institute for System Programming of the RAS (Moscow, Russian Federation) Alexander N. Tomilin, Dr. Sci. (Phys.–Math.), Professor, др , д ф , р ф р, ИСП РАН (Москва, Российская Федерация) Черных Андрей, д.ф.-м.н., профессор, Научно- исследовательский центр CICESE (Энсенада, Нижняя Калифорния, Мексика) Institute for System Programming of the RAS (Moscow, Russian Federation) Institute for System Programming of the RAS (Moscow, Russian Federation) Irina B. Virbitskaite, Dr. Sci. (Phys.–Math.), The A.P. Ershov Institute of Informatics Systems, Siberian Branch of the RAS (Novosibirsk, Russian Federation) Irina B. Virbitskaite, Dr. Sci. (Phys.–Math.), The A.P. Ershov Institute of Informatics Systems, Siberian Branch of the RAS (Novosibirsk, Russian Federation) Andrey Voronkov, Dr. Sci. Редколлегия Editor-in-Chief - Arutyun I. Avetisyan, Corresponding Member of RAS, Dr. Sci. (Phys.–Math.), Institute for System Programming of the RAS (Moscow, Russian Federation) Главный редактор - Аветисян Арутюн Ишханович, член-корр. РАН, д.ф.-м.н., ИСП РАН (Москва, Российская Федерация) Заместитель главного редактора - Кузнецов Сергей Дмитриевич, д.т.н., профессор, ИСП РАН (Москва, Российская Федерация) Deputy Editor-in-Chief - Sergey D. Kuznetsov, Dr. Sci. (Eng.), Professor, Institute for System Programming of the RAS (Moscow, Russian Federation) Igor B. Burdonov, Dr. Sci. (Phys.–Math.), Institute for System Programming of the RAS (Moscow, Russian Federation) Andrei Chernykh, Dr. Sci., Professor, CICESE Research Centre (Ensenada, Lower California, Mexico) Sergey S. Gaissaryan, PhD (Phys.–Math.), Institute for System Programming of the RAS (Moscow, Russian Federation) Leonid E. Karpov, Dr. Sci. (Eng.), Institute for System Programming of the RAS (Moscow, Russian Federation) Igor Konnov, PhD (Phys.–Math.), Vienna University of Technology (Vienna, Austria) Alexander S. Kossatchev, PhD (Phys.–Math.), Institute for System Programming of the RAS (Moscow, Russian Бурдонов Игорь Борисович, д.ф.-м.н., ИСП РАН (Москва, Российская Федерация) Бурдонов Игорь Борисович, д.ф.-м.н., ИСП РАН (Москва, Российская Федерация) Воронков Андрей Анатольевич, д.ф.-м.н., профессор, Университет Манчестера (Манчестер, Великобритания) Вирбицкайте Ирина Бонавентуровна, профессор, д.ф.- м.н., Институт систем информатики им. академика А.П. Ершова СО РАН (Новосибирск, Россия) Гайсарян Сергей Суренович, к.ф.-м.н., ИСП РАН (Москва, Российская Федерация) Евтушенко Нина Владимировна, профессор, д.т.н., ТГУ (Томск, Российская Федерация) Карпов Леонид Евгеньевич, д.т.н., ИСП РАН (Москва, Российская Федерация) Коннов Игорь Владимирович, к.ф.-м.н., Технический университет Вены (Вена, Австрия) (Москва, Российская Федерация) Воронков Андрей Анатольевич, д.ф.-м.н., профессор, Университет Манчестера (Манчестер, Великобритания) Вирбицкайте Ирина Бонавентуровна, профессор, д.ф.- м.н., Институт систем информатики им. академика А.П. Ершова СО РАН (Новосибирск, Россия) Гайсарян Сергей Суренович, к.ф.-м.н., ИСП РАН (Москва, Российская Федерация) Евтушенко Нина Владимировна, профессор, д.т.н., ТГУ (Томск, Российская Федерация) р р ф р ф р Университет Манчестера (Манчестер, Великобритания) Вирбицкайте Ирина Бонавентуровна, профессор, д.ф.- м.н., Институт систем информатики им. академика А.П. Ершова СО РАН (Новосибирск, Россия) р р ур , ф (Москва, Российская Федерация) у р , р ф р, , (Томск, Российская Федерация) Alexander S. Kossatchev, PhD (Phys.–Math.), Institute for System Programming of the RAS (Moscow, Russian Federation) Карпов Леонид Евгеньевич, д.т.н., ИСП РАН (Москва, Российская Федерация) р Российская Федерация) Nikolay N. Kuzyurin, Dr. Sci. (Phys.–Math.), Institute for System Programming of the RAS (Moscow, Russian Federation) Коннов Игорь Владимирович, к.ф.-м.н., Технический университет Вены (Вена, Австрия) Косачев Александр Сергеевич, к.ф.-м.н., ИСП РАН (Москва, Российская Федерация) (Москва, Российская Федерация) Alexey Lastovetsky, Dr. Sci. (Phys.–Math.), Professor, UCD School of Computer Science and Informatics (Dublin, Ireland) Кузюрин Николай Николаевич, д.ф.-м.н., ИСП РАН (Москва, Российская Федерация) Irina A. Lomazova, Dr. Sci. Заместитель главного редактора - Кузнецов Сергей Дмитриевич, д.т.н., профессор, ИСП РАН (Москва, Российская Федерация) (Phys.–Math.), Professor, University of Manchester (Manchester, UK) Nina V Yevtushenko Dr Sci (Eng ) Tomsk State University Шнитман Виктор Зиновьевич, д.т.н., ИСП РАН (Москва, Российская Федерация) Шнитман Виктор Зиновьевич, д.т.н., ИСП РАН (Москва, Российская Федерация) Andrey Voronkov, Dr. Sci. (Phys.–Math.), Professor, Шустер Ассаф, д.ф.-м.н., профессор, Технион — Израильский технологический институт Technion (Хайфа, Израиль) University of Manchester (Manchester, UK) University of Manchester (Manchester, UK) Nina V. Yevtushenko, Dr. Sci. (Eng.), Tomsk State University (Tomsk, Russian Federation) Nina V. Yevtushenko, Dr. Sci. (Eng.), Tomsk State University (Tomsk, Russian Federation) Адрес: 109004, г. Москва, ул. А. Солженицына, дом 25. Телефон: +7(495) 912-44-25 E-mail: info-isp@ispras.ru Сайт: http://www.ispras.ru/proceedings/ Адрес: 109004, г. Москва, ул. А. Солженицына, дом 25. Телефон: +7(495) 912-44-25 E-mail: info-isp@ispras.ru Сайт: http://www.ispras.ru/proceedings/ Адрес: 109004, г. Москва, ул. А. Солженицына, дом 25. Телефон: +7(495) 912-44-25 E-mail: info-isp@ispras.ru Сайт: http://www.ispras.ru/proceedings/ Address: 25, Alexander Solzhenitsyn st., Moscow, 109004, Russia. Address: 25, Alexander Solzhenitsyn st., Moscow, 109004 Russia. Tel: +7(495) 912-44-25 E-mail: info-isp@ispras.ru Web: http://www.ispras.ru/en/proceedings/ Address: 25, Alexander Solzhenitsyn st., Moscow, 109004, Russia. Tel: +7(495) 912-44-25 Tel: +7(495) 912-44-25 E-mail: info-isp@ispras.ru Web: http://www.ispras.ru/en/proceedings/  Институт Системного Программирования РАН, 2018  Институт Системного Программирования РАН, 2018 Т р у д ы И н с т и ту т а С и с т е м н о г о П р о г ра м м и ро в а н и я С о д е р ж а н и е С о д е р ж а н и е Обнаружение ошибок, возникающих при использовании динамической памяти после освобождения Асрян С.А., Гайсарян С.С., Курмангалеев Ш.Ф., Агабалян А.М., Овсепян Н.Г., Саргсян С.С. ........................................................................................... 7 Статический анализ для поиска переполнения буфера: актуальные направления развитияалгоритмов Дудина И.А. ...................................................................................................... 21 Извлечение архитектурной информации из исходного кода ARINC 653 совместимых приложений с использованием алгоритма CEGAR Лесовой С.Л. .................................................................................................... 31 Вопросы индустриального применения синхронизационных контрактов при динамическом поиске гонок в Java-программах Трифанов В.Ю. ................................................................................................ 47 Применение глубокого машинного обучения к синтезу цепочки вызовов C# Чебыкин А.Е., Кириленко Я.А. ...................................................................... 63 Скрытая отладка программ отладчиком WinDbg в эмуляторе Qemu Абакумов М.А., Довгалюк П.М. ..................................................................... 87 Конфигурируемый трассировщик системных вызовов в эмуляторе QEMU Иванов А.В., Довгалюк П.М., Макаров В.А. ............................................... 93 Анализ методов оценки надежности оборудования и систем. Практика применения методов Лаврищева Е.М., Пакулин Н.В., Рыжов А.Г., Зеленов С.В. ....................... 99 Эмуляция ввода-вывода оборудования с отображением в ОЗУ внутри ядер операционных систем Чепцов В.Ю., Хорошилов А.В. ...................................................................... 121 Построение модульного программного обеспечения на основе однородной компонентой модели Маллачиев К.А., Хорошилов А.В................................................................... 135 Методы защиты децентрализованных автономных организаций от системных отказов и атак А.А. Андрюхин ................................................................................................ 149 Нотация криптографической стековой машины версии один Прокопьев С.Е. .............................................................................................. 165 Построение модулей проверки на основе эталонных функциональных моделей при автономной верификации подсистемы связи Лебедев Д.А., Стотланд И.А. ...................................................................... 183 Верификация контроллеров связи в системах на кристалле Петроченков М.В., Муштаков Р.Е., Шпагилев Д.И. ................................. 195 Программные решения для динамического изменения пользовательского интерфейса на основе автоматически собранной информации о пользователе Зосимов В.В., Христодоров А.В., Булгакова А.С. ....................................... 207 Варианты задач китайского почтальона и их решения через преобразование в задачи маршрутизации Горденко М.К., Авдошин С.М. ...................................................................... 221 Анализ математических постановок задачи маршрутизации с ограничением по грузоподъемности и методов их решения. Береснева Е.Н., Авдошин С.М. ..................................................................... 233 Применение методов системного анализа к оцениванию работы учебных ассистентов Береснева Е.Н., Горденко М.К. .................................................................... 251 Статический анализ зависимостей для семантической валидации данных Ильин Д.В., Фокина Н.Ю., Семенов В.А. ..................................................... 271 Симуляция поведения мультиагентных систем с ациклически взаимодействующими агентами Нестеров Р.А., Мицюк А.А., Ломазова И.А. ............................................... 285 О верификации конечных автоматов-преобразователей над полугруппами Гнатенко А.Р., Захаров В.А. ......................................................................... 303 К проверке строго детерминированного поведения временных конечных автоматов Винарский Е.М., Захаров В.А........................................................................ 325 К построению модульной модели распределенного интеллекта Словохотов Ю.Л., Неретин И.С. ................................................................ 341 Proceedings of the Institute for System Programming of the RAS T a b l e o f C o n t e n t s Dynamic detection of Use After Free bugs Asryan S.A., Gaissaryan S.S., Kurmangaleev Sh. F., Aghabalyan A.М., Hovsepyan N.H., Sargsyan S.S. ........................................................................ 7 Buffer Overflow Detection via Static Analysis: Expectations vs. Reality Dudina I.A. ........................................................................................................ 21 Extracting architectural information from source code of ARINC 653- compatible application software using CEGAR-based approach Lesovoy S.L. ...................................................................................................... 31 Applying synchronization contracts approach for dynamic detection of data races in industrial applications Trifanov V.Yu. ................................................................................................... 47 Applying Deep Learning to C# Call Sequence Synthesis Chebykin A.E., Kirilenko I.A. ............................................................................ 63 Stealth debugging of programs in Qemu emulator with WinDbg debugger M.A. Abakumov., P.M. Dovgalyuk. ................................................................... 87 Configurable system call tracer in QEMU emulator Ivanov A.V., Dovgaluk P.M., Makarov V.A. ................................................... 93 Analysis of methods for assessing the reliability of equipment and systems. Practice of methods Pakulin N.V., Lavrischeva E.M., Ryzhov A.G. ................................................ 99 In-Kernel Memory-Mapped I/O Device Emulation Cheptsov V.Yu., Khoroshilov A.V. .................................................................. 121 Building Modular Real-time software from Unified Component Model Mallachiev K.A., Khoroshilov A.V. ................................................................. 135 Methods of protecting decentralized autonomous organizations from crashes and attacks A.A. Andryukhin .............................................................................................. 149 Cryptographic Stack Machine Notation One Prokopev S.E. .................................................................................................. 165 Construction of validation modules based on reference functional models in a standalone verification of communication subsystem Lebedev D.A., Stotland I.A. .......................................................................... 183 Verification of System on Chip Integrated Communication Controllers Petrochenkov M.V., Mushtakov R.E., Shpagilev D.I. ................................... 195 Dynamically changing user interfaces: software solutions based on automatically collected user information Zosimov V.V., Khrystodorov O.V., Bulgakova O.S. ..................................... 207 The Variants of Chinese Postman Problems and Way of Solving through Transformation into Vehicle Routing Problems Gordenko M.K., Avdoshin S.M........................................................................ 221 Analysis of Mathematical Formulations of Capacitated Vehicle Routing Problem and Methods for their Solution Beresneva E., Avdoshin S. ............................................................................... 233 Applying the methods of system analysis to teaching assistants evaluation Beresneva E., Gordenko M. ............................................................................ 251 Static dependency analysis for semantic data validation Ilyin D.V., Fokina N.Yu., Semenov V.A. .......................................................... 271 Simulating Behavior of Multi-Agent Systems with Acyclic Interactions of Agents Nesterov R.A., Mitsyuk A.A., Lomazova I.A. ................................................... 285 On the model checking of finite state transducers over semigroups Gnatenko A.R., Zakharov V.A. ........................................................................ 303 On the verification of strictly deterministic behaviour of Timed Finite State Machines Vinarskii E.M., Zakharov V.A. ........................................................................ 325 Toward construction of a modular model of distributed intelligence Slovokhotov Yu.L., Neretin I.S. ....................................................................... 341 * Работа поддержана грантом РФФИ № 17-01-00600 DOI: 10.15514/ISPRAS-2018-30(3)-1 Для цитирования: Асрян С.А., Гайсарян С.С., Курмангалеев Ш.Ф., Агабалян А.М., Овсепян Н.Г., Саргсян С.С. Обнаружение ошибок, возникающих при использовании динамической памяти после освобождения. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 7-20. DOI: 10.15514/ISPRAS-2018-30(3)-1 Для цитирования: Асрян С.А., Гайсарян С.С., Курмангалеев Ш.Ф., Агабалян А.М., Овсепян Н.Г., Саргсян С.С. Обнаружение ошибок, возникающих при использовании динамической памяти после освобождения. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 7-20. DOI: 10.15514/ISPRAS-2018-30(3)-1 Ключевые слова: динамический анализ программ; покрытие кода; use-after-free. Ключевые слова: динамический анализ программ; покрытие кода; use-after-free. Asryan S.A., Gaissaryan S.S., Kurmangaleev Sh. F., Aghabalyan A.М., Hovsepyan N.H., Sargsyan S.S. Dynamic detection of Use After Free bugs. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 7-20 Asryan S.A., Gaissaryan S.S., Kurmangaleev Sh. F., Aghabalyan A.М., Hovsepyan N.H., Sargsyan S.S. Dynamic detection of Use After Free bugs. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 7-20 Обнаружение ошибок, возникающих при использовании динамической памяти после её освобождения* 2 С.А. Асрян <asryan@ispras.ru> 1,3,5,6 С.С. Гайсарян <ssg@ispras.ru> 1 Ш.Ф. Курмангалеев <kursh@ispras.ru> 4А.М. Агабалян <anna.aghabalyan@ispras.ru> 4Н.Г. Овсепян <narekhnh@ispras.ru> 5С.С. Саргсян <sevaksargsyan@ispras.ru> 1 Институт системного программирования им. В.П. Иванникова РАН, 109004, Россия, г. Москва, ул. А. Солженицына, д. 25 2 Институт проблем информатики и автоматизации НАН РА Республика 2 С.А. Асрян <asryan@ispras.ru> 1,3,5,6 С.С. Гайсарян <ssg@ispras.ru> 1 Ш.Ф. Курмангалеев <kursh@ispras.ru> 4А.М. Агабалян <anna.aghabalyan@ispras.ru> 4Н.Г. Овсепян <narekhnh@ispras.ru> 5С.С. Саргсян <sevaksargsyan@ispras.ru> 1 Институт системного программирования им. В.П. Иванникова РАН, 109004, Россия, г. Москва, ул. А. Солженицына, д. 25 2 2 Институт проблем информатики и автоматизации НАН РА, Республика Армения, Ереван, 0014, ул. П.Севака, 1 Армения, Ереван, 0014, ул. П.Севака, 1 3 МГУ имени М.В. Ломоносова, факультет ВМК, 2119991 ГСП-1 Москва, Ленинские горы, 2- й учебный корпус 4̆̆ 4 Ереванский государственный университет Республика Армения, г Ереван 0025 ул Алека Манукяна 1 5 Московский физико-технический институт, 700, Московская облаcть, г. Долгопрудный, Институтский пер., 9 ̆̆ 141700, Московская облаcть, г. Долгопрудный, Институтский пер., 9 6 Национальный исследовательский университет «Высшая школа экономики» 101000, Россия, г. Москва, ул. Мясницкая, д. 20 Аннотация. Существенная часть программного обеспечения написана на языках программирования C/C++. Программы на этих языках часто содержат ошибки: использования памяти после освобождения (Use After Free, UAF), переполнения буфера (Buffer Overflow) и др. В статье предложен метод обнаружения ошибок UAF, основанный на динамическом анализе. Для каждого пути выполнения программы предлагаемый метод проверяет корректность операций создания и доступа, а также освобождения динамической памяти. Поскольку применяется динамический анализ, поиск ошибок производится только в той части кода, которая была непосредственно выполнена. Используется символьное исполнение программы с применением решателей SMT (Satisfiability Modulo Theories) [12]. Это позволяет сгенерировать данные, обработка которых приводит к обнаружению нового пути выполнения. * Работа поддержана грантом РФФИ № 17-01-00600 7 7 Asryan S.A., Gaissaryan S.S., Kurmangaleev Sh. F., Aghabalyan A.М., Hovsepyan N.H., Sargsyan S.S. Dynamic detection of Use After Free bugs. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 7-20 1. Введение В программном обеспечении могут содержаться такие ошибки, как:  использование динамической памяти после ее освобождения (Use After Free, UAF),  использование динамической памяти после ее освобождения (Use After Free, UAF),  переполнение буфера или кучи (buffer/heap overflow). Поскольку огромная часть программного обеспечения используется в критически-важных областях человеческой деятельности, наличие ошибок может привести к серьезным последствиям. Существует ряд инструментов, помогающих решить эту проблему, используя методы статического [1, 2] и динамического анализа [3, 4, 5, 6, 7]. Статический анализ предоставляет возможность исследования программного кода без его выполнения. Недостатком этого метода является отсутствие информации о состоянии программы (регистры, трасса программы, входные данные и т.д.) во время выполнения. Это приводит к большому количеству ложных срабатываний. Поэтому данный метод в большинстве случаев используется до применения динамического анализа для выявления фрагментов программы, содержащих потенциальные ошибки. Статический анализ предоставляет возможность исследования программного кода без его выполнения. Недостатком этого метода является отсутствие информации о состоянии программы (регистры, трасса программы, входные данные и т.д.) во время выполнения. Это приводит к большому количеству ложных срабатываний. Поэтому данный метод в большинстве случаев используется до применения динамического анализа для выявления фрагментов программы, содержащих потенциальные ошибки. Для поиска ошибок UAF инструмент [1] выполняет анализ, похожий на анализ доступных выражений (выражение x+y является доступным в точке p, если вдоль любого пути от входной точки до точки p данное выражение вычисляется, а между этими вычислениями значения x и y остаются неизменными [11]). Производится обход всех путей в программе, чтобы обеспечить выполнение условия «определение объектов до их использования». В случаи неудовлетворения данного условия считается, что было выполнено ошибочное использование памяти и выводится ошибка. Инструмент GUEB [2] основан на исследовании бинарного кода программы. Процесс анализа разделяется на два основных этапа. На первом этапе отслеживаются операции обращения к куче и присваивания адресов для проведения анализа набора данных (какой указатель к какому элементу кучи относится). На этом этапе информация {адрес, размер} сохраняется в множествах alloc_set и free_set при создании и освобождении памяти соответственно. На втором этапе выполняется поиск ошибок UAF. Используя собранную информацию для каждой точки программы, инструмент строит множество 8 Асрян С.А., Гайсарян С.С., Курмангалеев Ш.Ф., Агабалян А.М., Овсепян Н.Г., Саргсян С.С. Обнаружение ошибок, возникающих при использовании динамической памяти после освобождения. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 7-20 Асрян С.А., Гайсарян С.С., Курмангалеев Ш.Ф., Агабалян А.М., Овсепян Н.Г., Саргсян С.С. Обнаружение ошибок, возникающих при использовании динамической памяти после освобождения. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 1. Введение 9 Asryan S.A., Gaissaryan S.S., Kurmangaleev Sh. F., Aghabalyan A.М., Hovsepyan N.H., Sargsyan S.S. Dynamic detection of Use After Free bugs. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 7-20 Asryan S.A., Gaissaryan S.S., Kurmangaleev Sh. F., Aghabalyan A.М., Hovsepyan N.H., Sargsyan S.S. Dynamic detection of Use After Free bugs. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 7-20 В данной статье мы будем рассматривать подход, основанный на динамическом анализе программы с использованием динамической инструментации [8, 9]. В работе описывается метод обнаружения ошибок UAF, который проверяет корректность использования указателей для всех возможных путей выполнения программы. Этот метод основан на алгоритме покрытия кода, используемом в SAGE [10], и использует инфраструктуру динамического анализатора Triton [9]. В рамках данной работы была выполнена модификация алгоритма покрытия кода, используемого в Triton, что привело к значительному росту производительности, а также была добавлена поддержка анализа программ, работающих с файловыми входными данными, которая отсутствовала в реализации Triton. Второй раздел статьи посвящена описанию алгоритма покрытия кода программы в Triton и предлагаемой модификации. В третьем разделе рассматривается исходная реализация обнаружения ошибок UAF и ее объединение с динамическим покрытием кода. В четвертом разделе представлены результаты. 1. Введение 7-20 access_heap, которое содержит все элементы {адрес, размер} кучи, доступные в этой точке. Если пересечение access_heap и free_set является непустом множеством, считается, что найдена ошибка UAF. Одной из причин популярности динамического анализа является возможность исследования программ во время выполнения. Благодаря этому возможен доступ к значениям регистров и содержимому памяти. Инструмент Avalanche [3] реализует итеративный анализ исполняемого кода программы, основанный на динамической бинарной трансляции. В процессе анализа инструмент вычисляет входные данные анализируемой программы с целью автоматического обхода всех достижимых путей в программе и обнаружения аварийных завершений программы. access_heap, которое содержит все элементы {адрес, размер} кучи, доступные в этой точке. Если пересечение access_heap и free_set является непустом множеством, считается, что найдена ошибка UAF. Одной из причин популярности динамического анализа является возможность исследования программ во время выполнения. Благодаря этому возможен доступ к значениям регистров и содержимому памяти. Инструмент Avalanche [3] реализует итеративный анализ исполняемого кода программы, основанный на динамической бинарной трансляции. В процессе анализа инструмент вычисляет входные данные анализируемой программы с целью автоматического обхода всех достижимых путей в программе и обнаружения аварийных завершений программы. Инструменты DangNull [4] и FreeSentry [5] фокусируются на обнаружении и обнулении указателей на динамическую область программы после их освобождения, предотвращая появление ошибок. Оба инструмента используют статическую инструментацию программ. Undangle [6] также предотвращает ошибки использования памяти после освобождения. Этот инструмент помечает возвращаемые значения каждой функции распределения памяти и использует анализ помеченных данных для отслеживания этих меток. Далее при освобождении памяти проверяются, какие ячейки памяти ассоциированы с соответствующей меткой, и определяются висячие указатели (указатель с ненулевым значением, ссылающийся на освобожденный область памяти). Инструмент Mayhem [7] основан на методе поиска ошибок в бинарном коде, объединяющем офлайн- и онлайн-подходы к символьному выполнению программы. Офлайн-подход предполагает последовательное исследование путей программы: при каждом новом запуске инструмент покрывает только один путь выполнения. Недостатком является повторное выполнение общего начального фрагмента пути при каждом запуске программы. Онлайн-подход, в свою очередь, исследует все возможные пути выполнения программы одновременно, что приводит к нехватке памяти в определенный момент времени. Объединение этих двух подходов заключается в следующем: при достижении граничного значения расхода памяти создаются контрольные точки, исследование некоторых путей останавливается с сохранением информации о текущем состоянии выполнения, контекста символьного выполнения и конкретных входных данных. После освобождения ресурсов (завершились исследование некоторых путей), восстанавливается одна из контрольных точек (с использованием сохраненных данных воспроизводится конкретное выполнение до контрольной точки). Далее выполняется загрузка контекста символьного выполнения и начинается анализ нового пути. Данный подход позволяет избежать повторного символьного выполнения программы до места создания контрольной точки. 2.1 Покрытие кода в Triton В данной статье используется алгоритм увеличения покрытия кода программы, разработанный в компании Microsoft и используемый в инструменте SAGE [10]. Этот алгоритм частично реализован в Triton. Он состоит из двух этапов:  выбор начальных входных данных и сборка ограничений для каждого пути выполнения программы;  выбор начальных входных данных и сборка ограничений для каждого пути выполнения программы;  получение новых входных данных с помощью решения логических выражений, состоящих из ограничений, собранных на предыдущем этапе. Рассмотрим пример программы на рис. 1. Рис. 1. Пример программы из статьи [10] Fig. 1. An example of a program from [10] 10 Асрян С.А., Гайсарян С.С., Курмангалеев Ш.Ф., Агабалян А.М., Овсепян Н.Г., Саргсян С.С. Обнаружение ошибок, возникающих при использовании динамической памяти после освобождения. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 7-20 Асрян С.А., Гайсарян С.С., Курмангалеев Ш.Ф., Агабалян А.М., Овсепян Н.Г., Саргсян С.С. Обнаружение ошибок, возникающих при использовании динамической памяти после освобождения. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 7-20 Асрян С.А., Гайсарян С.С., Курмангалеев Ш.Ф., Агабалян А.М., Овсепян Н.Г., Саргсян С.С. Обнаружение ошибок, возникающих при использовании динамической памяти после освобождения. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 7-20 Асрян С.А., Гайсарян С.С., Курмангалеев Ш.Ф., Агабалян А.М., Овсепян Н.Г., Саргсян С.С. Обнаружение ошибок, возникающих при использовании динамической памяти после освобождения. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 7-20 Чтобы можно было исследовать все пути этой программы, на вход она должна получить строку “bad!”. Чтобы получить нужные данные, алгоритм начинает свою работу, запуская программу на начальной входной строке, которая помещается в список входных данных. После первого запуска программы получается набор ограничений < i0 ≠ b, i1 ≠ a, i2 ≠ d, i3 ≠ ! >, где i0, i1, i2, i3 представляют ячейки памяти input [0], input [1], input [2] и input [3] соответственно. Чтобы можно было исследовать все пути этой программы, на вход она должна получить строку “bad!”. Чтобы получить нужные данные, алгоритм начинает свою работу, запуская программу на начальной входной строке, которая помещается в список входных данных. После первого запуска программы получается набор ограничений < i0 ≠ b, i1 ≠ a, i2 ≠ d, i3 ≠ ! >, где i0, i1, i2, i3 представляют ячейки памяти input [0], input [1], input [2] и input [3] соответственно. 2.1 Покрытие кода в Triton В ходе работы алгоритма с помощью решения этих ограничений для каждого элемента из списка входных данных генерируются дочерние данные, удовлетворяющие этим ограничениям, которые далее помещаются в список входных данных. Для каждого элемента из этого списка программа заново запускается, и работа алгоритма возобновляется. Этот процесс продолжается до тех пор, пока все элементы из списка входных данных не будут поочередно рассмотрены (псевдокод алгоритма приведен на рис. 3). Применив алгоритм для программы на рис. 1 с начальной входной строкой “good”, мы получим набор решений, представленный на рис. 2. Рис. 2. Входные данные после каждой итерации алгоритма Fig. 2. Input data after each iteration of the algorithm Рис. 2. Входные данные после каждой итерации алгоритма Fig. 2. Input data after each iteration of the algorithm 11 Asryan S.A., Gaissaryan S.S., Kurmangaleev Sh. F., Aghabalyan A.М., Hovsepyan N.H., Sargsyan S.S. Dynamic detection of Use After Free bugs. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 7-20 Asryan S.A., Gaissaryan S.S., Kurmangaleev Sh. F., Aghabalyan A.М., Hovsepyan N.H., Sargsyan S.S. Dynamic detection of Use After Free bugs. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 7-20 Рис. 3. Псевдокод алгоритма покрытия кода программы Fig. 3. Pseudo code of a program code coverage algorithm Поскольку работа данного алгоритма требует неоднократного запуска программы, в Triton реализована возможность сохранения состояния программы. Это позволяет значительно улучшить производительность выполнения. Кроме того, в Triton отсутствует часть алгоритма SAGE [10], предназначенная для уменьшения набора входных данных. Поэтому инструмент неоднократно запускает анализируемую программы на входных данных, которые не открывают новых путей. В описываемой работе в алгоритм покрытия кода был добавлен новый функционал, который позволяет достичь значительного роста производительности. 2.2 Модификация алгоритма покрытия Input data after each iteration of the algorithm, after adding weight values 2.2 Модификация алгоритма покрытия В оригинальной реализации алгоритма SAGE [10] в Triton после каждой итерации программа всегда получала на вход последний элемент из списка входных данных, не учитывая при этом количество открытых базовых блоков программы с помощью данного элемента. Это приводило к тому, что вместе с обработкой входных данных, которые имеют воздействие на покрытие кода программы, рассматриваются и те входные данные, с помощью которых не были открыты никакие новые пути в программе. Поскольку в ходе работы алгоритма для каждого входного элемента генерируются ее дочерние данные, количество элементов в списке входных данных значительно увеличивается. Следовательно, для эффективного выполнения алгоритма требуется определение приоритетов для сгенерированных входных данных. В предлагаемой модификации алгоритма каждому элементу из списка входных данных мы присваиваем вес, который представляет из себя количество базовых блоков программы, открытых этим элементом. В начале работы алгоритма входным данным присваивается нулевой вес. Во время первой итерации алгоритма подсчитываются весовые значения начальных входных данных. После каждой итерации весовые значения обновляются следующим образом: весовые значения уже рассмотренных элементов передаются их дочерним элементам (входные данные, которые получились с помощью решения логических уравнений). Таким образом, применяется иерархический обход входных данных. Перед очередным запуском программы из списка выбирается элемент с наибольшим весом. Это позволяет значительно упростить дерево решений, как показано на рис. 4. На рисунке видно, что после добавления весовых значений количество рассматриваемых входных данных уменьшилось почти вдвое, что в свою На рисунке видно, что после добавления весовых значений количество рассматриваемых входных данных уменьшилось почти вдвое, что в свою 12 Асрян С.А., Гайсарян С.С., Курмангалеев Ш.Ф., Агабалян А.М., Овсепян Н.Г., Саргсян С.С. Обнаружение ошибок, возникающих при использовании динамической памяти после освобождения. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 7-20 очередь приводит к существенному увеличению производительности работы алгоритма (на некоторых тестах производительность выросла почти на 90%). Еще одним недостатком Triton была поддержка программ, принимающих на вход только аргументы командной строки. Для расширения набора анализируемых программ нами была добавлена поддержка программ, использующих файлы как источник входных данных. Кроме того, была добавлена возможность определения конкретных диапазонов входных данных, которые в ходе анализа будут помечены как символьные. Описанный подход к подсчету весовых значений не является единственным возможным, поэтому в дальнейших исследованиях будут рассматриваться и другие варианты определения этих значений. Рис. 4. Входные данные после каждой итерации алгоритма после добавления весов Fig. 4. Input data after each iteration of the algorithm, after adding weight values Рис. 4. Входные данные после каждой итерации алгоритма после добавления весов Fig. 4. 3. Поиск ошибок Динамический анализ программы основан на исследовании программного обеспечения в процессе выполнения. Это дает возможность исследования программ с учетом определенных условий выполнения, а также позволяет использовать конкретные значения указателей. Одним из недостатков динамического анализа является требование наличия качественного покрытия кода. Однако в большинстве случаев для найденных ошибок возможна генерация входных данных, которые позволяют воспроизвести ошибку. Ошибка UAF характеризуется возникновением двух последовательных событий:  создание висящих указателей (dangling pointers);  создание висящих указателей (dangling pointers);  доступ к памяти с использованием висящего указателя. 13 Asryan S.A., Gaissaryan S.S., Kurmangaleev Sh. F., Aghabalyan A.М., Hovsepyan N.H., Sargsyan S.S. Dynamic detection of Use After Free bugs. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 7-20 На рис. 5 приведён пример UAF. После проверки условия на строке 3 происходит освобождение памяти на который указывает переменная ptr (строка 4), затем управление переходит на строку 12, вследствие чего происходит повторное освобождение памяти. На рис. 5 приведён пример UAF. После проверки условия на строке 3 происходит освобождение памяти на который указывает переменная ptr (строка 4), затем управление переходит на строку 12, вследствие чего происходит повторное освобождение памяти. 3.1 Алгоритм поиска ошибок в Triton Используя инструментацию программы, алгоритм отслеживает функции выделения (malloc) и освобождения (free) памяти. В начале работы алгоритма создаются два множества (allocSET и freeSET) для отслеживания участков памяти, которые были выделены и освобождены во время выполнения программы. Элементами данных множеств являются парами, имеющие вид («адрес, размер»). Каждый раз при вызове malloc/free множества allocSET и freeSET обновляются путем добавления или удаления элементов с соответствующим адресом и размером выделенной памяти. При вызове функции malloc новый элемент («адрес_2, размер_2») добавляется в множество allocSET и удаляется из второго множества если данный элемент присутствует в множестве freeSET (т.е. есть совпадение как по адресу, так и по размеру). б д Рис. 5. Использование памяти после освобождения Рис. 5. Использование памяти после осв Рис. 5. Использование памяти после освобождения Fog. 5. An example of UAF Если данный элемент совпадает только по адресу, то перед обновлением множеств, выполняются дополнительные действия для обработки значений адреса и размера (если размер_2 < размер_1, то в freeSET добавляется элемент «адрес_2 + размер_2, размер_1 - размер_2»). При вызове функции free соответствующий элемент перемещается из множества allocSET в множество freeSET. Во время выполнения инструкций, осуществляющих доступ к памяти, проверяется наличие указателя в обоих множествах. Ошибки UAF фиксируются в двух случаях: элемент найден в множестве freeSET и его нет в множестве allocSet; один и тот же элемент встречается в freeSET больше одного раза. Работа алгоритма проиллюстрирована на рис. 6. 14 Асрян С.А., Гайсарян С.С., Курмангалеев Ш.Ф., Агабалян А.М., Овсепян Н.Г., Саргсян С.С. Обнаружение ошибок, возникающих при использовании динамической памяти после освобождения. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 7-20 Асрян С.А., Гайсарян С.С., Курмангалеев Ш.Ф., Агабалян А.М., Овсепян Н.Г., Саргсян С.С. Обнаружение ошибок, возникающих при использовании динамической памяти после освобождения. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 7-20 Асрян С.А., Гайсарян С.С., Курмангалеев Ш.Ф., Агабалян А.М., Овсепян Н.Г., Саргсян С.С. Обнаружение ошибок, возникающих при использовании динамической памяти после освобождения. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 7-20 3.2 Предлагаемый метод Results of comparison Предлагаемый подход Mayhem Triton UAF Использование покрытия кода + + - Офлайн-подход символьного выполнения + + + Онлайн-подход символьного выполнения - + - Приоритеты обрабатываемых входных данных + + - Поддержка файлов, в качестве входных данных + + - 4. Результаты Предлагаемый метод был протестирован на синтетических тестах, в том числе и на приведённых примерах (рис. 1, 7), результаты тестирования приведены на рис. 8. Данные результаты показывают, что на синтетических тестах, по сравнению с реализацией Triton, производительность выросла примерно на 80%. Запуск анализа на реальных программах показал, что в большинстве случаев количество символьных уравнений становится настолько большим, что алгоритм покрытия кода Triton не может решить полученные уравнения для всех путей. Для примеров, приведенных на рис. 7, ошибка UAF произойдет, если выполнение программ достигнет строк 21 и 23 для первой и второй программы соответственно. Использование предлагаемого метода позволяет найти входные данные для достижения нужного блока в коде (строки 20-21 и 22-23, рис. 7) и потом проверить данную часть на наличие ошибок UAF. 3.3 Сравнение подходов динамического анализа 3.3 Сравнение подходов динамического анализа В табл. 1 подход, описываемый в данной статье, сравнивается в подходами, применяемыми в Mayhem и Triton. Табл. 1. Результаты сравнения Table 1. Results of comparison Предлагаемый подход Mayhem Triton UAF Использование покрытия кода + + - Офлайн-подход символьного выполнения + + + Онлайн-подход символьного выполнения - + - Приоритеты обрабатываемых входных данных + + - Поддержка файлов, в качестве входных данных + + - В табл. 1 подход, описываемый в данной статье, сравнивается в п применяемыми в Mayhem и Triton. 3.2 Предлагаемый метод Для повышения эффективности поиска ошибок предлагается объединить два вышеописанных алгоритма. Объединенный метод позволяет искать ошибки UAF на разных путях программы, которые получаются из-за внедренных и нетривиальных проверок в коде. На рис. 7 приведены примеры программ, на которых обнаружение повторного освобождения памяти невозможно без использования информации о покрытии кода (из-за присутствия условных переходов, которые будут выполнены только при выполнении программы с определенными входными данными). Рис. 6. Пример работы алгоритма Fig. 6. An example of operation of the algorithm Рис. 6. Пример работы алгоритма Fig. 6. An example of operation of the algorithm Рис. 6. Пример работы алгоритма Рис. 6. Пример работы алгоритма Fig. 6. An example of operation of the algorithm р р р р Fig. 6. An example of operation of the algorithm р р р р Fig. 6. An example of operation of the algorithm 15 Asryan S.A., Gaissaryan S.S., Kurmangaleev Sh. F., Aghabalyan A.М., Hovsepyan N.H., Sargsyan S.S. Dynamic detection of Use After Free bugs. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 7-20 Рис. 7. Ошибки повторного освобождения памяти. Первый пример основан на ошибке UAF в программе libssh. На первом примере ошибка может возникнуть на строке 26, а на втором примере на строке 27 Fig. 7. Errors of double memory freeing. The first example is based on the UAF error in the libssh program. In the first example, an error may occur on line 26, and in the second example on line 27 Fig. 7. Errors of double memory freeing. The first example is based on the UAF error in the libssh program. In the first example, an error may occur on line 26, and in the second example on line 27 example on line 27 Для примеров, приведенных на рис. 7, ошибка UAF произойдет, если выполнение программ достигнет строк 21 и 23 для первой и второй программы соответственно. Использование предлагаемого метода позволяет найти входные данные для достижения нужного блока в коде (строки 20-21 и 22-23, рис. 7) и потом проверить данную часть на наличие ошибок UAF. 3.3 Сравнение подходов динамического анализа В табл. 1 подход, описываемый в данной статье, сравнивается в подходами, применяемыми в Mayhem и Triton. Табл. 1. Результаты сравнения Table 1. 4. Результаты Предлагаемый метод был протестирован на синтетических тестах, в том числе и на приведённых примерах (рис. 1, 7), результаты тестирования приведены на рис. 8. Данные результаты показывают, что на синтетических тестах, по сравнению с реализацией Triton, производительность выросла примерно на 80%. Запуск анализа на реальных программах показал, что в большинстве случаев количество символьных уравнений становится настолько большим, что алгоритм покрытия кода Triton не может решить полученные уравнения для всех путей. 16 Асрян С.А., Гайсарян С.С., Курмангалеев Ш.Ф., Агабалян А.М., Овсепян Н.Г., Саргсян С.С. Обнаружение ошибок, возникающих при использовании динамической памяти после освобождения. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 7-20 Асрян С.А., Гайсарян С.С., Курмангалеев Ш.Ф., Агабалян А.М., Овсепян Н.Г., Саргсян С.С. Обнаружение ошибок, возникающих при использовании динамической памяти после освобождения. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 7-20 Асрян С.А., Гайсарян С.С., Курмангалеев Ш.Ф., Агабалян А.М., Овсепян Н.Г., Саргсян С.С. Обнаружение ошибок, возникающих при использовании динамической памяти после освобождения. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 7-20 Для проверки предложенного подхода нами были специально внесены ошибки UAF в исходный код реальных проектов. Были исследованы проекты: gvgen из пакета graphviz, jasper из пакета libjasper-runtime и gif2rgb из пакета giflib. На данных проектах ошибки были найдены на различных уровнях встраивания. В случае программы gvgen встраивание было выполнено за пределами одной функции, максимальная глубина составляло три уровня (функции). Код встраивания в данном случае представлял с собой условное выражение, связанное с входными данными, и код самой ошибки. Выполнение этого условия приводило к воспроизведению данной ошибки. В проектах jasper и gif2rgb из-за сложности полученных символьных уравнений встраивание было выполнено в пределах только одной функции. Код встраивания был непосредственно кодом ошибки. Также были выделены отдельные участки кода из реальных программ содержащие UAF на которых данный подход смог найти соответствующие ошибки. Для проверки предложенного подхода нами были специально внесены ошибки UAF в исходный код реальных проектов. Были исследованы проекты: gvgen из пакета graphviz, jasper из пакета libjasper-runtime и gif2rgb из пакета giflib. На данных проектах ошибки были найдены на различных уровнях встраивания. В случае программы gvgen встраивание было выполнено за пределами одной функции, максимальная глубина составляло три уровня (функции). Код встраивания в данном случае представлял с собой условное выражение, связанное с входными данными, и код самой ошибки. Выполнение этого условия приводило к воспроизведению данной ошибки. Рис. 8. Результаты сравнения относительно времени выполнения анализа Fig. 8. Comparison results relative to the analysis execution time Рис. 8. 4. Результаты Результаты сравнения относительно времени выполнения анализа Fig. 8. Comparison results relative to the analysis execution time Рис. 8. Результаты сравнения относительно времени выполнения анализа Fig. 8. Comparison results relative to the analysis execution time 5. Заключение В статье представлен метод обнаружения ошибок UAF, возникающих при неправильной обработке указателей на динамическую память. Метод реализован с помощью инфраструктуры Triton [9] на базе алгоритма [10] и алгоритма обнаружения ошибок UAF. После проведенных модификаций и улучшений существующей реализации был получен прирост производительности выполнения анализа. 17 Asryan S.A., Gaissaryan S.S., Kurmangaleev Sh. F., Aghabalyan A.М., Hovsepyan N.H., Sargsyan S.S. Dynamic detection of Use After Free bugs. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 7-20 Dynamic detection of Use After Free bugs 2 S.A Asryan <asryan@ispras.ru> 1, 3, 5, 6 S.S. Gaissaryan <ssg@ispras.ru> 1 Sh. F. Kurmangaleev <kursh@ispras.ru> 4A.М. Aghabalyan <anna.aghabalyan @ispras.ru> 4 N.H. Hovsepyan <narekhnh@ispras.ru> 4S.S Sargsyan <sevaksargsyan@ispras.ru> 1 Ivannikov Institute for System Programming of the Russian Academy of Sciences, 25, Alexander Solzhenitsyn st., Moscow, 109004, Russia 2 Institute for Informatics and Automation Problems of NAS RA, 1, P. Sevak str., Yerevan, 0014, Republic of Armenia, 3 Lomonosov Moscow State University, GSP-1, Leninskie Gory, Moscow, 119991, Russia 4Yerevan State University 1, Alex Manoogian str., Yerevan, 0025, Republic of Armenia 5 Moscow Institute of Physics and Technology (State University), 9 Institutskiy per., Dolgoprudny, Moscow Region, 141700, Russia 6 National Research University Higher School of Economics (HSE) 11 Myasnitskaya str., Moscow, 101000, Russia 2 S.A Asryan <asryan@ispras.ru> 1, 3, 5, 6 S.S. Gaissaryan <ssg@ispras.ru> 1 Sh. F. Kurmangaleev <kursh@ispras.ru> 4A.М. Aghabalyan <anna.aghabalyan @ispras.ru> 4 N.H. Hovsepyan <narekhnh@ispras.ru> 4S.S Sargsyan <sevaksargsyan@ispras.ru> vannikov Institute for System Programming of the Russian Academy of Scien 25, Alexander Solzhenitsyn st., Moscow, 109004, Russia 2 Institute for Informatics and Automation Problems of NAS RA, 1, P. Sevak str., Yerevan, 0014, Republic of Armenia, 3 Lomonosov Moscow State University, GSP-1, Leninskie Gory, Moscow, 119991, Russia 4Yerevan State University 1, Alex Manoogian str., Yerevan, 0025, Republic of Armenia 5 Moscow Institute of Physics and Technology (State University), 9 Institutskiy per., Dolgoprudny, Moscow Region, 141700, Russia 6 National Research University Higher School of Economics (HSE) 11 Myasnitskaya str., Moscow, 101000, Russia Abstract. The article describes new method of use after free bug detection using program dynamic analysis. In memory-unsafe programming languages such as C/C++ this class of bugs mainly accurse when program tries to access specific area of dynamically allocated memory that has been already freed. This method is based on combination of two basic components. The first component tracks all memory operations through dynamic binary instrumentation and searches for inappropriate memory access. It preserves two sets of memory address for all allocation and free instructions. Using both sets this component checks whether current memory is accessible through its address or it has been already freed. It is based on dynamic symbolic execution and code coverage algorithm. It is used to maximize the number of execution paths of the program. Using initial input, it starts symbolic execution of the target program and gathers input constraints from conditional statements. Список литературы Обнаружение ошибок, возникающих при использовании динамической памяти после освобождения. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 7-20 Список литературы [1]. D. Dewey, B. Reaves, P. Trainor. Uncovering Use-After-Free Conditions in Compiled Code. In Proc of the 10th International Conference on Availability, Reliability and Security (ARES), 2015, pp. 90-99 [2]. J. Feist, L. Mounier, ML. Potet. Statically detecting use after free on binary code. Journal of Computer Virology and Hacking Techniques, vol. 10, issue 3, 2014, pp 211- 217 [3]. И.К. Исаев, Д.В. Сидоров, А.Ю. Герасимов, М.К. Ермаков. Avalanche: Применение динамического анализа для автоматического обнаружения ошибок в программах, использующих сетевые сокеты. Труды ИСП РАН, том 21, 2011 г., стр. 55-70. [4]. B. Lee, Ch. Song, Y. Jang, T. Wang. Preventing Use-after-free with Dangling Pointers Nullification. In Proc of the Network and Distributed System Security Symposium, 2015, https://www.ndss-symposium.org/ndss2015/ndss-2015-programme/preventing- use-after-free-dangling-pointers-nullification/, дата обращения 05.05.2018 [5]. Yves Younan. FreeSentry: Protecting Against Use-After-Free Vulnerabilities Due to Dangling Pointers. In Proc of the Network and Distributed System Security Symposium, 2015, https://www.ndss-symposium.org/ndss2015/ndss-2015-programme/freesentry- protecting-against-use-after-free-vulnerabilities-due-dangling-pointers/, дата обращения 05.05.2018 [6]. J. Caballero, G. Grieco, M. Marron, A. Nappa. Undangle: Early Detection of Dangling Pointers in Use-After-Free and Double-Free Vulnerabilities. In Proceedings of the 2012 International Symposium on Software Testing and Analysis, 2012, pp. 133-143 [7]. Sang Kil Cha, Thanassis Avgerinos, Alexandre Rebert and David Brumley. Unleashing MAYHEM on Binary Code. In Proceedings of the 2012 IEEE Symposium on Security and Privacy, 2012, pp. 380-394 [8]. Pin – A Dynamic Binary Instrumentation Tool, https://software.intel.com/en- us/articles/pin-a-dynamic-binary-instrumentation-tool, дата обращения 05.05.2018 [9]. Triton – Dynamic Binary Analysis Framework, https://triton.quarkslab.com/, дата обращения 05.05.2018 [10]. P. Godefroid, M. Y. Levin, D. Molnar. Automated Whitebox Fuzz Testing. In Proceedings of NDSS'2008 (Network and Distributed Systems Security), 2008, pp. 151- 166. [11]. A. Aho, J. Ullman, R. Sethi, M. S. Lam. Compilers: Principles, Techniques, and Tools. Addison Wesley; 2nd edition, September 10, 2006, 1000 p. [12]. Leonardo de Moura, Nikolaj Bjørner. Z3: an efficient SMT solver. In Proceedings of the 14th international conference on Tools and algorithms for the construction and analysis of systems, 2008, pp. 337-340 18 Асрян С.А., Гайсарян С.С., Курмангалеев Ш.Ф., Агабалян А.М., Овсепян Н.Г., Саргсян С.С. Обнаружение ошибок, возникающих при использовании динамической памяти после освобождения. Труды ИСП РАН, том 30, вып 3 2018 г стр 7 20 Асрян С.А., Гайсарян С.С., Курмангалеев Ш.Ф., Агабалян А.М., Овсепян Н.Г., Саргсян С.С. Обнаружение ошибок, возникающих при использовании динамической памяти после освобождения. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 7-20 Асрян С.А., Гайсарян С.С., Курмангалеев Ш.Ф., Агабалян А.М., Овсепян Н.Г., Саргсян С.С. Dynamic detection of Use After Free bugs The new inputs are generated by systematically solving saved constraints using constraint solver and then sorted by number of basic blocks they cover. Proposed method detects use after free bugs by applying first component each time when second one was able to open new path of the program. It was tested on our synthetic tests that were created based on well- known use after free bug patterns. The method was also tested on couple of real projects by injecting bugs on different levels of execution. Keywords: program dynamic analysis; use after free bug; dynamic symbolic execution; code coverage; instrumentation. DOI: 10.15514/ISPRAS-2018-30(3)-1 19 Asryan S.A., Gaissaryan S.S., Kurmangaleev Sh. F., Aghabalyan A.М., Hovsepyan N.H., Sargsyan S.S. Dynamic detection of Use After Free bugs. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 7-20 For citation: Asryan S.A., Gaissaryan S.S., Kurmangaleev Sh. F., Aghabalyan A.М., Hovsepyan N.H., Sargsyan S.S. Dynamic detection of Use After Free bugs. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 7-20 (in Russian). DOI: 10.15514/ISPRAS- 2018-30(3)-1 Asryan S.A., Gaissaryan S.S., Kurmangaleev Sh. F., Aghabalyan A.М., Hovsepyan N.H., Sargsyan S.S. Dynamic detection of Use After Free bugs. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 7-20 DOI: 10.15514/ISPRAS-2018-30(3)-2 DOI: 10.15514/ISPRAS-2018-30(3)-2 For citation: Dudina I.A. Buffer Overflow Detection via Static Analysis: Expectations vs. Reality. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 21-30. DOI: 10.15514/ISPRAS-2018-30(3)-2 References [1]. D. Dewey, B. Reaves, P. Trainor. Uncovering Use-After-Free Conditions in Compiled Code. In Proc of the 10th International Conference on Availability, Reliability and Security (ARES), 2015, pp. 90-99 [2]. J. Feist, L. Mounier, ML. Potet. Statically detecting use after free on binary code. Journal of Computer Virology and Hacking Techniques, vol. 10, issue 3, 2014, pp 211- 217 [3]. Ildar Isaev, Denis Sidorov, Alexander Gerasimov, Mikhail Ermakov. Avalanche: Using dynamic analysis for automatic defect detection in programs based on network sockets. Trudy ISP RAN/Proc. ISP RAS, vol. 21, 2011, pp. 55-70 (in Russian). [4]. B. Lee, Ch. Song, Y. Jang, T. Wang. Preventing Use-after-free with Dangling Pointers Nullification. In Proc of the Network and Distributed System Security Symposium, 2015, https://www.ndss-symposium.org/ndss2015/ndss-2015-programme/preventing- use-after-free-dangling-pointers-nullification/, accessed at 05.05.2018 [5]. Yves Younan. FreeSentry: Protecting Against Use-After-Free Vulnerabilities Due to Dangling Pointers. In Proc of the Network and Distributed System Security Symposium, 2015, https://www.ndss-symposium.org/ndss2015/ndss-2015-programme/freesentry- protecting-against-use-after-free-vulnerabilities-due-dangling-pointers/, accessed at 05.05.2018 [6]. J. Caballero, G. Grieco, M. Marron, A. Nappa. Undangle: Early Detection of Dangling Pointers in Use-After-Free and Double-Free Vulnerabilities. In Proceedings of the 2012 International Symposium on Software Testing and Analysis, 2012, pp. 133-143 [7]. Sang Kil Cha, Thanassis Avgerinos, Alexandre Rebert and David Brumley. Unleashing MAYHEM on Binary Code. In Proceedings of the 2012 IEEE Symposium on Security and Privacy, 2012, pp. 380-394 [8]. Pin – A Dynamic Binary Instrumentation Tool, https://software.intel.com/en- us/articles/pin-a-dynamic-binary-instrumentation-tool, accessed at 05.05.2018 [9]. Triton – Dynamic Binary Analysis Framework, https://triton.quarkslab.com/, accessed at 05.05.2018 [10]. P. Godefroid, M. Y. Levin, D. Molnar. Automated Whitebox Fuzz Testing. In Proceedings of NDSS'2008 (Network and Distributed Systems Security), 2008, pp. 151- 166. [11]. A. Aho, J. Ullman, R. Sethi, M. S. Lam. Compilers: Principles, Techniques, and Tools. Addison Wesley; 2nd edition, September 10, 2006, 1000 p. [12]. Leonardo de Moura, Nikolaj Bjørner. Z3: an efficient SMT solver. In Proceedings of the 14th international conference on Tools and algorithms for the construction and analysis of systems, 2008, pp. 337-340 20 Buffer Overflow Detection via Static Analysis: Expectations vs. Reality I.A. Dudina <eupharina@ispras.ru> Ivannikov Institute for System Programming of the Russian Academy of Sciences, 25, Alexander Solzhenitsyn st., Moscow, 109004, Russia. Lomonosov Moscow State University, GSP-1, Leninskie Gory, Moscow, 119991, Russia. Abstract. Over the last few decades buffer overflow remains one of the main sources of program errors and vulnerabilities. Among other solutions several static analysis techniques were developed to mitigate such program defects. We analyzed different approaches and tools that address this issue to discern common practices and types of detected errors. Also, we explored some popular sets of synthetic tests (Juliet Test Suite, Toyota ITC benchmark) and set of buggy code snippets extracted from real applications to define types of defects that a static analyzer is expected to uncover. Both sources are essential to understand the design goals of a production quality static analyzer. Test suites expose a set of features to support that is easy to understand, classify, and check. On the other hand, they don’t provide a real picture of a production code. Inspecting vulnerabilities is useful but provides an exploitation- biased sample. Besides, it does not include defects eliminated during the development process (probably with the help of some static analyzer). Our research has shown that interprocedural analysis, path-sensitivity and loop handling are essential. An analysis can really benefit from tracking affine relations between variables and modeling C-style strings as a very important case of buffers. Our goal is to use this knowledge to enhance our own buffer overrun detector. Now it can perform interprocedural context- and path-sensitive analysis to detect buffer overflow mainly for static and stack objects with approximately 65% true positive ratio. We think that promising directions are improving string manipulations handling and combining taint analysis with our approaches. Keywords: software error detection; static analysis; buffer overrun DOI: 10.15514/ISPRAS-2018-30(3)-2 1. Introduction Buffer overflow is a type of program defect caused by buffer access with index that exceeds buffer’s bounds. This can lead to a program crash or even to a security vulnerability. Defects of such kind are still common, despite all efforts made to 21 Dudina I.A. Buffer Overflow Detection via Static Analysis: Expectations vs. Reality. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 21-30 eliminate them. There are several techniques one can apply to detect buffer overflows. One approach is to employ testing and dynamic analysis. These methods don’t suffer from false positives, but in most cases, it’s impossible to check all execution paths, so some defects can remain undetected. Another approach is to analyze program code without executing it. In this way, one can find a defect on any path, even rarely executed. In this paper, we will focus on the latter approach known as static analysis. eliminate them. There are several techniques one can apply to detect buffer overflows. One approach is to employ testing and dynamic analysis. These methods don’t suffer from false positives, but in most cases, it’s impossible to check all execution paths, so some defects can remain undetected. Another approach is to analyze program code without executing it. In this way, one can find a defect on any path, even rarely executed. In this paper, we will focus on the latter approach known as static analysis. We are interested in building a buffer overflow detector that is applicable to large C/C++ programs with millions of lines of code while producing decent analysis performance and quality. Basic properties of the algorithms constituting such a detector are well-known and include among others interprocedural analysis, path sensitivity, and loop handling. However, after initial support for these features has been made and the quality goals achieved, it is unclear which direction to choose for the further improvement. The usual development pace that comes from the customer feedback and own code analysis may be not enough. In the following chapters, we’ll overview possible sources of inspiration for the buffer overflow detector development, present our short survey that is based on the buffer overflow-related vulnerabilities sample from the CVE database, then briefly describe our experience of developing an overrun detector as a part of the Svace tool, and present our conclusions from tools and vulnerabilities analysis. 2. Buffer overflow detection techniques and tools There exist many static analysis tools that can detect buffer overflows. In this section, we conduct a brief survey on the most popular methods. Some buffer overflows can be detected during the process of lexical analysis, like in the ITS4 tool [1]. Most common errors and bad patterns can be found at this level. This technique can work really fast and, as it doesn’t involve compilation, can be easily applied to any code, even if it is not complete. As a result, such analysis can be performed “on-the-fly” during the process of code development with IDE, so that erroneous patterns are eliminated on the very early coding stage. Of course, such a lightweight method is far from being sound, i.e. it misses many defects. Even changing the name of a variable can prevent such tools from detecting a defect. To detect more defects a deeper analysis of code is needed. To achieve this, many tools use the idea of abstract interpretation [2]. Some tools chose different numerical abstract domains to implement the analysis of integer index values, buffer sizes, and string lengths. These domains include intervals, zones, octagons, affine equalities, interval linear equalities, convex polyhedra, tropical polyhedra, etc. [3]. Tools based on these approaches derive sound relationship between integer values listed above in varying degrees of precision. Soundness is a major advantage of such tools, but less precise domains produce large number of false positives, while analysis with more precise domains doesn’t scale on many real-world programs. 22 Another popular approach is symbolic execution. The main idea of this method is performing analysis by traversing all paths in a function separately. This approach 22 Дудина И.А. Статический анализ для поиска переполнения буфера: актуальные направления развития. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 21-30 can be used to build a path-sensitive detector i.e. that can find errors that, at the same time, occur only on a certain feasible function path and are not inevitable for any single point from this path alone. While processing a particular path, the analyzer keeps track of variables values and relationships and computes a path predicate, i.e. a conjunction of all corresponding branch conditions that are taken along this path. This information is used to prune infeasible paths and check buffer access instructions. 2. Buffer overflow detection techniques and tools Analyzing all paths in a function can be a challenging task due to the path explosion, so a number of techniques are proposed to reduce this problem. A simple, but often effective approach is to abandon the idea of full path coverage and just to stop the analysis after some threshold or time limit reached. Another approach is to merge symbolic states at join points, preserving path- sensitivity of analysis by providing guard conditions for joined states. Third approach, first introduced in Marple, is employing demand-driven analysis [4], [5], i.e. reducing the set of analyzed paths by focusing only on those that end with buffers access. One of the main obstacles for all mentioned symbolic execution-based approaches is handling loops. Typical solution is to implement some heuristics to handle the most simple and common loops and ignore other loops. However, there are methods proposed to handle loops with multiple paths inside and summarize their effect on program values [6]. Many buffer overflow errors are caused by violations of function contracts. This can happen when a caller of a library or a user function provides unexpected data to a function, or, on the contrary, a function is not able to correctly handle all input cases implied by the contract. Interprocedural analysis is needed to detect such inconsistencies. On the lexical analysis level, formal and actual arguments matching can be based on similar variables names and usually happens only for the well-known library callees like memcpy. For more rigorous scan some tools analyze the whole program as a unified inter-procedural graph. The monomorphic analysis merges information for every call-site — efficient, but imprecise approach. The polymorphic analysis treats each call site individually, so this approach provides context-sensitivity but scales poorly. An alternative approach is using some approximation of a function’s behavior when analyzing its caller. These approximations can be provided in user’s annotations, but they are not always available. A tool can use its own findings obtained by the callee analysis as an approximation. This approached is called summary-based. By choosing the right function order, a tool can minimize the number of missing summaries, but handling recursion still requires additional tricks, e.g. making several analysis passes over strongly connected components of the call graph. 3. Buffer overflow detection tools benchmarking Each case is represented by a pair of a buggy test and a fixed test. These samples cover following features in varying combinations: (i) static, stack and heap buffers; (ii) different element types (char, int, float, struct, etc.); (iii) index calculations (constant, linear and non-linear expressions, passed as an argument or returned from a function, loop variables and array elements); (iv) obtaining buffer address (local/global variable, function argument, pointer arithmetic including loop variables and aliases); (v) buffer size (heap buffers only with constant sizes, pointer casting); (vi) access types (via index, pointer dereference, in a library function, in a string function). Toyota ITC Benchmark is a test suite created by Toyota InfoTechnology Center aimed at the static analysis tool evaluation [8]. It contains 1,276 simple tests (638 erroneous and 638 correct) divided into 9 types and 51 sub-types. Our interest is in the following tests: sub-types “static buffer overrun” (54 cases), “static buffer underrun” (13 cases) from the “static memory” type and sub-types “dynamic buffer overflow” (32 cases), “dynamic buffer underrun” (39 cases) from the “dynamic memory” type. Each case is represented by a pair of a buggy test and a fixed test. These samples cover following features in varying combinations: (i) static, stack and heap buffers; (ii) different element types (char, int, float, struct, etc.); (iii) index calculations (constant, linear and non-linear expressions, passed as an argument or returned from a function, loop variables and array elements); (iv) obtaining buffer address (local/global variable, function argument, pointer arithmetic including loop variables and aliases); (v) buffer size (heap buffers only with constant sizes, pointer casting); (vi) access types (via index, pointer dereference, in a library function, in a string function). 3. Buffer overflow detection tools benchmarking 23 For the past twenty years several studies have been published on evaluating and testing buffer overflow detectors. In addition, there exist different test suites, which 23 Dudina I.A. Buffer Overflow Detection via Static Analysis: Expectations vs. Reality. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 21-30 Dudina I.A. Buffer Overflow Detection via Static Analysis: Expectations vs. Reality. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 21-30 provide sets of synthetic buggy and correct code snippets to test the abilities and false positive rate of static analysis tools. One of the biggest and probably the most popular benchmark is Juliet Test Suite C/C++, created by NSA’s Center for Assured Software (CAS) [7]. For C/C++ code it contains 64,099 test cases tagged by CWE entries. Groups corresponding to buffer overflow defects are CWE 121 — “Stackbased Buffer Overflow” (4,968 tests), CWE 122 — “Heapbased Buffer Overflow” (5,922 tests), CWE 124 — “Buffer Underwrite” (2,048 tests), CWE 126 — “Buffer Over-read” (1,452 tests), and CWE 127 — “Buffer Under-read” (2048 tests). Tests in this suite are also tagged with a number called “flow variant” that represents the complexity of control and data flow in a particular test case. Control flow variants cover different types of conditionals (e.g. STATIC_CONST_FIVE==5, globalReturnsTrueOrFalse(), etc.) and different control statements (switch, while, etc.). Data flow variants describe many types of intraprocedural data flow and interprocedural interaction, e.g. data passing through function arguments (via pointer, C++ reference, array, container, etc.), return value, global variable, etc. There are many flow variants that represent C++-specific features and not applicable to C-tests. We noticed that the distribution of the flow variants is close to uniform in groups of our interest. Another observation is large number of tests involving wide characters. Many tests contain library function usage, e.g. memcpy-like functions, string manipulations, format string processing, etc. manipulations, format string processing, etc. Toyota ITC Benchmark is a test suite created by Toyota InfoTechnology Center aimed at the static analysis tool evaluation [8]. It contains 1,276 simple tests (638 erroneous and 638 correct) divided into 9 types and 51 sub-types. Our interest is in the following tests: sub-types “static buffer overrun” (54 cases), “static buffer underrun” (13 cases) from the “static memory” type and sub-types “dynamic buffer overflow” (32 cases), “dynamic buffer underrun” (39 cases) from the “dynamic memory” type. 4. Survey on overflow-related CVEs We believe that although evaluating with a test suite could give a good insight in a particular tool’s abilities, any test suite alone cannot perfectly represent the whole populations of buffer overflow defects in real code. One (but not the only one) noble 24 Дудина И.А. Статический анализ для поиска переполнения буфера: актуальные направления развития. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 21-30 goal for static analyzers is to prevent security vulnerabilities to sneak in the project source code. We wanted a better understanding of the features of a static analyzer that are more or less important for achieving this goal. Our survey technique was inspired by [9] and we mostly followed in their footsteps to produce a set of vulnerabilities to classify. We have to note that detection of exploitable vulnerabilities is not the only goal of a static analyzer. Still there are some types of defects that don’t lead to vulnerabilities or may not be exploited with ease, but it is undesirable to have those in the source code. Besides, we believe that nowadays developers more intensively use different (static and/or dynamic) analysis tools before releasing the product. For this reason, many simple defects are eliminated during the development process and don’t appear in the vulnerability databases. Consequently, we think that analysis of the vulnerabilities can reveal the weakest sides of modern static analysis and show potential improvement directions. First of all, we have randomly picked 100 entries from the “overflow” category from the CVE database [10]. For 25 of them we could find a source code of the vulnerable version to inspect. For each defect, we have studied its causes in the code and then classified the defect by several attributes. Our set of attributes is based on the taxonomy provided in [11] with some changes. Our first insight is that there are some trends in our sample that can be explained by the source of this sample (vulnerability database): (i) most of the overflows from our sample (72%) happened on write memory access, only few on read access; (ii) only the upper bounds of buffers are exceeded in the defects from our sample; (iii) almost all defects (92%) occurred when tainted data (unbounded data from network, file read, input parameters etc.) overflowed some buffer. We also noticed that simple errors (e.g. 4. Survey on overflow-related CVEs using unsafe functions like strcpy) are present in the old code (before 2010), but rarely in the late entries. We believe that this can be partially explained by the usage of code analysis tools. In our sample about a half of overflowed buffers (48%) reside on a stack, other half (48%) is allocated on a heap, and just a few are global variables. 40% of all defects have overflowed buffer accessed via index (e.g. buf[i]), 12% via pointer dereference, 44% via library calls, 24% of which are string functions. The latter requires C-strings modeling to properly analyze such patterns. When buffer is accessed in a library call, we think of size/limit argument as an index (when it’s reasonable) for further investigation. According to our data, 48% of all vulnerable buffers have constant size (all stack and static buffers and a few buffers on the heap). Another 16% have a size that is calculated as a linear combination of other variables. As a result, almost half of all inspected defects require deep analysis of integer variables relationship to detect them. Another feature that we have evaluated for every entry is whether buffer allocation is global or resides in the same function with buffer access. We have found that this 25 Dudina I.A. Buffer Overflow Detection via Static Analysis: Expectations vs. Reality. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 21-30 Dudina I.A. Buffer Overflow Detection via Static Analysis: Expectations vs. Reality. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 21-30 fer Overflow Detection via Static Analysis: Expectations vs. Reality. Trudy ISP RAN /Proc. ISP RAS, 2018, pp. 21-30 is true only for 24% of defects. On the other hand, all index calculations are in the same function with the access in 32% of defects. Both properties are true for 12% of defects. It follows from the foregoing that interprocedural analysis is essential for buffer overflow detection. Last thing that we have checked is whether there exists a program point that any path through this point will lead to a corresponding error. If there is no such point, then we assume that path-sensitive analysis is needed to detect this defect. Our sample contains only 28% of defects, for which such a program point exists. This means that path-sensitivity will provide the real advantage for a static analysis tool. [2]. P. Cousot and R. Cousot. Abstract Interpretation: a unified lattice model for static analysis of programs by construction or approximation of fixpoints. In Proceedings of the 4th ACM SIGACT-SIGPLAN symposium on Principles of programming languages, 1977, pp. 238–252. 5. Svace buffer overrun detector Svace is a static analysis tool that is designed to find as many defects of different types as possible with few false positives and acceptable analysis time [12]. The purpose of this work is to improve the Svace buffer overflow detector with the most needed features. Our detector implements the interprocedural path-sensitive detection algorithm based on symbolic execution with state merging [13]. For now, the analysis scope is limited to detection overflows of buffers with compile-time- known size. Our detector looks for faulty paths in a function, i.e. it reports a warning if it finds a path that for any input values is either infeasible or produces an error. Such a strict defect definition is chosen to prevent many false positives caused by unknown function preconditions. For a buffer access instruction, we collect a predicate that implies that there exists a faulty path through this instruction. We use an SMT solver to search a solution for this predicate if any. In case of this formula is satisfiable, we use its model provided by the solver to extract a faulty path. It follows from our experience that simply asking solver for any index value that exceeds buffer bounds in our case leads to many false positives. Reasons for that are unknown function precondition and symbolic path conditions being not precise enough (due to poor loop handling, calls of unknown or complex functions, etc.). Our interprocedural analysis is implemented using summaries. In the function summary, we save the information about relationships between integer values on function entry and exit points. We also save overflow conditions for those input- dependent buffer accesses whose correctness can only be checked in the caller context. Such facts can be propagated to the caller more than once, so the analysis can find an overflow of a buffer allocated in a function that is far away on the call stack from a function with the access instruction. We also implemented a heuristic to handle simple loops that have an inductive variable iterating over an arithmetic progression. Currently on Android 5.0.2 our detector emits 351 warnings with 65% true-positive ratio. 26 Дудина И.А. Статический анализ для поиска переполнения буфера: актуальные направления развития. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 21-30 Дудина И.А. Статический анализ для поиска переполнения буфера: актуальные направления развития. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 21-30 [3]. X. Allamigeon. Static analysis of memory manipulations by abstract interpretation – Algorithmics of tropical polyhedra, and application to abstract interpretation. PhD thesis, Ecole Polytechnique X, Nov. 2009. [Online]. Available: https://pastel.archives- ouvertes.fr/pastel-00005850, accessed: 2018-04-08. 6. Conclusion We have inspected a number of buffer overflow test suites, related CVE entries, and the source code of large production projects that our tool regularly analyzes. All three sources are essential to understand the design goals of a production quality static analyzer. Test suites expose a set of features to support that is easy to understand, classify, and check. On the other hand, they don’t provide a real picture of a production code. Inspecting vulnerabilities is useful but provides an exploitation-biased sample. Besides, it does not include defects eliminated during the development process (probably with the help of some static analyzer). Finally, while developing a static analyzer one always deals with false positives produced by the tool and reported by customers, but getting false negative samples is much more difficult. True positives reported by the other tools could be useful, but most of the state-of-the-art tools are proprietary and their results are closed. From what has been said above it follows that interprocedural analysis, path- sensitivity and loop handling are essential. An analysis can really benefit from tracking affine relations between variables and modeling C-style strings as a very important case of buffers. Our current goal is to improve the Svace buffer overflow detector to reduce the number of false negatives while preserving the moderate level of false positives. For the aforementioned reasons, we think that the most promising directions are handling buffers with dynamic size, C-string modeling, and tracking tainted values. We are working now on the extension of our detection technique described in Section 5 by tracking string length changes happening during string operations in much the same way as we track buffer indexes while calculating integer values. We believe that this will be sufficient for most of cases, but there are some promising works in the area of string solvers [14] that would additionally allow to track also string contents. As we have seen, static analysis detection of buffer overflows requires a number of techniques from vastly various fields to move on the road from expectations to real code, and there will always be a way to go. [1]. J. Viega, J. T. Bloch, Y. Kohno, and G. McGraw. Its4: A static vulnerability scanner for c and c++ code. In Proceedings of the 16th Annual Computer Security Applications Conference, 2000, pp. 257-269. References [1]. J. Viega, J. T. Bloch, Y. Kohno, and G. McGraw. Its4: A static vulnerability scanner for c and c++ code. In Proceedings of the 16th Annual Computer Security Applications Conference, 2000, pp. 257-269. [2]. P. Cousot and R. Cousot. Abstract Interpretation: a unified lattice model for static analysis of programs by construction or approximation of fixpoints. In Proceedings of the 4th ACM SIGACT-SIGPLAN symposium on Principles of programming languages, 1977, pp. 238–252. [3]. X. Allamigeon. Static analysis of memory manipulations by abstract interpretation – Algorithmics of tropical polyhedra, and application to abstract interpretation. PhD thesis, Ecole Polytechnique X, Nov. 2009. [Online]. Available: https://pastel.archives- ouvertes.fr/pastel-00005850, accessed: 2018-04-08. 27 Dudina I.A. Buffer Overflow Detection via Static Analysis: Expectations vs. Reality. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 21-30 Buffer Overflow Detection via Static Analysis: Expectations vs. Reality. Trudy ISP RAN /Proc. ISP RAS, 3, 2018, pp. 21-30 [4]. W. Le and M. L. Soffa. Marple: A Demand-Driven Path-Sensitive Buffer Overflow Detector. In Proceedings of the 16th ACM SIGSOFT International Symposium on Foundations of software engineering, 2008, p. 272-282. [4]. W. Le and M. L. Soffa. Marple: A Demand-Driven Path-Sensitive Buffer Overflow Detector. In Proceedings of the 16th ACM SIGSOFT International Symposium on Foundations of software engineering, 2008, p. 272-282. [5]. L. Li, C. Cifuentes, and N. Keynes. Practical and effective symbolic analysis for buffer overflow detection. In Proceedings of the Eighteenth ACM SIGSOFT International Symposium on Foundations of Software Engineering, 2010, pp. 317– 326. y p g g pp [6]. X. Xie, Y. Liu, W. Le, X. Li, and H. Chen. S-looper: automatic summarization for multipath string loops. In Proceedings of the 2015 International Symposium on Software Testing and Analysis, 2015, pp. 188–198. g y , , pp [7]. Juliet Test Suite v1.2 for C/C++. User Guide. Available: https://samate.nist.gov/SRD/around.php#juliet_documents, accessed: 2018-04-08. [8]. S. Shiraishi, V. Mohan, and H. Marimuthu. Test suites for benchmarks of static analysis tools. In Proceedings of the 2015 IEEE International Symposium on Software Reliability Engineering Workshops (ISSREW), Nov 2015, pp. 12–15. g g p pp [9]. T. Ye, L. Zhang, L. Wang, and X. Li. An Empirical Study on Detecting and Fixing Buffer Overflow Bugs. In Proceedings of the 2016 IEEE International Conference on Software Testing, Verification and Validation, 2016, pp. 91–101. [10]. CVE security vulnerability database. Security vulnerabilities, exploits, references and more. Available: https://www.cvedetails.com/index.php, accessed: 2018-04-08. [11]. K. Kratkiewicz and R. References Lippmann. A taxonomy of buffer overflows for evaluating static and dynamic software testing tools. In Proceedings of Workshop on Software Security Assurance Tools, Techniques, and Metrics, vol. 500, 2006, pp. 44-51. [12]. A. Borodin and A. Belevantcev. A static analysis tool Svace as a collection of analyzers with various complexity levels. Trudy ISP RAN /Proc. ISP RAS, vol. 27, issue 6, 2015, pp. 111–134. [13]. I.A. Dudina and A.A. Belevantsev. Using static symbolic execution to detect buffer overflows. Programming and Computer Software, vol. 43, no. 5, 2017, pp. 277–288. DOI: 10.1134/S0361768817050024. [14]. Y. Zheng, X. Zhang, and V. Ganesh. Z3-str: A z3-based string solver for web application analysis. In Proceedings of the 2013 9th Joint Meeting on Foundations of Software Engineering, 2013, pp. 114–124. DOI: 10.15514/ISPRAS-2018-30(3)-2 DOI: 10.15514/ISPRAS-2018-30(3)-2 Для цитирования: Дудина И.А. Статический анализ для поиска переполнения буфера: актуальные направления развития. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 21-30 (на английском языке). DOI: 10.15514/ISPRAS-2018-30(3)-2 Для цитирования: Дудина И.А. Статический анализ для поиска переполнения буфера: актуальные направления развития. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 21-30 (на английском языке). DOI: 10.15514/ISPRAS-2018-30(3)-2 Ключевые слова: анализ программ; статический анализ; переполнение буфера Ключевые слова: анализ программ; статический анализ; переполнение буфера Статический анализ для поиска переполнения буфера: актуальные направления развития И.А. Дудина <eupharina@ispras.ru> Институт системного программирования им. В.П. Иванникова РАН, 109004, Россия, г. Москва, ул. А. Солженицына, д. 25 Московский государственный университет им. М.В. Ломоносова, 119991, Россия, Москва, Ленинские горы, д. 1 И.А. Дудина <eupharina@ispras.ru> Институт системного программирования им. В.П. Иванникова РАН, 109004, Россия, г. Москва, ул. А. Солженицына, д. 25 Московский государственный университет им. М.В. Ломоносова, 119991, Россия, Москва, Ленинские горы, д. 1 28 Аннотация. В последние десятилетия переполнение буфера остаётся одним из главных источников программных ошибок и эксплуатируемых уязвимостей. Среди прочих подходов к устранению подобных дефектов активное развитие получили различные методы статического анализа. В работе рассматриваются основные подходы и инструменты, используемые для решения этой задачи, с целью выявить наиболее популярные методы и типы обнаруживаемых ошибок. Также исследованы наборы Дудина И.А. Статический анализ для поиска переполнения буфера: актуальные направления развития. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 21-30 синтетических тестов (Juliet Test Suite, Toyota ITC benchmark) и выборка фрагментов кода реальных приложений, содержащих эксплуатируемую ошибку переполнения буфера. Для понимания направлений развития промышленного статического анализатора важно рассматривать оба эти источника примеров ошибочных программ. Наборы тестов очерчивают круг ситуаций, которые необходимо поддержать в анализаторе, при этом их легко понять, классифицировать и проверить. С другой стороны, они не отражают распределение таких ситуаций в реальном коде. Выборка уязвимостей из промышленных проектов также представляет интерес для исследования, но оказывается смещённой в сторону эксплуатируемых ошибок и к тому же не включает ошибки, исправленные на стадии разработки (возможно, как раз с использованием статического анализатора). Полученные данные были использованы для выделения основных шаблонов дефектов, которые должен обнаруживать статической анализатор с точки зрения пользователя. В результате исследования к наиболее важным возможностям статического анализатора были отнесены межпроцедурный путе- и контекстно-чувствительный анализ, а также базовая поддержка циклов. Кроме того, полезными оказываются отслеживание аффинных отношений между переменными и моделирование строк как важного случая использования массивов. Результаты данного исследования используются для улучшения детектора переполнения буфера, реализованного в рамках инфраструктуры статического анализатора Svace. На данный момент используется межпроцедурный чувствительный к путям и контексту анализ, позволяющий обнаруживать переполнения буфера на стеке и в статической памяти с долей истинных срабатываний 65%. По результатам исследования наиболее перспективными направлениями представляются поддержка строковых операций и внедрение анализа помеченных данных в имеющиеся подходы. Список литературы [1]. J. Viega, J. T. Bloch, Y. Kohno, and G. McGraw. Its4: A static vulnerability scanner for c and c++ code. In Proceedings of the 16th Annual Computer Security Applications Conference, 2000, pp. 257-269. [1]. J. Viega, J. T. Bloch, Y. Kohno, and G. McGraw. Its4: A static vulnerability scanner for c and c++ code. In Proceedings of the 16th Annual Computer Security Applications Conference, 2000, pp. 257-269. [2]. P. Cousot and R. Cousot. Abstract Interpretation: a unified lattice model for static analysis of programs by construction or approximation of fixpoints. In Proceedings of the 4th ACM SIGACT-SIGPLAN symposium on Principles of programming languages, 1977, pp. 238–252. [2]. P. Cousot and R. Cousot. Abstract Interpretation: a unified lattice model for static analysis of programs by construction or approximation of fixpoints. In Proceedings of the 4th ACM SIGACT-SIGPLAN symposium on Principles of programming languages, 1977, pp. 238–252. [3]. X. Allamigeon. Static analysis of memory manipulations by abstract interpretation – Algorithmics of tropical polyhedra, and application to abstract interpretation. PhD thesis, Ecole Polytechnique X, Nov. 2009. [Online]. Available: https://pastel.archives- ouvertes.fr/pastel-00005850, accessed: 2018-04-08. [3]. X. Allamigeon. Static analysis of memory manipulations by abstract interpretation – Algorithmics of tropical polyhedra, and application to abstract interpretation. PhD thesis, Ecole Polytechnique X, Nov. 2009. [Online]. Available: https://pastel.archives- ouvertes.fr/pastel-00005850, accessed: 2018-04-08. 29 Dudina I.A. Buffer Overflow Detection via Static Analysis: Expectations vs. Reality. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 21-30 [4]. W. Le and M. L. Soffa. Marple: A Demand-Driven Path-Sensitive Buffer Overflow Detector. In Proceedings of the 16th ACM SIGSOFT International Symposium on Foundations of software engineering, 2008, p. 272-282. [4]. W. Le and M. L. Soffa. Marple: A Demand-Driven Path-Sensitive Buffer Overflow Detector. In Proceedings of the 16th ACM SIGSOFT International Symposium on Foundations of software engineering, 2008, p. 272-282. [5]. L. Li, C. Cifuentes, and N. Keynes. Practical and effective symbolic analysis for buffer overflow detection. In Proceedings of the Eighteenth ACM SIGSOFT International Symposium on Foundations of Software Engineering, 2010, pp. 317– 326. y p g g pp [6]. X. Xie, Y. Liu, W. Le, X. Li, and H. Chen. S-looper: automatic summarization for multipath string loops. In Proceedings of the 2015 International Symposium on Software Testing and Analysis, 2015, pp. 188–198. g y pp [7]. Juliet Test Suite v1.2 for C/C++. User Guide. Режим доступа: https://samate.nist.gov/SRD/around.php#juliet_documents, дата обращения: 2018-04- 08. [8]. S. Shiraishi, V. Extracting architectural information from source code of ARINC 653-compatible application software using CEGAR-based approach S.L. Lesovoy <lesovoy@ispras.ru> Ivannikov Institute for System Programming of RAS, 25, Alexander Solzhenitsyn st., Moscow, 109004, Russia. Abstract. It may be useful to analyze and reuse some components of legacy systems during development of new systems. By using a model-based approach it is possible to build an architecture model from the existing source code of the legacy system. The purpose of using architecture models is to analyze the system’s static and dynamic features during the development process. These features may include real-time performance, resources consumption, reliability etc. The architecture models can be used as for system analysis as well as for reusing some components of the legacy system in the new design. In many cases it will allow to avoid creation of a new system from scratch. For creation of the architectural models various modeling languages can be used. In the present work Architecture Analysis & Design Language (AADL) is used. The paper describes an algorithm of extracting architectural information from source code of ARINC 653-compatible application software. ARINC 653 specification defines the requirements for software components of Integrated Modular Avionics (IMA) systems. To access the various services of ARINC 653 based OS an application software uses function calls defined in the APplication/Executive (APEX) interface. Architectural information in source code of application software compliant with ARINC 653 specification includes different objects and their attributes such as processes in each partition, objects for interpartition and intrapartition communications, as well as global variables. To collect the architectural information, it is necessary to extract all APEX calls from source code of application software. The extracted architectural information can be further used for creation the architecture models of the system. For source code analysis an approach based on Counterexample-guided abstraction refinement (CEGAR) algorithm is used. CEGAR algorithm explores possible execution paths of the program using its representation in the form of Abstract Reachability Graph (ARG). In a classical CEGAR algorithm a path in a program to be explored is called a counterexample and it means a path to the error state. In CPAchecker tool the basic predicate-based CEGAR algorithm has been extended for explicit-value analysis. In this paper the extended for explicit-value analysis CEGAR algorithm is applied for the task of extracting architecture information from source code. Список литературы Mohan, and H. Marimuthu. Test suites for benchmarks of static analysis tools. In Proceedings of the 2015 IEEE International Symposium on Software Reliability Engineering Workshops (ISSREW), Nov 2015, pp. 12–15. [9]. T. Ye, L. Zhang, L. Wang, and X. Li. An Empirical Study on Detecting and Fixing Buffer Overflow Bugs. In Proceedings of the 2016 IEEE International Conference on Software Testing, Verification and Validation, 2016, pp. 91–101. [10]. CVE security vulnerability database. Security vulnerabilities, exploits, references and more. Режим доступа: https://www.cvedetails.com/index.php, дата обращения: 2018- 04-08. [11]. K. Kratkiewicz and R. Lippmann. A taxonomy of buffer overflows for evaluating static and dynamic software testing tools. In Proceedings of Workshop on Software Security Assurance Tools, Techniques, and Metrics, vol. 500, 2006, pp. 44-51. [12]. A. Borodin and A. Belevantcev. A static analysis tool Svace as a collection of analyzers with various complexity levels. Trudy ISP RAN /Proc. ISP RAS, vol. 27, issue 6, 2015, pp. 111–134. DOI: 10.15514/ISPRAS-2015-27(6)-8. [13]. I.A. Dudina and A.A. Belevantsev. Using static symbolic execution to detect buffer overflows. Programming and Computer Software, vol. 43, no. 5, 2017, pp. 277–288. DOI: 10.1134/S0361768817050024. [14]. Y. Zheng, X. Zhang, and V. Ganesh. Z3-str: A z3-based string solver for web application analysis. In Proceedings of the 2013 9th Joint Meeting on Foundations of Software Engineering, 2013, pp. 114–124. 30 1. Introduction The purpose of using architecture models is to analyze the system’s static and dynamic features during the development process. These features may include real- time performance, resources consumption, reliability etc. This aspect is extremely important while developing complex systems that include both software and hardware components produced by the different suppliers. Using model-based approach at the early stages of the development of the project will help to avoid a waste of time and money for correction of system defects when the system is created. For creation of the architectural models various modeling languages can be used. The most popular ones used for architecture modelling are SysML[1] and AADL[2,3]. The model-based development process includes two major project steps. At the first step, the system model is being created. There are different levels for representation of the system model. The primary focus of this paper is the architectural models. On the second step of the project the system model will be used as input for detailed design and system implementation. This step may also include the model transformations to some intermediate formats used in system design and implementation. In the ideal case, the system model can be transformed to the source code of the system. It may be useful to analyze and reuse some components of legacy systems during development of new systems. By using a model-based approach it is possible to build an architecture model from the existing source code of the legacy system. This model can be used as for system analysis as well as for reusing some components of the legacy system in the new design. In many cases, it will allow to avoid creation of a new system from scratch. A process of model creation for existing system is called a model-driven reverse engineering (MDRE). If the source code of a legacy system is available then it is possible to build a system model from its source code. This process contains two steps. The first step is source code analysis. The second step is model transformations to the target output format. This paper describes the first step – source code analysis for application software that is based on Integrated Modular Avionics (IMA) architecture and ARINC 653 specification. The goal of source code analysis is to extract architecture information that is necessary for creation of the architecture model of the system. DOI: 10.15514/ISPRAS-2018-30(3)-3 For citation: Lesovoy S.L. Extracting architectural information from source code of ARINC 653-compatible application software using CEGAR-based approach. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 31-46. DOI: 10.15514/ISPRAS-2018-30(3)-3 For citation: Lesovoy S.L. Extracting architectural information from source code of ARINC 653-compatible application software using CEGAR-based approach. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 31-46. DOI: 10.15514/ISPRAS-2018-30(3)-3 Extracting architectural information from source code of ARINC 653-compatible application software using CEGAR-based approach The main contribution of this paper is the application the ideas of counterexample and path feasibility check for the task of extracting the architectural information from source code. 31 Lesovoy S.L. Extracting architectural information from source code of ARINC 653-compatible application software using CEGAR-based approach. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 31-46 Keywords: architectural information, architecture models, ARINC 653, IMA, CEGAR Lesovoy S.L. Extracting architectural information from source code of ARINC 653-compatible application software using CEGAR-based approach. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 31-46 Keywords: architectural information, architecture models, ARINC 653, IMA, CE 2 IMA IMA architecture is widely used in avionics industry for implementation the safety critical applications. In IMA systems multiple avionics applications can share resources of a single hardware platform (core module) without any mutual influence. ARINC 653 [4] is a set of documents that define the requirements for software components of IMA systems. The key concept of ARINC 653 is a partition. ARINC 653 compatible Operating System (OS) provides a dedicated portion of memory and predefined time slot within a fixed schedule for each partition. It prevents any affect from software executing in one partition to software in other partitions. ARINC 653 specification defines that IMA system may include the following software components: core software, application partitions and system partitions. Core software consists of OS and APplication/EXecutive (APEX) interface. The APEX interface defines a set of services provided by the OS for application software. In each application, partition can be allocated only one application. System partitions contain system software that can directly interact with the OS without using APEX interface. Communications between applications allocated in different partitions is called the interpartition communication. The interpartition communication is only available via communication channels. To access a communication channel the application can use the ports created inside a partition. ARINC 653 supports two port types: sampling ports and queuing ports. An application software compliant with ARINC 653 specification has a typical structure. Such a software can be located in a single partition or in multiple partitions. An application software compliant with ARINC 653 specification has a typical structure. Such a software can be located in a single partition or in multiple partitions. To access the various services of ARINC 653 based OS an application software uses function calls defined in the APEX interface. For each partition several processes can be created. One process is responsible for partition initialization. This process creates other processes and various objects. Finally, when the initialization of partition has been finished this process sets the partition to NORMAL state using SET_PARTITION_MODE function call. Since this moment a scheduling for all processes created inside a partition is started. It is important to note that after the initialization of partition has been finished there is no way to create any new processes and objects. An ARINC 653 process is quite similar to a POSIX thread. 1. Introduction The rest of the paper is organized as follows. Section 2 provides an overview of IMA architecture and ARINC 653 specification. It also contains a simple example of source code to be used for further analysis. Section 3 describes the concept of 32 Лесовой С.Л. Извлечение архитектурной информации из исходного кода ARINC 653 совместимых приложений с использованием алгоритма CEGAR. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 31-46 Лесовой С.Л. Извлечение архитектурной информации из исходного кода ARINC 653 совместимых приложений с использованием алгоритма CEGAR. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 31-46 architectural information in source code, a general approach and a particular algorithm used for extracting architectural information from source code. Section 4 describes the results and outlines the future research and development tasks. 2 IMA To create a process, it is necessary to create a structure that contains the process’s attributes and pass it to CREATE_PROCESS function. ENTRY_POINT is an attribute of the process that contains the address of the function that will be called when the process is started. This function implements the application logic of the process and its communication procedures with other processes. 33 Lesovoy S.L. Extracting architectural information from source code of ARINC 653-compatible application software using CEGAR-based approach. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 31-46 Lesovoy S.L. Extracting architectural information from source code of ARINC 653-compatible application software using CEGAR-based approach. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 31-46 Communications between processes within a single partition is called intrapartition communication. Buffers and blackboards are used for communication between processes inside a partition. Semaphores, events and mutexes are used for process synchronization. Any objects for communication and synchronization can be created using function calls defined in the APEX interface. The processes located inside the same partition can also communicate via global variables. At the end of this section, a simple example of application software will be demonstrated and explained. The source code fragment of the application software compliant with ARINC 653 specification is shown in Fig.1. Source code fragment in Fig.1 includes three functions: Run_10_Hz, Run_Monitor and main. In function main two processes, one event and three sampling ports are created. while (1) { while (1) { SET_EVENT ( wakeup, ret ); ( ) { SET_EVENT ( wakeup, ret ); SET_EVENT ( wakeup, ret ); READ_SAMPLING_MESSAGE(port_raw_data, (MESSAGE_ADDR_TYPE)&sensor_data, &len, &validity, &ret); &len, &validity, &ret); // Some operations with data … WRITE_SAMPLING_MESSAGE(port_data_out, (MESSAGE_ADDR_TYPE)&output_data, &len2, &ret); &len2, &ret); PERIODIC_WAIT(&ret_pause); } Лесовой С.Л. Извлечение архитектурной информации из исходного кода ARINC 653 совместимых приложений с использованием алгоритма CEGAR. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 31-46 PROCESS_ATTRIBUTE_TYPE Proc_Monitor_Attributes; Proc_Monitor_Attributes.ENTRY_POINT = Run_Monitor; Proc_Monitor_Attributes.PERIOD = INFINITE_TIME_VALUE; strncpy(Proc_Monitor_Attributes.NAME, "Proc_Monitor", sizeof(PROCESS NAME TYPE)); strncpy(Proc_Monitor_Attributes.NAME, "Proc_Monitor", i f( )) sizeof(PROCESS_NAME_TYPE)); CREATE_PROCESS( &Proc_Monitor_Attributes, &pid_p1, &ret ); START( pid_p1, &ret ); EVENT_NAME_TYPE EventName; strncpy( EventName, "Wakeup", …); CREATE_EVENT ( EventName, wakeup, ret ); CREATE_EVENT ( EventName, wakeup, ret ); CREATE_SAMPLING_PORT( "RAW_DATA", CREATE_SAMPLING_PORT( "RAW_DATA", port_size, DESTINATION, period, &port_raw_data, …); port_size, DESTINATION, period, &port_raw_data, …); CREATE_SAMPLING_PORT( "DATA_OUT", port_size, SOURCE, period, &port_data_out, …); CREATE SAMPLING PORT( "STATUS" port size port_size, SOURCE, period, &port_data_out, …); port_size, SOURCE, period, &port_data_o CREATE_SAMPLING_PORT( "STATUS", port_size, CREATE_SAMPLING_PORT( "STATUS", port_size, SOURCE, period, &port_status …); SOURCE, period, &port_status …); SET_PARTITION_MODE ( NORMAL, &ReturnCode ); return 0; SET_PARTITION_MODE ( NORMAL, &ReturnCode ); return 0; A function Run_10_Hz is called periodically with period 10 milliseconds. This value for period was set in PERIOD attribute during the creation of the first process. Each time when the function Run_10_Hz is called, it activates the event Wakeup, reads a static void Run_Monitor(void) { static void Run_Monitor(void) { _ while (1) { WAIT_EVENT ( wakeup, TimeOut, ret ); RESET_EVENT ( wakeup, ret ); // Some operations with data … WRITE_SAMPLING_MESSAGE( port_status, (MESSAGE ADDR TYPE)&status data // Some operations with data … WRITE_SAMPLING_MESSAGE( port_status, (MESSAGE_ADDR_TYPE)&status_data, &len, &ret ); } void main(void) { PROCESS_ATTRIBUTE_TYPE Proc_10_Hz_Attributes; Proc_10_Hz_Attributes.ENTRY_POINT = Run_10_Hz; Proc_10_Hz_Attributes.PERIOD = 100000000LL; void main(void) { PROCESS_ATTRIBUTE_TYPE Proc_10_Hz_Attributes; Proc_10_Hz_Attributes.ENTRY_POINT = Run_10_Hz; Proc_10_Hz_Attributes.PERIOD = 100000000LL; strncpy(Proc_10_Hz_Attributes.NAME, "Proc_10_Hz", sizeof(PROCESS_NAME_TYPE)); CREATE_PROCESS( &Proc_10_Hz_Attributes, &pid_p0, &ret ); START( pid p0 &ret ); Proc_10_Hz_Attributes.PERIOD = 100000000LL; Proc_10_Hz_Attributes.PERIOD = 100000000LL; strncpy(Proc_10_Hz_Attributes.NAME, "Proc_10_Hz", sizeof(PROCESS_NAME_TYPE)); strncpy(Proc_10_Hz_Attributes.NAME, "Proc_10_Hz", sizeof(PROCESS_NAME_TYPE)); CREATE_PROCESS( &Proc_10_Hz_Attributes, &pid_p0, &ret ); START( pid_p0, &ret ); 34 Лесовой С.Л. Извлечение архитектурной информации из исходного кода ARINC 653 совместимых приложений с использованием алгоритма CEGAR. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 31-46 3.1 Architectural information in source code The main goal of source code analysis in the paper is to extract the architectural information from it. Architectural information in source code of application software compliant with ARINC 653 specification includes the processes in each partition and their attributes, all objects created for interpartition and intrapartition communications and their attributes. It also includes the ways of communications and synchronizations between processes located inside the same partition or in different partitions. If the global variables are used for communication between processes inside partition then these variables also should be considered as architectural information. The source code fragment in Fig.1 contains the following architectural information: two processes, one event and three sampling ports. Attributes of each process and each object (event, port) are also important architectural information. For synchronization between two processes the event object is used. In the first process APEX call SET_EVENT is used to activate an event. The second process uses APEX call WAIT_EVENT for receiving this event. Sampling ports are used in both processes to communicate with external environment, i.e. with processes allocated in other partitions or with external devices. The source code of real avionic application can contain hundreds of processes communicating with each other and with external environment via large number of the objects. Extracting such architectural information from source code can be time consuming task. This paper proposes a way to do it automatically. The next sections describe a general approach and a particular algorithm used for source code analysis. Fig. 1. Source code fragment with APEX calls. For process creation, APEX call CREATE_PROCESS is used. The first argument of CREATE_PROCESS has a type PROCESS_ATTRIBUTE_TYPE. It is a structure that contains attributes for the created process. The ENTRY_POINT attribute is equal to Run_10_Hz for the first process and is equal to Run_Monitor for the second one. Run_10_Hz and Run_Monitor are the function’s names that are called when the processes are started. Below in the main function, APEX call CREATE_EVENT is used to create an event object. An event object has a name Wakeup. Then APEX calls CREATE_SAMPLING_PORT are used to create three sampling ports. These ports have the following names: RAW_DATA, DATA_OUT and STATUS. In the end of the main function APEX call SET_PARTITION_MODE is used to set the partition to the NORMAL state. After that, OS will invoke functions Run_10_Hz and Run_Monitor. A function Run_10_Hz is called periodically with period 10 milliseconds. This value for period was set in PERIOD attribute during the creation of the first process. Each time when the function Run_10_Hz is called, it activates the event Wakeup, reads a 35 Lesovoy S.L. Extracting architectural information from source code of ARINC 653-compatible application software using CEGAR-based approach. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 31-46 message from sampling port RAW_DATA, performs some operations with data and writes a message to sampling port DATA_OUT. Function Run_Monitor belongs to the second process that is an aperiodic. This function waits for event Wakeup, resets it, performs some operations with data and message from sampling port RAW_DATA, performs some operations with data and writes a message to sampling port DATA_OUT. Function Run_Monitor belongs to the second process that is an aperiodic. This function waits for event Wakeup, resets it, performs some operations with data and writes a message to sampling port STATUS. 3.2 General approach for source code analysis For source code analysis, an approach based on Counterexample-guided abstraction refinement (CEGAR) algorithm is used. In CPAchecker tool [5] the basic predicate- based CEGAR algorithm has been extended for explicit-value analysis [7]. CPAchecker is a tool for configurable program analysis (CPA) [5,6] that combines the traditional program analyses and software model checking. In this paper the extended for explicit-value analysis CEGAR algorithm is applied for the task of extracting architecture information from source code. The algorithm is implemented in CPAchecker tool. The algorithm presented in this paper uses a Control-Flow Automata (CFA) as intermediate representations of the program to be analyzed. CFA is a directed graph 36 Лесовой С.Л. Извлечение архитектурной информации из исходного кода ARINC 653 совместимых приложений с использованием алгоритма CEGAR. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 31-46 containing nodes and edges. A node corresponds to a program location. An edge corresponds to a certain operation of the program, for example, an assignment statement, a conditional branch or a function call. During the analysis, the algorithm constructs an Abstract Reachability Graph (ARG) using a program CFA. ARG is also a directed graph but its nodes correspond to abstract states of the program. Each abstract state contains a program location, a data state and a call stack. A data state is a mapping between program variables and their values. In data state some program variables may not have the values. ARG represents possible execution paths of the program. It means that ARG can contain both feasible (real) program paths as well as the infeasible (spurious) paths. The program path is feasible if it can be executed at runtime otherwise it is infeasible. A path in ARG is a sequence of abstract states connected by edges. An abstract state is reachable if there is a feasible program path that contains this state. contain both feasible (real) program paths as well as the infeasible (spurious) paths. The program path is feasible if it can be executed at runtime otherwise it is infeasible. A path in ARG is a sequence of abstract states connected by edges. An abstract state is reachable if there is a feasible program path that contains this state. Target states. Some edges may correspond to a function call in source code. If this function is defined in APEX interface, the algorithm will need to collect additional information about this function call. An abstract state in ARG which immediately follows such a function call is called the target state. Any target state has an incoming edge with APEX call. For each target state there is a path in ARG from the initial state to it. The algorithm performs a feasibility check for these paths. 3.3 Extracting APEX calls from source code Before starting the algorithm description, it is necessary to explain some important concepts used by the algorithm. The algorithm constructs the ARG by sequentially adding the new abstract states to it. For the current state the algorithm gets the list of all its successors and adds each of them to ARG. There is an edge between the current state and each its successor. Precision. Explicit-value analysis tracks values for the program variables. In many cases it is enough to track only a part of program variables that are important for a particular analysis. A set of program variables that are being tracked for the current abstract state is called a precision. Different abstract states may have different precisions. The empty precision means that no variables are being tracked. The full precision means that all variables are being tracked. As described in [6] the value analysis algorithm implemented in CPAchecker can change a precision during the analysis depending on some conditions. It is called a precision adjustment. An edge in ARG can correspond to a program operation that changes a value of a program variable. For example, an assignment operation changes the value of the left- hand operand, for a function call the values of arguments are assigned to function's parameters, etc. When the algorithm handles an edge between the current state (predecessor) and next state (successor) it uses a precision of the predecessor. If 37 Lesovoy S.L. Extracting architectural information from source code of ARINC 653-compatible application software using CEGAR-based approach. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 31-46 Extracting architectural information from source code of ARINC 653-compatible application software R-based approach. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 31-46 precision of the predecessor contains the current variable, then the algorithm evaluates and stores its new value in abstract state. The algorithm of analysis can use the values of variables stored in abstract states for different purposes. Fig. 2 presents a pseudocode of the main algorithm for extracting the architectural information from source code. This algorithm implements a classical CEGAR cycle extended for explicit value analysis [7] and is applied for the task of extracting architectural information from source code. CFA of the program is used as an input data for the algorithm. The algorithm uses two variables to store the abstract states: “reached” and “waitlist”. A variable “reached” contains the set of abstract states that have been explored already. A variable “waitlist” contains the set of abstract states that have to be explored on the next steps of the algorithm. At the beginning, the algorithm takes the initial state from CFA and put it to “waitlist”. After that, the external loop of the algorithm begins. The algorithm takes and removes the current state from waitlist. END Fig. 2. The main algorithm for extracting the architectural information from source code Precision. Further the algorithm gets all reachable successors for the current state using function “getAbstractSuccessors”. A pseudocode for the function “getAbstractSuccessors” is shown on Fig.3. The first operation of the function gets all successors (CFA nodes) of the current state. Then the function consecutively handles the edges (function “handleEdge”) between the current state and each its successor. The function “handleEdge” takes two parameters. The first parameter is an edge to be explored. The second parameter is a precision. The precision is taken from the edge predecessor. Depending on the operation in source code that the edge corresponds to, the function “handleEdge” performs the following actions:  For an assignment operation, the algorithm evaluates a new value for this variable. The new value for a variable will be stored in abstract state if this variable is contained in the precision.  For a function call, the function’s arguments are assigned to function’s parameters.  For a function call, the function’s arguments are assigned to function’s parameters.  For a conditional branch, a logical value for a condition is evaluated. If the logical value of a conditional branch is equal to FALSE then the function “handleEdge” return FALSE. It means that this successor is not reachable. In all other cases the function returns TRUE and the current successor is added to the list of reachable successors. So, function “getAbstractSuccessors” returns for the current state a list of all its reachable successors. FUNCTION main INPUT CFA of the program; OUTPUT 38 The algorithm of function “isPathFeasible” is shown in Fig. 4. To check the path Architectural information // Traverse through all CFA nodes. LOOP WHILE waitlist ≠ 0 // External loop. curState = getAndRemoveStateFromWaitlist(); successors = getAbstractSuccessors(curState); // Traverse through all reachable successors. FOR EACH nextState IN successors // Internal loop. IF isTargetState(nextState) path = getPathToState(nextState); IF isPathFeasible(path) = FALSE // Refine the path. performRefinementForPath(path, reached, waitlist); BREAK // Go to external loop. merge(nextState, reached); update(reached); addStateToWaitlist(nextState); Лесовой С.Л. Извлечение архитектурной информации из исходного кода ARINC 653 совместимых приложений с использованием алгоритма CEGAR. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 31-46 Лесовой С.Л. Извлечение архитектурной информации из исходного кода ARINC 653 совместимых приложений с использованием алгоритма CEGAR. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 31-46 Лесовой С.Л. Извлечение архитектурной информации из исходного кода ARINC 653 совместимых приложени с использованием алгоритма CEGAR. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 31-46 Architectural information VARIABLES reached – a set of states that have been reached; waitlist – a set of states to be explored; BEGIN initState = getInitialState(CFA); addStateToWaitlist(initState); // Traverse through all CFA nodes. LOOP WHILE waitlist ≠ 0 // External loop. curState = getAndRemoveStateFromWaitlist(); // Get all reachable successors of the current state. successors = getAbstractSuccessors(curState); // Traverse through all reachable successors. FOR EACH nextState IN successors // Internal loop. IF isTargetState(nextState) path = getPathToState(nextState); IF isPathFeasible(path) = FALSE // Refine the path. performRefinementForPath(path, reached, waitlist); BREAK // Go to external loop. END IF END IF merge(nextState, reached); update(reached); addStateToWaitlist(nextState); END FOR EACH END LOOP END Fig. 2. The main algorithm for extracting the architectural information from source code Further in internal loop the main algorithm traverses through all reachable successo for the current state. At this part of algorithm, a successor is called as a “nextState с использованием алгоритма CEGAR. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 31-46 Architectural information VARIABLES reached – a set of states that have been reached; waitlist – a set of states to be explored; BEGIN initState = getInitialState(CFA); addStateToWaitlist(initState); // Traverse through all CFA nodes. LOOP WHILE waitlist ≠ 0 // External loop. curState = getAndRemoveStateFromWaitlist(); // Get all reachable successors of the current state. successors = getAbstractSuccessors(curState); // Traverse through all reachable successors. FOR EACH nextState IN successors // Internal loop. IF isTargetState(nextState) path = getPathToState(nextState); IF isPathFeasible(path) = FALSE // Refine the path. performRefinementForPath(path, reached, waitlist); BREAK // Go to external loop. END IF END IF merge(nextState, reached); update(reached); addStateToWaitlist(nextState); END FOR EACH END LOOP END Fig. 2. The main algorithm for extracting the architectural information from source code Further in internal loop the main algorithm traverses through all reachable successors for the current state. At this part of algorithm, a successor is called as a “nextState”. The algorithm checks whether a nextState is a target state If it is a target state the Architectural information reachableSuccessors // All reachable successors. BEGIN // Get all successors of the current state. allCFASuccessors = getAllSuccessors(curState); // Get all successors of the current state. allCFASuccessors = getAllSuccessors(curState); FOR EACH successor IN allCFASuccessors edge = getEdge(curState, successor); precision = getPrecisionForState(curState); precision = getPrecisionForState(curState); IF handleEdge(edge, precision) = TRUE // Add successor to reachableSuccessors. addToSet(reachableSuccessors, successor); END IF END FOR EACH RETURN reachableSuccessors; END Fig. 3. The algorithm of fun RETURN reachableSuccessors; Fig. 2. The main algorithm for extracting the architectural information from source code Further in internal loop the main algorithm traverses through all reachable successors for the current state. At this part of algorithm, a successor is called as a “nextState”. The algorithm checks whether a nextState is a target state. If it is a target state, the algorithm calculates a path in ARG from the initial state to the current target state and checks its feasibility using function “isPathFeasible”. In a classical CEGAR algorithm a path in a program to be explored is called a counterexample and it means a path to the error state. In the current algorithm it is just a path to the target state we need to explore. The algorithm of function “isPathFeasible” is shown in Fig. 4. To check the path 39 Lesovoy S.L. Extracting architectural information from source code of ARINC 653-compatible application software using CEGAR-based approach. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 31-46 feasibility the algorithm consecutively passes through all edges of the path, starting from the initial state. The algorithm analyses the operations for each edge. To track all program variables, for each state on the path the full precision is set, i.e. the algorithm performs the feasibility check for a path with the full precision. curState // Current state. RETURN reachableSuccessors // All reachable successors. BEGIN Fig. 3. The algorithm of function getAbstractSuccessors Fig. 3. The algorithm of function getAbstractSuccessors Each edge on the path is handled with the function “handleEdge” that was already described above. For a conditional branch the function “handleEdge” may return FALSE if logical condition is not satisfied. The path is not feasible if for any edge on the path the logical condition is not satisfied. In this case the function “isPathFeasible” returns FALSE. In all other cases the path is feasible. If the path is feasible then at the last state of the path the values for all program variables assigned on this path are known. The last edge and the last state of the path is passed to a function “handleApexCall”. FUNCTION isPathFeasible INPUT path RETURN TRUE – path is feasible; FUNCTION isPathFeasible INPUT path RETURN TRUE – path is feasible; 40 Лесовой С.Л. Извлечение архитектурной информации из исходного кода ARINC 653 совместимых приложений с использованием алгоритма CEGAR. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 31-46 FALSE – path is not feasible; BEGIN // Traverse through all edges. FOR EACH edge IN path precision = FULL; IF handleEdge(edge, precision) = FALSE RETURN FALSE END IF IF isLastEdge(edge, path) = TRUE lastState = getSuccessor(edge); handleApexCall(edge, lastState); END IF END FOR EACH RETURN TRUE END Fig. 4. The algorithm of function isPathFeasible Лесовой С.Л. Извлечение архитектурной информации из исходного кода ARINC 653 совместимых приложений с использованием алгоритма CEGAR. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 31-46 Fig. 4. The algorithm of function isPathFeasible The last edge contains the information about the APEX call. The last state contains values for all program variables on the path. The function “handleApexCall” extracts all architectural information including the function name for the last APEX call, values for its argument and call stack. It is important to note that the algorithm extracts architecture information only from the APEX calls that belong to a feasible paths. The algorithm collects the architectural information for each APEX call and uses it as output data. The format of the output data will be described in the next section. If the algorithm has detected that a path is infeasible, then it will refine this path. During refinement procedure the precision for some abstract states of the path are changed by adding variables for tracking. The refinement procedure is described in detail in [7]. Fig. 3. The algorithm of function getAbstractSuccessors Finally, the algorithm will update the ARG in such a way that will eliminate the infeasible path or its part for the further analysis. At the end of the internal loop the algorithm tries to merge the nextState with already reached states, updates reached states and adds the last explored state (nextState) to “waitlist”. These steps are described in details in [7] (see section “Reachability Algorithm for CPA”). The described above steps of internal loop are being repeated for each reachable successor of the current state. Then the algorithm leaves the internal loop and continues its execution by taking the first step on the main loop. It takes the next state from “waitlist” variable and repeats all steps already described above. The algorithm terminates when all ARG abstract states have been processed. 41 41 Lesovoy S.L. Extracting architectural information from source code of ARINC 653-compatible application software using CEGAR-based approach. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 31-46 Lesovoy S.L. Extracting architectural information from source code of ARINC 653-compatible application software using CEGAR-based approach. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 31-46 3.4 Output format The algorithm keeps the collected architectural information in the internal format. For further processing the architectural information has to be transformed to the external representation. The export format depends on the tool that is used for creation the architecture models. The architectural information can also be exported to human- readable format. In Fig. 5 the architectural information extracted by the algorithm from the source code fragment in Fig. 1 is presented in a human-readable format. The presented architectural information is divided onto sections. The first section contains information about ARINC653 processes. There are two processes with names Proc_10_Hz and Proc_Monitor. Below the process name there are the list of its attributes. On the Fig3 there are only three attributes are presented: PROCESS_ID, ENTRY_POINT and PERIOD. PROCESS_ID is a serial number of the process inside a partition. ENTRY_POINT is a name of the function that is being called when the process is started. PERIOD shows the period’s duration in milliseconds. INFINITE_TIME_VALUE in source code corresponds to aperiodic process. The next sections contain the information about other ARINC653 objects created in the source code. ARINC653_SAMPLING_PORTS section shows three sampling ports and its attributes. ARINC653_SAMPLING_PORTS section shows three sampling ports and its attributes. ARINC653_SAMPLING_MESSAGES section shows what processes are using sampling ports for sending (WRITE subsection) and for receiving (READ subsection) messages. For example the port DATA_OUT is used by the first process (function Run_10_Hz) for sending messages. ARINC653_EVENTS contains information about the events that have been created and used in the source code. ARINC653_EVENTS contains information about the events that have been created and used in the source code. ARINC653_EVENTS section has three subsections: SET_EVENT, WAIT_EVENT and RESET_EVENTS. The name of the subsection corresponds to the APEX call. For example, a subsection SET_EVENT corresponds to APEX call SET_EVENT that activate an event. ==ARINC653_PROCESSES== ==ARINC653_PROCESSES== Proc_10_Hz PROCESS_ID: 0 ENTRY_POINT: Run_10_Hz(0) PERIOD = 100 ms PERIOD = 100 ms Proc_Monitor PROCESS_ID: 1 ENTRY_POINT: Run_Monitor(1) APERIODIC Proc_Monitor ==ARINC653_SAMPLING_PORTS== ==ARINC653_SAMPLING_PORTS== 42 Лесовой С.Л. Извлечение архитектурной информации из исходного кода ARINC 653 совместимых приложений с использованием алгоритма CEGAR. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 31-46 Лесовой С.Л. Извлечение архитектурной информации из исходного кода ARINC 653 совместимых приложений с использованием алгоритма CEGAR. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 31-46 1) RAW_DATA MAX_MESSAGE_SIZE = 128 PORT_DIRECTION = DESTINATION REFRESH_PERIOD = 1000 1) RAW_DATA MAX_MESSAGE_SIZE = 128 PORT_DIRECTION = DESTINATION REFRESH_PERIOD = 1000 1) RAW_DATA MAX_MESSAGE_SIZE = 128 PORT_DIRECTION = DESTINATION REFRESH_PERIOD = 1000 2) DATA_OUT 3) STATUS … ==ARINC653_SAMPLING_MESSAGES== =WRITE= … ==ARINC653_SAMPLING_MESSAGES== =WRITE= 1) PORT_NAME=DATA_OUT; ENTRY_POINT=Run_10_Hz(0); 2) PORT_NAME=STATUS; ENTRY_POINT=Run_Monitor(1); 2) PORT_NAME=STATUS; ENTRY_POINT=Run_Mo =READ= 1) PORT_NAME=RAW_DATA; 1) PORT_NAME=RAW_DATA; ENTRY_POINT=Run_10_Hz(0); ENTRY_POINT=Run_10_Hz(0); Fig. 5. The architectural information in human-readable format. In the analyzed source call there is only one such a call for event with a name Wakeup. The ENTRY_POINT string contains a name of the ENTRY_POINT function where this call was made. In the real code the ENTRY_POINT function is determined using a call stack information. The serial number of the process is shown in parentheses. In the Fig.3 we can see that event Wakeup was set in function Run_10_Hz that belongs to the process with PROCESS_ID equal to 0 (Proc_10_Hz). From the section WAIT_EVENT, we can understand that the function Run_Monitor waits for the event Wakeup using APEX call WAIT_EVENT. The function Run_Monitor belongs to process Proc_Monitor. So, we can see that the event Wakeup is used by two processes for synchronization. The representation of architectural information in the human-readable format is presented only for explaining the content of such information and is useful mainly for debug purposes. As it was mentioned above for further processing the architectural 43 Lesovoy S.L. Extracting architectural information from source code of ARINC 653-compatible application software using CEGAR-based approach. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 31-46 information should be transformed to the format that is supported by the external tools. 4 Results and conclusions The algorithm presented in the paper allows extracting architectural information from source code of ARINC 653-compatible application software. The main contribution of this paper is the application the ideas of counterexample and path feasibility check for the task of extracting the architectural information from source code. In the presented algorithm the task of extracting architectural information from source code has been solved by transforming it into the task of path feasibility check. The work of the algorithm is demonstrated on the simple example. By this moment the algorithm has been tested on the several software applications that are compatible with ARINC 653 specification. These applications contained up to 50 ARINC 653 process and up to 30 objects for communications. The next task to be done is to extend the algorithm for extracting from source code the global variables that are used for communication between processes inside partition. It is also necessary to implement the algorithm of transformation of the architecture information to the architecture model. References [1]. OMG Systems Modeling Language (OMG SysML™) Version 1.5, 2017. [2]. [Online]. Available: http://www.omg.org/spec/SysML/1.5/ [3]. SAE International standard AS5506C, Architecture Analysis & Design Language (AADL), 2017. [Online]. Available: http://standards.sae.org/as5506c/ [4]. Feiler P., Gluch D. Model-Based Engineering with AADL: An Introduction to the SAE Architecture Analysis & Design Language. Addison-Wesley, 2012. [5]. ARINC Specification 653P1-4. Avionics Application Software Standard Interface Part 1 – Required Services. Published by SAE-ITC, Maryland, USA. August 21, 2015. [6]. [Online]. Available: https://cpachecker.sosy-lab.org/ [7]. D. Beyer, T. A. Henzinger and G. Theoduloz. Program Analysis with Dynamic Precision Adjustment. In Proc, of the 2008 23rd IEEE/ACM International Conference on Automated Software Engineering, 2008, pp. 29-38. [8]. Beyer D., Löwe S. Explicit-State Software Model Checking Based on CEGAR and Interpolation. Lecture Notes in Computer Science, vol. 7793, pp 146-162. 44 Лесовой С.Л. Извлечение архитектурной информации из исходного кода ARINC 653 совместимых приложений с использованием алгоритма CEGAR. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 31-46 DOI: 10.15514/ISPRAS-2018-30(3)-3 DOI: 10.15514/ISPRAS-2018-30(3)-3 Для цитирования: Лесовой С.Л. Извлечение архитектурной информации из исходного кода ARINC 653 совместимых приложений с использованием алгоритма CEGAR. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 31-46 (на английском языке). DOI: 10.15514/ISPRAS-2018-30(3)-3 Извлечение архитектурной информации из исходного кода ARINC 653 совместимых приложений с использованием алгоритма CEGAR С.Л. Лесовой <lesovoy@ispras.ru> Институт системного программирования им. В.П. Иванникова РАН, 109004, Россия, г. Москва, ул. А. Солженицына, д. 25 Аннотация. Модельно-ориентированный подход к разработке позволяет построить архитектурную модель существующей системы по ее исходному коду. Построенная архитектурная модель существующей системы позволяет проанализировать ее различные статические и динамические характеристики, включая производительность, требуемые аппаратные ресурсы, надежность и другие. Архитектурные модели могут использоваться как для анализа, так и для повторного использования некоторых компонентов существующей системы в новом проекте. Во многих случаях такой подход позволяет избежать построения новой системы с нуля. Для создания архитектурных моделей могут использоваться различные языки моделирования. В данной работе используется язык анализа и проектирования архитектуры (AADL). Данная статья описывает алгоритм извлечения архитектурной информации из исходного кода ARINC 653 совместимых программных приложений. Спецификация ARINC 653 определяет требования к программным компонентам для систем интегрированной модульной авионики (ИМА). Для доступа к различным сервисам операционной системы программные приложения используют прикладной исполняемый интерфейс. Архитектурная информация в исходном коде программных приложений совместимых с требованиями спецификации ARINC 653 включает процессы в каждом разделе, объекты для взаимодействия между процессами внутри и за пределами раздела, а также глобальные переменные. Для анализа исходного кода и получения архитектурной информации необходимо проанализировать все программные вызовы прикладного исполняемого интерфейса. Извлеченная архитектурная информация далее используется для построения архитектурных моделей системы. Для анализа исходного кода используется подход на основе алгоритма CEGAR (уточнение абстракции с помощью контрпримера), широко используемого при верификации программного обеспечения. Алгоритм CEGAR анализирует возможные пути исполнения программы, используя представление программы в виде абстрактного графа достижимости. В классическом алгоритме CEGAR исследуемый путь программы называется контрпримером и означает путь от начала программы до некоторого ошибочного состояния. Для подтверждения наличия ошибки в коде программы алгоритм CEGAR выполняет проверку достижимости для исследуемого пути. В программном инструменте CPAchecker базовый основанный на предикатах алгоритм CEGAR расширен для анализа явных значений переменных. В данной статье расширенный для анализа явных значений переменных алгоритм CEGAR используется для задачи извлечения архитектурной информации из исходного кода приложений. Основной вклад данной статьи заключается в применении идей контрпримера и проверки достижимости пути к задаче извлечения архитектурной информации из исходного кода приложений. 45 Lesovoy S.L. Extracting architectural information from source code of ARINC 653-compatible application software using CEGAR-based approach. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 31-46 Ключевые слова: архитектурная информация; архитектурные модели; ARINC 653; интегрированная модульная авионика (ИМА); алгоритм CEGAR Ключевые слова: архитектурная информация; архитектурные модели; ARINC 653; интегрированная модульная авионика (ИМА); алгоритм CEGAR Список литературы [1]. OMG Systems Modeling Language (OMG SysML™) Version 1.5, 2017. [2]. [Online]. Режим доступа: http://www.omg.org/spec/SysML/1.5/ [3]. SAE International standard AS5506C, Architecture Analysis & Design Language (AADL), 2017. [Online]. Режим доступа: http://standards.sae.org/as5506c/ [4]. Feiler P., Gluch D. Model-Based Engineering with AADL: An Introduction to the SAE Architecture Analysis & Design Language. Addison-Wesley, 2012. [5]. ARINC Specification 653P1-4. Avionics Application Software Standard Interface Part 1 – Required Services. Published by SAE-ITC, Maryland, USA. August 21, 2015. [6]. [Online]. Режим доступа: https://cpachecker.sosy-lab.org/ [7]. D. Beyer, T. A. Henzinger and G. Theoduloz. Program Analysis with Dynamic Precision Adjustment. In Proc, of the 2008 23rd IEEE/ACM International Conference on Automated Software Engineering, 2008, pp. 29-38. [8]. Beyer D., Löwe S. Explicit-State Software Model Checking Based on CEGAR and Interpolation. Lecture Notes in Computer Science, vol. 7793, pp 146-162. 46 46 DOI: 10.15514/ISPRAS-2018-30(3)-4 Для цитирования: Трифанов В.Ю. Вопросы индустриального применения синхронизационных контрактов при динамическом поиске гонок в Java-программах. Труды ИСП РАН, том 30, вып. 2, 2018 г., стр. 47-62. DOI: 10.15514/ISPRAS-2018-30(3)- 4 Для цитирования: Трифанов В.Ю. Вопросы индустриального применения синхронизационных контрактов при динамическом поиске гонок в Java-программах. Труды ИСП РАН, том 30, вып. 2, 2018 г., стр. 47-62. DOI: 10.15514/ISPRAS-2018-30(3)- 4 Вопросы индустриального применения синхронизационных контрактов при динамическом поиске гонок в Java- программах В.Ю. Трифанов <vitaly.trifanov@gmail.com>, Санкт-Петербургский государственный университет, 198504, Россия, Санкт-Петербург, Университетский пр., д. 28 В.Ю. Трифанов <vitaly.trifanov@gmail.com>, Санкт-Петербургский государственный университет, 198504, Россия, Санкт-Петербург, Университетский пр., д. 28 Аннотация. Состояния гонки (data race) возникает в многопоточной программе при одновременном обращении нескольких потоков к разделяемым данным. Существует два основных подхода к обнаружению гонок – статический анализ программы (без её запуска) и динамическое обнаружение гонок в процессе работы программы. Ранее авторами был предложен точный высокопроизводительный динамический подход к обнаружению гонок на основании специальным образом составленных синхронизационных контрактов – частичных спецификаций поведения классов и наборов методов целевого приложения в многопоточной среде. Данная статья рассматривает вопрос индустриального применения концепции синхронизационных контрактов на крупных нагруженных многопоточных приложениях. Предложены метод обработки контрактов и архитектура соответствующего модуля динамического детектора jDRD, выявлены основные проблемные места и потенциальные точки падения производительности, разработано техническое решение, лишённое подобных проблем. Ключевые слова: состояние гонки; многопоточность; динамический анализ; автоматическое обнаружение ошибок. 1. Введение Многопроцессорные и многоядерные системы начали своё развитие в конце XX века и на текущий момент фактически вытеснили однопроцессорные вычислительные машины. Основным преимуществом подобных систем является возможность одновременного выполнения различных 47 Trifanov V.Yu. Applying synchronization contracts approach for dynamic detection of data races in industrial applications. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 47-62 последовательностей инструкций параллельно. Наиболее популярной программной архитектурой таких систем является модель разделяемой памяти, в которой несколько потоков управления обмениваются данными через общую память. Такая архитектура лежит в основе таких широко распространённых в индустрии языков как C++ и Java [1]. Организация корректного взаимодействия нескольких потоков является одной из самых сложных задач в программировании, здесь возможны серьёзные ошибки, такие как взаимные блокировки (deadlocks), голодание (thread starvation) и гонки (data races). Гонка (или состояние гонки) возникает в многопоточной программе, когда несколько потоков одновременно обращаются к разделяемым данным, и хотя бы одно из этих обращений является записью данных [2]. Основная опасность гонок заключается в том, что они приводят к повреждению глобальных структур данных, но программа при этом зачастую продолжает работу, что приводит к сбоям и непредвиденным результатам. Ущерб от гонок может приводить к огромным финансовым потерям и к человеческим жертвам [3, 4]. Возникновение гонки зависит от чередования операций в потоках. При внешнем управлении потоками (а именно так устроено подавляющее большинство систем) переключение между ними происходит непредсказуемо, что затрудняет возможность воспроизведения гонок на стадии тестирования. Таким образом, задача автоматического обнаружения гонок актуальна, важна и находится в поле внимания исследователей на протяжении нескольких десятилетий. Принято выделять два основных подхода к автоматическому обнаружению гонок. Первый – это статический подход, который предполагает анализ исходного кода программы без её запуска. Например, возможно построение графа выполнения программы (control flow graph, CFG) и расчёт множества удерживаемых потоками блокировок. Такой алгоритм называется lockset [5, 6] и обладает существенным недостатком – с его помощью можно отслеживать только операции синхронизации, основанные на блокировках. В современном программировании это становится все менее актуально, поскольку разработаны неблокирующие операции синхронизации и типовые структуры данных, такие как очередь и хеш-таблица [7]. В простых случаях можно успешно доказать корректность программ без блокировок с помощью тяжеловесных методов – например, алгоритма Деккера или Петерсона. Анализ использования объектов и контекста [8-11] также позволяет корректно обрабатывать некоторые другие способы синхронизации. Также возможен анализ с помощью проверки моделей [12-13] или использования дополнительных типов [14] и их вывода с помощью аннотаций [15-16]. Второй подход – динамический, в рамках которого анализ программы осуществляется во время её выполнения [17-24]. 1. Введение Он анализирует лишь текущий путь выполнения программы, но в нём может обрабатывать любые операции синхронизации и обладать стопроцентной точностью, без ложных 48 Трифанов В.Ю. Вопросы индустриального применения синхронизационных контрактов при динамическом поиске гонок в Java-программах. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 47-62 срабатываний 1-го и 2-го рода. Однако на практике его точность сильно ограничена производительностью, поскольку обработка всех операций синхронизации и обращений к разделяемым данным требует в сотни раз больше ресурсов, чем выполнение самой анализируемой программы. Следует отметить, что последние разработки в сфере динамического поиска гонок сводятся к сокращению области анализа с помощью различных техник, в том или ином виде жертвующих точностью анализа [25-26]. Ранее авторами был предложен подход синхронизационных контрактов, который позволяет существенно повысить производительность динамического обнаружения гонок без потери точности [27-29]. Этот подход показал хорошие практические результаты, но в процессе его индустриального применения обнаружился ряд трудностей. Во-первых, созданные разработчиками контракты могут содержать неточности и ошибки, а во- вторых, контракты нужно эффективно применять в процессе анализа кода, чтобы достичь экономии накладных расходов на сбор информации и проверку гонок во время работы приложения. В данной статье предложен подход к анализу, верификации и динамическому применению контрактов, основанный на построении дерева контрактов срабатываний 1-го и 2-го рода. Однако на практике его точность сильно ограничена производительностью, поскольку обработка всех операций синхронизации и обращений к разделяемым данным требует в сотни раз больше ресурсов, чем выполнение самой анализируемой программы. Следует отметить, что последние разработки в сфере динамического поиска гонок сводятся к сокращению области анализа с помощью различных техник, в том или ином виде жертвующих точностью анализа [25-26]. срабатываний 1-го и 2-го рода. Однако на практике его точность сильно ограничена производительностью, поскольку обработка всех операций синхронизации и обращений к разделяемым данным требует в сотни раз больше ресурсов, чем выполнение самой анализируемой программы. Следует отметить, что последние разработки в сфере динамического поиска гонок сводятся к сокращению области анализа с помощью различных техник, в том или ином виде жертвующих точностью анализа [25-26]. Ранее авторами был предложен подход синхронизационных контрактов, который позволяет существенно повысить производительность динамического обнаружения гонок без потери точности [27-29]. Этот подход показал хорошие практические результаты, но в процессе его индустриального применения обнаружился ряд трудностей. Во-первых, созданные разработчиками контракты могут содержать неточности и ошибки, а во- вторых, контракты нужно эффективно применять в процессе анализа кода, чтобы достичь экономии накладных расходов на сбор информации и проверку гонок во время работы приложения. 1. Введение В данной статье предложен подход к анализу, верификации и динамическому применению контрактов, основанный на построении дерева контрактов (аналог графа потока управления), и продемонстрирована его целесообразность и практическая полезность. 2. Синхронизационные контракты и детектор jDRD Основным точным алгоритмом динамического обнаружения гонок в программах, основанных на модели разделяемой памяти, является алгоритм happens-before [18, 30], являющийся, фактически, поиском гонок «по определению». Алгоритм, анализируя операции в программе, выделяется два подмножества: (i) операции с данными внутри потоков исполнения и (ii) операции синхронизации, передающие изменения между потоками и таким образом синхронизирующие данные различных потоков. Основным точным алгоритмом динамического обнаружения гонок в программах, основанных на модели разделяемой памяти, является алгоритм happens-before [18, 30], являющийся, фактически, поиском гонок «по определению». Алгоритм, анализируя операции в программе, выделяется два подмножества: (i) операции с данными внутри потоков исполнения и (ii) операции синхронизации, передающие изменения между потоками и таким образом синхронизирующие данные различных потоков. Рассмотрим работу алгоритма на примере языка Java, у которого, наряду с C/C++, модель памяти максимально проработана [1]. На множестве всех операций синхронизации существует отношение полного порядка synchronized-with. Все операции внутри каждого конкретного потока также упорядочены – это порядок по времени. Объединение и последующее транзитивное замыкание этих двух отношений даёт отношение частичного порядка, называемое happens-before. Согласно определению гонки, две операции с одними и теми же данными находятся в состоянии гонки, если они не упорядочены с помощью отношения happens-before [1]. Традиционно выполнение отношения happens-before отслеживается с помощью векторных часов Лампорта [31]. Такой подход точен, но обладает очень большими накладными расходами скорость работы программы Рассмотрим работу алгоритма на примере языка Java, у которого, наряду с C/C++, модель памяти максимально проработана [1]. На множестве всех операций синхронизации существует отношение полного порядка synchronized-with. Все операции внутри каждого конкретного потока также упорядочены – это порядок по времени. Объединение и последующее транзитивное замыкание этих двух отношений даёт отношение частичного порядка, называемое happens-before. Согласно определению гонки, две операции с одними и теми же данными находятся в состоянии гонки, если они не упорядочены с помощью отношения happens-before [1]. Традиционно выполнение отношения happens-before отслеживается с помощью векторных часов Лампорта [31]. Такой подход точен, но обладает очень большими накладными расходами – скорость работы программы замедляется в 10-200 раз [32]. Для повышения производительности ранее 49 Trifanov V.Yu. Applying synchronization contracts approach for dynamic detection of data races in industrial applications. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 47-62 Trifanov V.Yu. Applying synchronization contracts approach for dynamic detection of data races in industrial applications. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 47-62 авторами был предложен подход синхронизационных контрактов, в основе которого лежит два наблюдения. 1. 2. Синхронизационные контракты и детектор jDRD Индустриальные приложения используют сторонние библиотеки и подсистемы, которые в совокупности могут превышать размер кода самого приложения в несколько десятков раз. Как правило, взаимодействие со сторонними библиотеками осуществляется через хорошо документированные интерфейсы. В частности, обычно хорошо документировано поведение методов, классов и интерфейсов в многопоточной среде. р 2. При поиске гонок обычно стоит задача обнаружения ошибок в собственном коде приложения, поскольку выбор библиотек обычно осуществляется экспертами на основании больших объёмов данных об их предыдущем использовании. Это позволяет сделать вывод о существенно меньшей надёжности самого разрабатываемого приложения, чем используемых им библиотек и подсистем, и сфокусироваться на анализе собственного кода приложения. На основе этих наблюдений был разработан метод синхронизационных контрактов, а также динамический детектор jDRD [27-29, 33]. Основная идея метода заключается в том, чтобы разделить весь программный код целевого приложения на две части – подлежащую анализу (обычно это собственный код приложения) и не подлежащую анализу, то есть считающуюся надёжной (обычно это сторонние библиотеки, подсистемы и модули). Далее необходимо определить интерфейсы и методы, посредством которых анализируемая часть взаимодействует с неанализируемой частью, и описать их поведение в многопоточной среде. Иными словами, нужно выделить используемые в приложении интерфейсы других компонент, проанализировать и классифицировать их классы и методы там в тех случаях, где про них что-то известно. В [33] предлагается следующая классификация: Основная идея метода заключается в том, чтобы разделить весь программный код целевого приложения на две части – подлежащую анализу (обычно это собственный код приложения) и не подлежащую анализу, то есть считающуюся надёжной (обычно это сторонние библиотеки, подсистемы и модули). Далее необходимо определить интерфейсы и методы, посредством которых анализируемая часть взаимодействует с неанализируемой частью, и описать их поведение в многопоточной среде. Иными словами, нужно выделить используемые в приложении интерфейсы других компонент, проанализировать и классифицировать их классы и методы там в тех случаях, где про них что-то известно. В [33] предлагается следующая классификация: 1. вызов пары методов (или несколько пар методов одного класса) обеспечивают передачу отношения happens-before согласно документации; 2. метод или класс потокобезопасен, но не вовлечён в передачу отношения happens-before; 3. метод объекта не предназначен для вызова в многопоточной среде (требует внешней синхронизации), но может рассматриваться как немодифицирующий состояние объекта-владельца (то есть, может рассматриваться как операция чтения данных). Единичное описание принадлежности метода (или всех методов класса) к какой-то категории и называется синхронизационным контрактом. Контракт для пары методов, обеспечивающих передачу отношения happens-before, называется happens-before контрактом. Листинг 1. Пример контракта Listing 1. Contract example Листинг 1. Пример контракта Listing 1. Contract example Листинг 1. Пример контракта В дальнейшем эти контракты используются на фазе динамического анализа: перед вызовом метода детектор определяет, есть ли контракт для этого метода, и если есть, то обрабатывает метод в соответствии с этим контрактом, что и даёт прирост производительности (в противном случае пришлось бы проводить анализ всего кода метода). В остальном схема устройства jDRD достаточно стандартна. На фазе запуска анализируемого приложения к нему посредством стандартной технологии java-agent подключается модуль jDRD, который модифицирует загружаемые классы с помощью техники инструментирования байт-кода. В код классов вставляются инструкции для обработки операций синхронизации и обращений к разделяемым данным. Во время выполнения таких операций управление передаётся в jDRD, который динамически обсчитывает векторные часы и обнаруживает гонки. 2. Синхронизационные контракты и детектор jDRD Точность и ограничения такого подхода описаны в работе [28], там же представлены основные виды Единичное описание принадлежности метода (или всех методов класса) к какой-то категории и называется синхронизационным контрактом. Контракт для пары методов, обеспечивающих передачу отношения happens-before, называется happens-before контрактом. Точность и ограничения такого подхода описаны в работе [28], там же представлены основные виды 50 Трифанов В.Ю. Вопросы индустриального применения синхронизационных контрактов при динамическом поиске гонок в Java-программах. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 47-62 Трифанов В.Ю. Вопросы индустриального применения синхронизационных контрактов при динамическом поиске гонок в Java-программах. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 47-62 контрактов и продемонстрировано повышение производительности динамического анализатора, увеличивающееся с сокращением анализируемой области и ростом базы контрактов. В последующей статье [29] представлен язык описания контрактов, позволяющий разрабатывать синхронизационные контракты отдельных методов, пар методов или классов. Кроме того, язык содержит директивы, позволяющие включить контракты из других файлов. Это обеспечивает переиспользуемость контрактов – достаточно описать один раз контракты библиотеки и внести файл с ними в комплект поставки. На листинге 1 представлен пример контракта, указывающего наличие отношения happens-before между записью в потокобезопасную хеш-таблицу o по некоторому ключу p1 и последующими чтениями из неё по тому же ключу. sync { key java.lang.Object=o, java.lang.Object=p1; send java.lang.Objectjava.util.concurrent.ConcurrentMap.put(java.lang.Object, java.lang.Object); receive receive java.lang.Objectjava.util.concurrent.ConcurrentMap.get(java.lang.Object); } java.lang.Objectjava.util.concurrent.ConcurrentMap.get(java.lang.Object); } java.lang.Objectjava.util.concurrent.ConcurrentMap.get(java.lang } 3. Проверка корректности контрактов 51 На этапе индустриального внедрения описанной выше технологии возник ряд сложностей, требующих правильных архитектурных и технических решений. В основной массе они связаны с тем, что в общем случае контракты создаются на основе различных источников (конфигурационные файлы, аннотации в коде, документация и т.д.), и зачастую различными специалистами. Это неизбежно приводит к возможности различных ошибок – например, неполноте, противоречивости или некорректности совокупного набора 51 ifanov V.Yu. Applying synchronization contracts approach for dynamic detection of data races in industria plications. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 47-62 Trifanov V.Yu. Applying synchronization contracts approach for dynamic detection of data races in industrial applications. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 47-62 контрактов – и поэтому после объединения всех контрактов воедино необходимо верифицировать полученный набор. Выходными данными модуля обработки контрактов является изменённый код целевого приложения (инструментированный байт-код), учитывающий контракты при обработке операций в приложении. Таким образом, задача обработки контрактов состоит из двух следующих частей (процедур). контрактов – и поэтому после объединения всех контрактов воедино необходимо верифицировать полученный набор. Выходными данными модуля обработки контрактов является изменённый код целевого приложения (инструментированный байт-код), учитывающий контракты при обработке операций в приложении. Таким образом, задача обработки контрактов состоит из двух следующих частей (процедур). 1. Загрузка, проверка корректности и размещение контрактов в памяти. 1. Загрузка, проверка корректности и размещение контрактов в памяти. 1. Загрузка, проверка корректности и размещение контрактов в памяти. 2. Применение контрактов и модификация кода целевого приложения. 2. Применение контрактов и модификация кода целевого приложения. Первая процедура осуществляется один раз при запуске приложения. Нет ограничений на производительность этой процедуры, её основной задачей является проверка корректности контрактов. Вторая процедура является критичной с точки зрения производительности и требует тщательной реализации: контракты будут применяться постоянно (сотни раз в секунду) на протяжении всего временя работы приложения. Рассмотрим подробнее первую процедуру. Язык описания контрактов [29] предоставляет синтаксические директивы для спецификации контрактов, поэтому на стадии чтения этих контрактов необходимо провести соответствующие процедуры по их разбору и проверке корректности. Здесь возможны следующие ситуации. 1. Контракты могут противоречить друг другу; например, в одном контракте указано, что метод A синхронизирован с B, а во втором – что, наоборот, B синхронизирован с A; в подобных случая необходимо сигнализировать об ошибке. 2. Контракты классов могут быть неполными, т.е. описывать лишь часть публичных методов класса. В этом случае нужно выдать предупреждение. 2. Контракты классов могут быть неполными, т.е. описывать лишь часть публичных методов класса. В этом случае нужно выдать предупреждение. 3. Трифанов В.Ю. Вопросы индустриального применения синхронизационных контрактов при динамическом поиске гонок в Java-программах. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 47-62 Трифанов В.Ю. Вопросы индустриального применения синхронизационных контрактов при динамическом поиске гонок в Java-программах. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 47-62 контракт, и выясняется, все ли его публичные методы упомянуты в контракте. Если это не так, то выдаётся предупреждение. контракт, и выясняется, все ли его публичные методы упомянуты в контракте. Если это не так, то выдаётся предупреждение. контракт, и выясняется, все ли его публичные методы упомянуты в контракте. Если это не так, то выдаётся предупреждение. Таким образом, процедура загрузки и проверки корректности контрактов имеет следующий вид. Таким образом, процедура загрузки и проверки корректности контрактов имеет следующий вид. Таким образом, процедура загрузки и проверки корректности контрактов имеет следующий вид. 1. Построить граф контрактов. 2. Для каждого контракта на пару методов (f, g) классов A и B добавить в граф контракты (f’, g’) для всех наследников классов А и B. 2. Для каждого контракта на пару методов (f, g) классов A и B добавить в граф контракты (f’, g’) для всех наследников классов А и B. 3. Выполнить проверку циклов в графе. 4. (Опционально). Для всех классов, упомянутых в контрактах, проверить наличие их публичных методов, не упомянутых в контрактах. 4. (Опционально). Для всех классов, упомянутых в контрактах, проверить наличие их публичных методов, не упомянутых в контрактах. Данная процедура реализована в виде компоненты, предварительно обрабатывающей контракты до запуска целевого приложения. Она выдаёт дерево контактов и другую информацию, используемую в дальнейших динамических проверках. 3. Проверка корректности контрактов Контракты могут дублироваться – в этом случае нужно убедиться в их семантической идентичности. Для проверки непротиворечивости контрактов необходимо убедиться в отсутствии циклических контрактных зависимостей между ними. Для обнаружения таких зависимостей строится граф контрактов. Вершинами графа являются методы контрактов, а ребро между парой вершин A и B существует тогда и только тогда, когда A предшествует B. Согласно принципу подстановки Лисков [34] при наследовании потомок должен удовлетворять контракту предка. Следовательно, для каждого класса целевого приложения необходимо получить список его потомков и добавить соответствующие рёбра в граф контрактов. Наличие цикла в таком графе означает наличие цепочки противоречащих друг другу контрактов, а отсутствие циклов – непротиворечивость всего набора контрактов. Для проверки контрактов классов на полноту на этапе загрузки контрактов анализируется каждый класс целевого приложения, для которого существует 52 Трифанов В.Ю. Вопросы индустриального применения синхронизационных контрактов при динамическом поиске гонок в Java-программах. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 47-62 Трифанов В.Ю. Вопросы индустриального применения синхронизационных контрактов при динамическом поиске гонок в Java-программах. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 47-62 4. Архитектура модуля применения контрактов Рассмотрим процедуру применения контрактов с точки зрения её оптимальной реализации. Точнее, опишем программный модуль, который реализует эту процедуру. Динамический детектор должен в режиме реального времени определять наличие контракта для определённого Java-метода. При этом скорость работы детектора не должна деградировать с ростом программы, увеличением числа контрактов или иерархии классов. Обратим внимание, что речь идёт о сотнях и тысячах операций в секунду, поскольку вызовы методов в Java-программах происходят регулярно. Соответственно, решение данной задачи требует нестандартных подходов как к организации структуры данных, управляющей контрактами, так и к техническим решениям по их проверке в режиме реального времени. В результате построения дерева контрактов для каждого метода становится известен набор контрактов, в которые он вовлечён. Поэтому во время вызова этого метода в процессе работы программы детектор jDRD может эффективно получить данную информацию. Для внедрения соответствующих инструкций в код целевого приложения используется техника инструментирования байт- кода. Механизм её работы заключается в следующем: на фазе загрузки классов целевого приложения в оперативную память, после загрузки очередного класса управление передаётся компоненте Instrumentator – специальному Java-агенту, который реализован в рамках jDRD и может модифицировать байт-код данного класса (см. рис. 1). Instrumentator анализирует список методов класса и, если для метода существуют контракты, встраивает в качестве первой инструкции метода обработку этого контракта. Во время работы модифицированное приложение автоматически обработает 53 Trifanov V.Yu. Applying synchronization contracts approach for dynamic detection of data races in industrial applications. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 47-62 контракты всех методов (компонента RaceDetector). В практическом плане наличие контракта метода означает следующее. контракты всех методов (компонента RaceDetector). В практическом плане наличие контракта метода означает следующее. контракты всех методов (компонента RaceDetector). В практическом плане наличие контракта метода означает следующее. 1. Исходный код метода анализировать не нужно. Иными словами, во время выполнения тела метода детектор не должен проводить никаких операций и проверок. Для этого сразу после входа в тело метода детектор ставит специальный флаг в состояние «contract», а на выходе – сбрасывает. Перед тем, как выполнить очередную операцию, детектор проверяет состояние флага для текущего потока, и если его состояние равно «contract», то игнорирует операцию. Таким образом, флаг хранится для каждого потока отдельно. Управление флагами вынесено в отдельную компоненту (класс) FlagManager, основанный на стандартной Java-структуре ThreadLocal, которая предоставляет эффективный способ хранения локальных данных потока (см. рис. 1). фф р д ( р ) 2. happens-before контракты подразумевают синхронизацию методов. Обработка этой информации реализована классическим способом – в виде векторных часов (компонента ClockStorage, см. рис. 1). 4. Архитектура модуля применения контрактов Для хранения часов используется потокобезопасная хеш-таблица. Однако с учётом высокой нагрузки и большой частоты обращений традиционные потокобезопасные хеш-таблицы здесь не подходят. После ряда экспериментов была выбрана неблокирующая хеш- таблица [35]. Остаётся вопрос в выборе ключа для хранения часов. Ключ представляет собой отдельный объект, состоящий из полей, указанных в контракте. Поскольку набор полей от контракта к контракту может отличаться, классы объектов типа «ключ» генерируются автоматически, «на лету», посредством инструментирования байт-кода. На практике таких ключей нужно не более 20-30, поэтому на производительность динамического анализа это не оказывает существенного влияния. Отметим, что основная работа по анализу гонок производится в компоненте RaceDetector, которая работает только с векторными часами. 2. 2. happens-before контракты подразумевают синхронизацию методов. Обработка этой информации реализована классическим способом – в виде векторных часов (компонента ClockStorage, см. рис. 1). Для хранения часов используется потокобезопасная хеш-таблица. Однако с учётом высокой нагрузки и большой частоты обращений традиционные потокобезопасные хеш-таблицы здесь не подходят. После ряда экспериментов была выбрана неблокирующая хеш- таблица [35]. Остаётся вопрос в выборе ключа для хранения часов. Ключ представляет собой отдельный объект, состоящий из полей, указанных в контракте. Поскольку набор полей от контракта к контракту может отличаться, классы объектов типа «ключ» генерируются автоматически, «на лету», посредством инструментирования байт-кода. На практике таких ключей нужно не более 20-30, поэтому на производительность динамического анализа это не оказывает существенного влияния. Отметим, что основная работа по анализу гонок производится в компоненте RaceDetector, которая работает только с векторными часами. Единственным существенным недостатком описанного выше подхода является необходимость постоянного обращения к компоненте FlagManager. В частности, это регулярно происходит при работе самого jDRD, поскольку внутри него используются структуры данных, которые активно используют контракты. Одно такое обращение занимает порядка 1 мкс, но с учётом большого числа обращений (десятки-сотни тысяч в секунду) это существенно снижает производительность jDRD. Возможным решением может быть полный отказ от использования стандартных Java-классов во внутренних структурах детектора jDRD. 54 Трифанов В.Ю. Вопросы индустриального применения синхронизационных контрактов при динамическом поиске гонок в Java-программах. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 47-62 Трифанов В.Ю. Вопросы индустриального применения синхронизационных контрактов при динамическом поиске гонок в Java-программах. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 47-62 Рис.1. Архитектура модуля применения синхронизационных контрактов Fig 1. Architecture of the contracts processing module. Рис.1. Архитектура модуля применения синхронизационных контрактов Fig 1. Architecture of the contracts processing module. 5. Экспериментальное исследование Предложенный подход был применён для анализа трёх приложений, имеющих разную специфику. В каждом содержалось 1000-2000 классов без учёта используемых библиотек, а также 10-30 потоков. Синхронизационные контракты для каждого приложения разрабатывались соответствующими командами программистов. Кроме того, использовался общий набор контрактов, разработанный ранее для стандартных средств синхронизации Java. Совокупный размер контрактов для каждого приложения не превышал долей процента от общего объёма кода приложения. В табл. 1 представлены результаты экспериментов. В столбце «Количество контрактов» указано совокупное количество методов, для которых были составлены контракты. Столбец «количество ошибок» указывает число ошибок, обнаруженных при проверке корректности контрактов. Последний столбец содержит количество публичных методов, которые нуждались в составлении контракта, но были пропущены составителями. Таблица 1. Результаты работы модуля обработки синхронизационных контрактов Table 1. The contract processing module work results Приложение Кол-во контрактов Кол-во ошибок Пропущенные методы А 80 0 75 B 135 3 95 C 120 1 80 1. Результаты работы модуля обработки синхронизационных контрактов The contract processing module work results Наборы контрактов в каждом случае описывали поведение порядка 500-1000 методов. Время работы модуля проверки занимает не более 10-20 секунд. Основная часть пропущенных методов приходилась на классы-структуры данных. Например, программист указывал метод get стандартной хеш- таблицы (HashMap) или списка (ArrayList) как немодифицирующий, но не указывал остальные немодифицирующие методы этого класса. Ещё в 55 Trifanov V.Yu. Applying synchronization contracts approach for dynamic detection of data races in industrial applications. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 47-62 нескольких случаях метод вспомогательного класса (utility class) указывался как потокобезопасный, в то время как таковым можно пометить весь класс. нескольких случаях метод вспомогательного класса (utility class) указывался как потокобезопасный, в то время как таковым можно пометить весь класс. Таким образом, внедрение модуля верификации контрактов оказалось полезным и позволило обнаружить несколько ошибок и ряд неточностей в составлении контрактов при достаточно малом времени работы. Таким образом, внедрение модуля верификации контрактов оказалось полезным и позволило обнаружить несколько ошибок и ряд неточностей в составлении контрактов при достаточно малом времени работы. Оценка производительности всего подхода синхронизационных контрактов на фазе динамического анализа является отдельной задачей, требующей, как минимум, разработки методики оценки. Дело в том, что, во-первых, динамический детектор jDRD совершает до ста тысяч операций в секунду1. Во-вторых, время, затрачиваемое на динамическую проверку контракта, может превосходить выигрыш от его применения. Тем не менее, на основе проведенных экспериментов могут быть сделаны некоторые выводы по производительности. 1Отметим, что задача тестирования производительности на микро-уровне (так называемый микробенчмаркинг) невероятно сложна. Так, команда разработчиков Core Java работала над утилитой, позволяющей надёжно измерять производительность операций на уровне микро- и нано-секунд, около 5 лет – см. саму утилиту (http://openjdk.java.net/projects/code-tools/jmh/) и выступление Алексея Шипилёва (https://www.youtube.com/watch?v=8pMfUopQ9Es). 5. Экспериментальное исследование Во-первых, два из трёх приложений являются клиентскими, и пользователи показали повышение скорости работы приложений в сравнении с анализом при помощи старой версии jDRD. Во-вторых, общее число обрабатываемых операций синхронизации в секунду сократилось, поскольку количество добавившихся (обработка контрактов) на несколько порядков меньше количества устранённых (обработка операций синхронизации Java внутри контрактных методов). Как следствие, сократился и расход памяти на содержание векторных часов. Полученные предварительные данные свидетельствуют о неухудшении производительности на анализируемых приложениях и о практической применимости предложенного подхода. Но требуются более детальные экспериментальные исследования. 6. Заключение Динамический анализ является одним из основных подходов к автоматическому обнаружению гонок, но его практическая применимость ограничена вопросами производительности. Подход к снижению накладных расходов, основанный на описании синхронизационных контрактов сторонних компонент и замене их анализа применением этих контрактов, показал ранее высокую точность и практическую применимость. Однако как разбор контрактов, так и их применение на фазе динамического анализа связано с множеством сложностей. Возникают задачи как валидации и нормализации 56 p Трифанов В.Ю. Вопросы индустриального применения синхронизационных контрактов при динамическом поиске гонок в Java-программах. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 47-62 контрактов при загрузке, так и оптимального их хранения и использования во время анализа целевого приложения. контрактов при загрузке, так и оптимального их хранения и использования во время анализа целевого приложения. Для решения этих задач предложен метод построения графа контрактов и проверки отсутствия циклов в нём. Далее в исходный код целевого приложения с помощью техники инструментирования байт-кода встраиваются соответствующие инструкции, проверяющие наличие контракта для метода. Хранение часов осуществляется в высокоскоростной хеш-таблице по генерируемым динамически синтетическим ключам. В рамках применения этого подхода для детектора jDRD разработана архитектура модуля контрактов и его техническая реализация. Доработанный таким образом jDRD был применён на трёх индустриальных приложениях. Измерение основных метрик показало как практическую пользу от построения графа контрактов (был выявлен ряд ошибок и несоответствий в разработанных контрактах), так и сокращение количества обрабатываемых операций синхронизации за единицу времени. Дальнейшие работы должны включать в себя постановку полноценного эксперимента по измерению времени работы детектора – от разработки методики эксперимента до его проведения. В качестве направления дальнейшего развития средств спецификации контрактов можно указать визуальное моделирование. Интерес представляет описание иерархии контрактов и соответствующих им Java-классов и Java-методов с помощью диаграмм классов UML, расширенных и настроенных подходящим образом [36-38]. Также возможны визуальные спецификации поведения контактов с помощью динамических моделей [39]. Интересно исследовать задачу автоматизированного извлечения описания контрактов из Java-кода и Java- документации и связь повторного использования контрактов с повторным использованием документации [40-41]. Список литературы [1] Java Language Specification, Third Edition. Threads and Locks. Happens-before Order. http://docs.oracle.com/javase/specs/jls/se7/html/jls-17.html#jls-17.4.5 [2] Netzer R., Miller B. What Are Race Conditions? Some Issues and Formalizations. ACM Letters on Programming Languages and Systems, 1(1), 1992, pp. 74–88. [3] Blackout Final Report, August 14, 2003, http://www.ferc.gov/industries/electric/indus- act/reliability/blackout/ch5.pdf [4] Leveson N., Turner C. S. An Investigation of the Therac-25 Accidents. In IEEE Computer, vol. 26, N 7, 1993, pp. 18–41. [5] Engler D., Ashcraft K. RacerX: Effective, Static Detection of Race Conditions and Deadlocks. Proceedings of the Nineteenth ACM Symposium on Operating Systems Principles, 2003, pp. 237–252. [6] Voung J., Jhala R., Lerner S. RELAY: Static Race Detection on Millions of Lines of Code. In ESEC/FSE, 2007, pp. 205–214. [7] Herlihy M., Shavit N. The Art of Multiprocessor Programming. Morgan Kaufmann Publishers Inc., San Francisco, CA, USA, 2008, 528 p. 57 Trifanov V.Yu. Applying synchronization contracts approach for dynamic detection of data races in industrial applications. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 47-62 Trifanov V.Yu. Applying synchronization contracts approach for dynamic detection of data races in industrial applications. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 47-62 [8] Kahlon V., Sinha N., Kruus E., Zhang Y.: Static data race detection for concurrent programs with asynchronous calls. In Proceedings of the 7th Joint Meeting of the European Software Engineering Conference and the Foundations of Software Engineering, 2009, pp. 13–22. [9] Naik M., Aiken A., Whaley J. Effective Static Race Detection for Java. In Proceedings of the 2006 ACM SIGPLAN Conference on Programming Language Design and Implementation, 2006, pp. 308–319. [10] Radoi C., Dig D. Practical static race detection for java parallel loops. In Proc. of the 13th International Symposium on Software Testing and Analysis, ISSTA ’13, 2013. P.178–190. [11] Xie X., Xue J., Zhang J. Acculock: Accurate and Efficient Detection of Data Races. Softw. Practice Experience, vol. 43, no. 5, May 2013, pp. 543–576. [12] Burckhardt S., Musuvathi M. Effective program verification for relaxed memory models. In Proceedings of the 20th international conference on Computer Aided Verification, Berlin, Heidelberg, 2008. pp. 107–120. [13] Huynh T., Roychoudhury A. Memory model sensitive bytecode verification. Form. Methods Syst. Des., 31(3), 2007, pp. 281–305. [14] Boyapati C., Lee R., Rinard M. Ownership types for safe programming: preventing data races and deadlocks. In Proceedings of the 17th ACM SIGPLAN conference on Object- oriented programming, systems, languages, and applications, 2002, pp. 211–230. Список литературы [15] Flanagan C., Freund S. Type inference against races. Sci. Comput. Program., Vol 64, January 2007, pp. 140–165. [16] Rose J., Swamy N., Hicks M. Dynamic inference of polymorphic lock types. Science of Computer Programming, 58(3), 2005, pp. 366–383. g g [17] Biswas S., Zhang M., Bond M., Lucia B. Valor: Efficient, Software-Only Region Conflict Exceptions. In OOPSLA, 2015, pp. 241–259. [18] Flanagan C., Freund S. FastTrack: Efficient and Precise Dynamic Race Detection. In ACM Conference on Programming Language Design and Implementation, 2009, pp. 121–133. [19] Kini D., Mathur U., Viswanathan M. Dynamic race prediction in linear time. SIGPLAN Not. 52(6), 2017, pp. 157–170. [20] Qi Y., Das R., Luo Z., Trotter M. MulticoreSDK: a practical and efficient data race detector for real-world applications. Proceedings Software Testing, Verification and Validation (ICST), IEEE, 21-25 March 2011, pp. 309–318. [21] Serebryany S., Iskhodzhanov T. ThreadSanitizer: Data race detection in practice. In Proceedings of the Workshop on Binary Instrumentation and Applications, 2009, pp. 62– 71. [22] Serebryany K., Potapenko A., Iskhodzhanov T., Vyukov D. Dynamic race detection with LLVM compiler - compile-time instrumentation for ThreadSanitizer. In RV, 2011, Lecture Notes in Computer Science, vol 7186, pp. 110–114. [23] Yu M., Bae D., SimpleLock+: Fast and Accurate Hybrid Data Race Detection. Comput. J., vol. 59, no. 6, 2016, pp. 793–809. [24] Zhang T., Jung C., Lee D. ProRace: Practical Data Race Detection for Production Use. In Proceedings of the International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS), 2017, pp. 149–162. [25] Bond M., Coons K., McKinley K. Pacer: Proportional Detection of Data Races. Proceedings of 2010 ACM SIGPLAN Conference on Programming Language Design and Implementation (PLDI 2010), Toronto, June 2010, pp. 255–268. 58 Трифанов В.Ю. Вопросы индустриального применения синхронизационных контрактов при динамическом поиске гонок в Java-программах. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 47-62 [26] Marino D., Musuvathi M., Narayanasamy S. LiteRace: Effective Sampling for Lightweight Data Race Detection. PLDI '09 Proceedings of the 2009 ACM SIGPLAN conference on Programming language design and implementation, Vol. 44, Issue 6, 2009, pp. 134–143. pp [27] Трифанов В.Ю. Обнаружение состояний гонки в Java-программах на основе синхронизационных контрактов. Компьютерные инструменты в образовании. №4, 2012, стр. 16-29. [28] Трифанов В.Ю, Цителов Д.И. Динамический поиск гонок в Java-программах на основе синхронизационных контрактов. Материалы конференции "Инструменты и методы анализа программ (TMPA-2013)", Кострома, 2013, стр. 273–285. [29] Трифанов В.Ю., Цителов Д.И. DOI: 10.15514/ISPRAS-2018-30(3)-4 For citation: Trifanov V.Yu. Applying synchronization contracts approach for dynamic detection of data races in industrial applications. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 47-62 (in Russian). DOI: 10.15514/ISPRAS-2018-30(3)-4 Applying synchronization contracts approach for dynamic detection of data races in industrial applications V.Yu. Trifanov <vitaly.trifanov@gmail.com> V.Yu. Trifanov <vitaly.trifanov@gmail.com> St. Petersburg State University, Universitetski pr., 28, 198504 St. Petersburg, Russia Abstract. Data race occurs in multithreaded program when several threads simultaneously access same shared data and at least of them writes. Two main approaches to automatic race detection – static and dynamic – have their pros and cons. Dynamic analysis can provide best precision on certain program execution but introduce enormous runtime overheads. Earlier we introduced high-performance approach that improves performance of dynamic race detection. The key idea is to define and exclude external trusted parts of code (e.g. libraries) from analysis and replace them with specifications of their behavior in multithreaded environment. Possible behavior was classified and corresponding language for describing contracts developed. Evaluation on lightweight applications confirmed performance boost but further industrial usage of detector revealed some problems. This article covers that problems, introduces method and architecture of contract processing module and some technical features that help to apply proposed approach on high load production systems. Ключевые слова: multithreading; data race; dynamic analysis; automatic error detection. Список литературы Язык описания синхронизационных контрактов для задачи поиска гонок в многопоточных приложениях. Программная инженерия. Т.8, N 6, 2017, стр. 250–257. [30] Elmas T., Qadeer S., Tasiran S. Goldilocks: A Race and Transaction-Aware Java Runtime. Proceedings of the 2007 ACM SIGPLAN Conference on Programming Language Design and Implementation (PLDI'07), 2007, pp. 245–255. [31] Lamport L. Time, Clocks and the Ordering of Events in a Distributed System. Communications of the ACM, Vol. 21, Issue 7, 1978, pp. 558–565. [32] Intel Thread Checker, http://software.intel.com/en-us/intel-thread-checker/ [33] Трифанов В.Ю. Динамическое обнаружение состояний гонки в многопоточных Java-программах. Дисс. на соискание степени канд. техн. наук. СПбГУ, 2013. р р у [34] Liskov B., Wing J. A behavioral notion of subtyping. ACM Trans. Program. Lang. Syst. 16 (6). November 1994, pp.1811–1841. [35] Click C. A lock-free wait-free hash table. https://web.stanford.edu/class/ee380/Abstr [35] Click C. A lock-free wait-free hash table. https://web.stanford.edu/class/ee380/Abstracts/070221_LockFreeHash.pdf [36] Гаврилова Т.А., Лещева И.А., Кудрявцев Д.В. Использование моделей инженерии знаний для подготовки специалистов в области информационных технологий. Системное программирование. 2012. Т. 7. № 1, pp. 90–105. [37] Кознов Д.В. Основы визуального моделирования. Интернет-Университет Информационных Технологий (ИНТУИТ). Москва, 2008. ф р ( ) [38] Ольхович Л.Б., Кознов Д.В. Метод автоматической валидации UML-спецификаций на основе языка OCL. Программирование. 2003. Т. 29. № 6, стр. 44–50. [39] Иванов А., Кознов Д., Мурашева Т. Поведенческая модель RTST++. Записки семинара Кафедры системного программирования "Case-средства RTST++". 1998. № 1, стр. 37–52. [40] Луцив Д.В., Кознов Д.В., Басит Х.А., Терехов А.Н. Задачи поиска нечётких повторов при организации повторного использования документации. Программирование. 2016. № 4, стр. 39–49. [41] Кознов Д.В., Романовский К.Ю. Автоматизированный рефакторинг документации семейств программных продуктов. Системное программирование. 2009. Т. 4, стр. 128–150. 59 59 Trifanov V.Yu. Applying synchronization contracts approach for dynamic detection of data races in industrial applications. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 47-62 References [1] Java Language Specification, Third Edition. Threads and Locks. Happens-before Order. http://docs.oracle.com/javase/specs/jls/se7/html/jls-17.html#jls-17.4.5 [2] Netzer R., Miller B. What Are Race Conditions? Some Issues and Formalizations. ACM Letters on Programming Languages and Systems, 1(1), 1992, pp. 74–88. [3] Blackout Final Report, August 14, 2003, http://www.ferc.gov/industries/electric/indus- act/reliability/blackout/ch5.pdf [4] Leveson N., Turner C. S. An Investigation of the Therac-25 Accidents. In IEEE Computer, vol. 26, N 7, 1993, pp. 18–41. [5] Engler D., Ashcraft K. RacerX: Effective, Static Detection of Race Conditions and Deadlocks. Proceedings of the Nineteenth ACM Symposium on Operating Systems Principles, 2003, pp. 237–252. [6] Voung J., Jhala R., Lerner S. RELAY: Static Race Detection on Millions of Lines of Code. In ESEC/FSE, 2007, pp. 205–214. [7] Herlihy M., Shavit N. The Art of Multiprocessor Programming. Morgan Kaufmann Publishers Inc., San Francisco, CA, USA, 2008, 528 p. [8] Kahlon V., Sinha N., Kruus E., Zhang Y.: Static data race detection for concurrent programs with asynchronous calls. In Proceedings of the 7th Joint Meeting of the 60 Трифанов В.Ю. Вопросы индустриального применения синхронизационных контрактов при динамическом поиске гонок в Java-программах. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 47-62 Трифанов В.Ю. Вопросы индустриального применения синхронизационных контрактов при динамическом поиске гонок в Java-программах. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 47-62 European Software Engineering Conference and the Foundations of Software Engineering, 2009, pp. 13–22. European Software Engineering Conference and the Foundations of Software Engineering, 2009, pp. 13–22. [9] Naik M., Aiken A., Whaley J. Effective Static Race Detection for Java. In Proceedings of the 2006 ACM SIGPLAN Conference on Programming Language Design and Implementation, 2006, pp. 308–319. [10] Radoi C., Dig D. Practical static race detection for java parallel loops. In Proc. of the 13th International Symposium on Software Testing and Analysis, ISSTA ’13, 2013. P.178–190. [11] Xie X., Xue J., Zhang J. Acculock: Accurate and Efficient Detection of Data Races. Softw. Practice Experience, vol. 43, no. 5, May 2013, pp. 543–576. [12] Burckhardt S., Musuvathi M. Effective program verification for relaxed memory models. In Proceedings of the 20th international conference on Computer Aided Verification, Berlin, Heidelberg, 2008. pp. 107–120. [13] Huynh T., Roychoudhury A. Memory model sensitive bytecode verification. Form. Methods Syst. Des., 31(3), 2007, pp. 281–305. [14] Boyapati C., Lee R., Rinard M. Ownership types for safe programming: preventing data races and deadlocks. References In Proceedings of the 17th ACM SIGPLAN conference on Object- oriented programming, systems, languages, and applications, 2002, pp. 211–230. [15] Flanagan C., Freund S. Type inference against races. Sci. Comput. Program., Vol 64, January 2007, pp. 140–165. [16] Rose J., Swamy N., Hicks M. Dynamic inference of polymorphic lock types. Science of Computer Programming, 58(3), 2005, pp. 366–383. [17] Biswas S., Zhang M., Bond M., Lucia B. Valor: Efficient, Software-Only Region Conflict Exceptions. In OOPSLA, 2015, pp. 241–259. [18] Flanagan C., Freund S. FastTrack: Efficient and Precise Dynamic Race Detection. In ACM Conference on Programming Language Design and Implementation, 2009, pp. 121–133. [19] Kini D., Mathur U., Viswanathan M. Dynamic race prediction in linear time. SIGPLAN Not. 52(6), 2017, pp. 157–170. [20] Qi Y., Das R., Luo Z., Trotter M. MulticoreSDK: a practical and efficient data race detector for real-world applications. Proceedings Software Testing, Verification and Validation (ICST), IEEE, 21-25 March 2011, pp. 309–318. [21] Serebryany S., Iskhodzhanov T. ThreadSanitizer: Data race detection in practice. In Proceedings of the Workshop on Binary Instrumentation and Applications, 2009, pp. 62– 71. [22] Serebryany K., Potapenko A., Iskhodzhanov T., Vyukov D. Dynamic race detection with LLVM compiler - compile-time instrumentation for ThreadSanitizer. In RV, 2011, Lecture Notes in Computer Science, vol 7186, pp. 110–114. [23] Yu M., Bae D., SimpleLock+: Fast and Accurate Hybrid Data Race Detection. Comput. J., vol. 59, no. 6, 2016, pp. 793–809. [24] Zhang T., Jung C., Lee D. ProRace: Practical Data Race Detection for Production Use. In Proceedings of the International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS), 2017, pp. 149–162. [25] Bond M., Coons K., McKinley K. Pacer: Proportional Detection of Data Races. Proceedings of 2010 ACM SIGPLAN Conference on Programming Language Design and Implementation (PLDI 2010), Toronto, June 2010, pp. 255–268. [26] Marino D., Musuvathi M., Narayanasamy S. LiteRace: Effective Sampling for Lightweight Data Race Detection. PLDI '09 Proceedings of the 2009 ACM SIGPLAN 61 Trifanov V.Yu. Applying synchronization contracts approach for dynamic detection of data races in industrial applications. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 47-62 conference on Programming language design and implementation, Vol. 44, Issue 6 2009, pp. 134–143. conference on Programming language design and implementation, Vol. 44, Issue 6, 2009, pp. 134–143. conference on Programming language design and implementation, Vol. 44, Issue 6, 2009, pp. 134–143. pp [27] Trifanov V.Yu. References Detecting data races in Java programs with synchronization contracts. Komp'juternye instrumenty v obrazovanii [Computer Tools in Education]. №4, 2012, pp. 16-29. (in Russian) [27] Trifanov V.Yu. Detecting data races in Java programs with synchronization contracts. Komp'juternye instrumenty v obrazovanii [Computer Tools in Education]. №4, 2012, pp. 16-29. (in Russian) [28] Trifanov V.Yu., Tsitelov D.I. Dynamic detection of data races in Java programs with synchronization contracts. Materialy konferencii "Instrumenty i metody analiza programm (TMPA-2013)" [Proc of Tools and Methods of Program Analysis conference TMPA-2013], Kostroma, 2013, pp. 273–285. (in Russian) [29] Trifanov V.Yu., Tsitelov D.I. Language for synchronization contracts creation to detect races in multithreaded applications. Programmnaja inzhenerija [Software Engineering], vol. 8, N 6, 2017, pp. 250–257. (in Russian) [30] Elmas T., Qadeer S., Tasiran S. Goldilocks: A Race and Transaction-Aware Java Runtime. Proceedings of the 2007 ACM SIGPLAN Conference on Programming Language Design and Implementation (PLDI'07), 2007, pp. 245–255. [31] Lamport L. Time, Clocks and the Ordering of Events in a Distributed System. Communications of the ACM, Vol. 21, Issue 7, 1978, pp. 558–565. [32] Intel Thread Checker, http://software.intel.com/en-us/intel-thread-checker/ [33] Trifanov V.Yu. Dynamic data race detection in multithreaded Java-programs. PhD thesis, SPbSU, 2013. (in Russian) [34] Liskov B., Wing J. A behavioral notion of subtyping. ACM Trans. Program. Lang. Syst. 16 (6). November 1994, pp.1811–1841. [35] Click C. A lock-free wait-free hash table. https://web.stanford.edu/class/ee380/Abstracts/070221_LockFreeHash.pdf [35] Click C. A lock-free wait-free hash table. https://web.stanford.edu/class/ee380/Abstra [36] Gavrilova T.A., Leshheva I.A., Kudrjavcev D.V. Using models of knowledge engineering for growing specialists in information technologies area. Sistemnoe programmirovanie [System programming], vol. 7, № 1, 2012, pp.. 90–105. (in Russian) [37] Koznov D.V. Basis of visual modeling. Internet-Universitet Informacionnyh Tehnologij (INTUIT) [Internet-University of Information Technologies], Moscow, 2008 (in Russian) [38] Ol'hovich L.B., Koznov D.V. OCL-Based Automated Validation Method for UML Specifications. Programming and Computer Software, vol. 29, № 6, 2003, pp. 44–50. DOI: 10.1023/B:PACS.0000004132.42846.11 [39] Ivanov A., Koznov D., Murasheva T. Behavioral model RTST++, Zapiski seminara Kafedry sistemnogo programmirovanija "Case-sredstva RTST++" [Notes of seminar “Case-tools RTST++” of system engeneering department], 1998, № 1, pp. 37–52. (in Russian) [40] Luciv D.V., Koznov D.V., Basit H.A., Terehov A.N. On fuzzy repetitions detection in documentation reuse. Programming and Computer Software, vol. 42, № 4, 2016, pp. 39– 49. DOI: 10.1134/S0361768816040046 [41] Koznov D.V., Romanovskij K.Ju. Automated documentation refactoring for lines of program products. Sistemnoe programmirovanie [System programming], vol. 4, 2009, pp. 128–150. (in Russian) 62 DOI: 10.15514/ISPRAS-2018-30(3)-5 For citation: Chebykin A.E., Kirilenko I.A. Applying Deep Learning to C# Call Sequence Synthesis. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 63-86. DOI: 10.15514/ISPRAS-2018-30(3)-5 Applying Deep Learning to C# Call Sequence Synthesis A.E. Chebykin <a.e.chebykin@gmail.com> I.A. Kirilenko <jake.kirilenko@gmail.com> Faculty of Mathematics and Mechanics, Saint Petersburg State University Universitetsky prospekt, 28, Peterhof, St. Petersburg, 198504, Russia Abstract. Many common programming tasks, like connecting to a database, drawing an image, or reading from a file, are long implemented in various frameworks and are available via corresponding Application Programming Interfaces (APIs). However, to use them, a software engineer must first learn of their existence and then of the correct way to utilize them. Currently, the Internet seems to be the best and the most common way to gather such information. Recently, a deep-learning-based solution was proposed in the form of DeepAPI tool. Given English description of the desired functionality, sequence of Java function calls is generated. In this paper, we show the way to apply this approach to a different programming language (C# over Java) that has smaller open code base; we describe techniques used to achieve results close to the original, as well as techniques that failed to produce an impact. Finally, we release our dataset, code and trained model to facilitate further research. Keywords: API; deep learning; code search; RNN; transfer learning. 1. Introduction When writing code, software developers often utilize various libraries via APIs. Since the problems being solved in this manner are usually similar for most users, their solutions form stable patterns of API invocations. API mining is a long-established line of research aimed at extracting these API usage trends from source code. The importance of the task lies in the fact that generally developers spend a lot of time trying to learn frameworks’ APIs in order to utilize them efficiently. A field study has found that developers often struggle to map a task from problem domain to the terminology of the API [1]. In another survey 67.6% of respondents identified that learning APIs is hindered by inadequate or absent resources [2]. 63 Chebykin A.E., Kirilenko I.A. Applying Deep Learning to C# Call Sequence Synthesis. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 63-86 Chebykin A.E., Kirilenko I.A. Applying Deep Learning to C# Call Sequence Synthesis. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 63-86 Usually, when facing such problems, developers turn to general web search engines. However, those are not optimized for programming-related queries and thus tend to be inefficient [3]. Usually, when facing such problems, developers turn to general web search engines. However, those are not optimized for programming-related queries and thus tend to be inefficient [3]. An alternative lies in various approaches based on statistical analysis of source code. They can provide sequences of API methods that are often used together [4], mine API specifications in the form of automata [5], synthesize relevant code snippets [6]. Deep API Learning [7] is a recent deep learning-based take on the problem that reports state-of-the-art results. The authors formulate the problem of providing API patterns satisfying users’ needs as a translation one. Input language, in which user describes desired functionality, is English, and the output language is one of API sequences: API calls are words of the language, ordered sequences of these calls form sentences. For example, English sentence “generate random int” could be translated to the language of Java API as “Random.new Random.nextInt”, which corresponds to the construction of an object of type Random and subsequent call of its nextInt method. DeepAPI tool targets exclusively Java programming language and reportedly performs well. Benefits of the approach come from the usage of deep recurrent neural networks. 1. Introduction 63-86 Our main contributions are:  reproduction of the DeepAPI experiment with a different dataset;  reproduction of the DeepAPI experiment with a different dataset;  modification of the approach via programming-language-independent data preprocessing which leads to results, comparable to original, despite lack of data;  collection of C# dataset of commented methods and publishing of it for the benefit of the future research in the area;  collection of C# dataset of commented methods and publishing of it for the benefit of the future research in the area;  employment of transfer learning techniques for additional improvement of the results. To the best of our knowledge, we are the first to investigate transfer learning in the area of API mining. The paper is organized as follows: in section 2 we outline DeepAPI model architecture. Next, in section 3 collection of the dataset needed for model training is discussed and additional preprocessing steps are introduced. We describe our application of transfer learning to the problem in section 4. Technical details of model training are reported in section 5, which is followed by section 6, where evaluation results are described. We finish the paper with section 7, where we report work done on related problems and discuss ways in which existing research differs from ours. 1. Introduction Thanks to them, trained model can distinguish synonyms and impact of word sequence (for example, it can distinguish queries convert string to int and convert int to string). However, the authors identify several threats to validity, including possible failure when extending the approach to other programming languages. Our main goals are to test this threat, thus appraising generality of the approach, and to consider possible improvements. We choose C# as a target language due to its general similarity to Java, aiming to make a first step towards more different — and therefore challenging — target languages. However even in our case simple copying of DeepAPI approach leads to bad results, and constructing well-working model proves to be far from trivial. In this paper, we describe our experience of extending the proposed approach to C#. To achieve our goals we collect dataset of 2,886,309 training samples from open source projects’ code and use it to first train a model with the architecture of DeepAPI (attaining the result of 10.94 BLEU), and then tune parameters to achieve BLEU 26.26. After that, we introduce data preprocessing, which reduces dataset size to 1,397,597, but improves its quality and increases BLEU metric to 46.99. Finally, we employ transfer learning on an alternative dataset of method names and achieve the best results of 50.14 BLEU, which is fairly close to the 54.42 reported by DeepAPI on Java dataset. Additionally we ask professional developers to evaluate output of our model on several queries, which shows that on average our model, DeepAPI#, performs as well as DeepAPI. 64 .Е., Кириленко Я.А, Применение глубокого машинного обучения к синтезу цепочки вызовов C#. П РАН, том 30, вып. 3, 2018 г., стр. 63-86 Чебыкин А.Е., Кириленко Я.А, Применение глубокого машинного обучения к синтезу цепочки вызовов C#. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 63-86 Чебыкин А.Е., Кириленко Я.А, Применение глубокого машинного обучения к синтезу цепочки вызовов C#. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 2. DeepAPI model We borrow general model structure from DeepAPI, which is itself based on recent advancements in neural machine translation. Here we will provide only an overview, for details please refer to the original paper [7] and our previous research-in-progress paper [8]. Since the goal is to generate one sequence of words based on another, the task falls in the category of Sequence-to-Sequence learning [9]. One of the best architectures for the task is an Encoder-Decoder network [10]. It consists of two recurrent neural networks (Recurrent neural network is a special class of neural networks where unit can be connected to itself, thus allowing its state to serve the role of memory). Encoder network reads input sequence, Decoder generates output one. The process goes as following. Encoder reads input word by word, embeds each one in a high-dimensional space and sequentially updates its hidden state, which by the end of the sentence contains language-independent idea of the input sentence. This state (also known as context vector) is then passed to the Decoder, which based on it and the last generated word generates words one by one until a special end-of-sequence token is outputted. An example of such model at work can be seen in Fig. 1. In the image states of networks are rolled out in time, so for example RNN1, RNN2, RNN3 is the RNN state at time steps 1, 2, 3. Note that Encoder and Decoder consist of different RNNs and work in different time windows: at first, Encoder RNN makes 3 steps in time and then Decoder RNN makes 3 steps in time. 65 Chebykin A.E., Kirilenko I.A. Applying Deep Learning to C# Call Sequence Synthesis. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 63-86 Fig. 1. RNN Encoder-Decoder workflow Fig. 1. RNN Encoder-Decoder workflow The benefits of this model include synonym handling (words used in the similar contexts get embedded near each other), successful processing of long inputs thanks to the memorizing ability of the recurrent networks, and finally appreciation of word sequence impact. One major downside of such a model is the need for a large amount of sentence pairs describing the same functionality in two languages (“generate random number”, “Random.new Random.Next”). Format of the API language description is reported in the point 3.1.3. 3.1 Dataset collection To train the model, we need to gather large amount of pairs (English description of functionality, API description). One way to do it is to process open source projects, looking for methods with documentation comments, extracting summary sections and linearizing interesting parts of ASTs (i.e. API calls). The processing of individual methods is described in section 3.1.3. GitHub1 is one of the most popular open source project hostings. Following DeepAPI authors, we construct our dataset from data published there. We attempted to augment GitHub data with data from alternative sources. In our previous paper [8] we proposed using Nuget2 – a repository of compiled C# packages. However we eventually found out that compared to GitHub it does not provide much data, and what samples it provides often duplicate ones collected from GitHub. So we discontinued using Nuget as data source. There are other sites with published open source projects, for example, Codeplex3 and SourceForge4. Unfortunately, we found there only a small amount of C# projects, many of which gradually migrate to GitHub, or have already done so. These hosting sites also lack search APIs that are essential for the automatic collection of our dataset. So the potentially small amount of additional data is nontrivial to collect, and therefore we choose to ignore these alternative sources. We collect dataset from GitHub in several steps: We collect dataset from GitHub in several steps: 1) obtain a list of repositories relevant to us; 2) download these repositories; 3) process them, extracting from methods with documentation comments these comments, linearized in a special way API calls, types and names of method parameters. 3) process them, extracting from methods with documentation comments these comments, linearized in a special way API calls, types and names of method parameters. The architecture overview can be seen in Fig. 2. Let us discuss every step in detail. 2. DeepAPI model Source of such data can be methods’ documentation comments (that in C# are XML- based and contain summary section, in which brief description of the method’s functionality should be supplied) and corresponding API calls made in the method body. Details of the dataset collection are described in section 3. There are several improvements of the Encoder-Decoder architecture that were shown to reliably improve results.  Using Bidirectional Encoder leads to input being processed twice: in normal order and in reverse, resulting in 2 context vectors, which are then concatenated to get final context vector [11].  Attention mechanism [12] allows decoder to focus on different input words when generating different output ones. In the original DeepAPI paper an additional improvement is introduced in the form of a regularization term punishing generation of the most widespread and therefore probably problem-irrelevant API calls, such as logging ones. We have not tried such regularization since its reported impact on BLEU score is minimal. We leave testing of this enhancement for future research. In the original DeepAPI paper an additional improvement is introduced in the form of a regularization term punishing generation of the most widespread and therefore probably problem-irrelevant API calls, such as logging ones. We have not tried such regularization since its reported impact on BLEU score is minimal. We leave testing of this enhancement for future research. 66 Чебыкин А.Е., Кириленко Я.А, Применение глубокого машинного обучения к синтезу цепочки вызовов C#. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 63-86 Чебыкин А.Е., Кириленко Я.А, Применение глубокого машинного обучения к синтезу цепочки вызовов C#. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 63-86 3.1.1 Obtaining list of relevant repositories We are interested in repositories in C# language. Similar to the original paper, we would like to consider only projects that have at least one star in order to filter unused or toy projects. Both these requirements can be satisfied when setting specific parameters of GitHub Search API. 1 github.com 2 nugget.org 3 archive.codeplex.com 4 sourceforge.net 1 github.com 2 nugget.org 3 archive.codeplex.com 4 sourceforge.net 67 Chebykin A.E., Kirilenko I.A. Applying Deep Learning to C# Call Sequence Synthesis. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 63-86 Fig. 2. Dataset gathering workflow Fig. 2. Dataset gathering workflow Using this API via Octokit.rb5 library, we retrieve 140,990 URLs of relevant projects created from 2012 to 2017. This contrasts to the original paper that reports working with 442,928 Java repositories. Therefore, we initially have approximately 3 times less projects to work with. This lack of data can potentially be a significant obstacle when transferring the approach to other languages with smaller open code bases. Search API also poses several technical difficulties. Firstly, it returns no more than 1,000 results for any search request. To go around this restriction, we set additional parameter limiting repository creation date to a short span of time, for example, “2016-01-01 .. 2016-01-08”. Every our requests covers 8 days, which we find short enough a period that no more than 1,000 repositories are created during it. Secondly, Search API limits number of requests per minute by 30. In order not to exceed this limit, our script sleeps for 2 seconds after each request. We store repositories list and the rest of our data in a SQLite database6. 5 github.com/octokit/octokit.rb 6 sqlite.org 3.1.2 Downloading repositories Having gathered repository list, we can start cloning them with git. We set clone depth to 1 to speed up the process. 68 Чебыкин А.Е., Кириленко Я.А, Применение глубокого машинного обучения к синтезу цепочки вызовов C#. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 63-86 After download, we search for solution files — special files that encompass source code files, as well as store project dependencies. We process these files in the next step. 7 github.com/dotnet/roslyn 3.1.3 Extracting data C# type system is problematic for our purposes compared to Java because of the implicit type “var” introduced in version 3.0. As a consequence of its existence, code needs to be compiled in order for the type of a variable to be determined correctly, as opposed to Java where name of the variable’s type or supertype is evident from its declaration. This need for compilation limits number of projects we can process. For compilation and syntax tree processing, we use Roslyn7 — an open source C# compiler developed by Microsoft. To compile a project we need it to satisfy two requirements: 1) no manual actions are necessary for its build and compilation; 2) a solution file, encompassing source code files, must exist. In order to compile more projects, we employ Nuget to restore project dependencies prior to compilation. In order to compile more projects, we employ Nuget to restore project dependencies prior to compilation. About 80.6% percent of repositories contain solution files, and of those 47.1% could be compiled. After compilation, we process projects in the following fashion After compilation, we process projects in the following fashion: 1) find methods with documentation comments; 2) store whole comment and summary section; 3) walk syntax tree of the method body, collecting API call sequence; 4) store method name; 5) store parameter types and names, which we think can potentially provide valuable information, but are not used in this work. An example of extracting data from method with documentation comment is provided in Fig. 3. We construct API sequence similarly to the original paper. We traverse the tree in the way an interpreter might traverse it during execution, e.g. depth-first post order, processing method call’s arguments before processing the call itself, and so on. When encountering constructor invocation new C(), we add C.new to the API sequence. Wh t i th d ll () h i i t f l C dd C We construct API sequence similarly to the original paper. We traverse the tree in the way an interpreter might traverse it during execution, e.g. depth-first post order, processing method call’s arguments before processing the call itself, and so on. When encountering constructor invocation new C(), we add C.new to the API sequence. When encountering method call o.m() where o is an instance of a class C, we add C.m to the API sequence. 3.1.3 Extracting data Additionally, when encountering if-else statement, we firstly process condition expression, then if-branch statements and finally else-branch g (), q When encountering method call o.m() where o is an instance of a class C, we add C.m to the API sequence. Additionally, when encountering if-else statement, we firstly process condition expression, then if-branch statements and finally else-branch statements. When encountering method call o.m() where o is an instance of a class C, we add C.m to the API sequence. Additionally, when encountering if-else statement, we firstly process condition expression, then if-branch statements and finally else-branch statements. We introduce one additional step to this scheme: when encountering try-with- resources node, we save the class C of an object being created in the try node and 69 Chebykin A.E., Kirilenko I.A. Applying Deep Learning to C# Call Sequence Synthesis. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 63-86 Chebykin A.E., Kirilenko I.A. Applying Deep Learning to C# Call Sequence Synthesis. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 63-86 lenko I.A. Applying Deep Learning to C# Call Sequence Synthesis. Trudy ISP RAN /Proc. ISP RAS, 8 pp 63-86 after processing everything inside try branch we add C.Dispose to the API sequence. While it is easier for a programmer to rely on the language feature of try-with- resources block to take care of finalization of the resources, this construct is not always used, and we think that our model should know that certain sequences of API calls end with finalization call. Eventually we obtain 2,886,309 pairs of English descriptions and API sequences. However, this number is not directly comparable to the 7,519,907 methods reported in the DeepAPI paper. The authors explained to us (in an e-mail) that 7,519,907 is the amount of data after filtering out-of-vocabulary words, the step which in our experience removes certain samples entirely, significantly reducing size of the dataset. Our preprocessing and the final size of dataset is discussed in the further section. Fig. 3. Example of data extraction 3.2 Data preprocessing Upon inspecting the gathered data we conclude that it can be improved prior to being used for model training. By introducing following preprocessing steps we aim to make the training easier and the results consequently better - a notion supported by our experiments (see section 6). 70 70 Чебыкин А.Е., Кириленко Я.А, Применение глубокого машинного обучения к синтезу цепочки вызовов C#. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 63-86 Чебыкин А.Е., Кириленко Я.А, Применение глубокого машинного обучения к синтезу цепочки вызовов C#. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 63-86 3.2.2 Leaving only distinct pairs The percent of unique pairs is about 86.6%. Note that we consider two pairs distinct even if English descriptions coincide while API descriptions do not, and vice versa. We could identify several reasons for occurrence of repetitions:  auto-generated code and comments (Windows Forms are especially ubiquitous);  auto-generated code and comments (Windows Forms are especially ubiquitous);  libraries being copied to the project sources instead of being linked as dependencies.  libraries being copied to the project sources instead of being linked as dependencies. This step reduces amount of training instances from 2,606,424 to 2,259,653. This step reduces amount of training instances from 2,606,424 to 2,259,653. 3.2.1 Language detection We consider our model to work with English language as input, however, many comments are not in it. Therefore, we try to filter out non-English comments using a language detection package8. We find, however, that some English sentences are recognized as non-English. In our opinion, most likely reasons are extreme shortness of sentences used for language detection and uncommon profession-specific programmers’ vocabulary. We do not want to decrease dataset size by filtering out comments incorrectly recognized as non- English, and so we change our filtering approach. Instead of leaving only sentences recognized as English, we remove ones that are reported to be in a set of well-recognizable languages (which we deduce by hand examination) that occur in our dataset most often. Languages, sentences in which we remove, are Chinese, Korean, Japanese, Russian, German and Polish (reported in the order of decreasing frequency). As a side note, the reason for good recognition of said languages probably lies in them having alphabets different from the English one. Such filtering leads to vocabulary containing mostly English words. It reduces training size from 2,886,309 pairs to 2,606,424. 8 github.com/Mimino666/langdetec 72 9 kaggle.com/awesomelemon/csharp-commented-methods-github 10 github.com/AwesomeLemon/api-extraction 11 github.com/AwesomeLemon/api-extraction-scripts 3.2.3 Repetition contraction In some API sequences an API call is repeated several times in a row. This could happen as a result of our AST linearization in a situation where, for example, an API call is made with different parameters in branches of an if-else statement. Since we do not record call parameters and when linearizing if-else statement save API calls from both branches, this may lead to an API call repeating twice in the resulting sequence. End user would not care about such repetitions in the output of the model, so we remove them before training, leaving only one copy of API call in a row. This step does not influence amount of data but rather is intended to improve quality This step does not influence amount of data, but rather is intended to improve quality of the existing training samples 71 Chebykin A.E., Kirilenko I.A. Applying Deep Learning to C# Call Sequence Synthesis. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 63-86 Kirilenko I.A. Applying Deep Learning to C# Call Sequence Synthesis. Trudy ISP RAN /Proc. ISP RAS, 2018, pp. 63-86 Chebykin A.E., Kirilenko I.A. Applying Deep Learning to C# Call Sequence Synthesis. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 63-86 4. Transfer learning for API mining Broadly speaking, transfer learning is utilizing knowledge gained in one problem to solve another. It is often used in NLP [13] and neural machine translation, especially in the contexts where data is scarce [14]. Since our situation is one of lacking data (as shown by an experiment in section 6), we decided to investigate this idea. 3.2.5 Stemming Additionally we experiment with, but eventually discard a preprocessing step of stemming. Stemming is the process of reducing inflected words to their bases. We intended to use it, as is usual, to decrease vocabulary by replacing multiple word forms with the root. In our case it fails to provide improvement and instead makes results worse. A possible explanation may lie in the fact that stemming model was trained on regular words, not ones specific for software development and therefore works badly with this unusual vocabulary. We discuss impact of the preprocessing steps in section 6. The final size of our dataset is 1,397,597 pairs, which is more than 5 times smaller than 7,519,907 pairs used for training in the original paper. Even if only preprocessing from the original paper is used (i.e. vocabulary filtering and nothing else), dataset size is 1,692,898 (of which 1,434,805 pairs are unique). We consider this a significant problem that very probably makes achieving comparable results harder and takes a great toll on the model performance. For easy reproduction of our research and for conduction of new experiments in the area, we provide our dataset9, as well as the code used to collect10 and preprocess11 it. 3.2.4 Vocabulary filtering Similar to the original paper, we create vocabularies of 10,000 most popular words in each language, and filter out the rest. If after filtering no words are left in either English description or API one, we remove the pair altogether. This step reduces training dataset size significantly, from 2,259,653 to 1,397,597. 4.1 Alternative dataset To apply transfer learning to our problem of generating API calls given English description, we need to train a model for a task that is different, yet very similar. As already mentioned, the DeepAPI paper proposes method body as a source of API description of the functionality and method comment as a source of the English one. But there is another description for a method functionality beside its comment — its 72 g Чебыкин А.Е., Кириленко Я.А, Применение глубокого машинного обучения к синтезу цепочки вызовов C#. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 63-86 .Е., Кириленко Я.А, Применение глубокого машинного обучения к синтезу цепочки вызовов C#. П РАН, том 30, вып. 3, 2018 г., стр. 63-86 name. Combined with class name, it seems descriptive of the method’s contents While not forming proper natural language sentences, these names could provide crude approximations. name. Combined with class name, it seems descriptive of the method’s contents. While not forming proper natural language sentences, these names could provide crude approximations. Examples of correspondence between comments and names of the methods are provided in Table 1. It can be seen that generally tokenized names are very similar to summary sections of documentation comments. However, this is not always the case. Examples of correspondence between comments and names of the methods are provided in Table 1. It can be seen that generally tokenized names are very similar to summary sections of documentation comments. However, this is not always the case. In the last two examples despite similarity between comment and name, essential information is missing from the tokenized name. In the first of these samples key word is “Matches”, without it tokenized method name loses meaning. In the second one “Dword” is separated to “d” and “word” due to the tokenizing technique. When we tokenize method name, we assume that naming guidelines are followed and therefore first letter of the method name and first letters of every word in the name are capitalized. Here this leads to a wrong division of words and thus vital information disappears, making description senseless. However, in most cases method names tokenized in this way are similar to comments and thus provide relatively good description of method contents. We start exploration of this alternative dataset by simply training a model on it with the best parameters and our preprocessing. 4.2 Applying transfer learning for model improvement We hypothesize that the alternative method names dataset contains valuable information about correspondence between English words and API calls. In terms of transfer learning, we can consider both our source task and target task to be the same, namely to generate API call sequence given English description of it. The difference lies in the datasets. When training for the source task, we can use the alternative dataset of pairs (Tokenized method name, API call sequence). Then we can utilize gained knowledge when training the model for the target task, which makes use of the original dataset of pairs (Documentation comment summary, API call sequence). Therefore, we train a model on the alternative dataset, and then use learned weights for initializing the model to be trained on the standard dataset, which is a technique known as pretraining. In addition, we wonder if we can similarly bootstrap learning without using an alternative dataset. We perform an experiment by training the model on the comments dataset and using it for initialization and training on the same dataset. We evaluate impact of both approaches in section 6. Chebykin A.E., Kirilenko I.A. Applying Deep Learning to C# Call Sequence Synthesis. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 63-86 Chebykin A.E., Kirilenko I.A. Applying Deep Learning to C# Call Sequence Synthesis. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 63-86 E., Kirilenko I.A. Applying Deep Learning to C# Call Sequence Synthesis. Trudy ISP RAN /Proc. ISP RAS 3, 2018, pp. 63-86 lenko I.A. Applying Deep Learning to C# Call Sequence Synthesis. Trudy ISP RAN /Proc. ISP RAS, 8 63 86 74 12 github.com/pascanur/GroundHog 4.1 Alternative dataset Results are not very good (model №4 in Table 2; the table is discussed minutely in section 6). We conclude that comments indeed seem to be more descriptive of method contents than method names. But can we utilize this new dataset nonetheless? than method names. But can we utilize this new dataset nonetheless? Table 1. Comparison between method names and comments Full method name Tokenized method name Summary section of documentation comment Method name corresponds to comment well ManagedFusion. Serialization. JsonSerializer.Serialize json serializer serialize Serializes to JSON MathNet.Symbolics. Packages.Standard. Structures. ComplexValue.Cosine complex value cosine Trigonometric Cosine of an angle in radians StickyDesk. Utilities.ResizeBitmap utilities resize bitmap Resizes a bitmap image Nini.Config. IniConfigSource. RemoveSections ini config source remove sections Removes all INI sections that were removed as configs Method name corresponds to comment badly Spark.Parser. CharGrammar. StringOf char grammar string of Matches a string of characters TagLib.Asf. DescriptionRecord. ToDWord description record to d word Gets the DWORD value contained in the current instance. 73 5. Model training Per description in section 2, original authors use Encoder-Decoder architecture. As implementation of RNN they choose GRU [10]. They use 1-layered model with 1,000 hidden units and 120 dimensions for word embedding. To train the model, GroundHog12 is used. GroundHog since then has been discontinued, instead we use popular modern framework OpenNMT [15] that is designed specifically to train neural translation models. We start training from the architecture reported in the original paper. After getting bad results we go on and empirically tune parameters, eventually arriving at following values. As RNN implementation we use LSTM [16] — a more complex model than GRU, with on-par performance, which is highly dependent on the problem. In our task it performs better. We find that 1 layer makes model not complex enough to work with C#, and since it is known that adding more layers increases model’s learning ability [17], we introduce additional layers to the total of 3, which impacts results positively. We leave number of hidden units at 1,000 and word embedding at 120 dimensions. For training, Stochastic Gradient Descent [18] is used with batch size of 32 and exponential learning rate decay. We initialize learning rate to be 1.0 and start multiplying it by 0.7 after every epoch, starting from the sixth one. Every model is trained for approximately 25 epochs on the server equipped with one Nvidia GTX 1070 GPU. 74 Чебыкин А.Е., Кириленко Я.А, Применение глубокого машинного обучения к синтезу цепочки вызовов C#. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 63-86 Чебыкин А.Е., Кириленко Я.А, Применение глубокого машинного обучения к синтезу цепочки вызовов C#. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 63-86 For model testing we separate 12,000 random pairs of descriptions from the dataset; the rest is used for training. We publish our trained model for easy reproduction of the results13. For model testing we separate 12,000 random pairs of descriptions from the dataset; the rest is used for training. We publish our trained model for easy reproduction of the results13. After training, when translating queries to API sequences we follow original authors in using beam search [19], a heuristic search algorithm popular in statistical translation. Instead of generating only the most probable word on every step, we generate multiple, and then keep only several most probable sequences. 5. Model training This approach solves the problem of discarding good translation sequences because of some sub-optimal words. 6.1 Metrics In the area of API mining there are no universally adopted metrics. For better comparison to the original paper we follow in its steps and calculate BLEU score [20] for intrinsic evaluation, FRank score [6] and Precision@N for extrinsic one. 6.1.1 BLEU BLEU is a standard metric used in machine translation to evaluate how closely generated translation resembles reference one. It does not consider grammar or others high-level features, instead calculating corrected geometric mean of n-gram precision on the whole test set [20]. Since we expect the model to generate sequences of API calls similar to the ones extracted from human-written source code, n-gram approach is applicable to our situation. The theoretical foundations of the metric stand in our case, despite target language being language of API calls rather than natural language. BLEU is reported on the scale from 0 to 100, where higher score corresponds with bigger similarity between generated and reference sequences. 13 public-resources.ml-labs.aws.intellij.net.s3.amazonaws.com/deep-api-sharp/deep-api- sharp-model.t7 6.1.3 Precision@N Precision@N measures percentage of the relevant results in the first N outputs produced by the system. Following DeepAPI, we report Precision@5 and Precision@10 (note that the term used in the DeepAPI paper is “relevancy ratio N”, which does not seem to be an established term). This metric is reported on the scale from 0 to 100, where higher is better. 6.1.2 FRank FRank metric value is the position of the first relevant result in the ranked list, as decided by a human evaluator. Such a metric is justified by two facts. Firstly, good scores of it show that the model has solved exactly the problem we intended for it, i.e. the problem of translating from English to relevant API calls. It was possible for the model to learn a target function uninteresting for us, in which case human evaluators would not find in model output API calls, relevant to the input. Secondly, it is known that humans scan through ranked results from top to bottom [21], thus making it a desired trait for a model to rank relevant output higher. 75 Chebykin A.E., Kirilenko I.A. Applying Deep Learning to C# Call Sequence Synthesis. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 63-86 E., Kirilenko I.A. Applying Deep Learning to C# Call Sequence Synthesis. Trudy ISP RAN /Proc. ISP RAS 3, 2018, pp. 63-86 In our case FRank is measured on the scale from 1 to 10 (since similar to the DeepAPI paper, our model generates 10 outputs for every query), where lower is better. Where models fail to provide relevant results, FRank is considered to be 11. 6.2 BLEU evaluation In Table 2 we report results of our experiments in terms of BLEU score. We start experiments with model architecture reported in the original paper and achieve surprisingly bad results of 10.94 BLEU, which is significantly worse than 54.42 BLEU reported in the paper. Since Java and C# are fairly similar, we expected original model to work better. Possible explanation may lie in the size of our dataset, which is more than 5 times smaller. Table 2. BLEU scores for various models № Parameters Dataset Preprocessing Transfer learning from model № BLEU Parameter tuning 1 original comments - - 10.94 2 tuned comments - - 26.26 Data preprocessing 3 tuned comments yes - 46.99 Different datasets 4 tuned names yes - 28.57 5 tuned comments (part) yes - 36.63 6 tuned comments and names yes - 44.31 Transfer learning 7 tuned comments yes 3 46.18 8 tuned comments yes 4 50.14 Model with tuned parameters achieves higher BLEU score of 26.26, which is still far from the original results. 76 Чебыкин А.Е., Кириленко Я.А, Применение глубокого машинного обучения к синтезу цепочки вызовов C#. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 63-86 Чебыкин А.Е., Кириленко Я.А, Применение глубокого машинного обучения к синтезу цепочки вызовов C#. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 63-86 After introduction of our preprocessing steps a 94% increase in BLEU is obtained, and the resulting score of 46.99 comes fairly close to the reported performance of the DeepAPI model. The best result is achieved by model №8, where we employ transfer learning techniques and pretrain the model on the alternative dataset of method names. Additional analysis of transfer learning application is presented in section 6.3. Model №4 was trained on the alternative dataset of method names (with the size of 1,967,414 pairs) and yielded not outstanding BLEU score of 28.57. So our model performs worse on the alternative dataset, which is logical, given that descriptions there are not in grammatically correct English and sometimes do not provide good descriptions of functionality, as already discussed in section 4. To measure if number of training instances indeed impacts model result, as we hypothesized, we try to train the model on 800,000 samples as opposed to the usual 1,397,597. This is the model №5, and it achieves 36.63 BLEU, which is worse than 46.99 achieved under the same parameters, but bigger dataset size. 6.2 BLEU evaluation This leads us to the conclusion that dataset size is vital for model performance. 6.3 Transfer learning evaluation several questions regarding our application of transfer learning techniques: 1) Does it improve our results? 1) Does it improve our results? 2) Can we use the model itself for pretraining, without utilizing model trained on the alternative dataset? 2) Can we use the model itself for pretraining, without utilizing model trained on the alternative dataset? 3) Is transfer learning necessary for performance improvement, or are instead our two datasets so similar that they could be merged and considered as one big dataset? We answer these questions with several experiments, and come up with following nswers. 1) Transfer learning leads to the best results achieved by us (model №8 with BLEU score of 50.14). 2) A model with sub-optimal parameters (which we do not include in the table in order not to clutter it) is improved by approximately 2.5 BLEU when pretrained on itself. However, best model is not, as shown by performance of model №7, that achieves only 46.18 BLEU, which approximately equals the result of the model №3 used for pretraining. So bootstrapping with the dataset itself may make sense sometimes, but not always. Presumably, model №3 was trained so well that there was no room for improvement. 3) We try to merge comments and names in one dataset, which we use for training model №6. Resulting BLEU score of 44.31 is better than using only names (28.57 BLEU, model №4), but worse than using only comments (46.99 BLEU, model №3). Thus we conclude that datasets are fairly different and should not be used together in a straightforward way. 77 Chebykin A.E., Kirilenko I.A. Applying Deep Learning to C# Call Sequence Synthesis. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 63-86 Chebykin A.E., Kirilenko I.A. Applying Deep Learning to C# Call Sequence Synthesis. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 63-86 Kirilenko I.A. Applying Deep Learning to C# Call Sequence Synthesis. Trudy ISP RAN /Proc. ISP RAS, 2018 pp 63-86 Chebykin A.E., Kirilenko I.A. Applying Deep Learning to C# Call Sequence Synthesis. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 63-86 6.4 Human evaluation DeepAPI reports FRank, Precision@5, Precision@10 on two types of queries: popular ones, that often occur in Bing search log [6], and ones designed to showcase abilities of the proposed approach, including handling of semantically similar requests, longer input handling, combination of several tasks. We would like to address a potential problem in evaluating the model on queries from the first group. While the DeepAPI paper reports that they do not occur in the training dataset, it seems unlikely since they were chosen for the perk of being popular, i.e. widespread, and authors do not mention filtering them out. We test the hypothesis that such popular queries occur in the dataset by searching for them in ours. In our training data most of these popular inputs occur multiple times as exact matches. For example, “copy file” occurs 14 times, “reverse string” occurs 7 times, “execute sql statement” occurs 14 times. We conduct this search after filtering out non-unique pairs, so for these occurrences API calls do not coincide; however, they are very similar. Therefore, we believe that testing on such inputs makes little sense, because it essentially means testing on the training set, which speaks only about the model’s ability to memorize. That is expected from any model, and consequently is not very interesting. However, to show that our model is capable of that, we test on 5 of these queries (the first 5 queries in Table 3). However, more interesting is the inspection of the model’s ability to generalize, i.e. use gained knowledge to work with novel data. The model should be able to handle combined or semantically similar requests that are not included in the training data. We evaluate our model on 4 new queries, constructed for this exact purpose, and one such query from the DeepAPI paper. Since DeepAPI paper does not report results on 4 new queries, we used online demo of the tool14 to generate corresponding sequences. To avoid conflict of interest, we ask 5 professional developers to evaluate extrinsic metrics for our model. Since the correspondence between query and model output is viewed differently by every developer and is up to debate, we consider relevant only those answers that were marked as relevant by at least 2 developers. In the Table 3 we report results of extrinsic evaluation. 6.5 Limitations As already discussed in the previous section, our model can be inconsistent and sensitive to query wording. While DeepAPI# is capable of understanding synonymous queries and generating similar relevant output, it does not generate exactly the same sequences. In addition, our model is data-hungry. While we do not artificially limit our vocabulary with standard C# library, as DeepAPI does with Java and JDK, we still observe that the model cannot take into account APIs with low amount of usages. It can work with extremely popular Math.NET and Json.NET, but not with many other frameworks, even though their APIs are included in the model dictionary. It remains an open problem for the further research to find ways to make model less data-hungry, or to fine-tune it for use of specific not very popular libraries. 6.4 Human evaluation 30, issue 3, 2018, pp. 63-86 E., Kirilenko I.A. Applying Deep Learning to C# Call Sequence Synthesis. Trudy ISP RAN /Proc. ISP RAS 3, 2018, pp. 63-86 DeepAPI, while DeepAPI# produces low amount of relevant answers, and on the contrary, query “write text using socket” perplexes DeepAPI, that generates no socket-related calls in the top 10 results, while DeepAPI# generates only relevant output. Additionally it should be noted that while models’ outputs are not directly comparable due to different target languages, both models should still be able to correctly answer queries we are testing them on, since these tasks are fairly common and programming- language-independent. 6.4 Human evaluation In general our model performs approximately the same as the original, which, having established importance of data and our lack of it, we consider an achievement. Table 3. Extrinsic model evaluation Query DeepAPI DeepAPI# DeepAPI# output FRank P@5 P@10 FRank P@5 P@10 convert int to string 2 40 90 1 80 50 CultureInfo.InvariantCulture Int64.ToString 14 211.249.63.55 14 211.249.63.55 78 Чебыкин А.Е., Кириленко Я.А, Применение глубокого машинного обучения к синтезу цепочки вызовов C#. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 63-86 convert string to int 1 100 100 3 40 60 Int32.TryParse get current time 10 10 10 1 60 40 DateTime.Now get files in folder 3 40 50 1 80 90 DirectoryInfo.GetFiles FileInfo.Name List.Add generate md5 hash code 1 100 100 1 80 60 MD5.Create Encoding.GetBytes MD5.ComputeHash Byte[].Length StringBuilder.Append StringBuilder.ToString copy a file and save it to a destination path 1 100 100 2 40 40 File.Exists String.Equals File.Exists IO.File.Copy create socket and then send text 1 100 90 3 20 10 AddressFamily.InterNetwork SocketType.Stream ProtocolType.Tcp Socket.Connect Encoding.GetBytes Socket.Send SocketShutdown.Both Socket.Shutdown Socket.Close write text using socket - 0 0 1 100 100 ASCIIEncoding.GetBytes Socket.Send connect to database and execute statement 1 80 50 6 0 30 IDbConnection.Open IDbConnection.CreateCommand IDbCommand.CommandText IDbCommand.ExecuteScalar Convert.ToInt32 Exception.ToString Console.WriteLine IDbConnection.Close download from url and save image 3 20 20 1 60 50 String.IsNullOrEmpty WebClient.DownloadFile Average scores 3.4 59 61 2.0 56 52 Our model produces slightly less amount of relevant outputs (as shown by Precision@N scores), but ranks these outputs slightly higher (as shown by FRank). Good performance on the first 5 queries demonstrate that our model is capable or memorizing correct answers, and outputs to the second 5 queries show that it can manage long requests, that require performing several action, as well as semantically similar requests. However, both models are not very stable to slight semantic variations in the input. For example, query “create socket and then send text” is understood very well by q However, both models are not very stable to slight semantic variations in the input. For example, query “create socket and then send text” is understood very well by However, both models are not very stable to slight semantic variations in the input. For example, query “create socket and then send text” is understood very well by 79 Chebykin A.E., Kirilenko I.A. Applying Deep Learning to C# Call Sequence Synthesis. Trudy ISP RAN /Proc. ISP RAS, vol. 7.1 API usage pattern mining This group of projects is primarily concerned with extracting common usages of the library. The first algorithm to mine API patterns was MAPO [22]. It starts with clustering API sequences, then for every cluster finds API calls that are the most frequent there and passes those to an API usage recommender, that ranks API calls according to their similarity to the code context. UP-Miner [23] improves upon MAPO by using API call sequence n-grams as a clustering metric and an additional clustering step. A near parameter-free approach PAM [4] significantly outperforms both MAPO and UP-Miner, introducing a probability model constructed in the form of a joint probability distribution over API calls observed in code and the underlying unobserved API patterns, used by developer. Acharya et al. [24] extract API patterns as partial orders, and unfortunately do not compare results to those of previous approaches. The differences of these projects from our work are twofold. Firstly, these models do not allow user to specify their exact needs (MAPO and UP-Miner take API call as input, but an API call can be utilized in more than one scenario, therefore using it as input can be ambiguous; PAM and framework of Acharya et al. do not ask for input). This leads to the output containing many samples irrelevant to user, while not 80 Чебыкин А.Е., Кириленко Я.А, Применение глубокого машинного обучения к синтезу цепочки вызовов C#. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 63-86 Чебыкин А.Е., Кириленко Я.А, Применение глубокого машинного обучения к синтезу цепочки вызовов C#. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 63-86 guaranteeing to provide those he was wishing for. Secondly, to use such models one needs to know beforehand which API calls (in case of MAPO and UP-Miner) or libraries (in case of PAM and framework of Acharya et al.) he is interested in. Our approach allows for recommendation of APIs to use, as well as the specifics of the usage. 7.2 Generating source code from natural language Code generation based on natural language input is one of the holy grails of Computer Science. It could be seen as a more promising alternative to our problem: after all, rather than generate API call sequence and leave it to the software developer to write code utilizing it, it would be better to just generate code in the first place. However, current research in the area seems to be far from this dream. It mostly focuses on Domain Specific Languages [25], [26], which are simpler than general- purpose programming languages and have by definition limited usage domain. Recent developments in generating code in general-purpose languages include works by Ling et al. [27] and Yin et al. [28]. The first paper proposes a novel approach of Latent Predictor Networks that allows for better copying of relevant key words from input to output. The second paper introduces a special version of Encoder-Decoder model, where Decoder is tailored to generate syntax trees as opposed to sequences. The main difference between these works and ours lies in the datasets. Ling et al. and Yin et al. report results on two datasets: code of Hearthstone cards and annotated Django code (Ling et al. also report results on the dataset of code of Magic the Gathering cards, but this dataset is semantically very similar to the Hearthstone one). The target code for the Hearthstone dataset is rather homogenous and limited to small subset of the wide variety that is the Python language, thus resembling code in DSL more than code in general-purpose language. And while Django dataset covers various usage scenarious, it contains impractically sesquipedalian natural language descriptions of every line of code. For example description of the line “for i in range(0, len(result)):” is “for every i in range of integers from 0 to length of result, not included”. The generation of code from descriptions several times longer than itself seems impractical. Our dataset, on the other hand, contains wide variety of API usages, described by reasonably long sentences like “Serializes to JSON”, which resemble real queries written by programmers in order to look up interesting APIs. 8. Conclusion In this paper, we applied deep learning approach for recommendation of C# API calls, removing one of the threats to the validity of the paper that originally proposed this approach for Java. To achieve this goal, we collected massive dataset, introduced several data preprocessing steps, and finally employed transfer learning techniques. Extending DeepAPI approach turned out to be nontrivial even for a similar language. Nonetheless, its main idea of modelling API sequences with RNN Encoder-Decoder stands. Data preprocessing steps, suggested by us, are not dependent on C# and should therefore be applicable to any programming language, thus they should make extending the approach even to very different languages much easier. By releasing data, code and trained model we hope to allow for repeatability of the experiments and to inspire further research in the area. Acknowledgment The authors would like to thank JetBrains Research15 for providing a GPU-equipped server for fast machine learning models training, as well as for the Young Researcher stipend granted to our team. Additionally we would like to thank Kirsanov Alexander and other friendly developers from the JetBrains ReSharper team for their input in evaluating FRank and Precision@N metrics. 7.3 Deep neural machine translation and source code Deep API Learning paper [7] itself was published in 2016, is widely cited, but little work has followed from it. The authors went on to successfully apply the neural machine translation approach to code migration between Java and C# [29], which shows that the proposed architecture can model both languages of API sequences well. 81 Chebykin A.E., Kirilenko I.A. Applying Deep Learning to C# Call Sequence Synthesis. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 63-86 Chebykin A.E., Kirilenko I.A. Applying Deep Learning to C# Call Sequence Synthesis. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 63-86 lenko I.A. Applying Deep Learning to C# Call Sequence Synthesis. Trudy ISP RAN /Proc. ISP RAS, 8, pp. 63-86 Lin et al. [30] similarly to us apply the Encoder-Decoder approach to a different target language, specifically Bash. They succeed, but it should be noted that their research problem is a simpler one in terms of target language used, since only 17 commands were selected from Bash. Together with command flags, types of open-vocabulary constants and logical connectives (&&, \|\|, parentheses) total output dictionary size does not exceed 300. To contrast that, our work is concerned with the same API dictionary size as original paper, which is 10,000 and therefore requires vastly bigger dataset and more complex model. [4]. J. Fowkes and C. Sutton. Parameter-free probabilistic api mining across github. In Proceedings of the 2016 24th ACM SIGSOFT International Symposium on Foundations of Software Engineering, 2016, pp. 254–265. 82 15 research.jetbrains.org [3]. J. Stylos and B. A. Myers. Mica: A web-search tool for finding api components and examples. In Proc. of the IEEE Symposium on Visual Languages and Human-Centric Computing, 2006, pp. 195–202. Robillard. What makes apis hard to learn? Answers from developers. IEEE software 26, no. 6, 2009, pp, 27-34. References [1]. M. P. Robillard and R. Deline. A field study of api learning obstacles. Empirical Software Engineering, vol. 16, no. 6, 2011, pp. 703–732. [2]. M. P. Robillard. What makes apis hard to learn? Answers from developers. IEEE software, vol. 26, no. 6, 2009, pp, 27-34. [3]. J. Stylos and B. A. Myers. Mica: A web-search tool for finding api components and examples. In Proc. of the IEEE Symposium on Visual Languages and Human-Centric Computing, 2006, pp. 195–202. [4]. J. Fowkes and C. Sutton. Parameter-free probabilistic api mining across github. In Proceedings of the 2016 24th ACM SIGSOFT International Symposium on Foundations of Software Engineering, 2016, pp. 254–265. 82 Чебыкин А.Е., Кириленко Я.А, Применение глубокого машинного обучения к синтезу цепочки вызовов C#. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 63-86 [5]. S. Shoham, E. Yahav, S. J. Fink, and M. Pistoia. Static specification mining using automata-based abstractions, IEEE Transactions on Software Engineering, vol. 34, no. 5, 2008, pp. 651–666. [6]. M. Raghothaman, Y. Wei, and Y. Hamadi. Swim: Synthesizing what i mean-code search and idiomatic snippet synthesis, In Proc. of the IEEE/ACM 38th International Conference on Software Engineering (ICSE), 2016, pp. 357–367. [7]. X. Gu, H. Zhang, D. Zhang, and S. Kim. Deep api learning In Proceedings of the 2016 24th ACM SIGSOFT International Symposium on Foundations of Software Engineering, 2016, pp. 631–642. pp [8]. A. Chebykin, M. Kita, and I. Kirilenko. Deepapi#: Clr/c# call sequence synthesis from text query. In Proceedings of the Second Conference on Software Engineering and Information Management, vol. 1864. CEUR-WS.org, 2017, pp. 6–11. (in Russian) [Online]. Available: http://ceur-ws.org/Vol-1864/paper 5.pdf [9]. I. Sutskever, O. Vinyals, and Q. V. Le. Sequence to sequence learning with neural networks. In Advances in neural information processing systems, 2014, pp. 3104–3112. [10]. K. Cho, B. Van Merri¨ enboer, C. Gulcehre, D. Bahdanau, F. Bougares, H. Schwenk, and Y. Bengio. Learning phrase representations using rnn encoder-decoder for statistical machine translation, arXiv preprint arXiv:1406.1078, 2014. [11]. M. Schuster and K. K. Paliwal. Bidirectional recurrent neural networks. IEEE Transactions on Signal Processing, vol. 45, no. 11, 1997, pp. 2673–2681. [12]. D. Bahdanau, K. Cho, and Y. Bengio. Neural machine translation by jointly learning to align and translate. arXiv preprint arXiv:1409.0473, 2014. [13]. P. H. Calais Guerra, A. Veloso, W. Meira Jr, and V. Almeida. From bias to opinion: a transfer-learning approach to real-time sentiment analysis. References In Proceedings of the 10th Working Conference on Mining Software Repositories, 2013, pp. 319–328. [24]. M. Allamanis and C. Sutton. Mining idioms from source code. In Proceedings of the 22nd ACM SIGSOFT International Symposium on Foundations of Software Engineering, 2014, pp. 472–483. pp [25]. A. Desai, S. Gulwani, V. Hingorani, N. Jain, A. Karkare, M. Marron, S. Roy. Program synthesis using natural language. In Proceedings of the 38th International Conference on Software Engineering, 2016, pp. 345–356. [26]. S. Gulwani and M. Marron. Nlyze: Interactive programming by natural language for spreadsheet data analysis and manipulation. In Proceedings of the 2014 ACM SIGMOD international conference on Management of data, 2014, pp. 803–814. [27]. W. Ling, E. Grefenstette, K. M. Hermann, T. Koˇ cisk` y, A. Senior, F. Wang, and P. Blunsom. Latent predictor networks for code generation. arXiv preprint arXiv:1603.06744, 2016. [28]. P. Yin and G. Neubig. A syntactic neural model for general-purpose code generation. arXiv preprint arXiv:1704.01696, 2017. [29]. X. Gu, H. Zhang, D. Zhang, and S. Kim. Deepam: Migrate apis with multi-modal sequence to sequence learning. arXiv preprint arXiv:1704.07734, 2017. [30]. X. V. Lin, C. Wang, D. Pang, K. Vu, and M. D. Ernst. Program synthesis from natural language using recurrent neural networks. Technical Report UW-CSE-17-03-01, University of Washington, Department of Computer Science and Engineering, 2017. References In Proceedings of the 17th ACM SIGKDD international conference on Knowledge discovery and data mining, 2011, pp. 150–158. pp [14]. B. Zoph, D. Yuret, J. May, and K. Knight. Transfer learning for low-resource neural machine translation. arXiv preprint arXiv:1604.02201, 2016. [15]. G. Klein, Y. Kim, Y. Deng, J. Senellart, and A. M. Rush. Opennmt: Open-source toolkit for neural machine translation. arXiv preprint arXiv:1701.02810, 2017. [16]. F. A. Gers, J. Schmidhuber, and F. Cummins. Learning to forget: Continual prediction with lstm. Neural Computation, vol. 12, issue 10, 2000, pp. 2451-2471 [17]. A. Graves, A.-r. Mohamed, and G. Hinton. Speech recognition with deep recurrent neural networks. In Proceedings of the IEEE international conference on Acoustics, speech and signal processing, 2013, pp. 6645–6649. [18]. J. Kiefer and J. Wolfowitz. Stochastic estimation of the maximum of a regression function. The Annals of Mathematical Statistics, vol. 23, 1952, pp. 462–466. [19]. P. Koehn. Pharaoh: a beam search decoder for phrase-based statistical machine translation models. In Proceedings of the Conference of the Association for Machine Translation in the Americas, 2004, pp. 115–124. [20]. K. Papineni, S. Roukos, T. Ward, and W.-J. Zhu. Bleu: a method for automatic evaluation of machine translation. In Proceedings of the 40th annual meeting on association for computational linguistics, 2002, pp. 311–318. [21]. L. A. Granka, T. Joachims, and G. Gay. Eye-tracking analysis of user behavior in www search. In Proceedings of the 27th annual international ACM SIGIR conference on Research and development in information retrieval, 2004, pp. 478–479. [22]. T. Xie and J. Pei. Mapo: Mining api usages from open source repositories. In Proceedings of the 2006 international workshop on mining software repositories, 2006, pp. 54–57. 83 Chebykin A.E., Kirilenko I.A. Applying Deep Learning to C# Call Sequence Synthesis. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 63-86 Chebykin A.E., Kirilenko I.A. Applying Deep Learning to C# Call Sequence Synthesis. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 63-86 [23]. J. Wang, Y. Dang, H. Zhang, K. Chen, T. Xie, and D. Zhang. Mining succinct and high- coverage api usage patterns from source code. In Proceedings of the 10th Working Conference on Mining Software Repositories, 2013, pp. 319–328. [23]. J. Wang, Y. Dang, H. Zhang, K. Chen, T. Xie, and D. Zhang. Mining succinct and high- coverage api usage patterns from source code. DOI: 10.15514/ISPRAS-2018-30(3)-5 Для цитирования: Чебыкин А.Е., Кириленко Я.А. Применение глубокого машинного обучения к синтезу цепочки вызовов C#. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 63-86 (на английском языке). DOI: 10.15514/ISPRAS-2018-30(3)-5 Применение глубокого машинного обучения к синтезу цепочки вызовов C# А.Е. Чебыкин <a.e.chebykin@gmail.com> Я.А. Кириленко <jake.kirilenko@gmail.com> Математико-механический факультет, Санкт-Петербургский государственный университет Университетский пр., дом 28, Санкт-Петербург, 198504, Россия А.Е. Чебыкин <a.e.chebykin@gmail.com> Я.А. Кириленко <jake.kirilenko@gmail.com> Математико-механический факультет, Санкт-Петербургский государственный университет Университетский пр., дом 28, Санкт-Петербург, 198504, Россия А.Е. Чебыкин <a.e.chebykin@gmail.com> Я.А. Кириленко <jake.kirilenko@gmail.com> Математико-механический факультет, Санкт-Петербургский государственный университет Университетский пр., дом 28, Санкт-Петербург, 198504, Россия Аннотация. Большая часть стандартных для программирования задач — например, соединение с базой данных, отображение картинки, чтение файла — давно реализована в различных библиотеках и доступна через соответствующие Application Programming Interfaces (APIs). Однако чтобы воспользоваться ими, разработчик должен сначала узнать, что они существуют, а затем — как правильно с ними работать. В настоящее время Интернет кажется наилучшим и самым популярным источником подобной информации. Недавно был предложен другой подход, основанный на глубоком машинном обучении и реализованный в виде инструмента под названием DeepAPI. По описанию желаемой функциональности на английском языке он генерирует цепочку вызовов Java функций. В данной статье мы показываем, как подход может быть перенесен на другой язык программирования (C# вместо Java), на котором доступно меньше открытого кода; мы описываем техники, позволившие достичь результата, близкого к оригинальному, а также техники, которые не улучшили производительность. 84 Чебыкин А.Е., Кириленко Я.А, Применение глубокого машинного обучения к синтезу цепочки вызовов C#. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 63-86 Чебыкин А.Е., Кириленко Я.А, Применение глубокого машинного обучения к синтезу цепочки вызов руды ИСП РАН, том 30, вып. 3, 2018 г., стр. 63-86 Наконец, чтобы облегчить будущие исследования в области, мы публикуем наши набор данных, код и обученную модель. Ключевые слова: API; глубокое обучение; поиск кода; рекуррентная нейронная сеть; обучение с подкреплением. Список литературы Applying Deep Learning to C# Call Sequence Synthesis. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 63-86 [14]. B. Zoph, D. Yuret, J. May, and K. Knight. Transfer learning for low-resource neural machine translation. arXiv preprint arXiv:1604.02201, 2016. [15]. G. Klein, Y. Kim, Y. Deng, J. Senellart, and A. M. Rush. Opennmt: Open-source toolkit for neural machine translation. arXiv preprint arXiv:1701.02810, 2017. [16]. F. A. Gers, J. Schmidhuber, and F. Cummins. Learning to forget: Continual prediction with lstm. Neural Computation, vol. 12, issue 10, 2000, pp. 2451-2471 [17]. A. Graves, A.-r. Mohamed, and G. Hinton. Speech recognition with deep recurrent neural networks. In Proceedings of the IEEE international conference on Acoustics, speech and signal processing, 2013, pp. 6645–6649. g p g pp [18]. J. Kiefer and J. Wolfowitz. Stochastic estimation of the maximum of a regression function. The Annals of Mathematical Statistics, vol. 23, 1952, pp. 462–466. [19]. P. Koehn. Pharaoh: a beam search decoder for phrase-based statistical machine translation models. In Proceedings of the Conference of the Association for Machine Translation in the Americas, 2004, pp. 115–124. pp [20]. K. Papineni, S. Roukos, T. Ward, and W.-J. Zhu. Bleu: a method for automatic evaluation of machine translation. In Proceedings of the 40th annual meeting on association for computational linguistics, 2002, pp. 311–318. [21]. L. A. Granka, T. Joachims, and G. Gay. Eye-tracking analysis of user behavior in www search. In Proceedings of the 27th annual international ACM SIGIR conference on Research and development in information retrieval, 2004, pp. 478–479. [22]. T. Xie and J. Pei. Mapo: Mining api usages from open source repositories. In Proceedings of the 2006 international workshop on mining software repositories, 2006, pp. 54–57. [23]. J. Wang, Y. Dang, H. Zhang, K. Chen, T. Xie, and D. Zhang. Mining succinct and high- coverage api usage patterns from source code. In Proceedings of the 10th Working Conference on Mining Software Repositories, 2013, pp. 319–328. [24]. M. Allamanis and C. Sutton. Mining idioms from source code. In Proceedings of the 22nd ACM SIGSOFT International Symposium on Foundations of Software Engineering, 2014, pp. 472–483. [25]. A. Desai, S. Gulwani, V. Hingorani, N. Jain, A. Karkare, M. Marron, S. Roy. Program synthesis using natural language. In Proceedings of the 38th International Conference on Software Engineering, 2016, pp. 345–356. [26]. S. Gulwani and M. Marron. Nlyze: Interactive programming by natural language for spreadsheet data analysis and manipulation. Список литературы [1]. M. P. Robillard and R. Deline. A field study of api learning obstacles. Empirical Software Engineering, vol. 16, no. 6, 2011, pp. 703–732. [2]. M. P. Robillard. What makes apis hard to learn? Answers from developers. IEEE software, vol. 26, no. 6, 2009, pp, 27-34. [3]. J. Stylos and B. A. Myers. Mica: A web-search tool for finding api components and examples. In Proc. of the IEEE Symposium on Visual Languages and Human-Centric Computing, 2006, pp. 195–202. [4]. J. Fowkes and C. Sutton. Parameter-free probabilistic api mining across github. In Proceedings of the 2016 24th ACM SIGSOFT International Symposium on Foundations of Software Engineering, 2016, pp. 254–265. [5]. S. Shoham, E. Yahav, S. J. Fink, and M. Pistoia. Static specification mining using automata-based abstractions, IEEE Transactions on Software Engineering, vol. 34, no. 5, 2008, pp. 651–666. pp [6]. M. Raghothaman, Y. Wei, and Y. Hamadi. Swim: Synthesizing what i mean-code search and idiomatic snippet synthesis, In Proc. of the IEEE/ACM 38th International Conference on Software Engineering (ICSE), 2016, pp. 357–367. [7]. X. Gu, H. Zhang, D. Zhang, and S. Kim. Deep api learning In Proceedings of the 2016 24th ACM SIGSOFT International Symposium on Foundations of Software Engineering, 2016, pp. 631–642. [8]. A. Chebykin, M. Kita, and I. Kirilenko. Deepapi#: Clr/c# call sequence synthesis from text query. In Proceedings of the Second Conference on Software Engineering and Information Management, vol. 1864. CEUR-WS.org, 2017, pp. 6–11. (in Russian) [Online]. Режим доступа: http://ceur-ws.org/Vol-1864/paper 5.pdf [9]. I. Sutskever, O. Vinyals, and Q. V. Le. Sequence to sequence learning with neural networks. In Advances in neural information processing systems, 2014, pp. 3104–3112. [10]. K. Cho, B. Van Merri¨ enboer, C. Gulcehre, D. Bahdanau, F. Bougares, H. Schwenk, and Y. Bengio. Learning phrase representations using rnn encoder-decoder for statistical machine translation, arXiv preprint arXiv:1406.1078, 2014. [11]. M. Schuster and K. K. Paliwal. Bidirectional recurrent neural networks. IEEE Transactions on Signal Processing, vol. 45, no. 11, 1997, pp. 2673–2681. [12]. D. Bahdanau, K. Cho, and Y. Bengio. Neural machine translation by jointly learning to align and translate. arXiv preprint arXiv:1409.0473, 2014. [13]. P. H. Calais Guerra, A. Veloso, W. Meira Jr, and V. Almeida. From bias to opinion: a transfer-learning approach to real-time sentiment analysis. In Proceedings of the 17th ACM SIGKDD international conference on Knowledge discovery and data mining, 2011, pp. 150–158. 85 Chebykin A.E., Kirilenko I.A. Chebykin A.E., Kirilenko I.A. Applying Deep Learning to C# Call Sequence Synthesis. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 63-86 Список литературы In Proceedings of the 2014 ACM SIGMOD international conference on Management of data, 2014, pp. 803–814. [27]. W. Ling, E. Grefenstette, K. M. Hermann, T. Koˇ cisk` y, A. Senior, F. Wang, and P. Blunsom. Latent predictor networks for code generation. arXiv preprint arXiv:1603.06744, 2016. [28]. P. Yin and G. Neubig. A syntactic neural model for general-purpose code generation. arXiv preprint arXiv:1704.01696, 2017. [29]. X. Gu, H. Zhang, D. Zhang, and S. Kim. Deepam: Migrate apis with multi-modal sequence to sequence learning. arXiv preprint arXiv:1704.07734, 2017. [30]. X. V. Lin, C. Wang, D. Pang, K. Vu, and M. D. Ernst. Program synthesis from natural language using recurrent neural networks. Technical Report UW-CSE-17-03-01, University of Washington, Department of Computer Science and Engineering, 2017. 86 DOI: 10.15514/ISPRAS-2018-30(3)-6 For citation: Abakumov M.A., Dovgalyuk P.M. Stealth debugging of programs in Qemu emulator with WinDbg debugger. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 87-92. DOI: 10.15514/ISPRAS-2018-30(3)-6 For citation: Abakumov M.A., Dovgalyuk P.M. Stealth debugging of programs in Qemu emulator with WinDbg debugger. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 87-92. DOI: 10.15514/ISPRAS-2018-30(3)-6 Stealth debugging of programs in Qemu emulator with WinDbg debugger M.A. Abakumov <mikhail.abakumov@ispras.ru> P.M. Dovgalyuk <dovgaluk@ispras.ru> Yaroslav-the-Wise Novgorod State University, 41, Great St. Petersburg st., Velikiiy Novgorod, 173003, Russia M.A. Abakumov <mikhail.abakumov@ispras.ru> P.M. Dovgalyuk <dovgaluk@ispras.ru> Yaroslav-the-Wise Novgorod State University, 41, Great St. Petersburg st., Velikiiy Novgorod, 173003, Russia Abstract. When programs are analyzed for the presence of vulnerabilities and malicious code, there is a need for a quality isolation of the analysis tools. There are two reasons for this. At first, the program can influence the tool environment. This problem is solved by using the emulator. At second, the tool environment can influence behavior of the analyzed program. So, the programmer will think that the program is harmless, but in fact it is not. This problem is solved by the mechanism of stealth debugging. The WinDbg debugger has the possibility of connecting to a remote debug service (Kdsrv.exe) in the Windows kernel. Therefore, it is possible to connect to the guest system running in the QEMU emulator. Interaction between WinDbg client and server occurs through packets by protocol KDCOM. However, kernel debugging is possible only with the enabled debugging mode in boot settings. And it reveals the debugging process. We developed special module of WinDbg debugger for Qemu emulator. It is an alternative of the remote debugging service in the kernel. Thus, the debugger client tries to connect to the WinDbg server, but module intercepts all packets, generates all the necessary information from the Qemu emulator and sends response to the client. Module completely simulates the behavior of the server, so the client does not notice the spoofing and perfectly interacts with it. At the same time for debugging there is no need to enable debugging mode in the kernel. This leads to stealth debugging. Our module supports all features of WinDbg regarding remote debugging, besides interception of events and exceptions. Keywords: WinDbg; Qemu; Windows; remote debugging; stealth debugging 1. Introduction When performing a dynamic analysis of binary (executable) code, the problem arises of qualitatively isolating the code and the instrumentation on which this 87 Abakumov M.A., Dovgalyuk P.M. Stealth debugging of programs in Qemu emulator with WinDbg debugger. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 87-92 analysis is performed. The need for isolation is dual. On the one hand, it is necessary to limit the impact of the code being studied, since it is able to affect the state of the instrument machine, which is especially important in the study of malicious software. On the other hand, analysis tools can indirectly change the behavior of the program being studied. The most indicative are the situations when errors in working with dynamic memory and race conditions cease to appear in the debugging mode. analysis is performed. The need for isolation is dual. On the one hand, it is necessary to limit the impact of the code being studied, since it is able to affect the state of the instrument machine, which is especially important in the study of malicious software. On the other hand, analysis tools can indirectly change the behavior of the program being studied. The most indicative are the situations when errors in working with dynamic memory and race conditions cease to appear in the debugging mode. The search for undocumented features in a binary code encounters a similar problem. Various techniques and techniques are known [1], with the help of which malware reveals that its execution takes place in a controlled environment, and does not fulfill its objective functions. To find the debugger to be connected, check the int 3 handler and hardware debug registers, evaluate the behavior of certain API functions, and track the progress of the system time. It is possible to divide potential sources of information, which makes it possible to identify the fact of working in a controlled environment into three disjoint groups. The first is the interaction with external, uncontrolled components, the program being studied, such as remote servers. To the same category, it is necessary to include speed checks. Successful struggle with such sources allows the mechanism of deterministic reproduction [2]. If you write the progress of the system in advance, when debugging and analyzing it during playback, there will be no effect on the guest's state because all time characteristics are fixed during recording. 1. Introduction The second group of sources is the discrepancy in the behavior of the equipment [3]. The implementation of virtual equipment in software emulators is not always ideal. Known inaccuracies can be used to determine the emulator in which the program runs. The third group is the analysis tools present in the runtime. This kind of facility occurs even when the debugger is running in conjunction with a virtual machine. It is possible to divide potential sources of information, which makes it possible to identify the fact of working in a controlled environment into three disjoint groups. The first is the interaction with external, uncontrolled components, the program being studied, such as remote servers. To the same category, it is necessary to include speed checks. Successful struggle with such sources allows the mechanism of deterministic reproduction [2]. If you write the progress of the system in advance, h d b i d l i it d i l b k th ill b ff t th 3. WinDbg The WinDbg debugger is one of the most advanced debuggers for Windows operating systems. WinDbg is claimed by developers, because it extracts symbolic information from applications and libraries, displays the contents of internal Windows data structures, performs remote debugging of a physical or virtual machine. WinDbg can be used for debugging user applications, device drivers, the operating system itself in kernel mode, to analyze memory dumps in kernel mode created after the so-called Blue Screen of Death, which occurs when an error is issued. It can also be used for debugging custom mode crash dumps. WinDbg supports several debugging modes: debugging the local process, debugging the kernel, and remote debugging. Target applications can easily detects local debugging process. Remote debugging requires enabled debugging mode in kernel. In this mode kernel uses the debugging server (KdSrv.exe) for interacting with remotely client. But It is also reveals system control (Fig. 1). Fig. 1. Direct kernel debugging Fig. 1. Direct kernel debugging 2. Related work In the Qemu emulator at the moment there is only a module of the GDB debugger, which allows debugging the kernel of the system, but in itself it has relatively small functionality and does not have a GUI. You can use IDA Pro Disassembler [4] ore to connect to the emulator via the GDB interface, but this will not extend the range of the GDB's features, but will only increase the ease of use. In addition, there is a utility called Winbagility [5], which allows the debugger WinDbg to connect to the kernel without debugging mode of the operating system. It is utility for the VirtualBox emulator and is the intermediary between the debugger and the emulator. There is the FDP protocol between Winbagility and the emulator - the introspection API for VirtualBox. It is a minus in this implementation, since the number of provided functions limits the interface. 88 Абакумов М.А., Довгалюк П.М. Скрытая отладка программ отладчиком WinDbg в эмуляторе Qemu. Труды ИСП РАН, том 30, вып. 3, 2018 г,, стр. 87-92 Абакумов М.А., Довгалюк П.М. Скрытая отладка программ отладчиком WinDbg в эмуляторе Qemu. Труды ИСП РАН, том 30, вып. 3, 2018 г,, стр. 87-92 4. Stealth debugging We developed a mechanism for stealth debugging for the QEMU emulator, which allows WinDbg to be remotely connected. The mechanism is an analysis module built into the emulator, and turns out to be an external tool in relation to the guest system. The needs of the KdSrv service in the kernel of the debugging system is not required - the analysis module itself extracts the necessary data from the system and transfers it to the remote client debugger (Fig. 2). The programs running in the guest system cannot track the presence of the connected debugger through functions such as IsDebuggerPresent or through the state of the hardware registers. One way to remotely kernel debugging using the WinDbg debugger is to debug through the COM port. Interaction between the computers takes place via a private KDCOM protocol, the specification of which has been restored. One of the computers in this case is represented by a virtual machine. The second is an instrumental computer with Windows OS where this machine is started. Running 89 Abakumov M.A., Dovgalyuk P.M. Stealth debugging of programs in Qemu emulator with WinDbg debugger. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 87-92 Abakumov M.A., Dovgalyuk P.M. Stealth debugging of programs in Qemu emulator with WinDbg debug RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 87-92 WinDbg client connects to the emulator via a named pipe, through which the COM- port of the virtual machine is forwarded. WinDbg client connects to the emulator via a named pipe, through which the COM- port of the virtual machine is forwarded. The developed module for the emulator fully implements the KDCOM protocol, within the framework of the restored specification, so the debugger WinDbg interacts with it, as with the debugging module of the Windows kernel, without The developed module for the emulator fully implements the KDCOM protocol, within the framework of the restored specification, so the debugger WinDbg interacts with it, as with the debugging module of the Windows kernel, without noticing the substitution. It should be noted that the use of the QEMU emulator as a runtime opens the possibility of debugging during deterministic playback of the virtual machine. The recorded scenarios can be debugged many times in the emulator, which would not be possible if the Windows debug module running inside the guest system were used. Fig. 2. 4. Stealth debugging Kernel debugging through the special module Fig. 2. Kernel debugging through the special module 5. Results and contributons The developed module supports almost all features of WinDbg regarding remote debugging, besides interception of events and exceptions. It is open source project placed in: github.com/ispras/qemu/tree/windbg. The official community recognized the module as useful. In addition, patches have already been prepared for inclusion in the official repository. The work was supported by the Ministry of Education and Science of Russia, research project No. 2.6146.2017/8.9. 6. Acknowledment The work was supported by the Ministry of Education and Science of Russia, research project No. 2.6146.2017/8.9. 90 Абакумов М.А., Довгалюк П.М. Скрытая отладка программ отладчиком WinDbg в эмуляторе Qemu. Труды ИСП РАН, том 30, вып. 3, 2018 г,, стр. 87-92 References [1]. Timothy Vidas and Nicolas Christin. Evading android runtime analysis via sandbox detection. In Proceedings of the 9th ACM Symposium on Information, Computer and Communications Security, 2014, pp. 447- 458. [1]. Timothy Vidas and Nicolas Christin. Evading android runtime analysis via sandbox detection. In Proceedings of the 9th ACM Symposium on Information, Computer and Communications Security, 2014, pp. 447- 458. [2]. P. Dovgalyuk. Deterministic Replay of System's Execution with Multi-target QEMU Simulator for Dynamic Analysis and Reverse Debugging. In Proceedings of the 2012 16th European Conference on Software Maintenance and Reengineering, 2012, pp. 553- 556. [3]. Roberto Paleari, Lorenzo Martignoni, Giampaolo Fresi Roglia, and Danilo Bruschi. A fistful of red-pills: how to automatically generate procedures to detect CPU emulators. In Proceedings of the 3rd USENIX conference on Offensive technologies (WOOT'09). 2009. [4]. IDA ProDisassembler. Available at: https://www.hex- rays.com/products/ida/index.shtml, accessed 19.06.2018. [5]. Winbagility. Available at: https://winbagility.github.io/, accessed 19.06.2018. Скрытая отладка программ отладчиком WinDbg в эмуляторе Qemu М.А. Абакумов <mikhail.abakumov@ispras.ru> П.М. Довгалюк <dovgaluk@ispras.ru> DOI: 10.15514/ISPRAS-2018-30(3)-6 Для цитирования: Абакумов М.А., Довгалюк П.М. Скрытая отладка программ отладчиком WinDbg в эмуляторе Qemu. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 87-92 (на английском языке). DOI: 10.15514/ISPRAS-2018-30(3)-6 М.А. Абакумов <mikhail.abakumov@ispras.ru> П.М. Довгалюк <dovgaluk@ispras.ru> П.М. Довгалюк <dovgaluk@ispras.ru> Новгородский государственный университет имени Ярослава Мудрого, 173003, Россия, Великий Новгород, Большая Санкт-Петербургская, д. 41 Новгородский государственный университет имени Ярослава Мудрого, 173003, Россия, Великий Новгород, Большая Санкт-Петербургская, д. 41 Аннотация. При анализе программ на наличие уязвимостей и вредоносного кода бывают ситуации, в которых возникает необходимость качественной изоляции инструментов анализа. Этому есть две причины. Во-первых, анализируемая программа может влиять на инструментальную среду. Эта проблема решается использованием эмулятора. Во-вторых, инструменты анализа могут влиять на программу. Так, программист может подумать, что программа безопасна, хотя на самом деле это может быть не так. Эта проблема может быть решена механизмом скрытой отладки. Отладчик WinDbg имеет функцию подключения к удаленному отладочному серверу (Kdsrv.exe), запущенному в ядре Windows. Поэтому есть возможность подключиться к гостевой системе, запущенной в эмуляторе QEMU. Клиент взаимодействует с сервером через пакеты по протоколу KDCOM. Однако отлаживать ядро можно лишь с включенным режимом отладки в настройках запуска, что раскрывает процесс отладки. Мы разработали специальный модуль отладчика WinDbg для QEMU, который является альтернативой удаленному отладочному сервису в ядре. Модуль перехватывает пакеты при взаимодействии клиента отладчика WinDbg с сервером, самостоятельно генерирует всю необходимую отладочную информацию, используя возможности эмулятора Qemu, и отправляет ответ клиенту. Модуль полностью эмулирует поведение отладочного сервера, поэтому клиент на замечает подмены и успешно взаимодействует с ним. При этом отпадает необходимость в отладочном режиме ядра. Тем самым происходит скрытая отладка. При использовании модуля работоспособны все 91 Abakumov M.A., Dovgalyuk P.M. Stealth debugging of programs in Qemu emulator with WinDbg debugger. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 87-92 Abakumov M.A., Dovgalyuk P.M. Stealth debugging of programs in Qemu emulator with WinDbg debugger. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 87-92 возможности WinDbg, которые он представляет для удаленной отладки, кроме перехвата событий и исключений. Список литературы [1]. Timothy Vidas and Nicolas Christin. Evading android runtime analysis via sandbox detection. In Proceedings of the 9th ACM Symposium on Information, Computer and Communications Security, 2014, pp. 447- 458. [1]. Timothy Vidas and Nicolas Christin. Evading android runtime analysis via sandbox detection. In Proceedings of the 9th ACM Symposium on Information, Computer and Communications Security, 2014, pp. 447- 458. [2]. P. Dovgalyuk. Deterministic Replay of System's Execution with Multi-target QEMU Simulator for Dynamic Analysis and Reverse Debugging. In Proceedings of the 2012 16th European Conference on Software Maintenance and Reengineering, 2012, pp. 553- 556. [3]. Roberto Paleari, Lorenzo Martignoni, Giampaolo Fresi Roglia, and Danilo Bruschi. A fistful of red-pills: how to automatically generate procedures to detect CPU emulators. In Proceedings of the 3rd USENIX conference on Offensive technologies (WOOT'09). 2009. [4]. IDA Pro Disassembler. Режим доступа: https://www.hex- rays.com/products/ida/index.shtml, дата обращения 19.06.201 [4]. IDA Pro Disassembler. Режим доступа: https://www.hex- rays.com/products/ida/index.shtml, дата обращения 19.06.2018. [5]. Winbagility. Режим доступа: https://winbagility.github.io/, дата обращения 19.06.2018. [5]. Winbagility. Режим доступа: https://winbagility.github.io/, дата обращения 19.06.2018. 92 92 A.V. Ivanov <alexey.ivanov@ispras.ru> P.M. Dovgaluk <pavel.dovgaluk@ispras.ru> V.A. Makarov <vladimir.makarov@ispras.ru> Yaroslav-the-Wise Novgorod State University, 41, Great St. Petersburg st., Velikiiy Novgorod, 173003, Russia Abstract. Sometimes programmers face the task of analyzing the work of a compiled program. To do this, there are many different tools for debugging and tracing written programs. One of these tools is the analysis of the application through system calls. With a detailed study of the mechanism of system calls, you can find a lot of nuances that you have to deal with when developing a program analyzer using system calls. This paper discusses the implementation of a tracer that allows you to analyze programs based on system calls. In addition, the paper describes the problems that I had to face in its design and development. Now there are a lot of different operating systems and for each operating system must be developed its own approach to implementing the debugger. The same problem arises with the architecture of the processor, under which the operating system is running. For each architecture, the analyzer must change its behavior and adjust to it. As a solution to this problem, the paper proposes to describe the operating system model, which we analyze. The model description is a configuration file that can be changed depending on the needs of the operating systems. When a system call is detected the plugin collects the information downloaded from the configuration file. In a configuration file, arguments are expressions, so we need to implement a parser that needs to recognize input expressions and calculate their values. After calculating the values of all expressions, the tracer formalizes the collected data and outputs it to the log file. Keywords: QEMU; configurable system calls; debugging; plugin; system calls; tracing. DOI: 10.15514/ISPRAS-2018-30(3)-7 DOI: 10.15514/ISPRAS-2018-30(3)-7 For citation: Ivanov A.V., Dovgaluk P.M., Makarov V.A. Configurable system call tracer in QEMU emulator. Trudy ISP RAN/Proc. ISP RAS, Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 93-98. DOI: 10.15514/ISPRAS-2018-30(3)-7 For citation: Ivanov A.V., Dovgaluk P.M., Makarov V.A. Configurable system call tracer in QEMU emulator. Trudy ISP RAN/Proc. ISP RAS, Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 93-98. DOI: 10.15514/ISPRAS-2018-30(3)-7 Configurable system call tracer in QEMU emulator A.V. Ivanov <alexey.ivanov@ispras.ru> P.M. Dovgaluk <pavel.dovgaluk@ispras.ru> V.A. Makarov <vladimir.makarov@ispras.ru> Yaroslav-the-Wise Novgorod State University, 41, Great St. Petersburg st., Velikiiy Novgorod, 173003, Russia A.V. Ivanov <alexey.ivanov@ispras.ru> P.M. Dovgaluk <pavel.dovgaluk@ispras.ru> V.A. Makarov <vladimir.makarov@ispras.ru> Yaroslav-the-Wise Novgorod State University, 41, Great St. Petersburg st., Velikiiy Novgorod, 173003, Russia 1. Introduction Sometimes programmers face the task of analyzing the work of a compiled program to find its flaws, defects, and even search for malicious code in it. To analyze the work of such applications, we have to study their binary code or try to decompile the code, which is a laborious task. In order to simplify the analysis of applications, 93 Ivanov A.V., Dovgaluk P.M., Makarov V.A. Configurable system call tracer in QEMU emulator. Trudy ISP RAN/Proc. ISP RAS, Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 93-98 Ivanov A.V., Dovgaluk P.M., Makarov V.A. Configurable system call tracer in QEMU emulator. Trudy ISP RAN/Proc. ISP RAS, Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 93-98 we can use the system calls of this application. System calls provide an essential interface between a program and the operating system. It is possible to track which system calls the application makes, and draw conclusions about the behavior of the program. This method allows us to debug the application without delving into the level of instructions and architecture features, thereby reducing the time required to find the problem. Debugging applications using system tracing can be done inside the operating system, but still a number of problems arise:  strong dependence of the debugger on the operating system;  impossibility to run several debuggers at the same time;  inaccessibility to the privileged execution;  inaccessibility to the privileged execution;  necessity to secure the operating system when analyzing programs that have harmful effects. To solve these problems, we can use the virtual machine tools. In this way, we can debug applications in a wide range of different operating systems running under different processor architectures. 2. Approach and uniqueness The two mechanisms developed by AMD and Intel are independent of each other, but, in fact, perform the same functions. These are SYSCALL / SYSRET or SYSENTER / SYSEXIT statements. They allow us to transfer control to the operating system without interrupts. Each operating system supports values returned from the system call, which are passed as reference types when the system call handler is called. During the execution of the system call, the service procedure records the required values if necessary by the available links, after which the system call is exited. One of the main tasks that we had to face was the task of supporting the plugin of different operating systems and processor architectures. The solution to this problem was the interface with the configuration file. The configuration file makes the debugger more flexible and customizable. With its help, we can disconnect a certain mechanism of system calls from the trace or disable unnecessary system calls. In addition, such a mechanism makes it easier to add support for new operating systems and processor architectures. To implement the interface with the configuration file, it was necessary to study a wide range of different operating systems and processor architectures. After gathering the necessary information, we can determine the information necessary for debugging: what type of system call is supported by SYSCALL / SYSRET or SYSENTER / SYSEXIT and their opcodes; location of system call arguments; a list of system calls, with the name of each system call, its code, and the list of arguments. Thus, by developing an interface for debugger and configuration file interfacing, we can add support for operating systems without going into the debugger code. When implementing the debugger interaction interface with the configuration file, it became necessary to recognize the various expressions read from the file. For this task, we used the generator of the bison parser and developed the corresponding grammar [3]. 2. Approach and uniqueness To date, several debuggers allow us to trace an application using system calls. All these debuggers have a drawback - they do not provide enough portability of the debugger within different operating systems and processor architectures. We offer a new approach to implementing the debugger through system calls, by loading all the information necessary for tracing from the configuration file. The configuration files will allow us to easily configure and change the parameters needed for debugging, and to simplify the addition of support for new operating systems and architectures without recompiling the program and learning the debugger code. It was decided to implement the debugger under the virtual machine QEMU [1], using the plugin mechanism. QEMU is an open source virtual machine that emulates the hardware of various platforms. This virtual machine supports the emulation of a large number of processor architectures such as x86, PowerPC, ARM, MIPS, SPARC, m68k. In addition, this simulator supports the launch of a large number of different operating systems. Now, there is a plugin mechanism for QEMU implemented by ISP RAS [2], which allows us to connect developed plugins to a virtual machine during its both startup and operation. The implementation of the plugin mechanism enables each additional translation of the instruction to be substituted by an additional code for execution, when this instruction is called. This mechanism is suitable for debugging through system calls, so it was decided to use it. In addition, various mechanisms of the system call play an important role. The classical way of implementation is the use of interrupts. With the help of interrupts, control is transferred to the kernel of the operating system, with the application 94 Иванов А.В., Довгалюк П.М., Макаров В.А. Конфигурируемый трассировщик системных вызовов в эмуляторе QEMU. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 93-98 Иванов А.В., Довгалюк П.М., Макаров В.А. Конфигурируемый трассировщик системных вызовов в эмуляторе QEMU. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 93-98 having to enter the number of the system call and the necessary arguments into the corresponding registers of the processor. having to enter the number of the system call and the necessary arguments into the corresponding registers of the processor. For many RISC processors, this method is the only one; however, the CISC architecture has additional methods. 3. Background and related work Now, there are several debuggers to solve existing problems. Nitro [4] allows us to trace system calls, but it works only under Intel x86 architecture. Another debugger – Panda [5], can also trace system calls, supporting such operating systems as Linux, Windows 7, Windows XP and two architectures of the i386 processor and Now, there are several debuggers to solve existing problems. Nitro [4] allows us to trace system calls, but it works only under Intel x86 architecture. Another debugger – Panda [5], can also trace system calls, supporting such operating systems as Linux, Windows 7, Windows XP and two architectures of the i386 processor and ARM. The description of all system calls is found in the code of this debugger, because of which this approach makes it difficult to add support for new operating systems and processor architectures, and worsens the flexibility in configuring the plugin, since the system debugger settings mechanism is not provided. 95 Ivanov A.V., Dovgaluk P.M., Makarov V.A. Configurable system call tracer in QEMU emulator. Trudy ISP RAN/Proc. ISP RAS, Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 93-98 Ivanov A.V., Dovgaluk P.M., Makarov V.A. Configurable system call tracer in QEMU emulator. Trudy ISP RAN/Proc. ISP RAS, Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 93-98 4. Conclusion and discussion Based on the results of the work done, the plugin was developed in the QEMU virtual machine, with which we can trace and debug an application using system calls. As input to the plugin, the configuration file corresponding to the operating system running in the QEMU virtual machine and corresponding to the selected processor architecture is used. The structure of the configuration file consists of 4 parts. The first part provides information about the operating system, its name and bit capacity. The second part is responsible for the supported mechanisms of system calls. The next part contains the location of the system call arguments. The last part includes a list of all available system calls and service information about the arguments of the system calls. Because of the plugin’s work, a log file containing all the system calls that the plugin has intercepted is created. Each system call displays detailed information: the name and value of each system call argument, the number of the thread of execution from which this system call was made and the value that returned the system call after execution. Fig. 1 presents a small fragment of the output file that was created by the implementation of the plugin launched in the windows XP operating system and the i386 processor architecture and the i386 processor architecture. Fig. 1. Part of the output file of the plugin Fig. 1. Part of the output file of the plugin Upon the information gathered in the log file, we can analyze the operation of the debugged application running inside the virtual machine. The operating system load 96 Иванов А.В., Довгалюк П.М., Макаров В.А. Конфигурируемый трассировщик системных вызовов в эмуляторе QEMU. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 93-98 time when using the developed plugin is increased 20% slowdown compared to the time of the operating system loading without this plugin. time when using the developed plugin is increased 20% slowdown compared to the time of the operating system loading without this plugin. References [1]. F. Bellard. QEMU, a fast and portable dynamic translator. In Proceedings of the Annual Conference on USENIX Annual Technical Conference, 2005. [2]. Vasiliev I.A., Fursova N.I., Dovgaluk P.M., Klimushenkova M.A., Makarov V.A. Modules for instrumenting the executable code in QEMU. Problemy informacionnoj bezopasnosti. Komp'juternye sistemy [Journal of Information Security Problems. Computer Systems], no. 4, 2015, pp. 195-203 (in Russian). [3]. GNU Bison [HTML] (https://www.gnu.org/software/bison/) [4]. Nitro [HTML] (http://nitro.pfoh.net/index.html) [5]. Panda. Plugin: syscalls2. [HTML] (https://github.com/panda- re/panda/blob/master/panda/plugins/syscalls2/USAGE.md) Acknowledgments The work was supported by the Russian Foundation of Basic Research (research grant 18-07-00900 А) The work was supported by the Russian Foundation of Basic Research (research grant 18-07-00900 А) DOI: 10.15514/ISPRAS-2018-30(3)-7 Для цитирования: Иванов А.В., Довгалюк П.М., Макаров В.А. Конфигурируемый трассировщик системных вызовов в эмуляторе QEMU. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 93-98 (на английском языке). DOI: 10.15514/ISPRAS-2018-30(3)-7 Для цитирования: Иванов А.В., Довгалюк П.М., Макаров В.А. Конфигурируемый трассировщик системных вызовов в эмуляторе QEMU. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 93-98 (на английском языке). DOI: 10.15514/ISPRAS-2018-30(3)-7 Конфигурируемый трассировщик системных вызовов в эмуляторе QEMU А.В. Иванов <alexey.ivanov@ispras.ru> П.М. Довгалюк <pavel.dovgaluk@ispras.ru> В.А. Макаров <vladimir.makarov@ispras.ru> Новгородский государственный институт имени Ярослава Мудрого, 173003, Россия, г. Великий Новгород, ул. Б. Санкт-Петербургская, д. 41 А.В. Иванов <alexey.ivanov@ispras.ru> П.М. Довгалюк <pavel.dovgaluk@ispras.ru> В.А. Макаров <vladimir.makarov@ispras.ru> Новгородский государственный институт имени Ярослава Мудрого, 173003, Россия, г. Великий Новгород, ул. Б. Санкт-Петербургская, д. 41 В.А. Макаров <vladimir.makarov@ispras.ru> 97 Аннотация. Разработчики программ часто сталкиваются с проблемой анализа работы различных приложений. Для этого существует большое множество различных средств отладки, отслеживания, трассировки написанных программ. Одним из таких средств является анализ работы приложения через системные вызовы. При детальном изучении механизма системных вызовов, можно обнаружить большое количество нюансов, с которыми приходится столкнуться при разработке анализатора программ с использованием системных вызовов. В статье рассматривается реализация трассировщика, который позволяет анализировать программы на основе системных вызовов, и проблемы, с которыми пришлось столкнуться при его проектировании и разработке. На данный момент существует большое количество различных операционных систем и для каждой операционной системы должен быть разработан свой подход в реализации отладчика. Такая же проблема возникает и с архитектурой процессора, под которой запущена операционная система. Для каждой архитектуры, анализатор должен менять своё поведение и подстраиваться под неё. В качестве решения данной проблемы, в статье предлагается описать модель операционной системы, которую мы анализируем. Описание модели представляет собой конфигурационный файл, который может быть изменён в зависимости от потребностей 97 Ivanov A.V., Dovgaluk P.M., Makarov V.A. Configurable system call tracer in QEMU emulator. Trudy ISP RAN/Proc. ISP RAS, Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 93-98 операционных систем. При обнаружении системного вызова, в его обработчик передаются аргументы и вся сопутствующая информация, загруженная из конфигурационного файла. Изначально, в конфигурационном файле, все аргументы представляют собой выражения, поэтому возникает необходимость также реализовать синтаксический анализатор, которому необходимо распознать входные выражения и посчитать их значения. После просчёта значений всех выражений, трассировщик формализует собранные данные и выводит их в лог файл. Ключевые слова: QEMU; конфигурируемые системные вызовы; настраиваемые системные вызовы; отладка; отладчик; плагин; системные вызовы; трассировка; трассировщик. Список литературы [1]. F. Bellard. Qemu, a fast and portable dynamic translator. In Proceedings of the Annual Conference on USENIX Annual Technical Conference, 2005. [1]. F. Bellard. Qemu, a fast and portable dynamic translator. In Proceedings of the Annual Conference on USENIX Annual Technical Conference, 2005. [1]. F. Bellard. Qemu, a fast and portable dynamic translator. In Proceedings of the Annual Conference on USENIX Annual Technical Conference, 2005. [2]. Васильев И.А., Фурсова Н.И., Довгалюк П.М., Климушенкова М.А., Макаров В.А. Модули для инструментирования исполняемого кода в симуляторе QEMU. Проблемы информационной безопасности. Компьютерные системы, no, 4, 2015г., стр. 195-203 [2]. Васильев И.А., Фурсова Н.И., Довгалюк П.М., Климушенкова М.А., Макаров В.А. Модули для инструментирования исполняемого кода в симуляторе QEMU. Проблемы информационной безопасности. Компьютерные системы, no, 4, 2015г., стр. 195-203 р [3]. GNU Bison [HTML] (https://www.gnu.org/software/bison/) [4]. Nitro. [HTML] (http://nitro.pfoh.net/index.html) [4]. Nitro. [HTML] (http://nitro.pfoh.net/index.html) [5]. Panda. Plugin: syscalls2. [HTML] (https://github.com/panda- re/panda/blob/master/panda/plugins/syscalls2/USAGE.md) 98 98 Анализ методов оценки надежности оборудования и систем. Практика применения методов 1,2Е.М. Лаврищева <lavr@ispras.ru> 1,2,3 Н.В. Пакулин<npak@ispras.ru> 1А.Г. Рыжов <ryzhov@ispras.ru> 1,3С.В. Зеленов <zelenov@ispras.ru> 1Институт системного программирования им. В.П. Иванникова РАН, 109004, Россия, г. Москва, ул. А. Солженицына, д. 25 2 Московский физико-технический институт, 141701, Россия, Московская обл., г. Долгопрудный, Институтский пер., д. 9 3Национальный исследовательский университет «Высшая школа экономики» 101000, Россия, г. Москва, ул. Мясницкая, д. 20 Анализ методов оценки надежности оборудования и систем. Практика применения методов 1,2Е.М. Лаврищева <lavr@ispras.ru> 1,2,3 Н.В. Пакулин<npak@ispras.ru> 1А.Г. Рыжов <ryzhov@ispras.ru> 1,3С.В. Зеленов <zelenov@ispras.ru> 1Институт системного программирования им. В.П. Иванникова РАН, 109004, Россия, г. Москва, ул. А. Солженицына, д. 25 2 Московский физико-технический институт, 141701, Россия, Московская обл., г. Долгопрудный, Институтский пер., д. 9 3Национальный исследовательский университет «Высшая школа экономики» 101000, Россия, г. Москва, ул. Мясницкая, д. 20 1,2Е.М. Лаврищева <lavr@ispras.ru> 1,2,3 Н.В. Пакулин<npak@ispras.ru> 1А.Г. Рыжов <ryzhov@ispras.ru> 1,3С.В. Зеленов <zelenov@ispras.ru> 1Институт системного программирования им. В.П. Иванникова РАН, 109004, Россия, г. Москва, ул. А. Солженицына, д. 25 2 Московский физико-технический институт, 141701, Россия, Московская обл., г. Долгопрудный, Институтский пер., д. 9 3Национальный исследовательский университет «Высшая школа экономики» 101000, Россия, г. Москва, ул. Мясницкая, д. 20 Аннотация. Проводится анализ моделей и методов оценки надежности технических и программных средств. Определяются основные понятия методов надежности и безопасности таких систем и ситуаций, приводящих к ошибкам, дефектам и отказам. Дано определение надежности и безопасности технических систем и программного обеспечения (ПО) систем. Приведена классификация моделей надежности: прогнозирующего, измерительного и оценочного типов. Описаны оценочные модели, которые применяются на практике. Определен стандарт жизненного цикла ПО (ISO 15288:2002), ориентированный на разработку и контроль компонентов систем на ошибки, начиная с требований к системе. Представлены результаты применения моделей надежности (Мусы, Гоэла-Окомото и др.) к малым, средним и большим проектам и дана сравнительная их оценка. Описан технологический модуль (ТМ) оценки надежности сложных комплексов программ ВПК (1989). Показана модель качества стандарта ISO 9126 (1-4):2002-2004 с показателями функциональность, надежность, эффективность и др., которые используются при определении зрелости и сертификата качества продукта. Ключевые слова: надежность; ошибка; дефект; отказ; плотность дефектов; случайный процесс; безопасность; гарантоспособность; восстанавливаемость; отказоустойчивость; завершенность; оценка надежности; сертификат качества DOI: 10.15514/ISPRAS-2018-30(3)-8 DOI: 10.15514/ISPRAS-2018-30(3)-8 DOI: 10.15514/ISPRAS-2018-30(3)-8 Для цитирования: Лаврищева Е.М., Пакулин Н.В., Рыжов А.Г., Зеленов С.В. Анализ методов оценки надежности оборудования и систем. Практика применения методов. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 99-120. DOI: 10.15514/ISPRAS-2018- 30(3)-8 99 Lavrischeva E.M., Pakulin N.V., Ryzhov A.G., Zelenov S.V. Analysis of methods for assessing the reliability of equipment and systems. Practice of methods. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 99-120 Lavrischeva E.M., Pakulin N.V., Ryzhov A.G., Zelenov S.V. Analysis of methods for assessing the reliability of equipment and systems. Practice of methods. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 99-120 1. Введение Рассматривается теория надежности и безопасности технических и программных систем [1-10]. Теория надежности техники возникла в рамках теории массового обслуживания систем и повлияла на становление и развитие теории надежности компьютерных систем и программного обеспечения (ПО) систем. Теоретики, изучая природу ошибок в системах, разработали более 100 математических моделей надежности, основанных на учете различных ситуаций, возникающих в технических и программных системах на первых и последующих поколениях ЭВМ. Методы надежности обеспечивают повышение надежности систем путем исправления разного рода обнаруженных ошибок в процессе разработки и их эксплуатации. Надежность систем сформировалась как самостоятельная теоретическая и прикладная наука, способствующая определению качественных показателей систем (функциональность, точность, отказоустойчивость, гарантоспособность, завершенность и др.). Методы оценки надежности систем позволяют прогнозировать, измерять и оценивать качество продукта с учетом возникающих ошибок, количества и интенсивности отказов, а также процессов разработки отдельных компонентов систем в жизненном цикле (ЖЦ). В работе рассматриваются все аспекты обеспечения надежности, безопасности и качества технических и программных систем [1-11]. 2. Методы надежности оборудования и систем Методы оценки надежности технических систем (аппаратуры, устройства, оборудование и др.) были разработаны значительно ранее компьютерных систем и основывались на вероятностных Марковских процессах с множеством числа состояний по теории массового обслуживания [12-16]. Эти методы обеспечивали проверку надежности функционирования техники, приборов и устройств в различных областях (машиностроение, энергетика, космос, медицина и др.). На их работоспособность влияли неисправности и разные недоработки в конструкции, приводящие к разрушительным последствиям и к ущербу системы в целом. На оценку надежности компьютерных систем и ПО существенным образом влияют следующие особенности. На оценку надежности компьютерных систем и ПО существенным образом влияют следующие особенности. 1. Большое количество кода в ПО, зачастую превышающее емкость физических элементов ЭВМ, и способов взаимодействия отдельных элементов ПО между собой; 2. Нематериальный характер элементов ПО, которые не деградируют, но стареют во времени, и в их процессах, программах и конструкциях могут случаться разного рода непредвиденные ситуации; 100 Лаврищева Е.М., Пакулин Н.В., Рыжов А.Г. , Зеленов С.В. Анализ методов оценки надежности оборудования и систем. Практика применения методов. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 99-120 Лаврищева Е.М., Пакулин Н.В., Рыжов А.Г. , Зеленов С.В. Анализ методов оценки надежности оборудования и систем. Практика применения методов. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 99-120 Лаврищева Е.М., Пакулин Н.В., Рыжов А.Г. , Зеленов С.В. Анализ методов оценки надежности оборудования и систем. Практика применения методов. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 99-120 3. Возникновение ошибочных ситуаций, разных дефектов и отказов как в задании формальной спецификации отдельных элементов, так и в их выходном коде; 3. Возникновение ошибочных ситуаций, разных дефектов и отказов как в задании формальной спецификации отдельных элементов, так и в их выходном коде; 4. Элементы ПО трудно поддаются визуализации, обнаружению и коррекции найденных ошибок, поэтому измерение надежности ПО требует анализа и проверки данных об ошибках в большей степени, чем для аппаратуры; 5. Системы ПО могут изменяться при функционировании и выходить из рабочего состояния от разных ситуаций внешней среды (вирусов, атак и др.), которые не предусмотрены в соответствующих моделях надёжности и обеспечиваются методами безопасности информации и систем. Надежность технических систем зависит от двух факторов: Надежность технических систем зависит от двух факторов:  качества отдельных технических конструктивных элементов системы;  отсутствия неисправностей в конструктивных элементах и их способность работать надежно и качественно.  качества отдельных технических конструктивных элементов системы;  отсутствия неисправностей в конструктивных элементах и их способность работать надежно и качественно. 2. Методы надежности оборудования и систем  отсутствия неисправностей в конструктивных элементах и их способность работать надежно и качественно. Надежность программных систем зависит от этих же факторов и от случайных изменений данных и маршрутов исполнения программ, которые могут привести к неверным результатам или отказам. Надежность программных систем зависит от этих же факторов и от случайных изменений данных и маршрутов исполнения программ, которые могут привести к неверным результатам или отказам. Между надежностью аппаратуры и ПО систем имеется сходство, состоящее в возникновении случайных явлений в процессах и системах, которые должны анализироваться методами теории вероятности, надежности и безопасности. 2.1. Определение термина надежности и безопасности систем Под надежностью систем понимается способность системы сохранять свои свойства (безотказность, восстанавливаемость, защищенность и др.) на заданном уровне в течение фиксированного промежутка времени при определенных условиях эксплуатации. Термин надежность (reliability) обозначает способность системы обладать свойствами, обеспечивающими выполнение функций системы в соответствии с заданными требованиями. Термин гарантоспособность (dependability) означает способность системы гарантировать выполнение услуг, для которых она предназначена, и состоит в: Термин гарантоспособность (dependability) означает способность системы гарантировать выполнение услуг, для которых она предназначена, и состоит в:  безотказности выполнения;  готовности к работе;  достоверности результатов;  приспособленности к обслуживанию или ремонту (maintainability);  информационной безопасности (security);  конфиденциальности (confidentiality), секретности и целостности информации (integrity); приспособленности к обслуживанию или ремонту (maintainability);  информационной безопасности (security);  конфиденциальности (confidentiality), секретности и целостности информации (integrity); конфиденциальности (confidentiality), секретности и целостност ф (i t it ) 101 Lavrischeva E.M., Pakulin N.V., Ryzhov A.G., Zelenov S.V. Analysis of methods for assessing the reliability of equipment and systems. Practice of methods. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 99-120 Lavrischeva E.M., Pakulin N.V., Ryzhov A.G., Zelenov S.V. Analysis of methods for assessing the reliability of equipment and systems. Practice of methods. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 99-120  безопасности (safety) работы системы без катастрофических последствий;  безопасности (safety) работы системы без катастрофических последствий;  эксплуатационной завершенности ПО и способности к восстановлению работоспособности системы.  эксплуатационной завершенности ПО и способности к восстановлению работоспособности системы.  эксплуатационной Отказоустойчивость (fault-tolerance) обозначает способность системы автоматически за ограниченное время прогнозировать, предупреждать и восстанавливать функциональность системы после отказов с помощью механизмов поддержки всех составляющих гарантоспособности. Системы или процессы, которые обладают таким комплексным свойством, называют гарантоспособными. Им присущи традиционные надежностные свойства (безотказность, готовность, безопасность, целостность, конфиденциальность, восстанавливаемость и др.). Вопросы разработки и использования гарантии качества систем обсуждаются более 25 лет на международных форумах и конференциях (Conference on Dependable Systems and Networks (DSN), European Dependable Computing Conference, (EDCC), International Conference on Computer Safety, Reliability and Security (SAFECOM), Probabilistic Safety Assessment and Management Conference (PSAM), Dependable Systems, Services and Technologies (DESSERT), Conference on Dependability of Complex Systems (DepCoS) и др.). С 2004 году ассоциация IEEE издает журнал Dependable and Security Computing. В нем обсуждаются бизнес-критические приложения, электронная коммерция, банковские технологии и др. [16, 17]. С точки зрения гарантоспособности надежность является целевой функцией реализации системы. К ней предъявляются высокие требования (недопустимость ошибок, отказов, дефектов и других аварийных ситуаций). Лаврищева Е.М., Пакулин Н.В., Рыжов А.Г. , Зеленов С.В. Анализ методов оценки надежности оборудования и систем. Практика применения методов. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 99-120 Лаврищева Е.М., Пакулин Н.В., Рыжов А.Г. , Зеленов С.В. Анализ методов оценки надежности оборудования и систем. Практика применения методов. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 99-120 2.1. Определение термина надежности и безопасности систем Надежность систем зависит от числа оставшихся и не устраненных ошибок в отдельных программах и компонентах системы. Чем интенсивнее проводится эксплуатация системы, тем интенсивнее выявляются ошибки, быстрее растет надежность системы и соответственно ее качество. Надежность, по существу, является функцией от ошибок, оставшихся в системе после ввода ее в эксплуатацию. Системы без ошибок считаются абсолютно надежными. Для оценки надежности систем используются собранные статистические данные – время безотказной работы, дефекты и частота (интенсивность) отказов. Исследование надежности систем проводится с помощью методов теории вероятностей, математической статистики, теории массового обслуживания и математических методов надежности и безопасности. Главным источником информации для оценки надежности являются процессы тестирования, эксплуатации и испытания системы и данные, полученные при разработке систем в соответствии со стандартами жизненного цикла (ЖЦ) (ISO/IEC 15846-1998, 15939:2002) системной инженерии [15-23]. 102 2.2. Базовые понятия моделей надежности и безопасно Они сравниваются с требованиями заказчика на систему и используются при сертификации продукта. Для оценки надежности и функциональной безопасности используются стандарты ISO/IEC 12207 для ПО и ISO 15288 -2006 систем. На работоспособность системы влияют отказы, дефекты и ошибки проектирования, которые приводят к длительности восстановления и к необходимости устранять в программе ошибки средствами программной и информационной избыточности. Согласно стандарта ISO 9126 (1-4) определяются характеристики надежности с учетом обнаруженных дефектов и ошибок при функционировании гарантоспособного ПО систем. Степень работоспособности/гарантоспособности зависит от соответствия характеристик требований, заданных в проекте, выявленным ошибкам и отказам в ПО и возможным неисправностям в конструктивных элементах систем. 2.2. Базовые понятия моделей надежности и безопасно Analysis of methods for assessing the reliability of equipment and systems. Practice of methods. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 99-120 equipment and systems. Practice of methods. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 99 120 В характеристиках безопасности учитываются только те отказы, которые могут привести к катастрофическим последствиям и ущербам (например, пожар, взрыв, разрушение здания и др.). Оценка безопасности системы базируется на надежности функционирования ПО и БД. Оценка надежности зависит от метрик стандарта качества (внешние, внутренние, эксплуатационные). Они сравниваются с требованиями заказчика на систему и используются при сертификации продукта. Для оценки надежности и функциональной безопасности используются стандарты ISO/IEC 12207 для ПО и ISO 15288 -2006 систем. На работоспособность системы влияют отказы, дефекты и ошибки проектирования, которые приводят к длительности восстановления и к необходимости устранять в программе ошибки средствами программной и информационной избыточности. Согласно стандарта ISO 9126 (1-4) определяются характеристики надежности с учетом обнаруженных дефектов и ошибок при функционировании гарантоспособного ПО систем. В характеристиках безопасности учитываются только те отказы, которые могут привести к катастрофическим последствиям и ущербам (например, пожар, взрыв, разрушение здания и др.). Оценка безопасности системы базируется на надежности функционирования ПО и БД. Оценка надежности зависит от метрик стандарта качества (внешние, внутренние, эксплуатационные). Они сравниваются с требованиями заказчика на систему и используются при сертификации продукта. Для оценки надежности и функциональной безопасности используются стандарты ISO/IEC 12207 для ПО и ISO 15288 -2006 систем. На работоспособность системы влияют отказы, дефекты и ошибки проектирования, которые приводят к длительности восстановления и к необходимости устранять в программе ошибки средствами программной и информационной избыточности. Согласно стандарта ISO 9126 (1-4) определяются характеристики надежности с учетом обнаруженных дефектов и ошибок при функционировании гарантоспособного ПО систем. Степень работоспособности/гарантоспособности зависит от соответствия В характеристиках безопасности учитываются только те отказы, которые могут привести к катастрофическим последствиям и ущербам (например, пожар, взрыв, разрушение здания и др.). Оценка безопасности системы базируется на надежности функционирования ПО и БД. Оценка надежности зависит от метрик стандарта качества (внешние, внутренние, эксплуатационные). Они сравниваются с требованиями заказчика на систему и используются при сертификации продукта. Для оценки надежности и функциональной безопасности используются стандарты ISO/IEC 12207 для ПО и ISO 15288 -2006 систем. В характеристиках безопасности учитываются только те отказы, которые могут привести к катастрофическим последствиям и ущербам (например, пожар, взрыв, разрушение здания и др.). Оценка безопасности системы базируется на надежности функционирования ПО и БД. Оценка надежности зависит от метрик стандарта качества (внешние, внутренние, эксплуатационные). 2.2. Базовые понятия моделей надежности и безопасно К базовым понятиям, которые используются в моделях надежности систем, относятся следующие [1-5]. К базовым понятиям, которые используются в моделях надежности систем, относятся следующие [1-5].  Отказ (failure) – это переход системы из рабочего состояния в нерабочее.  Отказ (failure) – это переход системы из рабочего состояния в нерабочее.  Дефект (fault) – это последствие выполнения элемента программы, приводящее к некоторому непредвиденному событию (неверной интерпретации компьютером); невыявленные дефекты – источник потенциальных ошибок и отказов системы.  Ошибка (error) может быть следствием недостатка в спецификациях любой из программ или при принятии неверных действий в процессе испытания системы.  Ошибка (error) может быть следствием недостатка в спецификациях любой из программ или при принятии неверных действий в процессе испытания системы.  Интенсивность отказов – это частота появления отказов или дефектов в системе при ее тестировании, эксплуатации и сопровождении системы. Одним из подходов к исследованию надежности на основе отказов систем является классическая теория вероятностей, согласно которой отказы в системе считаются случайными и зависят от ошибок, внесенных при разработке системы. Все модели оценки надежности базируются на статистике отказов и интенсивности выявленных отказов в процессе верификации, тестирования и испытания системы для обеспечения ее работоспособности и гарантоспособности [16, 17]. Модели надежности основываются на нахождении случайной величины в системе, числом и интенсивностью возникновения отказов в системе. Для случайной величины вычисляется математическое ожидание и дисперсия (среднее отклонение). Если случайная величина дискретна, т.е. принимает конечное число значений х1, х2, …, xn , то ее распределение описывается вероятностью Р( = хi), и в общем случае F(х) = P(  хi) является функцией распределения случайной величины. Случайный процесс с непрерывным временем, который описывается однородными событиями, называется пуассоновским процессом. Если случайные величины, полученные на всем ЖЦ системы, распределены по показательному, эрланговскому или гиперэрланговскому законам, то поведение системы описывается Марковским процессом без непрерывных компонентов. Надежность по существу очень близка задачам безопасности. При разработке систем научными и некоммерческими институтами их трудно оценить на надежность и безопасность из-за того, что они делаются, как правило, не по стандартам. В то время как системы управления авиацией, атомной энергетикой и оборонной промышленностью разрабатываются по стандартам. В них надежность и безопасность определяют работоспособность системы в соответствии с требованиями и с минимум отказов и дефектов. 103 Lavrischeva E.M., Pakulin N.V., Ryzhov A.G., Zelenov S.V. Analysis of methods for assessing the reliability of equipment and systems. Practice of methods. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 99-120 Lavrischeva E.M., Pakulin N.V., Ryzhov A.G., Zelenov S.V. 3. Классификация моделей надежности ПО Большинство моделей надежности исходят из предположения, что найденные ошибки и дефекты устраняются немедленно или определяются временем их устранения и новые дефекты не вносятся. В результате количество дефектов в системе уменьшается, а надежность возрастает, такие модели получили название моделей роста надежности. Shick G. [6, 16, 17] предложил следующую классификацию моделей надежности. Прогнозирующие модели надежности основаны на измерении технических характеристик создаваемой программы: длина, сложность, число циклов и степень их вложенности, количество ошибок на страницу операторов программы и др. Модель Мотли–Брукса основывается на длине и сложности структуры программы (количество ветвей и циклов, вложенность циклов), количестве и типах переменных, а также интерфейсов. Модель Холстеда дает прогнозирование количества ошибок в программе в зависимости от ее объема и таких данных, как число операций (n1) и операндов (n2), а также общее число обращений к ним (N1 + N2). Измерительные модели предназначены для измерения надежности ПО, Измерительные модели предназначены для измерения надежности ПО, работающего с заданной внешней средой и следующими ограничениями: Измерительные модели предназначены для измерения надежности ПО, работающего с заданной внешней средой и следующими ограничениями:  ПО не модифицируется во время периода измерений свойств надежности;  ПО не модифицируется во время периода измерений свойств надежности;  обнаруженные ошибки не исправляются; 104 Лаврищева Е.М., Пакулин Н.В., Рыжов А.Г. , Зеленов С.В. Анализ методов оценки надежности оборудования и систем. Практика применения методов. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 99-120 Лаврищева Е.М., Пакулин Н.В., Рыжов А.Г. , Зеленов С.В. Анализ методов оценки надежности оборудования и систем. Практика применения методов. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 99-120  измерение надежности проводится для зафиксированной конфигурации ПО.  измерение надежности проводится для зафиксированной конфигурации ПО.  измерение надежности проводится для зафиксированной конфигурации ПО. Примером таких моделей является модель Нельсона, Рамамурти–Бастани и др.[3]. Примером таких моделей является модель Нельсона, Рамамурти–Бастани и др.[3]. Примером таких моделей является модель Нельсона, Рамамурти–Бастани и др.[3]. Модель Нельсона основывается на выполнении k прогонов программы при тестировании и позволяет определить надежность по формуле: Модель Нельсона основывается на выполнении k прогонов программы при тестировании и позволяет определить надежность по формуле: Модель Нельсона основывается на выполнении k прогонов программы при тестировании и позволяет определить надежность по формуле: где tj – время выполнения j-го прогона, (tj) = –ln(1–qj)/tj и при qi  1 интерпретируется как функция интенсивности отказов. где tj – время выполнения j-го прогона, (tj) = –ln(1–qj)/tj и при qi  1 интерпретируется как функция интенсивности отказов. 3. Классификация моделей надежности ПО Оценочные модели основываются на серии тестовых прогонов и проводятся на этапах тестирования системы. В тестовой среде определяется вероятность отказа программы при ее тестировании или выполнении. Эти типы моделей могут применяться на этапах ЖЦ и могут быть следующих видов [7-11]. Модели без подсчета ошибок основаны на измерении интервала времени между отказами и позволяют спрогнозировать количество ошибок, оставшихся в программе. К этим моделям относятся модели Джелински и Моранды, Шика Вулвертона, и Литвуда–Вералла. Модели с подсчетом отказов базируются на количестве ошибок, обнаруженных на заданных интервалах времени. К этому классу моделей относятся модели Шика–Вулвертона, Шумана, Пуассоновская модель и др. Модели с подсевом ошибок основаны на количестве устраненных ошибок и подсеве, внесенном рограмму искусственных ошибок, тип и количество которых заранее известны. При внесении изменений в программу проводится повторное тестирование и оценка надежности. Этот подход базируется на тестировании и редко используется из-за дополнительного объема работ для покрытия тестами компонентов системы. Модели с выбором области входных значений основываются на генерации множества тестовых выборок из входного набора. К этому типу моделей относится модель Нельсона и др. На процессах выявления отказов и их интенсивности используются также: Модели с выбором области входных значений основываются на генерации множества тестовых выборок из входного набора. К этому типу моделей относится модель Нельсона и др. На процессах выявления отказов и их интенсивности используются также: 1) модели, рассматривающие интенсивность отказов, как Марковский и пуассоновский процесс; 1) модели, рассматривающие интенсивность отказов, как Марковский и пуассоновский процесс; 2) пуассоновский процесс; 2) модели роста надежности. Четкой границы между этими моделями провести нельзя, однако по фактору интенсивности отказов и моделей поведения их можно разделить на экспоненциальные, логарифмические, геометрические, байесовские и др. Для практической оценки надежности более всего представляет интерес оценочная модель Мусы, Мусы-Окомото и др. Рассмотрим их. Оценочная модель Мусы [8] основана на следующих положениях:  тесты адекватно представляют среду функционирования;  происходящие отказы учитываются (оценивается их количество) 105 Lavrischeva E.M., Pakulin N.V., Ryzhov A.G., Zelenov S.V. Analysis of methods for assessing the reliability of equipment and systems. Practice of methods. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 99-120 Lavrischeva E.M., Pakulin N.V., Ryzhov A.G., Zelenov S.V. Analysis of methods for assessing the reliability of equipment and systems. Practice of methods. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 99-120  интервалы между отказами независимы;  интервалы между отказами независимы;  время между отказами распределено по экспоненциальному закону;  интенсивность отказов пропорциональна числу ошибок;  скорость исправления ошибок (относительно времени функционирования)  интервалы между отказами независимы;  время между отказами распределено по экспоненциальному закон  время между отказами распределено по экспоненциальному зако  интенсивность отказов пропорциональна числу ошибок;  время между отказами распределено по экспоненциальному закону;  интенсивность отказов пропорциональна числу ошибок;  скорость исправления ошибок (относительно времени функционирования) р рц у ;  скорость исправления ошибок (относительно времени функционирования)  пропорциональна интенсивности их появления. Эта модель учитывает интервалы между отказами, которые распределяется по экспоненциальному закону, а интенсивность отказов пропорциональна числу обнаруженных ошибок. Исходя из этой модели, можно установить зависимость: Исходя из этой модели, можно установить зависимость: 1) среднего числа отказов от времени функционирования τ (рис.1), которое задается в виде: m=M0 [1 – ехр ( )], 0 0T M c где М0 – общее число ошибок; Т0 – начальная наработка на отказ; с – коэффициент времени испытаний; τ – время функционирования. 2) средней наработки на отказ T от времени функционирования τ (рис.2): T = T0 ехр ( ), где М0, Т0, с – зависят от наработки на отказ. 2) средней наработки на отказ T от времени функционирования τ (рис. T = T0 ехр ( ), где М0, Т0, с – зависят от наработки на отказ. c T = T0 ехр ( T = T0 ехр ( ), где М0, Т0, с – зависят от наработки на отказ. c T0 ехр ( ), где М0, Т0, с зависят от наработки на отказ. Рис. 1. Зависимость числа отказов от времени τ Fig. 1. The dependence of the number of failures on time of τ Рис. 2. Зависимость средней наработки на отказ от времени τ Fig. 2. The dependence mean time to failure rate from the time τ фик этой зависимости представлен областью 1 (рис. 3), для которой М1 = ... – номера наблюдений, а τ1, τ2 … τM1 – время между отказами. Область 2 3) соответствует достижению средней наработки Тр на отказ за время Δτ. 0 0T M c р р 0 0T M c Рис. 1. Зависимость числа отказов от времени τ Fig. 1. The dependence of the number of failures on time of τ Рис. 2. Зависимость средней наработки на отказ от времени τ Fig. 2. The dependence mean time to failure rate from the time τ фик этой зависимости представлен областью 1 (рис. 3), для которой М1 = ... – номера наблюдений, а τ1, τ2 … τM1 – время между отказами. Область 2 .3) соответствует достижению средней наработки Тр на отказ за время Δτ. 0 0T M c Рис. 1. Зависимость числа отказов от времени τ Fig. 1. The dependence of the number of failures on time of τ Рис. 2. Зависимость средней наработки на отказ от времени τ Fig. 2. The dependence mean time to failure rate from the time τ Рис. 2. Зависимость средней наработки на отказ от времени τ Fig. 2. The dependence mean time to failure rate from the time τ График этой зависимости представлен областью 1 (рис. 3), для которой М1 = 1, 2, ... – номера наблюдений, а τ1, τ2 … τM1 – время между отказами. Область 2 (рис.3) соответствует достижению средней наработки Тр на отказ за время Δτ. График этой зависимости представлен областью 1 (рис. 3), для которой М1 = 1, 2, ... – номера наблюдений, а τ1, τ2 … τM1 – время между отказами. Область 2 (рис.3) соответствует достижению средней наработки Тр на отказ за время Δτ. 106 Лаврищева Е.М., Пакулин Н.В., Рыжов А.Г. , Зеленов С.В. Анализ методов оценки надежности оборудования и систем. Практика применения методов. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 99-120 Рис. 3. Области отказов и наработки на отказ от времени τ Pic. 3. The field of failures, and mean time to failure rate from τ Т ТF Т0 … … … 1 2  Рис. 3. Области отказов и наработки на отказ от времени τ Pic. 3. The field of failures, and mean time to failure rate from τ По собранным данным об ошибках оцениваются параметры T0 и M0, с помощью которых определяются дополнительное число ошибок по формуле: Δm = M0T0 [ Т 1 - 0 1 Т ]. 2. Модель Мусы–Окумото (логарифмическая) допускает, что некоторые ошибки имеют большую вероятность проявления в виде отказов, снижают интенсивность отказов с каждой устраненной ошибкой и дают экспоненциальное распределение. Процесс оценки надежности включает:  протоколирование отказов в ходе функционирования системы, измерение надежности по отказам и использование результатов измерений для определения потерь надежности в период времени эксплуатации;  анализ частоты и серьезности отказов при определении порядка устранения соответствующих ошибок;  анализ частоты и серьезности отказов при определении порядка устранения соответствующих ошибок;  оценка влияния времени функционирования системы на надежность с помощью инструментов измерения надежности.  оценка влияния времени функционирования системы на надежность с помощью инструментов измерения надежности. Рис. 1. Зависимость числа отказов от времени τ Fig. 1. The dependence of the number of failures on time of τ Рис. 2. Зависимость средней наработки на отказ от времени τ Fig. 2. The dependence mean time to failure rate from the time τ фик этой зависимости представлен областью 1 (рис. 3), для которой М1 = ... – номера наблюдений, а τ1, τ2 … τM1 – время между отказами. Область 2 .3) соответствует достижению средней наработки Тр на отказ за время Δτ. 0 0T M c 99-120 – m/N –1 + exp (–bT) = 0 – m/N –1 + exp (–bT) = 0 4. S-образная модель. Функция интенсивности (t) выявления ошибок в зависимости от времени работы имеет вид: 4. S-образная модель. Функция интенсивности (t) выявления ошибок в зависимости от времени работы имеет вид:  (t) = a 2 t exp (–t), где a – общее количество дефектов, обнаруженных от начала и до конца тестирования; где a – общее количество дефектов, обнаруженных от начала и до конца тестирования;  – скорость изменения функции интенсивности выявления отказов.  – скорость изменения функции интенсивности выявления отказов. Введение в формулу параметра в степени 1 модели Мусы и Гоэла–Окомото дает изменение формы кривой так, что она сначала растет, а потом спадает. Практика применения этих моделей в автоматизированных системах привела к уточнению функции интенсивности при введении дополнительного параметра n:  (t) = a n+1 tn exp (–t),  (t) = a n+1 tn exp (–t), где n отражает сложность и размер проекта некоторой системы. Это позволяет более точно определить форму кривой интенсивности с учетом получаемых практических результатов. где n отражает сложность и размер проекта некоторой системы. Это позволяет более точно определить форму кривой интенсивности с учетом получаемых практических результатов. Процесс оценки надежности включает: Рис. 1. Зависимость числа отказов от времени τ Fig. 1. The dependence of the number of failures on time of τ Рис. 2. Зависимость средней наработки на отказ от времени τ Fig. 2. The dependence mean time to failure rate from the time τ фик этой зависимости представлен областью 1 (рис. 3), для которой М1 = ... – номера наблюдений, а τ1, τ2 … τM1 – время между отказами. Область 2 .3) соответствует достижению средней наработки Тр на отказ за время Δτ. 0 0T M c Мат.ожидание найденных ошибок m(t) имеет вид: m (t) = (1/) ln (0t + 1), где 0 – исходная интенсивность отказов,  – скорость спада интенсивности отказов с каждым устраненным дефектом, а функция интенсивности отказов (t) имеет вид: m (t) = (1/) ln (0t + 1), где 0 – исходная интенсивность отказов,  – скорость спада интенсивности отказов с каждым устраненным дефектом, а функция интенсивности отказов (t) имеет вид: m (t) = (1/) ln (0t + 1), где 0 – исходная интенсивность отказов,  – скорость спада интенсивности отказов с каждым устраненным дефектом, а функция интенсивности отказов (t) имеет вид: (t) = 0 / (0  t + 1). 3. Модель Гоело–Окумото (экспоненциального роста) описывает обнаружение ошибок с помощью неоднородного пуассоновского процесса. В ней интенсивность отказов зависит от времени, а количество выявленных ошибок при тестировании трактуется как случайная величина. Исходные данные m, Xi и Т аналогичны данным предыдущих моделей. Функция среднего числа отказов, обнаруженных к моменту t, имеет вид: m (t ) = N (1 – e –bt ), где b – интенсивность обнаружения отказов; q(t )=b - показатель роста надежности. Функция интенсивности  (t) зависит от времени работы системы до отказа:  (t) = Nbе –bt , t  0, где N и b решаются с помощью уравнения: 3. Модель Гоело–Окумото (экспоненциального роста) описывает обнаружение ошибок с помощью неоднородного пуассоновского процесса. В ней интенсивность отказов зависит от времени, а количество выявленных ошибок при тестировании трактуется как случайная величина. Исходные данные m, Xi и Т аналогичны данным предыдущих моделей. Функция среднего числа отказов, обнаруженных к моменту t, имеет вид: m (t ) = N (1 – e –bt ), где b – интенсивность обнаружения отказов; q(t )=b - показатель роста надежности. где b – интенсивность обнаружения отказов; q(t )=b - показатель роста надежности. Функция интенсивности  (t) зависит от времени работы системы до отказа:  (t) = Nbе –bt , t  0, где N и b решаются с помощью уравнения: 107 Lavrischeva E.M., Pakulin N.V., Ryzhov A.G., Zelenov S.V. Analysis of methods for assessing the reliability of equipment and systems. Practice of methods. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 99-120 Lavrischeva E.M., Pakulin N.V., Ryzhov A.G., Zelenov S.V. Analysis of methods for assessing the reliability of equipment and systems. Practice of methods. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 3.1. Оценка надежности систем реального времени Некоторые типы систем реального времени с обеспечением безопасности требуют высокой надежности (недопустимость ошибок, точность, достоверность и др.), которая в значительной степени зависит от количества оставшихся и не устраненных ошибок в процессе ее разработки на этапах ЖЦ. В ходе эксплуатации системы также могут обнаруживаться и устраняться ошибки. Если при их исправлении не вносятся новые ошибки или их меньше, чем устраняется, то в ходе эксплуатации надежность системы непрерывно растет. Чем интенсивнее проводится эксплуатация, тем интенсивнее выявляются ошибки и быстрее растет надежность. На надежность ПО влияют, с одной стороны, угрозы, приводящие к неблагоприятным последствиям, риск нарушения безопасности системы, с другой стороны, способность совокупности компонентов системы сохранять устойчивость в процессе ее эксплуатации. Риск уменьшает свойства 108 Лаврищева Е.М., Пакулин Н.В., Рыжов А.Г. , Зеленов С.В. Анализ методов оценки надежности оборудования и систем. Практика применения методов. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 99-120 Лаврищева Е.М., Пакулин Н.В., Рыжов А.Г. , Зеленов С.В. Анализ методов оценки надежности оборудования и систем. Практика применения методов. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 99-120 надежности, особенно если обнаруженные ошибки могут быть результатом проявления угрозы извне [16]. надежности, особенно если обнаруженные ошибки могут быть результатом проявления угрозы извне [16]. Методы и модели надежности постоянно развиваются и уточняются, поскольку надежность является одной из ключевых проблем измерения качества современных распределенных по Интернет систем. Появилось новое направление – инженерия надежности ПО (Software reliability engineering – SRE), которое ориентировано на количественное изучение операционного поведения компонентов системы по отношению к пользователю, ожидающему получить надежную работу системы [16-20]. Надежность обеспечивается путем: Появилось новое направление – инженерия надежности ПО (Software reliability engineering – SRE), которое ориентировано на количественное изучение операционного поведения компонентов системы по отношению к пользователю, ожидающему получить надежную работу системы [16-20]. Надежность обеспечивается путем: 1) измерения надежности, т.е. проведения количественной ее оценки методами предсказаний, собранными данными о поведении системы в процессе тестирования и эксплуатации системы; 2) оценки стратегии и применения метрик для готовых компонентов, созданных в процессе разработки компонентов системы в заданной среде и стандартов на измерение надежности системы; 3) применения современных методов инспектирования, верификации, валидации и тестирования в процессе разработки отдельных компонентов и системы в целом. 3.2. Обеспечение надежности на этапах ЖЦ Для получения высокой надежности системы требуется наблюдать за достижением показателей надежности и качества на всех этапах ЖЦ согласно рекомендациям стандарта ISO/IEC 12207: ЖЦ [16]. К основным процессам стандарта ЖЦ относятся:  спецификация требований,  проектирование,  реализация,  тестирование,  испытание,  сопровождение.  сопровождение. На этапе спецификации требований определяются задачи и внешние спецификации основных (целевых) требований к системе с заданием метрик для оценки надежности, в терминах интенсивности отказов или вероятности безотказного его функционирования. Разработчики системы формируют: приоритеты функций системы по критерию важности их реализаци приоритеты функций системы по критерию важности их реализации;  параметры среды и интенсивности использования функций и их отказов;  входные и выходные данные для каждой функции системы;  категорий отказов и их интенсивности при выполнении функций в единицу календарного времени. На этапе проектирования определяются: На этапе проектирования определяются: 109 Lavrischeva E.M., Pakulin N.V., Ryzhov A.G., Zelenov S.V. Analysis of methods for assessing the reliability of equipment and systems. Practice of methods. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 99-120 Lavrischeva E.M., Pakulin N.V., Ryzhov A.G., Zelenov S.V. Analysis of methods for assessing the reliability of equipment and systems. Practice of methods. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 99-120 Lavrischeva E.M., Pakulin N.V., Ryzhov A.G., Zelenov S.V. Analysis of methods for assessing the reliability of equipment and systems. Practice of methods. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 99-120  размеры информационной и алгоритмической сложности всех типов проектируемых компонентов;  размеры информационной и алгоритмической сложности всех типов проектируемых компонентов;  размеры информационной и алгоритмической сложности всех типов проектируемых компонентов;  категории дефектов, свойственные всем типам компонентов си  категории дефектов, свойственные всем типам компонентов системы;  стратегии функционального тестирования компонентов по принципу «черного ящика» с помощью тестов для выявления ошибок в данных. Для обеспечения надежности продукта проводится:  категории дефектов, свойственные всем типам компонентов системы;  стратегии функционального тестирования компонентов по принципу «черного ящика» с помощью тестов для выявления ошибок в данных. Для обеспечения надежности продукта проводится:  стратегии функционального тестирования компонентов по принципу «черного ящика» с помощью тестов для выявления ошибок в данных. Д б ля обеспечения надежности продукта проводится: Для обеспечения надежности продукта проводится:  анализ вариантов архитектуры системы на соответствие требованиям к надежности;  анализ рисков, режимов отказов, деревьев ошибок для критических компонентов с целью обеспечения отказоустойчивости и восстанавливаемости системы;  анализ рисков, режимов отказов, деревьев ошибок для критических компонентов с целью обеспечения отказоустойчивости и восстанавливаемости системы;  прогнозирование показателей размера системы, чувствительности к ошибкам, степени тестируемости, оценки риска и сложности системы.  прогнозирование показателей размера системы, чувствительности к ошибкам, степени тестируемости, оценки риска и сложности системы.  сопровождение. На этапе реализации и тестирования системы проектные спецификации переводятся в коды и подготавливаются наборы тестов для автономного и комплексного их тестирования. При проведении автономного тестирования обеспечение надежности состоит в предупреждении появления дефектов в компонентах и создание эффективных методов защиты от них. Все последующие этапы разработки не могут обеспечить надежность систем, а лишь способствуют повышению уровня надежности за счет обнаружения ошибок с помощью тестов различных категорий. На этапе испытаний проводится системное тестирование для соответствия внешних спецификаций функций целям проекта. Испытание проводится или в реальной среде функционирования, или на испытательном стенде, который имитирует работу аппаратуры. При подготовке к испытаниям изучается "история" тестирования на процесса ЖЦ в целях использования ранее разработанных тестов, а также составления специальных тестов испытаний. На этом этапе осуществляется:  управление ростом надежности путем неоднократного исправления и регрессионного тестирования ПО;  управление ростом надежности путем неоднократного исправления и регрессионного тестирования ПО;  принятие решения о степени готовности ПО и возможности его передачи в эксплуатацию;  принятие решения о степени готовности ПО и возможности его передачи в эксплуатацию;  оценка надежности по результатам системного тестирования и испытаний по соответствующим моделям надежности, подходящих для заданных целей. На этапе сопровождения оценка надежности ПО проводится: На этапе сопровождения оценка надежности ПО проводится:  протоколирование отказов в ходе работы системы, измерения надежности функционирования и использования результатов измерений для определения потерь надежности в период времени эксплуатации;  анализ частоты и серьезности отказов для определения порядка их устранения; 110 Лаврищева Е.М., Пакулин Н.В., Рыжов А.Г. , Зеленов С.В. Анализ методов оценки надежности оборудования и систем. Практика применения методов. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 99-120 Лаврищева Е.М., Пакулин Н.В., Рыжов А.Г. , Зеленов С.В. Анализ методов оценки надежности оборудования и систем. Практика применения методов. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 99-120  оценка влияния функционирования системы на надежность в условиях усовершенствования технологии и применения новых инструментов разработки.  оценка влияния функционирования системы на надежность в условиях усовершенствования технологии и применения новых инструментов разработки.  оценка влияния функционирования системы на надежность в условиях усовершенствования технологии и применения новых инструментов разработки. 3.3. Применение моделей для оценки надежности ПО Практика применения моделей показывает, что среди названных моделей наиболее перспективными являются модели оценочного типа, которые базируются на пуассоновских процессах (модели Мусы, Гоэла–Окомото, S- образные и др.). По этим моделям надежность стремиться к 1. Одним из недостатков является форма кривой интенсивности выявленных отказов (экспоненциальная), которая строго спускается при t>0 вблизи t=0. Это говорит о том, что при тестировании проведено недостаточно экспериментов или мало найдено ошибок, когда интенсивность отказов была близка 0. В системе остаются ошибки и их поиск требует больше времени. В таблице 1 представлены практические значения функций интенсивности отказов  (t) и количество отказов µ(t) для базовых и общих моделей. В них значения a и  находятся в следующих соотношениях: N= a,  = a, b = , 1 = , a0 = a. N= a,  = a, b = , 1 = , a0 = a. N= a,  = a, b = , 1 = , a0 = a. Табл. 1. Характеристика моделей надежности Пуассоновского типа Table 1. Characteristics of Poisson type reliability models Название модели Функции интенсивности отказов  (t) Функции кумулятивного количества отказов µ(t) Mодель Гоэла-Окумото  (t) = Nb exp (–bt) µ(t)= N (1- exp (–bt)) Модель Мусы  (t) = 0 1 exp (–1t) µ(t)= 0 (1- exp (–1t)) S –подобная модель  (t) = a2 t exp (–t)  (t) = a{1-(1+t) exp ( –t) Модель Шнайдевинда  (t) = a0 exp (–t) µ(t)= a0/ (1- exp (–t)) Общая модель пуассо- новского процесса  (t) = a n+1 tn exp (–t) µ(t)= a (n1–  nn-1 /(n-1)1 tn exp(-t))) Для метода максимального правдоподобия задаются данные a, , n, решим й Табл. 1. Характеристика моделей надежности Пуассоновского типа Table 1. Characteristics of Poisson type reliability models Для метода максимального правдоподобия задаются данные a, , n, решим систему уравнений: Для метода максимального правдоподобия задаются данные a, , n, решим систему уравнений: Для метода максимального правдоподобия задаются данные a, , n, решим систему уравнений: 0 1 1 ! 1 exp( ) ( )! exp( ) 1 ! 3.3. Применение моделей для оценки надежности ПО 1 exp( ) n i n i n m m i n n m m m k n i n i n k m m n t t n i m t t n m t n t t                                    0 1 1 0 ! 1 exp( ) ( )! exp( ) 1 ! 1 exp( ) ( )! n i n i n m m i n n m m m k n i n i n k m m i m n t t n i m t t n m t n t t n i                                       1 1 0 exp( ) 1 ! 1 exp( ) ( )! n n m m m k n i n i n k m m i m t t n m t n t t n i                         111 111 Lavrischeva E.M., Pakulin N.V., Ryzhov A.G., Zelenov S.V. Analysis of methods for assessing the reliability of equipment and systems. Practice of methods. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 99-120 Lavrischeva E.M., Pakulin N.V., Ryzhov A.G., Zelenov S.V. Analysis of methods for assessing the reliability of equipment and systems. Practice of methods. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 3.3. Применение моделей для оценки надежности ПО 99-120 где параметр n зависит от процесса тестирования и его рекомендуемых значений: n=0 – для небольшого проекта, в котором разработчик является также тестером (модель Мусы, Гоэло-Окомото и др.); n=0 – для небольшого проекта, в котором разработчик является также тестером (модель Мусы, Гоэло-Окомото и др.); n=1 – для среднего проекта, в котором тестирование и проектирование ПО исполняются несколькими разработчиками из одной рабочей группы (S- образная модель); n=1 – для среднего проекта, в котором тестирование и проектирование ПО исполняются несколькими разработчиками из одной рабочей группы (S- образная модель); n=2 – для большого проекта, в котором группы тестирования и проектирования работают параллельно; n=3 – для очень большого проекта, в котором группы тестирования и разработки работают независимо друг от друга. На основе экспериментальных данных получены функции о количестве отказов µ(t) и интенсивности отказов  (t) на выходных данных и значениях параметра n (рис. 4). Этот рисунок показывает вид функций µ(t) при разных значений n=0,1, 2, 3. Наибольшее приближение достигается при n=3, а наименьшее при n=0 (модель Мусы, Гоэло-Окомото и др.). Это подтверждается соответствующими статистическими данными (табл.2), которые задают разницу между выходными данными (t_2) и соответствующими значениями функции µ(t) при значениях n = 0, 1, 2, 3. Табл. 2. Статистические данные функции µ(t) при n=3, 2, 1 и данных t_2 Table 2. Statistical data of the function μ(t) for n = 3, 2, 1 and data t_2 Табл. 2. Статистические данные функции µ(t) при n=3, 2, 1 и данных t_2 Table 2. Statistical data of the function μ(t) for n = 3, 2, 1 and data t_2 Статистические показатели Разница функций t_2 - µ_3 Разница функций t_2 - µ_2 Разница функций t_2 - µ_1 Разница функций t_2 - µ Среднее отклонение 16.13522 16.22889 19.88387 58.93807 Медианное отклонение 15.27700 14.11600 16.0000 60.89700 Максимум отклонение 33.58100 54.23600 49.10800 88.80200 Минимум отклонение 4. 848000 -1.280000 4.175000 15.96200 Среднеквадратическое отклонение 8.374089 17.37143 14.07056 23.63765 л. 2. Статистические данные функции µ(t) при n=3, 2, 1 и данных t_2 e 2. Statistical data of the function μ(t) for n = 3, 2, 1 and data t_2 На основе экспериментальных данных a, , n, (табл. 2) приведены значения функций µ(t) и  (t) при n = 3, 2, 1, полученные при использовании методов оценки надежности Мусы, Мусы-Окомото и Шнайдевинда. Функции µ(t) для этих методов приведены на графике (рис. 4). Им соответствуют кривые экспоненциального типа. Графики этих функций близки друг другу из-за близких значений, полученных по заданным моделям. 3.3. Применение моделей для оценки надежности ПО 112 Лаврищева Е.М., Пакулин Н.В., Рыжов А.Г. , Зеленов С.В. Анализ методов оценки надежности оборудования и систем. Практика применения методов. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 99-120 Рис. 4. Графики функций µ(t) по моделям Мусы, Окомото и Шнайдевинда Fig. 4. The graphs of function µ(t) on models of Musa, Okumoto and Shneidewind Рис. 4. Графики функций µ(t) по моделям Мусы, Окомото и Шнайдевинда Fig. 4. The graphs of function µ(t) on models of Musa, Okumoto and Shneidewind Для более эффективного применения приведеннях моделей надежности требуется значительное количество статистических данных о количестве и распределении отказов. А это требует увеличения количества экспериментов на процесах тестирования, системного тестирования для покрытия тестами всех компонентов и маршрутов прохождения путей в ПО. . Технологический модуль (ТМ) оценки надежности систем ТМ разработан в рамках работ по проекту ПРОТВА ВПК (1986-1989). В состав ТМ надежности входит четыре программных модуля (ПТМ) [15, 16, 22 c.283-296]: распределение надежности, прогнозирование плотности дефектов, прогнозирование надежности и оценки надежности. I. ПТМ «Распределения надежности» реализует метод распределения надежности по компонентам системы путем парного их сравнения и построения квадратной матрицы A размером n n из элементов вида: 1 ... 22 11     nn a a a , ji ij a a 1  , и, j = 1,…,n; i j; n=k, l, m, 1 ... 22 11     nn a a a , ji ij a a 1  , и, j = 1,…,n; i j; n=k, l, m, где n – количество сравниваемых компонентов, k, l, m – количество функций и модулей соответственно. Матрица включает относительный вес i-го компонента и вычисляется по формулам: 113 Lavrischeva E.M., Pakulin N.V., Ryzhov A.G., Zelenov S.V. Analysis of methods for assessing the reliability of equipment and systems. Practice of methods. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 99-120 Lavrischeva E.M., Pakulin N.V., Ryzhov A.G., Zelenov S.V. Analysis of methods for assessing the reliability of equipment and systems. Practice of methods. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 99-120 Lavrischeva E.M., Pakulin N.V., Ryzhov A.G., Zelenov S.V. Analysis of methods for assessing the reliability of equipment and systems. Practice of methods. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 99-120       n i n j ij n j ij i a a w 1 1 1 , 1 1    n i i w       n i n j ij n j ij i a a w 1 1 1 , 1 1    n i i w В случае больших размеров матрицы в целях получения более точных относительных оценок компонентов иерархии, вычисляются, так называемые, собственный вектор и собственные значения матрицы согласно известным уравнениям [24]. В них используются следующие данные: max – максимальное собственное значение матрицы А n–порядка, wi – коэффициент относительного веса элементов матрицы А, W = (w1, w2, … wn) – собственный вектор, которому соответствует max. Общность решения задачи сравнения устанавливается соотношением    n i i w 1  и значением 1 1    n i i w . . Технологический модуль (ТМ) оценки надежности систем Если матрица A имеет n-1 собственных значений , равных нулю и max = n, то она является согласованной. Определение индекса согласованности CI и коэффициента согласованности CR проводится по формулам: 1 max    n n CI  , ) (CI E CI CR  , где – математическое ожидание для матрицы парных сравнений A (n n). Критерий приемлемости парного сравнения элементов в матрицах размером n 3 получен такой: 05 . 0  CR и CR < 0.1 для n>5. По результатам сравнения формируется квадратная матрица F(k  k). Аналогично проводится сравнение приложений ПС. В результате сравнения получают k матриц. Возможный порядок каждой матрицы – l, а максимальный порядок каждой из них – m. Инструмент для поддержки метода сравнения – ExpertChoice для входной матрицы A автоматически получает собственный вектор W, собственное значение max и коэффициент согласованности CR. Для вычисления max и W используются соответствующие функции пакета Matlab [15]. Результаты сравнений заносятся в форму, содержащую перечень весовых коэффициентов программ, критерии, индексы и коэффициенты согласованности. Они предоставляются в виде готовых результатов обработки матриц. Полученные весовые коэффициенты синтезируются с помощью пакета MATLAB 6.5. Результаты отображаются в виде отчета о распределении надежности по объектам системы. Результаты сравнений заносятся в форму, содержащую перечень весовых коэффициентов программ, критерии, индексы и коэффициенты согласованности. Они предоставляются в виде готовых результатов обработки матриц. Полученные весовые коэффициенты синтезируются с помощью пакета MATLAB 6.5. Результаты отображаются в виде отчета о распределении Результаты сравнений заносятся в форму, содержащую перечень весовых коэффициентов программ, критерии, индексы и коэффициенты согласованности. Они предоставляются в виде готовых результатов обработки матриц. Полученные весовые коэффициенты синтезируются с помощью пакета MATLAB 6.5. Результаты отображаются в виде отчета о распределении надежности по объектам системы. II. ПТМ «Прогнозирование плотности дефектов» реализует набор моделей надежности для заданного класса программ системы обработки данных [15]. II. ПТМ «Прогнозирование плотности дефектов» реализует набор моделей надежности для заданного класса программ системы обработки данных [15]. II. ПТМ «Прогнозирование плотности дефектов» реализует набор моделей надежности для заданного класса программ системы обработки данных [15]. 114 Лаврищева Е.М., Пакулин Н.В., Рыжов А.Г. , Зеленов С.В. Анализ методов оценки надежности оборудования и систем. Практика применения методов. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. . Технологический модуль (ТМ) оценки надежности систем 99-120 Прогнозирование плотности дефектов проводится по модели RLM (Rome Laboratory Model) и состоит в оценке влияния на плотность дефектов согласно следующих действий: Прогнозирование плотности дефектов проводится по модели RLM (Rome Laboratory Model) и состоит в оценке влияния на плотность дефектов согласно следующих действий: 1) анализ значений параметров модели прогнозирования надежности, включая остаток дефектов от предыдущего этапа работ с ПО, используется для целевого распределенного значения надежности ПО; 2) сравнение прогнозируемого значения надежности с целевым распределенным значением; 3) корректировки переменных параметров для учета текущего состояния проекта ПО; 4) оценка параметров модели прогнозирования надежности; 5) прогнозирование плотности дефектов; 6) определение пороговых значений (допусков) для оценок результатов прогнозирования и анализа альтернатив; 7) расчет прогнозного значения надежности для ПО. Полученная оценка является модификатором базового значения плотности дефектов для определенного класса ПО. Полученная оценка является модификатором базового значения плотности дефектов для определенного класса ПО. Расчет плотности дефектов делается по модели RLM (Rome Laboratory Model). Сначала выполняется однократное прогнозирование плотности дефектов по формуле: Расчет плотности дефектов делается по модели RLM (Rome Laboratory Model). Сначала выполняется однократное прогнозирование плотности дефектов по формуле: 𝐷𝑔= ∏𝐾𝑖 9 𝑖=1 , 𝐷𝑔= ∏𝐾𝑖 9 𝑖=1 , где Кi – модификаторы плотности дефектов D0, с учетом пороговых значений данных о плотности дефектов. Затем для каждого ПО результаты сравниваются с полученными по модели RLM. Проверка показала, что для ПО объемом 10 - 25 KSLOC погрешность прогнозировании плотности дефектов – примерно составляет 30-35%. Это объясняется некоторыми ограничениями системы Hugin Lite 6.5. Полученные результаты по определению плотности дефектов используются при прогнозировании надежности ПО. III. ПТМ «Прогнозирование надежности» реализует метод прогнозирования значения надежности по каждому модулю системы по следующей модели надежности [15, 23]: ))] exp( 1 ( exp[ t I K I D R i i i i i i          , где i – параметр среды эксплуатации і-го модуля, i – характеристика среды ее разработки, Іі – оцененный размер начального кода, а Di – прогнозируемая плотность дефектов в системе. Коэффициент дефектов K – константа, предвиденная для всех объектов ПС, а значения i и i – известны на момент 115 Lavrischeva E.M., Pakulin N.V., Ryzhov A.G., Zelenov S.V. Analysis of methods for assessing the reliability of equipment and systems. Practice of methods. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 99-120 первоначального прогнозирования надежности, они не изменяются во врем разработки компонентов системы. первоначального прогнозирования надежности, они не изменяются во время разработки компонентов системы. IV. . Технологический модуль (ТМ) оценки надежности систем ПТМ «Оценка надежности системы» согласно классификации дефектов (Orthogonal Defect Classification), в соответствии с которой для каждого выявленного дефекта определяются параметры: тип дефекта, триггер дефекта, влияние дефекта. Эти параметры используется одной или двумя подходящими моделями надежности из выше приведенного в целях проведения оценки прогнозного значения надежности отдельных модулей и системы в целом. Результаты оценки сравниваются, и выбирается из них наиболее правдоподобная модель. Табл. 3. Характеристики качества в стандарте ISO/IEC 9126 Table 3. Characteristics of quality in ISO/IEC 9126 standard № Наименование характеристики Определение характеризуемых свойств ПС q1 Функциональность (functionality) Свойства ПП, обусловливающие способность выполнения функций в соответствии требованиям в процессе тестирования и испытания системы в заданной среде q2 Надежность (reliability) Свойства ПП, обусловливающие ее способность сохранять уровень функционирования и низкую вероятность отказов в процессе выполнения q3 Применимость (usability) Свойства ПП, обусловливающие ее способность быть понимаемой и удобной для использования в указанных условиях q4 Эффективность (efficiency) Свойства ПП для рационального использования выделенных ресурсов при работе системы в установленных условиях q5 Сопровождаемость (maintainability) Свойства ПП, которые обеспечивают модификацию, усовершенствование или адаптацию системы к изменениям среды, требований и функциональности. q6 Переносимость (portability) Свойства ПП, обусловливающие ее способность быть перенесенным из одной среды в другую. На основе полученных данных о надежности и других показателях качества (функциональность, эффективность и др.) рассчитывается целевое значение завершенности и полезности системы (ПС), адекватных потребностям заказчика. При этом мера эксплуатационного качества системы определяется функцией полезности вида: k i k i i пс R a Q   1 116 Лаврищева Е.М., Пакулин Н.В., Рыжов А.Г. , Зеленов С.В. Анализ методов оценки надежности оборудования и систем. Практика применения методов. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 99-120 Лаврищева Е.М., Пакулин Н.В., Рыжов А.Г. , Зеленов С.В. Анализ методов оценки надежности оборудования и систем. Практика применения методов. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 99-120 где ai – мера важности i-й функции системы для процесса, Ri – надежность выполнения функций в заданном периоде t эксплуатации системы. Данные по всем показателям качества (q-quality) q1-q6 в табл. 3 оцениваются где ai – мера важности i-й функции системы для процесса, Ri – надежность выполнения функций в заданном периоде t эксплуатации системы. Данные по всем показателям качества (q-quality) q1-q6 в табл. . Технологический модуль (ТМ) оценки надежности систем 3 оцениваются по формуле: j j j j w m a q 1 1 6 1 1 1    где аi - атрибуты каждого показателя качества (i=1-6); mij – метрики qi показателя с j-атрибутами качества; wij - вес i -показателя качества системы с j-атрибутами. Полученные данные по показателям (характеристикам) модели качества и Ri – надежность выполнения функций входят в сертификат качества [15]. 4. Заключение В работе рассмотрены подходы к оценке надежности технических и программных систем с применением моделей надежности из множества существующих моделей разных видов и типов. Определены основные базовые понятия надежности, обеспечивающие оценку надежности по соответствующим моделям надежности ПС, основанным на времени функционирования и/или количестве отказов (ошибок), получаемых в компонентах в процессах ЖЦ тестирования, системного тестирования и эксплуатации системы. Согласно приведенной классификации моделей надежности процессы обнаружения ошибок в программах носят случайный Марковский и пуассоновский характер и обеспечивают поиск ошибок, дефектов и отказов. Некоторые модели надежности позволяют прогнозировать число ошибок в процессе тестирования, другие оценивать надежность с помощью функций надежности по данным, собранным на этапах ЖЦ разработки системы и испытания. Для примера приведены экспериментальные данные для оценки интенсивности отказов  (t) и количества отказов µ(t) с помощью базовых (Мусы, Гоэла-Окомото и др.) и общей модели надежности, собранных данных на этапах ЖЦ и приведены сравнительные оценки результатов оценки. Дано описание инструментального комплекса модулей ПТМ, обеспечивающих распределение надежности, прогнозирование плотности дефектов и оценки надежности. Приведены показатели качества в стандартной модели ISO 9126 (1-4) и оценки качества, включая измерение показателя надежности, а также других показателей качества, которые входят в сертификат готового продукта. Некоторые модели надежности позволяют прогнозировать число ошибок в процессе тестирования, другие оценивать надежность с помощью функций надежности по данным, собранным на этапах ЖЦ разработки системы и испытания. Для примера приведены экспериментальные данные для оценки интенсивности отказов  (t) и количества отказов µ(t) с помощью базовых (Мусы, Гоэла-Окомото и др.) и общей модели надежности, собранных данных на этапах ЖЦ и приведены сравнительные оценки результатов оценки. Дано описание инструментального комплекса модулей ПТМ, обеспечивающих распределение надежности, прогнозирование плотности дефектов и оценки надежности. Приведены показатели качества в стандартной модели ISO 9126 (1-4) и оценки качества, включая измерение показателя надежности, а также других показателей качества, которые входят в сертификат готового продукта. р ур [1]. Липаев В.В. Надежность программного обеспечения. М.: СИНТЕГ, 1998, 231 стр. Список литературы Список литературы [1]. Липаев В.В. Надежность программного обеспечения. М.: СИНТЕГ, 1998, 231 стр. 117 Lavrischeva E.M., Pakulin N.V., Ryzhov A.G., Zelenov S.V. Analysis of methods for assessing the reliability of equipment and systems. Practice of methods. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 99-120 [2]. Липаев В.В. Методы обеспечения качества крупномасштабных программных систем. М.: СИНТЕГ, 2003, 510 стр. 3]. Майерс Г. Надежность программного обеспечения. М.: Мир, 1980, 360 стр [4]. Мороз Г.Б., Лаврищева Е.М. Модели роста надежности программного обеспечения. Киев: Институт кибернетики АНУ, препринт 92–38, 1992, 23 стр. [5]. Липаев В.В.. Надежность и функциональная безопасность комплексов программ реального времени. Москва, ЗАО «Светлица», 2013, 193 стр. [6]. Shick G.J., Wolverton R.W. An analysis of computing software reliability models. IEEE Transactions on Software Engineering, vol. SE–4, № 2, 1978, pp. 104–120. [7]. Shanthikumar J.G. Software reliability models: A Review. Microelectronics Reliability, vol. 23, № 5, 1983, pp. 903–943. [8]. Goel Amrit L. Software reliability models: Assumptions, limitations, and applicability. IEEE Transactions on Software Engineering, vol. SE–11, № 12, 1985, pp. 1411–1423. [9]. Musa J.D. Okumoto K. A. Logarithmic Poisson Time Model for Software Reliability Measurement. In Proc. of the 7th International Conference on Software Engineering, 1984, pp. 230–238. [10]. Yamada S., Ohba M., Osaki S. S–shaped software reliability grows modeling for software error detection. IEEE Transactions on Reliability, vol. R–32, № 5, pp. 475– 478. [11]. Chulani S. Constructive quality modeling for defect density prediction: COQUALMO. In Proc. of the International Symposium on Software Reliability Engineering (ISSRE'99), 1999. [12]. Гнеденко Б.В., Коваленко И.Н. Введение в теорию массового обслуживания. М., Наука, 1966, 432 стр. [13]. Коваленко И.Н., Шпак В.Д. Вероятностные характеристики сложных систем с иерархическим управлением. Известия АН СССР. Техническая кибернетика, no. 6, 1972, стр. 30-34. [14]. Duval P., Matyas R., Grover A. Continuous integration improving Software quality and reducing risk. Addison Wesley, 2009, 691 p. [15]. Коваль Г.И. Модели и методы инженерии качества систем на ранних этапах ЖЦ. Реф. дис. ИК НАНУ, 2005, 20 стр. [16]. Андон Ф.И., Коваль Г.И. и др. Основы инженерии качества программных систем. К.: Наукова думка, 2007, 670 стр. [17]. Горбенко А.В., Засуха С.А., Рубан В.И., Тарасюк О.М., Харченко В.С. Безопасность ракетно-космической техники и надежность компьютерных систем: 2000-е годы. Авиационно-космическая техника и технология, №1(78), 2011, стр. 9- 20. [18]. A. Avizienis, J.-C. Laprie, B. Randell, C. Landwehr. Basic Concepts and Taxonomy of Dependable and Secure Computing. IEEE Transactions on Dependable and Secure Computing, vol. DOI: 10.15514/ISPRAS-2018-30(3)-8 For citation: Lavrischeva E.M., Pakulin N.V., Ryzhov A.G., Zelenov S.V. Analysis of methods for assessing the reliability of equipment and systems. Practice of methods. Trudy ISP RAN/Proc. ISP RAS, том 30, вып. 3, 2018 г., стр. 99-120. DOI: 10.15514/ISPRAS- 2018-30(3)-8 For citation: Lavrischeva E.M., Pakulin N.V., Ryzhov A.G., Zelenov S.V. Analysis of methods for assessing the reliability of equipment and systems. Practice of methods. Trudy ISP RAN/Proc. ISP RAS, том 30, вып. 3, 2018 г., стр. 99-120. DOI: 10.15514/ISPRAS- 2018-30(3)-8 Analysis of methods for assessing the reliability of equipment and systems. Practice of methods q p y 1,2E.M. Lavrischeva <lavr@ispras.ru> 1,2,3N.V. Pakulin<npak@ispras.ru> 1A.G. Ryzhov <ryzhov@ispras.ru> 1,3S.V. Zelenov <zelenov@ispras.ru> 1 Institute for System Programming of the Russian Academy of Sciences, 25, Alexander Solzhenitsyn st., Moscow, 109004, Russia, 2Moscow Institute of physics and technology (MIPT) 141700, Russia, Moscow region, Dolgoprudny, Campus per., 9. 3 National Research University Higher School of Economics (HSE) 20 Myasnitskaya Ulitsa, Moscow, 101000, Russia 1,2E.M. Lavrischeva <lavr@ispras.ru> 1,2,3N.V. Pakulin<npak@ispras.ru> 1A.G. Ryzhov <ryzhov@ispras.ru> 1,3S.V. Zelenov <zelenov@ispras.ru> 1 Institute for System Programming of the Russian Academy of Sciences, 25, Alexander Solzhenitsyn st., Moscow, 109004, Russia, 2Moscow Institute of physics and technology (MIPT) 141700, Russia, Moscow region, Dolgoprudny, Campus per., 9. 3 National Research University Higher School of Economics (HSE) 20 Myasnitskaya Ulitsa, Moscow, 101000, Russia 1,2E.M. Lavrischeva <lavr@ispras.ru> 1,2,3N.V. Pakulin<npak@ispras.ru> 1A.G. Ryzhov <ryzhov@ispras.ru> 1,3S.V. Zelenov <zelenov@ispras.ru> 1 Institute for System Programming of the Russian Academy of Sciences, 25, Alexander Solzhenitsyn st., Moscow, 109004, Russia, 2Moscow Institute of physics and technology (MIPT) 141700, Russia, Moscow region, Dolgoprudny, Campus per., 9. 3 National Research University Higher School of Economics (HSE) 20 Myasnitskaya Ulitsa, Moscow, 101000, Russia Abstract. The analysis of models and methods of reliability evaluation of hardware and software is carried out. The basic concepts of reliability and safety methods of such systems and situations leading to errors, defects and failures are defined. The definition of reliability and safety of technical systems and software systems is given. The classification of reliability models: predictive, measuring and evaluation types. Evaluation models that are used more in practice are described. The standard of Software life cycle (ISO 15288:2002) is defined, focused on the development and control of system components for errors, starting with the system requirements. The results of application of reliability models (Moussa, Goel- Okomoto, etc.) to small, medium and large projects are presented and their comparative assessment is given. The technological module (TM) of reliability evaluation of complex software systems VPK (1989) is described. The quality model of the standard ISO 9126 (1- 4): 2002-2004 with indicators of functionality, reliability, efficiency, etc., which are used in determining the maturity and certificate of the product is shown. Keywords: reliability, model, method, error, defect, failure, random process, safety, dependability, recoverability, fault tolerance, completeness, reliability assessment, quality certificate. Список литературы 1, no. 1, 2004, pp. 11-33. [19]. IEC 62628. Guidance on software aspects of dependability. Geneva: IEC, 20 [20]. ISO 15288:2002. Systems Engineering. Cycle Life Processes of Systems. [21]. Лаврищева Е.М. Методы программирования. Теория, инженерия, практика. К.: Наукова думка, 2006, 452 стр. [22]. Лаврищева Е.М., Грищенко В.Н. Сборочное программирование. Основы индустрии программных продуктов. К.: Наукова думка, 2009, 372 стр. [23]. Лаврищева Е.М. Software Engineering компьютерных систем. Парадигмы, технологии, СASE-средства. К.: Наукова думка, 2014, 284 стр. 118 Лаврищева Е.М., Пакулин Н.В., Рыжов А.Г. , Зеленов С.В. Анализ методов оценки надежности оборудования и систем. Практика применения методов. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 99-120 Лаврищева Е.М., Пакулин Н.В., Рыжов А.Г. , Зеленов С.В. Анализ методов оценки надежности оборудования и систем. Практика применения методов. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 99-120 [24]. Саати Т. Принятие решений. Метод анализа иерархий. М.: Радио и связь, 1993, 315 стр. [1] Lipaev V. V. Software Reliability. M.: SINTEG, 1998, 231 p. (in Russian) [2] Lipaev V. V. Methods of quality assurance of large-scale software systems. M.: SINTEG, 2003, 510 p. (in Russian). [1] Lipaev V. V. Software Reliability. M.: SINTEG, 1998, 231 p. (in Russian) [2] Lipaev V. V. Methods of quality assurance of large-scale software systems. M.: SINTEG, 2003, 510 p. (in Russian). [3] Myers G. Software Reliability, M.: Mir, 1980, 360 p. (in Russian). p ( ) 3] Myers G. Software Reliability, M.: Mir, 1980, 360 p. (in Russian). References [1] Lipaev V. V. Software Reliability. M.: SINTEG, 1998, 231 p. (in Russian) [2] Lipaev V. V. Methods of quality assurance of large-scale software systems. M.: SINTEG, 2003, 510 p. (in Russian). p ( ) [3] Myers G. Software Reliability, M.: Mir, 1980, 360 p. (in Russian). 119 Lavrischeva E.M., Pakulin N.V., Ryzhov A.G., Zelenov S.V. Analysis of methods for assessing the reliability of equipment and systems. Practice of methods. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 99-120 [4] Moroz G. B., lavrisheva E. M. Models of software reliability growth. K.: V.M. Glushkov Institute of Cybernetics of NAS of Ukraine, preprint 92-38, 1992, 23p. (in Russian). [5] Lipaev V. V. Reliability and functional safety of software systems real time. Moscow, Svetlitsa, 2013, 193 p. [6] Shick G.J., Wolverton R.W. An analysis of computing software reliability models. IEEE Transactions on Software Engineering, vol. SE–4, № 2, 1978, pp. 104–120. [7] Shanthikumar J.G. Software reliability models: A Review. Microelectronics Reliability, vol. 23, № 5, 1983, pp. 903–943. [8] Goel Amrit L. Software reliability models: Assumptions, limitations, and applicability. IEEE Transactions on Software Engineering, vol. SE–11, № 12, 1985, pp. 1411–1423. [9] Musa J.D. Okumoto K. A. Logarithmic Poisson Time Model for Software Reliability Measurement. In Proc. of the 7th International Conference on Software Engineering, 1984, pp. 230–238. [10] Yamada S., Ohba M., Osaki S. S–shaped software reliability grows modeling for software error detection. IEEE Transactions on Reliability, vol. R–32, № 5, pp. 475–478. [11] Chulani S. Constructive quality modeling for defect density prediction: COQUALMO. In Proc. of the International Symposium on Software Reliability Engineering (ISSRE'99), 1999. ( ) [12] Gnedenko B.V., Kovalenko I.N. Introduction to the queueing theory. M.: Science, 1966, 432 p. (in Russian). [13] I. N. Kovalenko and V. D. Shpak. Probabilistic characteristics of complex systems with hierarchical control. Izv. Akad. Nauk SSSR, Tekhn. Kibern., no. 6, 1972, pp. 30–34 (in Russian). [14] Duval P., Matyas R., Grover A. Continuous integration improving Software quality and reducing risk. Addison Wesley, 2009, 691 p. [15] Koval G.I., Models and methods for engineering quality systems at early stages of life cycle. IK NANU, Extended abstract of PhD Thesis, 2005, 20 p. (in Russian) [16] Andon F. I. et al. Foundation of quality engineering software system. K.: Naukova Dumka, 2007, 670 p. (in Russian) [17] Gorbenko A.V.et al. the Safety of rocket-space engineering and reliability of computer systems. References Aerospace technics and technology, №1 (78), 2011, pp. 9-20, 2011. [18] A. Avizienis, J.-C. Laprie, B. Randell, C. Landwehr. Basic Concepts and Taxonomy of Dependable and Secure Computing. IEEE Transactions on Dependable and Secure Computing, vol. 1, no. 1, 2004, pp. 11-33. p g, , , , pp [19] IEC 62628. Guidance on software aspects of dependability. Geneva: IEC, 2011, 63 p. [20] ISO 15288:2002. Systems Engineering. Cycle Life Processes of Systems. [21] Lavrischeva E. M. Programming Methods. Theory, engineering, practice. K.: Naukova Dumka, 2006, 452 p. (in Russian). [22] Lavrisheva E. M., Grishchenko V. N. Assembly programming. Foundation of software industries. K.: K.: Naukova Dumka, 2009, 372 p. (in Russian). [23] Lavrisheva E. M. Software Engineering of computer systems. Paradigms, technologies, CASE-means. K.: K.: Naukova Dumka, 2014, 284 p. (in Russian). [24] Saati T. Decision-Making. Method of hierarchy analysis. M.: Radio and communications, 1993. 315p. (in Russian). 120 For citation: Cheptsov V.Yu., Khoroshilov A.V. In-Kernel Memory-Mapped I/O Device Emulation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 121-134. DOI: 10.15514/ISPRAS-2018-30(3)-9 In-Kernel Memory-Mapped I/O Device Emulation 1,4V.Yu. Cheptsov<cheptsov@ispras.ru> 1,2,3,4 A.V. Khoroshilov<khoroshilov@ispras.ru> 1 Ivannikov Institute for System Programming of RAS 25 Alexander Solzhenitsyn Str., Moscow, 109004, Russia 2 Lomonosov Moscow State University, GSP-1, Leninskie Gory, Moscow, 119991, Russia 3 Moscow Institute of Physics and Technology, 9 Institutskiy per., Dolgoprudny, Moscow Region, 141700, Russia 4 Higher School of Economics. 20 Myasnitskaya Ulitsa Moscow 101000 Russia 20, Myasnitskaya Ulitsa, Moscow 101000, Russia Abstract. Device emulation is a common necessity that arises at various steps of the development cycle, hardware migration, or reverse-engineering. While implementing the algorithms behind the device may be a nontrivial task by itself, connecting the emulator to an existing environment, such as drivers intended to work with the actual hardware, may be no less complex. Devices relying on memory-mapped input/output are of a particular interest, because unlike port-mapped input/output there is much less of a chance that the target platform provides a direct interface to intercept the transmissions. A well-known approach used in various virtual machine software is to put the entire operating system under a hypervisor and build the emulator externally. This may not be desirable for reasons like hypervisor complexity, performance loss, and additional requirements for the host hardware. In this paper we extend this approach to the kernel and explain how it may be possible to build the emulator by relying on the existing interfaces provided by an operating system. Given the common availability of an MMU unit as well as memory protection mechanisms, allowing the handling of page or segment traps at read or write access, we presume that a suggested technique of intercepting memory-mapped input/output could be implemented in a broad number of target platforms. To illustrate the specifics and show potential issues we provide the ways to simplify the implementation and optimize it in speed depending on the target capabilities, the protocol emulated, and the project requirements. As a working proof we created a SMC emulator for an x86 target, which makes use of this approach. Keywords: device emulation; memory-mapped i/o; kernel modules 1. Introduction One of the common engineering demands is device emulation. It may arise during the software development cycle, for example, in testing or driver verification, at hardware migration, when there is no easy way to rewrite the existing software. Other than that, in the world of proprietary hardware and software it is not rare that the only way to understand and document the device abilities is to reverse-engineer it, and the ability to dynamically debug or reverse-engineer the code could be the key in security analysis or adding the device support to a virtual machine. Speaking of virtual machines, or rather hypervisors, building the entire virtual stack for a single device one needs to emulate is often an overkill due to performance reasons, although it could be partially mitigated by hardware-assisted virtualization and software compatibility. The latter may involve working on completely unrelated parts of the driver stack and result in unnecessary costs for continuous support. However, while the development of full platform emulators is a considerably common topic with abundance of existing papers and products like qemu, bochs, iOS simulator, etc., peripheral emulation is much less widespread. In some cases, virtual machine guest tools do try to mimic certain hardware, but even that is usually implemented as a part of a full scale platform emulation. The problem with the peripherals is not just in implementing the algorithms behind the device, which may be a nontrivial task by itself, but also connecting the emulator to an existing environment, such as other drivers above in the stack intended to work with the real hardware. Since one of the important aspects of using any peripherals is the ability for the CPU to communicate to them, the common demand for a device emulator is to provide a way to do it. Presently there are two common low-level approaches to perform input and output operations: port-mapped I/O (PMIO) and memory-mapped I/O (MMIO). While there are other ways such as involving some dedicated hardware, they are relatively less widespread. High-level communications operating on a packet basis (like USB bus) usually go through the dedicated abstraction layer, and thus may be implemented with the standard APIs offered by the operating system without any special effort. DOI: 10.15514/ISPRAS-2018-30(3)-9 DOI: 10.15514/ISPRAS-2018-30(3)-9 121 For citation: Cheptsov V.Yu., Khoroshilov A.V. In-Kernel Memory-Mapped I/O Device Emulation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 121-134. DOI: 10.15514/ISPRAS-2018-30(3)-9 121 121 Cheptsov V.Yu., Khoroshilov A.V. In-Kernel Memory-Mapped I/O Device Emulation. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 121-134 1. Introduction It is fairly easy to implement communication protocols with a hypervisor, the standard approach is to ensure that accessing certain memory exits the virtual machine context (vmexit), which is later handled by the implementation. However, as we mentioned previously, the use of a hypervisor may be impractical, and we have to look for other means of intercepting memory access. Since direct memory access is very common, yet quite problematic to intercept, in this paper we explain how one could implement a considerably portable MMIO emulator in the kernel and cover the details of emulating device communication protocols on common platforms. 122 Чепцов В.Ю., Хорошилов А.В. Эмуляция ввода-вывода оборудования с отображением в ОЗУ внутри ядер операционных систем. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 121-134 Чепцов В.Ю., Хорошилов А.В. Эмуляция ввода-вывода оборудования с отображением в ОЗУ внутри ядер операционных систем. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 121-134 2. State of the art We admit to not being the first to experiment with peripheral device emulation. Every single year several published papers in the field of hardware virtualisation cover this topic to a certain level. Articles published by VMware Inc. researchers [1] [2] provide an in-depth coverage of x86- compatible hardware emulation. They explain the existing obstacles and necessary actions to be taken to implement a complete virtual stack from the CPU to network adapters. In their works they pay a lot of attention to performance optimization, hardware-assisted virtualization and show a visible performance penalty reduction over the new CPU generations in Virtualization nanobenchmarks section of the first referenced paper. As a result of continuous contribution from different parties and competitive product development, the general hypervisor performance has dramatically improved. While GPU emulation is out of the scope of this paper, it should be admitted that there are several works which do manage to provide a complete GPU emulation at a reasonable performance [3] [4]. These works feature an open GPUvm platform in the Xen hypervisor. Another related direction involves security analysis or reverse-engineering. While less frequently found in academic writing, there are several products, tools, and patches for Linux intended to log execution details from the Linux kernel for later analysis. One of the most well-known toolsets is Linux Trace Toolkit, and one of the most prominent cases of applying the approach in practice is for Nouveau driver development for NVIDIA GPUs. Enabling OS Research by Inferring Interactions in the Black-Box GPU Stack by Konstantinos Menychtas, Kai Shen, and Michael L. Scott [5] provides a good coverage in detail. 3. Basic I/O Introduction Port-mapped I/O is usually more demanding to the CPU instruction architecture and requires a number of so-called ports the devices will be mapped to, and perhaps a dedicated instruction set to access these ports as well. Because the device memory is accessed indirectly, another name for PMIO is detached I/O. As an example, one of the most popular architectures to implement PMIO is x86. It can be utilized by means of two dedicated instructions: in and out, which enable one to receive and send 8, 16, or 32 bits of data to a port from 0 to 65535. Since there are faster ways to perform I/O on x86 and PMIO is not recommended for use nowadays, in some literature it may be referred to as legacy I/O. This may not be the case for other architectures found in micro-controllers, but in general MMIO support is increasing. Memory-mapped I/O involves direct mapping of the device memory to the host memory, enabling the software to access the device just like a normal chunk of noncacheable RAM with the use of the native instruction set. Since MMIO implementation is often faster than PMIO and sometimes simpler to use, it will be the one to opt for when implementing a device communication protocol. For example, on 123 Khoroshilov A.V. In-Kernel Memory-Mapped I/O Device Emulation. Trudy ISP RAN /Proc. ISP RAS, 2018, pp. 121-134 Cheptsov V.Yu., Khoroshilov A.V. In-Kernel Memory-Mapped I/O Device Emulation. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 121-134 x86 various devices installed as PCI extension cards or system management controllers make a use of it. Virtual devices are not supposed to be functionally different from real hardware. For this reason, emulators have difficulties supporting I/O communication protocols. The taken approach varies depending on the demands and available resources, but usually one of the following is used:  Custom device development  Driver reimplementation  Building a hypervisor Sadly, each of these has serious limitations, and most of them create obstacles for generic peripheral emulation, as observed in Table. 1.  Custom device development  Driver reimplementation  Building a hypervisor Sadly, each of these has serious limitations, and most of them create obstacles for generic peripheral emulation, as observed in Table. 1. Table 1. 4. Intercepting the I/O As a result I/O interception comes out as a pragmatic way to achieve the goal. Despite not being very common, software and hardware have enough capabilities to intercept raw device communication without touching the higher-level drivers themselves. raw device communication without touching the higher level drivers themselves. For example, for the past 8 years the recent x86 firmwares contain a dedicated UEFI System Management Mode [6] protocol to intercept PMIO. This protocol originally existed as a EFI_SMM_IO_TRAP_DISPATCH_PROTOCOL protocol1, but later on was extended with an additional IO_TRAP_EX_DISPATCH_PROTOCOL protocol2. Both protocols allow you to create direct handlers to intercept the portmapped access. By design, the management mode affects the operating system code as well, so it works throughout the boot process and is fully transparent to the higher level software implementations like OS kernel or drivers. However, aside from not being very well documented, third-party code execution in the System Management Mode is generally prohibited. So even if one is to reimplement the SMI handler similar to what Intel offers with the open source platform code, it will be of no use for anyone but UEFI firmware developers. Fortunately, most of PMIO interface code is usually well abstracted in the kernel, and when it comes to intercepting you could just replace the underlying low level function implementation within the emulator context. However, devices relying on MMIO are of a particular interest, because unlike PMIO there is a much less chance that the target platform provides a direct interface to intercept the transmissions. For embedded devices it may well be sufficient to statically analyze the firmware, find the instructions responsible for I/O, and either dynamically or statically overwrite them to jump to prepared thunks that will handle them accordingly. This approach is common for security analysis especially when very little is known not only about the explored peripherals but the whole controller. However, since the firmware or the driver may receive updates in the future, this approach is not very effective outside of security or code coverage analysis, and the like. One of the first ideas that comes to mind due to the nature of MMIO writes is relying on CPU debug registers. These registers (e.g. DR on Intel or BP_CTRL/BP_COM on ARM Cortex) allow you to implement hardware breakpoints or rather watchpoints, which may trap read and write access. 3. Basic I/O Introduction 121-134 likely become a blocker if the rest of the environment is not generic and well-known. It is unfortunate, but even the mainstream operating systems may be unwilling to expose new interfaces for virtual machines (like most of the graphical stack on Apple macOS). 1 GUID: 58DC368D-7BFA-4E77-ABBC-0E29418DF930 2 GUID: 5B48E913-707B-4F9D-AF2E-EE035BCE395D 3. Basic I/O Introduction Pros and cons summary Device Driver Hypervisor Software independency + - ± Low costs - ± + Legal issues + - - Infrastructure dependency - + - Forward compatibility + - - Performance + ± ± Other device support + + - Table 1. Pros and cons summary Device Driver Hypervisor Software independency + - ± Low costs - ± + Legal issues + - - Infrastructure dependency - + - Forward compatibility + - - Performance + ± ± Other device support + + - Developing a new device by extending a microcontroller to offer a required interface or creating an entire chip mostly works for very simple devices when a single copy is going to be used for some kind of deep debugging or instrumentation. A good example could be removable BIOS chips for debugging or HDMI to VGA adapters with HDCP decoding. While this solution is very reliable for creating a test device, the results of mass-producing a customised device will likely be not worth the effort. It will be either more expensive or worse in quality. In addition, it is important to have the legal part of the question in mind and avoid patent infringement. However, this method could be most reliable when it comes to stability. Reimplementing the driver to support another communication interface for the virtual device is very useful when working with performance-critical hardware such as GPUs. For them each extra communication layer may heavily affect the performance due to high bandwidth usage, and that is why virtualization software implements extended GPU support (like DirectX or OpenGL) in such a way. However, in our case it defeats the entire purpose of creating a virtual device. If the point is to test the driver, it will no longer stay the same. If the reason is to support a proprietary driver, one will have to reverse-engineer it and have issues every time it gets updated. Bringing in a virtual machine with a hypervisor is a way to overdo it. While a decent virtual machine has a wide range of supported hardware, it adds a lot of downsides as well. In particular there will always be potential performance issues, even with hardware-assisted virtualization support. More than that, compatibility issues will 124 Чепцов В.Ю., Хорошилов А.В. Эмуляция ввода-вывода оборудования с отображением в ОЗУ внутри ядер операционных систем. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 4. Intercepting the I/O However, these registers are very few, and their scope area is small (i.e. a 32-bit or 64-bit word). Other than that, the kernel, 125 Cheptsov V.Yu., Khoroshilov A.V. In-Kernel Memory-Mapped I/O Device Emulation. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 121-134 Yu., Khoroshilov A.V. In-Kernel Memory-Mapped I/O Device Emulation. Trudy ISP RAN /Proc. ISP RAS 3, 2018, pp. 121-134 debuggers, or other software may use these registers for their own needs, which leads to them being simply impractical for this kind of work. In general-purpose operating systems with defined kernel APIs there are much better ways, such as a page protection mechanism, which is used to implement watchpoints in software. While this is suitable for doing MMIO emulation, most of the known works relying on this technique either use it for tracing or just for debugging backends. The notable example is MMIO trace in Linux, which was originally In general-purpose operating systems with defined kernel APIs there are much better ways, such as a page protection mechanism, which is used to implement watchpoints in software. While this is suitable for doing MMIO emulation, most of the known works relying on this technique either use it for tracing or just for debugging backends. The notable example is MMIO trace in Linux, which was originally developed to reverse-engineer proprietary NVIDIA drivers by tracing the register access [7]. Other than that, there are very few examples of how it can be utilized for device I/O emulation. developed to reverse-engineer proprietary NVIDIA drivers by tracing the register access [7]. Other than that, there are very few examples of how it can be utilized for device I/O emulation. 5. Proposed approach 121-134 Чепцов В.Ю., Хорошилов А.В. Эмуляция ввода-вывода оборудования с отображением в ОЗУ внутри ядер операционных систем. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 121-134 2. host performs the I/O operation (by reading or writing at a defined addre space); 2. host performs the I/O operation (by reading or writing at a defined address space); p ) 3. host ensures that the operation is complete and repeats the process. 3. host ensures that the operation is complete and repeats the process. The 1st and 3rd steps are usually implemented as a write-and-poll, a write-and- interrupt or just as a poll. Another advantage comes out from common differences in frequencies between the host and the peripheral. Since communications usually happen between the devices with different clock bases, most of the protocols are synchronous, and the host generally does not overwrite the areas it has just written to without making a read to confirm it was successful. Even more, most of the protocols are stateful, and it is uncommon to see subsequent reads from the same place expecting the value to change more than once. A write operation will most likely appear in-between. The 1st and 3rd steps are usually implemented as a write-and-poll, a write-and- interrupt or just as a poll. Another advantage comes out from common differences in frequencies between the host and the peripheral. Since communications usually happen between the devices with different clock bases, most of the protocols are synchronous, and the host generally does not overwrite the areas it has just written to without making a read to confirm it was successful. Even more, most of the protocols are stateful, and it is uncommon to see subsequent reads from the same place expecting the value to change more than once. A write operation will most likely appear in-between. Under these assumptions we use a simple satisfactory transaction model as an example: Under these assumptions we use a simple satisfactory transaction model as an example: 1. write operation type (read or write); 2. read acknowledge status until status ready; 3. handle the values: 3. handle the values: 3.1. read the value for read operations; 3.2. write the value for write operations and read acknowledge status until status ready. 3.2. write the value for write operations and read acknowledge status until status ready. 5. Proposed approach The idea of general purpose I/O interception is very simple: catch reads and writes, make sure that the values read are correct, and the values written are accounted for. To apply it to MMIO we could limit page protection of the target area, and trap the faults as they happen. Due to bandwidth limitations and architecture simplicity the I/O sequences are generally serialized, even if they happen from different threads. It may not be the case for GPUs, yet GPUs likely will not need this kind of emulation due to performance reasons. Still, in general if serialized I/O is not guaranteed even within a single memory page (which is rare) one could always implement it manually by utilizing the synchronization primitives. Therefore, the most obvious approach will be: 1. mark the relevant page as neither writable nor readable (not present in x86 terms); 1. mark the relevant page as neither writable nor readable (not present in x86 terms); 2. catch a fault and decode the fault address and the direction (in or out); 3. disassemble the instruction that caused the fault and obtain its operands from the frame; 4. handle the operands for the emulation; 5. update the destination registers or memory for the reads as necessary; 6. return to the location after the instruction, which caused the fault. While it indeed solves the problem and looks very straightforward, the implementation itself could be very convoluted. While the saved context is likely to contain the fault and return addresses, bringing a full-scale disassembling framework to the kernel is inflexible due to extra architecture dependencies and considerable amounts of code required for instruction emulation. Even more, it may impose additional performance penalties, which are already tough enough. For these reasons we tried to alter the algorithm in a way that would be simpler, less platform-dependent, and similarly performant. After examining several real-world examples, we consider the following model of a MMIO-based I/O protocol, which could be applied to quite a number of devices: 1. host ensures that the target is ready for an I/O operation; 126 Чепцов В.Ю., Хорошилов А.В. Эмуляция ввода-вывода оборудования с отображением в ОЗУ внутри ядер операционных систем. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 121-134 цов В.Ю., Хорошилов А.В. Эмуляция ввода-вывода оборудования с отображением в ОЗУ внутри яде ационных систем. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 5.2 Without write-only page support When write-only pages are not available, we may still be able to work out a simpler approach, and this is where memory patching comes in hand. The idea is to let the original instruction perform the I/O just as normal, but to encode a jump-back instruction right afterwards to ensure that page protection is limited again to trap the next I/O operation. Initially this approach may appear to have too many issues to be considered in practice, however, they could all be solved with enough effort, and some of them could even be turned into benefits. The first issue to solve is the length of the faulted instruction. A number of architectures provide fixed-length instruction sets, so the next instruction address to encode our jumpback instruction could be calculated even without knowing anything about the current instruction. For others one could write or find simple instruction fetchers, to only decode the length without operand or operation details. Such software may also go under the name of length disassemblers, and various implementations exist for popular platforms [11]. It may become a little more involved when the I/O instruction results in non-linear control flow, but in general I/O and branching instructions belong to separate classes and are not mixed together. The second issue occurs when the device memory is mapped to userspace and the communication happens in userspace as well. In this case a direct jump to protection restoration code is not possible, and a breakpoint or similar instruction will have to be encoded to trigger the context switch, return to the kernel and pass the control to our handler. The third and probably the most serious issue happens when I/O operations are performed through shared code. By assuming serialized I/O we consider no cases of simultaneous code execution from the same area (unless there are multiple devices). Therefore, we could safely patch it. However, nothing prohibits the driver from utilizing generic memory primitives like memcpy or memset to bulk-write or read the dedicated area. These primitives generally have no effect on the I/O itself, and we do not need to intercept every byte they touch. To avoid the issue one could examine the stack trace and modify the instruction at the return address. 5.1 With write-only page support If write-only pages are supported, in a number of cases one may implement a flip- flop approach that will switch page protection from read-only to write-only and backwards as the process goes. To emulate the proposed transaction we could start the communication process with the page marked as read-only, which will then trap on operation type. Here we will initiate the transaction and switch the protection to write-only. After the operation is written the trap on the status read will trigger, where we will read the written operation type, update the value for read operations and set its status. Afterwards the page protection is returned to read-only and the control is transferred back to the driver. For read operations that is all of it, for write operations the driver will read the status and attempt to perform the actual write, which should trigger the trap again. From there on one could repeat the process as described for the operation type. In the end for both reads and writes page protection returns back to read-only, eliminating any platform-specific disassembling and relying on generic approach. Expectedly one does not have any easy access to write-only pages on popular architectures such as ARM [8] or x86. Perhaps, if these architectures were originally designed at present, when the demand for better memory protection management is much higher and when features like WˆX memory and execute-only memory have already become commonplace, we would have had finer memory management that would support write-only pages. However, nowadays write-only pages are not very 127 Cheptsov V.Yu., Khoroshilov A.V. In-Kernel Memory-Mapped I/O Device Emulation. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 121-134 Khoroshilov A.V. In-Kernel Memory-Mapped I/O Device Emulation. Trudy ISP RAN /Proc. ISP RAS, 2018, pp. 121-134 common in both hardware and software implementations. Certain PowerPC implementations [9] or processor extensions may provide access to them, so it remains a good idea to check CPU manuals before abandoning the try. For example, Intel x86 processors starting from Nehalem technically support write-only memory via EPT (Extended Page Tables [10]), yet it can hardly be used for anything but virtualization. 6. Evaluation To apply the proposed solution in practice we created a software-based emulator for the 2nd generation Apple SMC in a form of a kernel extension for Apple macOS. System Management Controller (SMC) is a chip commonly found in Intel-based Apple Macintosh computers or certain Google Chromebooks. This chip is responsible for computer power management, display backlight control, HDD monitoring, thermal control, hybrid sleep and hibernation support, external device current regulation (AirPort, USB, FireWire), charging the battery, trackpad controls, screen mirroring, etc. This chip is not essential for computer functioning, and could be viewed as a convenience feature for a vendor to rely on to centralize and simplify hardware management. There are two main generations of SMC controllers in Apple computers. The 1st generation was built on a 16-bit Renesas H8S/2117 controller and exposed port- mapped I/O interfaces to communicate with the operating system. The 2nd and subsequent generations are based on 32-bit ARMv7-A processors, and expose memory-mapped and port-mapped I/O interfaces. Both approaches are used to implement the same functionality within a single synchronous stateful protocol. Initially the communication happens via the PMIO protocol, and then a switch to MMIO protocol happens if the device supports it. The whole communication process happens within the kernel and the existing drivers for the 2nd generation hardware are closed-source. Fortunately, due to side researchers the communication protocols are mostly documented [12]. The reasons for taking this particular device into consideration was not only because it is a challenging task compared to devices with open specifications and decent documentation, but also for the importance of having better control of the hardware you use. Apple SMC has complete access to every device in the system and could monitor the bus communications. Other than that it stores temporary encryption keys for hibernation images or user action free restarts (authenticated restarts), when full disk encryption is enabled. Apple SMC drivers expose a dedicated protocol to userspace. This protocol provides a way to obtain SMC data and configure both SMC and onboard devices. Given its direct connection to the hardware, it may be possible to inflict damage on the computer by overheating or causing power surges. Moreover, previous researches discovered that it was very easy to modify SMC firmware, which is also a very serious concern [13]. The actual implementation follows the proposed approach without write-only page support with all the suggested optimizations and certain platform-specific adjustments. 5.2 Without write-only page support Not only this does not require disassembling but also reduces the penalty from trapping extra I/O operations, so a quick stack unwinding that can often be implemented with compiler intrinsics easily pays off. With all the pieces put together it creates a solid approach for a large chunk of I/O protocols. In addition to these general improvements platform-specific optimizations could be applied. For example, extra page protection changes may be avoided for 128 Чепцов В.Ю., Хорошилов А.В. Эмуляция ввода-вывода оборудования с отображением в ОЗУ внутри ядер операционных систем. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 121-134 write operations, if the hardware may ignore interrupts caused by write protection violation (CR0 WP bit on x86). It should be noted that one is to pay extra attention to the scheduler (e.g. disable preemption) not to let it switch the task to another core, where write protection is on. 6. Evaluation SMC MMIO protocol covers a 64 KB area, which we split into pages 129 Cheptsov V.Yu., Khoroshilov A.V. In-Kernel Memory-Mapped I/O Device Emulation. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 121-134 Yu., Khoroshilov A.V. In-Kernel Memory-Mapped I/O Device Emulation. Trudy ISP RAN /Proc. ISP RAS 3, 2018, pp. 121-134 with the dedicated handlers based on the page index. Since the access to each page is serialized, no additional I/O wrapping is necessary. In the XNU kernel, which powers all modern Apple hardware including Macs, Intel CPU exceptions are routed through a dedicated kernel_trap function. To let the driver communicate with the emulated device we added a SMC nub via the standard I/O Kit APIs with mapped memory regions with restricted protection and extended the kernel_trap function in EXC_I386_PGFLT handling code specifically for our memory. A simplified version of this code is shown in Listing 1. ioRegionStart and ioRegionEnd locate the emulated I/O area starting and ending addresses, appleSmcStart and appleSmcEnd point to the AppleSMC driver address range. instrSize function calculates the instruction length at the return address to later write the jump-back code via writeTrampoline function, which not only writes the trampoline code (by disabling the WP bit and interrupts) but additionally disables CPU preemption to avoid the scheduler switch. p p auto faultAddr = state->cr2; if (faultAddr >= ioRegionStart && faultAddr < ioRegionEnd) { auto retAddr = state->rip; if (retAddr >= appleSmcStart && retAddr < appleSmcEnd) { // Simple case (from AppleSMC) retAddr += instrSize(retAddr, 1); } else { // Complex case (from e.g. memcpy) retAddr = unwindToSMC(state->rsp); } auto faultType = FaultTypeRead; if (state->err & T_PF_WRITE) { faultType = FaultTypeWrite; } updateProtection(faultType, faultAddr); saveOrgCode(retAddr, TrampolineSize); writeTrampoline(faultType, faultAddr); return; } 7. Conclusion Emulating peripheral devices within the existing operating system is not a new problem. Different solutions and approaches have appeared over the years. The industrial demand for full-stack operating system virtualization brought their performance to a completely different level, and the needs for better customization resulted in operating system developers providing more flexible interfaces with the possibility to create virtual hardware out of the box. Programmable microcontrollers made the process of building a device clone with the necessary features a much simpler task to accomplish. However, there are numerous cases, where in-kernel peripheral emulation is highly anticipated, such as driver development needs, testing and verification, hardware migration, security analysis, etc. As we stated, it is often not possible or extremely impractical to attempt to incorporate virtual machines due to development costs or performance penalties. While virtual machines succeed in emulating CPUs of the same architecture at almost the same speed with hardware assisted virtualization, the performance of other CPUs without the use of JITs, commonly used in video game console emulators but rarely found in generic virtualization software, is often much worse. And in terms of I/O emulation, which is the primary concern of this paper, the situation is no better. Furthermore, all the solutions heavily depend on the target architecture. While it was possible to think of x86 as the main architecture for personal computers in the beginning of 2000- s, today the concept of personal computers has shifted away, and other major players, e.g. ARM, appeared on the market. With this in mind the classical approach to virtualizing the whole operating system could face severe issues in the future. The idea of using page protection faults to handle device I/O events without a hypervisor may be known but not widespread anywhere out of I/O tracing. In this paper we described a way to implement a complete MMIO protocol emulator in the kernel with the use of a generic approach that has few dependencies on the target architecture and relies on platform features such as MMU and paging. We showed that certain target architecture capabilities and device protocol specifics may affect the implementation, and effectively allow or disallow a broad range of optimizations. We believe that a suggested device I/O protocol model is applicable to various hardware, and give examples on how to simplify and optimize its implementation. Listing 1. Sample code Listing 1. Sample code To transfer the control flow to the protocol emulator updateProtection is performing the actual protection upgrade of the emulated I/O area and invokes the read access handler. It should be noted that a dedicated procedure may be needed for platforms with delayed physical mapping update. For example, with XNU it is necessary to trigger virtual memory fault twice when the page is not present. Similarly, the 130 Чепцов В.Ю., Хорошилов А.В. Эмуляция ввода-вывода оборудования с отображением в ОЗУ внутри ядер операционных систем. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 121-134 Чепцов В.Ю., Хорошилов А.В. Эмуляция ввода-вывода оборудования с отображением в ОЗУ внутри ядер операционных систем. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 121-134 protection restoration routine invoked from the trampoline preserves the registers and calls the write handler. As a result, it was possible to emulate all the existing SMC protocols at no issue and avoid the use of the original device. 7. Conclusion After exploring the existing hardware, we built a SMC emulator in the XNU kernel to illustrate the suggested approach. 131 oroshilov A.V. In-Kernel Memory-Mapped I/O Device Emulation. Trudy ISP RAN /Proc. ISP RAS, 8, pp. 121-134 Cheptsov V.Yu., Khoroshilov A.V. In-Kernel Memory-Mapped I/O Device Emulation. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 121-134 Cheptsov V.Yu., Khoroshilov A.V. In-Kernel Memory-Mapped I/O Device Emulation. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 121-134 Acknowledgements The authors thank ISP RAS and SYRCoSE staff for review and comment, Nikita Golovliov for aid in SMC emulator development, and Marvin Häuser for reverse- engineering Apple SMC UEFI drivers. Чепцов В.Ю., Хорошилов А.В. Эмуляция ввода-вывода оборудования с отображением в ОЗУ внутри ядер операционных систем. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 121-134 цов В.Ю., Хорошилов А.В. Эмуляция ввода-вывода оборудования с отображением в ОЗУ внутри яде ационных систем. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 121-134 DOI: 10.15514/ISPRAS-2018-30(3)-9 Для цитирования: Чепцов В.Ю., Хорошилов А.В. Эмуляция ввода-вывода оборудования с отображением в ОЗУ внутри ядер операционных систем. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 121-134 (на английском языке). DOI: 10.15514/ISPRAS- 2018-30(3)-9 References [1]. Jeremy Sugerman, Ganesh Venkitachalam, Beng-Hong Lim. Virtualizing I/O Devices on VMware Workstation’s Hosted Virtual Machine Monitor. In Proceedings of the General Track: 2001 USENIX Annual Technical Conference, 2001, pp. 1-14. Available at: http://static.usenix.org/legacy/publications/library/proceedings/usenix01/sugerman/suger man.ps, accessed 12.06.18 [2]. Keith Adams, Ole Agesen. A Comparison of Software and Hardware Techniques for x86 Virtualization. In Proceedings of the 12th international conference on Architectural support for programming languages and operating systems, 2006, pp. 2-13. Available at: https://www.vmware. com/pdf/asplos235 adams.pdf, accessed 09.06.18 p p p p [3]. Yusuke Suzuki, Shinpei Kato, Hiroshi Yamada, and Kenji. GPUvm: Why Not Virtualizing GPUs at the Hypervisor? In Proceedings of the 2014 USENIX Annual Technical Conference, 2014, pp. 109-120. Available at: https://www.usenix.org/system/files/conference/ atc14/atc14-paper-suzuki.pdf, accessed 12.06.18 [4]. Hangchen Yu, Christopher J. Rossbach. Full Virtualization for GPUs Reconsidered. In Proceedings of the Annual Workshop on Duplicating, Deconstructing, and Debunking, 2017. [5]. Konstantinos Menychtas, Kai Shen, Michael L. Scott. Enabling OS Research by Inferring Interactions in the Black-Box GPU Stack. In Proceedings of the 2013 USENIX conference on Annual Technical Conference, 2013, pp. 291-296. Available at: https://www.usenix.org/system/files/ conference/atc13/atc13-menychtas.pdf, accessed accessed 12.06.18 [6]. Unified EFI, Inc. Platform Initialization (PI) Specification. Version 1.6. 2017. Available at: http://www.uefi.org/sites/default/files/resources/PI Spec 1 6.pdf, accessed 09.06.18 [7]. Jeff Muizelaar, Pekka Paalanen. In-kernel memory-mapped I/O tracing Available at: https://www.kernel.org/doc/Documentation/trace/mmiotrace.txt, accessed 12.06.18 [8]. Arm Holdings. ARM1176JZ-S Technical Reference Manual. Available at: http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.ddi0333h/ Caceaije.html, accessed 12.06.18 [9]. NXP Semiconductors. e500mc Core Reference Manual. Available at: http://cache. freescale.com/files/32bit/doc/ref manual/E500MCRM.pdf, accessed 09.06.18 [10]. Intel. Intel Virtualization Technology: Hardware Support for Efficient Processor Virtualization. Available at: http://www.ece.cmu.edu/_ece845/sp17/ docs/vt-overview- itj06.pdf, accessed 12.06.18 j p [11]. BeaEngine. Length Disassembler Engine for Intel 64-bit processors. Available at: https://github.com/BeaEngine/lde64, accessed 12.06.18 [12]. CupertinoNet. EfiPkg, AppleSmcIo protocol. Available at: https://github.com/ CupertinoNet/EfiPkg, accessed 12.06.18 [13]. Crowdstrike. Alex Ionescu. ”Spell”unking in Apple SMC Land. 2013. Available at: http://www.nosuchcon.org/talks/2013/D1 02 Alex Ninjas and Harry Potter.pdf, accessed 09.06.18 132 Эмуляция ввода-вывода оборудования с отображением в ОЗУ внутри ядер операционных систем 1,4 В. Ю. Чепцов <cheptsov@ispras.ru> 1,2,3,4 А. В. Хорошилов <khoroshilov@ispras.ru> 1 Институт системного программирования им. В.П. Иванникова РАН, 109004, Россия, г. Москва, ул. А. Солженицына, д. 25; 2 Московский государственный университет имени М.В. Ломоносова, 119991, Россия, Москва, Ленинские горы, д. 1, 3 101000, Россия, г. Москва, ул. Мясницкая, д. 20. Аннотация. Необходимость эмуляции оборудования часто возникает на различных стадиях цикла разработки, миграции оборудования или обратной разработки. Реализация алгоритмов, связанных с конкретным устройством, сама по себе является нетривиальной задачей, но интеграция эмулятора с существующей средой, например, драйверами, предназначенными для работы с реальным оборудованием, зачастую оказывается не менее сложной. Устройства, полагающиеся на ввод-вывод с отображением в оперативную память, представляют особый интерес, так как в этих случаях, в отличие от использования портов ввода-вывода, гораздо меньше вероятность, что целевая платформа предоставит интерфейс для перехвата операций. Один из распространённых подходов, широко используемый в ПО виртуальных машин, состоит в том, чтобы поместить всю операционную систему под гипервизор и создать внешний эмулятор. Однако это может быть нежелательно по причинам сложности гипервизора, потери производительности, дополнительных требований к аппаратному обеспечению и пр. В данной статье такой подход распространяется на ядро, и предлагается описание возможности построить эмулятор, прибегая лишь к существующим интерфейсам, предоставляемым операционной системой. Ввиду частой доступности MMU и механизмов защиты страниц, позволяющих перехватывать доступ записи и чтения, предполагается, что предлагаемый подход может быть использован на значительном количестве целевых платформ. В статье приводится подробное рассмотрение проблем, возникающих при написании конкретной реализации, и приводятся способы её упрощения и оптимизации в зависимости от возможностей целевой платформы, эмулируемого протокола и иных требований к задаче. В качестве экспериментального доказательства работоспособности предлагаемого подхода приводится реализация эмулятора SMC для платформы x86. Ключевые слова: эмуляция оборудования; ввод-вывод с отображением в ОЗУ; модули ядра Ключевые слова: эмуляция оборудования; ввод-вывод с отображением в ОЗУ; модули 133 Cheptsov V.Yu., Khoroshilov A.V. In-Kernel Memory-Mapped I/O Device Emulation. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 121-134 Khoroshilov A.V. In-Kernel Memory-Mapped I/O Device Emulation. Trudy ISP RAN /Proc. ISP RAS, 2018, pp. 121-134 Cheptsov V.Yu., Khoroshilov A.V. In-Kernel Memory-Mapped I/O Device Emulation. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 121-134 Список литературы [1]. Jeremy Sugerman, Ganesh Venkitachalam, Beng-Hong Lim. Virtualizing I/O Devices on VMware Workstation’s Hosted Virtual Machine Monitor. In Proceedings of the General Track: 2001 USENIX Annual Technical Conference, 2001, pp. 1-14. Режим доступа: http://static.usenix.org/legacy/publications/library/proceedings/usenix01/sugerman/suger man.ps, дата обращения 12.06.18 [2]. Keith Adams, Ole Agesen. A Comparison of Software and Hardware Techniques for x86 Virtualization. In Proceedings of the 12th international conference on Architectural support for programming languages and operating systems, 2006, pp. 2-13. Режим доступа: https://www.vmware. com/pdf/asplos235 adams.pdf, дата обращения 09.06.18 [3]. Yusuke Suzuki, Shinpei Kato, Hiroshi Yamada, and Kenji. GPUvm: Why Not Virtualizing GPUs at the Hypervisor? In Proceedings of the 2014 USENIX Annual Technical Conference, 2014, pp. 109-120. Режим доступа: https://www.usenix.org/system/files/conference/ atc14/atc14-paper-suzuki.pdf, дата обращения 12.06.18 [4]. Hangchen Yu, Christopher J. Rossbach. Full Virtualization for GPUs Reconsidered. In Proceedings of the Annual Workshop on Duplicating, Deconstructing, and Debunking, 2017. [5]. Konstantinos Menychtas, Kai Shen, Michael L. Scott. Enabling OS Research by Inferring Interactions in the Black-Box GPU Stack. In Proceedings of the 2013 USENIX conference on Annual Technical Conference, 2013, pp. 291-296. Режим доступа: https://www.usenix.org/system/files/ conference/atc13/atc13-menychtas.pdf, дата обращения 12.06.18 [6]. Unified EFI, Inc. Platform Initialization (PI) Specification. Version 1.6. 2017. Режим доступа: http://www.uefi.org/sites/default/files/resources/PI Spec 1 6.pdf, дата обращения 09.06.18 [7]. Jeff Muizelaar, Pekka Paalanen. In-kernel memory-mapped I/O tracing Режим доступа: https://www.kernel.org/doc/Documentation/trace/mmiotrace.txt, дата обращения 12.06.18 [8]. Arm Holdings. ARM1176JZ-S Technical Reference Manual. Режим доступа: http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.ddi0333h/ Caceaije.html, дата обращения 12.06.18 [9]. NXP Semiconductors. e500mc Core Reference Manual. Режим доступа: http://cache. freescale.com/files/32bit/doc/ref manual/E500MCRM.pdf, дата обращения 09.06.18 [10]. Intel. Intel Virtualization Technology: Hardware Support for Efficient Processor Virtualization. Режим доступа: http://www.ece.cmu.edu/_ece845/sp17/ docs/vt- overview-itj06.pdf, дата обращения 12.06.18 [11]. BeaEngine. Length Disassembler Engine for Intel 64-bit processors. Режим доступа: https://github.com/BeaEngine/lde64, дата обращения 12.06.18 [12]. CupertinoNet. EfiPkg, AppleSmcIo protocol. Режим доступа: https://github.com/ CupertinoNet/EfiPkg, дата обращения 12.06.18 [13]. Crowdstrike. Alex Ionescu. ”Spell”unking in Apple SMC Land. 2013. Режим доступа: http://www.nosuchcon.org/talks/2013/D1 02 Alex Ninjas and Harry Potter.pdf, дата обращения 09.06.18 134 DOI: 10.15514/ISPRAS-2018-30(3)-10 For citation: Mallachiev K.A., Khoroshilov A.V. Building Modular Real-time software from Unified Component Model. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 135- 148. DOI: 10.15514/ISPRAS-2018-30(3)-10 Building Modular Real-time software from Unified Component Model 1,2 K.A. Mallachiev <mallachiev@ispras.ru> 1,2,3,4 A.V. Khoroshilov <khoroshilov@ispras.ru> 1 Ivannikov Institute for System Programming of the Russian Academy of Sciences, 25, Alexander Solzhenitsyn st., Moscow, 109004, Russia. 2 Lomonosov Moscow State University, GSP-1, Leninskie Gory, Moscow, 119991, Russia 3 Moscow Institute of Physics and Technology, 9 Institutskiy per., Dolgoprudny, Moscow Region, 141700, Russia 4 Higher School of Economics. 20, Myasnitskaya Ulitsa, Moscow 101000, Russia 1,2 K.A. Mallachiev <mallachiev@ispras.ru> 1,2,3,4 A.V. Khoroshilov <khoroshilov@ispras.ru> 1 Ivannikov Institute for System Programming of the Russian Academy of Sciences, 25, Alexander Solzhenitsyn st., Moscow, 109004, Russia. 2 Lomonosov Moscow State University, GSP-1, Leninskie Gory, Moscow, 119991, Russia 3 Moscow Institute of Physics and Technology, 9 Institutskiy per., Dolgoprudny, Moscow Region, 141700, Russia 4 Higher School of Economics. 20, Myasnitskaya Ulitsa, Moscow 101000, Russia Abstract. Modern real-time operating systems are complex embedded product made by many vendors: OS vendor, board support package vendor, device driver developers, etc. These operating systems are designed to run on different hardware; the hardware often has limited memory. Embedded OS contains many features and drivers to support different hardware. Most of the drivers are not needed for correct OS execution on a specific board. OS is statically configured to select drivers and features for each board. Modularity of OS simplifies both configuration and development. Splitting OS to isolated modules with well- specified interfaces reduces developers’ needs to interact during joint development. The configurator, in turn, can easily compose isolated components without component developers. We use formal models to specify components and their composition. Formal model describes the behavior of components and their interaction. Usage of formal models has many benefits. Models contain enough information to generate source code in C language. Our model is executable; this allows configurator to quickly verify the correctness of component configurations. Moreover, model contains constraints on its parameters. These constraints are internal consistency or some external properties. Constraints are translated into asserts in generated source code. Therefore, we can check these constraints both at model simulation and at source code execution. This paper presents our approach to describe such models at Scala language. We successfully tested the approach in RTOS JetOS. Abstract. Modern real-time operating systems are complex embedded product made by many vendors: OS vendor, board support package vendor, device driver developers, etc. These operating systems are designed to run on different hardware; the hardware often has limited memory. Building Modular Real-time software from Unified Component Model Embedded OS contains many features and drivers to support different hardware. Most of the drivers are not needed for correct OS execution on a specific board. OS is statically configured to select drivers and features for each board. Modularity of OS simplifies both configuration and development. Splitting OS to isolated modules with well- specified interfaces reduces developers’ needs to interact during joint development. The configurator, in turn, can easily compose isolated components without component developers. Keywords: components; modularity; RTOS; formal models; code generation 1. Introduction Modern embedded operating systems support several CPU architectures and a lot of peripheral devices. OS contains many drivers to support numerous different hardware. Embedded OS are often designed for execution in a restricted environment, for example, with limited memory. Most of the drivers are not needed for correct OS execution on some specific board and spend valuable resources. Therefore, OS must support configuration to select drivers, which will execute on the target hardware. Static OS configuration is used in cases when it is known in advance, on which hardware the OS image is going to be executed. Static means that configuration is performed on the host machine before OS loading to the target machine. The result of static OS configuration is the final image, which can be run on the target. Static configuration allows keeping final image small. Typically, there are two roles taking part in the process of OS image building. The first role is a developer of whole OS or some driver. Developer implements his part in some programming language, writes documentation and provides support of source code and documentation. The second one is a system integrator who is responsible for correct OS configuration for specific task of specific board. Usually the system integrator does not change OS source code. Besides simple selecting, which driver will be in the final OS image, many operating systems support finer tuning. For example, configuration allows selecting file system for each hard drive, or set IP address that will be used by network stack. These details are configured statically because for embedded OS and especially for safety-critical systems simplicity is more important than generality. It is a natural desire to divide the operating system into isolated components, but not every part of the OS can be isolated. For example, OS core often is strongly coupled and might be divided into isolated components only if the core will be fully redesigned to support new architecture. If we investigate configurations of the same OS on different boards, then we will see that there is the most variable part in the OS. We call this part OS drivers. OS drivers contain device drivers and some services such as network stack, file system, logging, etc. Our work aimed to support flexible configuration of OS drivers. It is common that there are many vendors involved in building of OS drivers. DOI: 10.15514/ISPRAS-2018-30(3)-10 DOI: 10.15514/ISPRAS-2018-30(3)-10 DOI: 10.15514/ISPRAS-2018-30(3)-10 For citation: Mallachiev K.A., Khoroshilov A.V. Building Modular Real-time software from Unified Component Model. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 135- 148. DOI: 10.15514/ISPRAS-2018-30(3)-10 135 Mallachiev K.A., Khoroshilov A.V. Building Modular Real-time software from Unified Component Model. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 135-148 Mallachiev K.A., Khoroshilov A.V. Building Modular Real-time software from Unified Component Model. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 135-148 Mallachiev K.A., Khoroshilov A.V. Building Modular Real-time software from Unified Component Model. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 135-148 2. Related Works Classical distributed component models like Enterprise JavaBeans, CORBA and CORBA Component Model [2, 3] define components and interfaces between them. These models allow substituting one component with another one if both have the same interfaces. Brokers dynamically change components configuration. This dynamic configuration is not suitable for embedded systems with static configuration. Ideas to separate OS appeared long ago in microkernels. Microkernel architecture’s [4, 5] primary goal is to separates OS into independent servers that could be isolated from each other. Servers interact through inter-process communication (IPC). IPC calls are typed and servers with the same interface can substitute one another. But there cannot be two servers with the same interface; therefore, this model is not suitable for our tasks too. Classical distributed component models like Enterprise JavaBeans, CORBA and CORBA Component Model [2, 3] define components and interfaces between them. These models allow substituting one component with another one if both have the same interfaces. Brokers dynamically change components configuration. This dynamic configuration is not suitable for embedded systems with static configuration. dynamic configuration is not suitable for embedded systems with static configuration. Ideas to separate OS appeared long ago in microkernels. Microkernel architecture’s [4, 5] primary goal is to separates OS into independent servers that could be isolated from each other. Servers interact through inter-process communication (IPC). IPC calls are typed and servers with the same interface can substitute one another. But there cannot be two servers with the same interface; therefore, this model is not suitable for our tasks too. OS-Kit [6] and eCos [7] apply modularity benefits into OS development process. They provide a set of OS components, which are used as building blocks to configure an OS. For configuration, eCos uses the Component Definition Language (CDL), an extension of the existing Tool Command Language (Tcl) scripting language. Configuration is represented as feature tree with internal dependencies, group and feature constraints. Enabling of one component can lead to enable of whole components subtree. Components can have calculated value in configuration, which are calculated based on other configuration parameters. However, this is not enough for our task. Configurator cannot manage component connections and cannot add copies of the same component. μC/OS-II kernel uses THINK component framework [8, 9]. THINK is an implementation of the FRACTAL component model that aims to take into account the specific constraints of embedded systems development. Component describes through its interface. 1. Introduction When services or drivers are strongly coupled, their developers have to interact a lot. Therefore, splitting OS drivers into independent isolated components helps to simplify and accelerate development. Component should interact with each other. Appearance of fixed interface between components would make component development easier. Moreover, fixed interface can make system flexible. Only connected components can interact, and only component with the same interfaces can be connected. System integrator is responsible for connection of the components. 136 Маллачиев К.А., Хорошилов А.В Построение модульного программного обеспечения на основе однородной компонентой модели. Труды ИСП РАН, Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 135-148 Маллачиев К.А., Хорошилов А.В Построение модульного программного обеспечения на основе однородной компонентой модели. Труды ИСП РАН, Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 135-148 Suppose that system integrator created a composition of the components, which describes how each component is configured and how components are connected Suppose that system integrator created a composition of the components, which describes how each component is configured and how components are connected. We call component-based system flexible if the system integrator can: Suppose that system integrator created a composition of the components, which describes how each component is configured and how components are connected. We call component based system flexible if the system integrator can: describes how each component is configured and how components are connected We call component-based system flexible if the system integrator can:  modify configuration of the single component without modifying others  substitute component with another one of the same interface without modifying other components,  add a new component between two other connected components without modifying any component configuration except the new one.  add to composition a copy of existing component, and they should not disturb each other. We are developing an embedded real-time operating system for civil aircraft computers called JetOS [1]. JetOS is ARINC-653 compliant and statically configured. Approaches presented in this paper are designed for JetOS. Since JetOS is a RTOS, we are focused on minimizing the overhead added by component-based system. 2. Related Works Interaction between components is possible after establishment 137 Mallachiev K.A., Khoroshilov A.V. Building Modular Real-time software from Unified Component Model. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 135-148 Mallachiev K.A., Khoroshilov A.V. Building Modular Real-time software from Unified Component Model. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 135-148 of bindings between their interfaces. Binding is a communication channel between two or more components. Binding can be created between components of a distributed system (RPC binding). This concept also does not allow having several copies of the same component in the composition. VxWorks is a popular embedded operating system. VxWorks board support package (BSP) is divided into components. Components interfaces are declared in Component Description Language (CDL). Note that this CDL is different from the CDL used in eCos. BSP developer can construct BSP from existing component and can add their own components. However, this system is not flexible. For example, each component has fixed list of component names, with which it can interact. We are not aware of any component-based model with the following set of features:  static configuration; We are not aware of any component-based model with the following set of f  static configuration;  low overhead;  flexible configuration (in all aspects described in the introduction);  flexible configuration (in all aspects described in the introduction);  type checking of the connection, i.e. checking that connected components have the same interface.  type checking of the connection, i.e. checking that connected components have the same interface. 3.2 System integrator view System integrator gets specification of all component types in the system. System integrator decides how many instances of each component should be created and how they should be connected for solution of the specific problem. For each instance, integrator sets its configuration values. 3. Component-based Model Instances have unique values of state and configuration. It is easy to see that concepts of component type and component instance are similar to terms “class” and “class object” respectively. Instances have unique values of state and configuration. It is easy to see that concepts of component type and component instance are similar to terms “class” and “class object” respectively. 3.1 Component Developer View Component developer designs component state structure, how it should be initialized base on configuration and how it is changed during execution. Developer chooses types of configuration parameters. Developer does not aware of specific configuration parameters values, but he can add constraints on the values. He designs component input and output ports and implements handlers for input ports. Component’s input and output ports restrict component developer’s knowledge about “outside world”. He does not know how many instances of his component will be created or how they will be connected. Component developer’s definition of component types consists of two parts: component type specification and implementation. Specification contains:  component type name  component type name  component input and output port names and their types  structure of component configuration  component’s purpose description: how it should be configured and in which environment its input ports should be called.  component’s purpose description: how it should be configured and in  component’s purpose description: how it should be c which environment its input ports should be called. p p p p which environment its input ports should be called. The rest of the information is private for component and is considered a implementation part. 3. Component-based Model Our model is component-based. Component has state, which is changed during model execution, and configuration, which is immutable. Components can communicate with other components via ports. Port is a set of functions; there are two kinds of ports: input ports and output ports. Output port can be connected with input port. Set of port function signatures is called port type. Only input and output port of the same port type can be connected. Each function of a component input port has an assigned handler inside the component. Call of output port function leads to the call of connected input port, which, in turn, calls the assigned handler. These calls are standard function call, or in other words synchronous call inside the same thread. Therefore, component loses control during output port call. Thus, port call keeps the current thread. Threads cannot be created dynamically during model execution. Threads count is constant during execution. If component needs an additional thread, then this should be explicitly specified in the model. These components are called active. Active components have special handlers, which are called periodically or once in the context of the new thread. We call these handlers the activity handlers. In order to facilitate component reuse we introduce the concepts of a component type and a component instance. Each component type can have any number of instances. The components described above are close to component instances. Component type contains types of component state and configuration, but not their values. Component type contains types and names of input and output ports, but not their connection. In addition, component type contains implementation of: 138 Маллачиев К.А., Хорошилов А.В Построение модульного программного обеспечения на основе однородной компонентой модели. Труды ИСП РАН, Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 135-148 К.А., Хорошилов А.В Построение модульного программного обеспечения на основе однородной ой модели. Труды ИСП РАН, Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 135-148  component initialization function, which is called at start and is used to initialize state based on the configuration;  component initialization function, which is called at start and is used to initialize state based on the configuration;  component initialization function, which is called at start and is used to initialize state based on the configuration;  handlers assigned with input ports, if component has any;  activity handlers if component is active. 3.2 Simple example Suppose that component developer created Amplifier component type. Amplifier has single input port “in” and single output port “out”. In addition, it has single configuration parameter “factor”. Components aim is to amplify input signal from “in” port by factor “factor” and put output to “out” port. Suppose that the system integrator wants to pass signal from two sensors to a single actuator, but he should amplify signal from first sensor by factor of 2 and from 139 Mallachiev K.A., Khoroshilov A.V. Building Modular Real-time software from Unified Component Model. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 135-148 Mallachiev K.A., Khoroshilov A.V. Building Modular Real-time software from Unified Component Model. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 135-148 second one by factor of 10. System integrator decides to use Amplifier component type. He does not worry about implementation, only interfaces matters to him. For simplicity, let us assume that all ports have the same type. Amplifier component type as seen by system integrator can be seen at Fig. 1 second one by factor of 10. System integrator decides to use Amplifier component type. He does not worry about implementation, only interfaces matters to him. For simplicity, let us assume that all ports have the same type. Amplifier component type as seen by system integrator can be seen at Fig. 1 Fig. 1. Graphical representation of Amplifier component type specification Fig. 1. Graphical representation of Amplifier component type specification System integrator creates two instances of Amplifier component type: “amp1” with configuration value “factor” equal to 2 and “amp2” with configuration value “factor” equal to 10. Then connects them accordingly to sensors and to actuator. Scheme of the result can be seen at Fig. 2 eme of the result can be seen at Fig. 2 Fig. 2. Amplifier instances connection scheme Fig. 2. Amplifier instances connection scheme 5. Model-Based approach We decided to go further along the path of abstraction and use abstract models of components and their composition. We use formal executable models. This has many benefits. Model contains more information than source code, thus source code can be generated based on the model. In addition, executable model allows simulating instances behaviour and their interaction. This is very useful for system integrator to quickly verify the correctness of configurations. Moreover, formal model can be used to formally verify its internal consistency. We use Scala language to model components. Scala is a functional object-oriented language that suits us well. 4. Prototype In previous work [10], we implemented component-based approach in C language with some YAML code. We used common approach to apply object-oriented ideas in C language. Component state and configuration is presented as C structure, which explicitly passed to all component functions. Wrappers hid calls to output ports. There was a lot of boilerplate code used to create component instances, describe their configuration, and their connections, in component type specification and its wrappers implementation. To reduce amount of handwork we started to use YAML — simple declarative language. In the YAML developer specifies component type state, configuration, input and output ports, names of functions-handler for input ports. System integrator describes in the YAML component instances, their configuration and connections. We generated C code based on these YAML specifications. This approach has some disadvantages.  Component developer has to manually keep consistent two files (in YAML and C languages). Change in one file leads to change in another one.  Component developer has to manually keep consistent two files (in YAML and C languages). Change in one file leads to change in another one. 140 Маллачиев К.А., Хорошилов А.В Построение модульного программного обеспечения на основе однородной компонентой модели. Труды ИСП РАН, Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 135-148 Маллачиев К.А., Хорошилов А.В Построение модульного программного обеспечения на основе однородной компонентой модели. Труды ИСП РАН, Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 135-148  Component developer’s workflow is not comfortable: after change in YAML code generation should be processed and only then C code should be updated accordingly.  System integrator can connect instances incorrectly (this does not apply to type checking, which is performed during compilation) and cannot see the problem until final OS image is prepared and executed in target hardware.  System integrator can connect instances incorrectly (this does not apply to type checking, which is performed during compilation) and cannot see the problem until final OS image is prepared and executed in target hardware. Fig. 3. «Counter» component type 5.1.1 Component Developer View Component type is presented as Scala class inherited from interface (trait) «Component». Component configuration and state are the class fields with fixed names «config» and «state» respectively. Active components have functions, which are called periodically or once. If component type inherits trait «RunOnce» then it should implement function «start», which will be called once after component initialization. If component type inherits class «Periodically», then it should implement function «periodically»; the frequency of the call is determined by the configuration. For example, consider “Counter” component type, at Fig. 3, which has a state but no configuration. State contains value «callCount», which is initialized with zero. Function «periodically» increases «callCount» on every call. Fig. 3. «Counter» component type 141 141 Mallachiev K.A., Khoroshilov A.V. Building Modular Real-time software from Unified Component Model. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 135-148 Mallachiev K.A., Khoroshilov A.V. Building Modular Real-time software from Unified Component Model. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 135-148 Port types are declared as interfaces (traits). Input ports are defined inside component type class as objects, which inherited port type. Output port are class fields with type of port type. Output ports values are passed as component type constructor parameters. It is worth noting that output ports can be passed by name to constructor, this allows initializing component instances with cycle connections among them. Example of input/output ports for “Amplifier” component type (defined in previous sections) can be seen at Fig. 4 Model can have constraints on state and configuration parameters values. These constraints are defined using Scala require function. Example of require statement for “Amplifier” component type can be seen at Fig. 5. Fig. 4. Port type SignalProcessor and ports of «Amplifier» component type. The component type has input port «in» and output port «out», both of them have type SignalProcessor. here is an implementation of function processSignal of «in» port. Port «out» passed-by-name. Scala syntax may be confusing, here function processSignal returns result of out port call Fig. 5. Configuration constraint for «Amplifier» component type; «factor» can take values only in the interval from 1 to 49 Fig. 5. Configuration constraint for «Amplifier» component type; «factor» can take values only in the interval from 1 to 49 Fig. 6. «Amplifier» instances connection scheme Fig. 6. «Amplifier» instances connection scheme Маллачиев К.А., Хорошилов А.В Построение модульного программного обеспечения на основе однородной компонентой модели. Труды ИСП РАН, Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 135-148 Маллачиев К.А., Хорошилов А.В Построение модульного программного обеспечения на основе однородной компонентой модели. Труды ИСП РАН, Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 135-148 5.1.2 System Integrator View System integrator creates instances of component type and connects them. For each instance, he defines its configuration parameters values. As an example of component instances and their connections, consider model of the scheme depicted in the Fig. 2. This model can be seen at Fig 6. 142 Маллачиев К.А., Хорошилов А.В Построение модульного программного обеспечения на основе однородной компонентой модели. Труды ИСП РАН, Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 135-148 5.1.3 Preconfigured components There is often component which have configuration parameters that have the same value in different configuration. To simplify configuration process for system integrator, we can define new component type, in which these parameters are fixed and cannot be configured. New component type class constructor calls constructor of the original one with values of these parameters. For example, it is possible to define “AmplifierBy2” which amplifies signal by fixed factor of 2. It is more interesting to define new component, which is a composition of existing components. This is useful if some compositions are used often. Our approach assumes unified modeling of components and their composition. This allows using component-composition transparently for system integrator. As an example, assume that there are component type «Amplifier» and «Filter», that are often connected. We create a new component type «AmplifyAndFilter» that is the composition of «Amplifier» and «Filter» Graphical representation of the «AmplifyAndFilter» component type can be seen at Fig. 7 and implementation at Fig. 8. Fig. 7. Graphical representation of «AmplifyAndFilter» component type. Fig. 7. Graphical representation of «AmplifyAndFilter» component type. 143 Mallachiev K.A., Khoroshilov A.V. Building Modular Real-time software from Unified Component Model. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 135-148 , , , , pp Fig. 8. Implementation of «AmplifyAndFilter» component type Fig. 10. C_code annotation example This C_code annotation allows iteratively develop generator tool. At start, when tool supports only a few Scala statements, almost all code has C annotations. When support for new Scala statements adds to the tool, C annotations for these statements are no longer needed. Therefore, during tool development number of C_code annotations decreases. Маллачиев К.А., Хорошилов А.В Построение модульного программного обеспечения на основе однородной компонентой модели. Труды ИСП РАН, Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 135-148 Fig. 9. Port type SignalProcessor and ports of «Amplifier» component type with annotations. Scala language has rich syntax and not every statement can be easily translated to C. We allow annotating blocks of Scala code or Scala functions with C code. Fig. 10 contains partial example. Fig. 10. C_code annotation example 5.2 Model Usage We use model to simulate instances behaviour and their interaction. We can verify that constraints are hold during simulation. In addition, we can write tests (unit and integration) to check that component model is correct. We use model to generate C code, which gets into JetOS. We statically parse Scala code, extract needed information and translate it into C code. Generated C code structurally looks much like code generated by prototype based on YAML files. We use same approach to model OOP in C language. Some parts of the model can be translated into C without modifications, for example, simple operations and function calls. Some parts modified automatically during translation, but some can not be automatically translated without human help. JetOS has strict coding style and, for instance, function can not have more than one return statement. We can generate code according this code style and, for example, we can automatically substitute several return statements in the model with a single one in the generated code. As was mentioned, there are also statements, which cannot be easily translated into C. In addition, there are situations when generator tool cannot get enough information statically analysing Scala code. To solve these problems we add annotations to Scala code. Annotations does not change behaviour of model, they used only to provide additional information for the generator tool. We use annotations to highlight input and output ports and their type interfaces. Annotations are «inport», «outport» and «interface» for input ports, output ports and port types respectively. As an example, Fig, 9 shows «Amplifier» component type with annotations. 144 Маллачиев К.А., Хорошилов А.В Построение модульного программного обеспечения на основе однородной компонентой модели. Труды ИСП РАН, Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 135-148 6 Future Work First, we still do not support many Scala statements and have a lot of C_code in our models. We are going to fix this in the new versions of generator tool. For now, system developer should write Scala code by hand. This Scala code is very simple and matches a simple pattern. Thus, we can generate this Scala code from some GUI interface. Configuration constraints of the model can be extracted and added to this tool. This is one of optional future works. Furthermore, formal model is a powerful tool and allows much more than C code generation. Formal model can be used for model checking and formal verifying internal consistency, preconditions or state invariants. Tests and requirements can be generated based on the model and requirement generation is our next task. Requirement is the most important part of safety-critical system certification. Requirement writing is a hard handwork and automation (at least partial) will be very helpful. 145 Mallachiev K.A., Khoroshilov A.V. Building Modular Real-time software from Unified Component Model. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 135-148 Mallachiev K.A., Khoroshilov A.V. Building Modular Real-time software from Unified Component Model. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 135-148 Mallachiev K.A., Khoroshilov A.V. Building Modular Real-time software from Unified Component Model. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 135-148 7 Conclusion The paper presents continuation of the work on modularity of RTOS. OS drivers are decomposed into isolated components. System integrator carries out component composition, and it can be done without contacting component developers and without writing C code. We use a unified formal model to specify both components and their composition. Model, which is written in Scala language, is used to generated C code. Also, model is executable, this allows system integrator to quickly verify correctness of composition. Model contains constraints on the model parameters. These constraints are tested during model simulation, also constraints can be translated into asserts in the generated C code. Model-based approach still has disadvantage since the model is divided in two parts written in two languages, which have to be manually kept consistent. However, C code for some Scala statement is placed right before the statement, we hope that this will stimulate developers to update parts synchronously. Maturing of the generator tool decreases amount of C code in the model and reduces the importance of the problem. The approach has been successfully tested on OS drivers of JetOS — ARINC-653 compliant RTOS. ARINC-653 has restrictions on the code executed in OS. For instance, resources (like buffers, semaphores, threads, etc.) can be requested only during initialization stage. Model restriction on threads creation apply well to ARINC-653 restrictions. Moreover, constructor code of the component type class is executed during initialization stage. Thus, component can request resources in the constructor. References [1]. K.M. Mallachiev, N.V. Pakulin, and A.V. Khoroshilov. Design and architecture of real- time operating system. Trudy ISP RAN / Proc. ISP RAS, vol. 28, no. 2, 2016, pp. 181– 192. DOI: 10.15514/ISPRAS-2016-28(2)-12 [2]. J. Siegel and D. Frantz. CORBA 3 fundamentals and programming. John Wiley & Sons New York, NY, USA, 2000, vol. 2. [3]. N. Wang, D. C. Schmidt, and C. O’Ryan. Overview of the corba component model. In Component-Based Software Engineering. Addison-Wesley Longman Publishing Co., Inc., 2001, pp. 557–571. [4]. A. Gefflaut, T. Jaeger, Y. Park, J. Liedtke, K. J. Elphinstone, V. Uhlig, J. E. Tidswell, L. Deller, and L. Reuther. The sawmill multiserver approach. In Proceedings of the 9th workshop on ACM SIGOPS European workshop: beyond the PC: new challenges for the operating system, 2000, pp. 109–114. [5]. I. Boule, M. Gien, and M. Guillemont. Chorus distributed operating systems. Computing Systems, vol. 1, no. 4, 1988, pp. 305-370. [6]. B. Ford, G. Back, G. Benson, J. Lepreau, A. Lin, and O. Shivers. The flux oskit: A substrate for kernel and language research. ACM SIGOPS Operating Systems Review, vol. 31, no. 5, 1997, pp. 38–51. 146 Маллачиев К.А., Хорошилов А.В Построение модульного программного обеспечения на основе однородной компонентой модели. Труды ИСП РАН, Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 135-148 [7]. A. Massa, Embedded software development with eCos. Prentice Hall Professional Technical Reference, 2002. [8]. J.-P. Fassino, J.-B. Stefani, J. L. Lawall, and G. Muller. Think: A software framework for component-based operating system kernels. In Proceedings of the USENIX Annual Technical Conference, General Track, 2002, pp. 73–86. [9]. F. Loiret, J. Navas, J.-P. Babau, and O. Lobry. Component-based real-time operating system for embedded applications. In Proceedings of the International Symposium on Component-Based Software Engineering. Springer, 2009, pp. 209–226. [10]. K. Mallachiev, N. Pakulin, A. Khoroshilov, and D. Buzdalov. Using modularization in embedded OS. Trudy ISP RAN / Proc. ISP RAS, vol. 29, issue. 4, 2017, pp. 283–294. DOI: 10.15514/ISPRAS-2017-29(4)-19 Построение модульного программного обеспечения на основе однородной компонентой модели DOI: 10.15514/ISPRAS-2018-30(3)-10 Для цитирования: Маллачиев К.А., Хорошилов А.В Построение модульного программного обеспечения на основе однородной компонентой модели. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 135-148 (на английском языке). DOI: 10.15514/ISPRAS-2018-30(3)-10 Построение модульного программного обеспечения на основе однородной компонентой модели А.В. Хорошилов <khoroshilov@ispras.ru> 1 Институт системного программирования им. В.П. Иванникова РАН, 109004, Россия, г. Москва, ул. А. Солженицына, д. 25, 2 Московский государственный университет имени М.В. Ломоносова, 119991, Россия, Москва, Ленинские горы, д. 1 3Московский физико-технический институт, 141700, Московская облаcть, г. Долгопрудный, Институтский пер., 9 4 Высшая школа экономики, 101000, Россия, г. Москва, ул. Мясницкая, д. 20 101000, Россия, г. Москва, ул. Мясницкая, д. 20 Аннотация. Современные операционные системы реального времени являются сложным продуктом, разрабатываемым многими поставщиками: непосредственными разработчиками ОС, поставщиками пакета поддержки аппаратуры, разработчиками драйверов устройств и т.д. Такие ОС спроектированы так, чтобы иметь возможность запускаться на различном оборудовании, часто имеющем ограниченные ресурсы. Встраиваемые ОС содержат множество настроек и драйверов для поддержки разной аппаратуры. Большинство из этих драйверов являются излишними для запуска ОС на каком-то конкретном оборудовании. ОС статически конфигурируется для выбора набора драйверов и настроек для каждого типа аппаратуры. Модульность ОС упрощает как разработку ОС, так и ее конфигурирование. Разделение ОС на изолированные модули с фиксированными интерфейсами уменьшает необходимость взаимодействия между разработчиками в ходе совместной разработки. Мы используем формальные модели для описания компонентов и их взаимодействия. Использование формальных моделей приносит большую пользу. Описываемые модели содержат достаточно информации для генерации исходного кода компонента на языке Си. Предоставляемые модели являются исполняемыми, что позволяет человеку, отвечающему за конфигурацию, быстро проверить правильность заданной конфигурации. Кроме того, модель содержит ограничения на конфигурационные параметры. Примером таких ограничений являются ограничения на внутреннюю согласованность модели. При генерации исходного кода такие ограничения транслируются в специальные проверки 147 Mallachiev K.A., Khoroshilov A.V. Building Modular Real-time software from Unified Component Model. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 135-148 Mallachiev K.A., Khoroshilov A.V. Building Modular Real-time software from Unified Component Model. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 135-148 на уровне исходного кода. Следовательно, ограничениями могут быть проверены как во время симуляции модели, так и во время исполнения исходного кода. В данной работе представлен подход к описанию таких моделей на языке программирования Scala. Мы успешно апробировали данный подход на основе ОС реального времени JetOS. на уровне исходного кода. Следовательно, ограничениями могут быть проверены как во время симуляции модели, так и во время исполнения исходного кода. В данной работе представлен подход к описанию таких моделей на языке программирования Scala. Мы успешно апробировали данный подход на основе ОС реального времени JetOS. Ключевые слова: компоненты; модульность; ОСРВ; формальные модели; генерация кода Ключевые слова: компоненты; модульность; ОСРВ; формальные модели; генерация кода Список литературы [11]. K.M. Mallachiev, N.V. Pakulin, and A.V. Khoroshilov. Design and architecture of real- time operating system. Trudy ISP RAN / Proc. ISP RAS, vol. 28, no. 2, 2016, pp. 181– 192. DOI: 10.15514/ISPRAS-2016-28(2)-12 [1]. J. Siegel and D. Frantz. CORBA 3 fundamentals and programming. John Wiley & Sons New York, NY, USA, 2000, vol. 2. [2]. N. Wang, D. C. Schmidt, and C. O’Ryan. Overview of the corba component model. In Component-Based Software Engineering. Addison-Wesley Longman Publishing Co., Inc., 2001, pp. 557–571. [3]. A. Gefflaut, T. Jaeger, Y. Park, J. Liedtke, K. J. Elphinstone, V. Uhlig, J. E. Tidswell, L. Deller, and L. Reuther. The sawmill multiserver approach. In Proceedings of the 9th workshop on ACM SIGOPS European workshop: beyond the PC: new challenges for the operating system, 2000, pp. 109–114. [4]. I. Boule, M. Gien, and M. Guillemont. Chorus distributed operating systems. Computing Systems, vol. 1, no. 4, 1988, pp. 305-370. [5]. B. Ford, G. Back, G. Benson, J. Lepreau, A. Lin, and O. Shivers. The flux oskit: A substrate for kernel and language research. ACM SIGOPS Operating Systems Review, vol. 31, no. 5, 1997, pp. 38–51. pp [6]. A. Massa, Embedded software development with eCos. Prentice Hall Professional Technical Reference, 2002. [7]. J.-P. Fassino, J.-B. Stefani, J. L. Lawall, and G. Muller. Think: A software framework for component-based operating system kernels. In Proceedings of the USENIX Annual Technical Conference, General Track, 2002, pp. 73–86. [8]. F. Loiret, J. Navas, J.-P. Babau, and O. Lobry. Component-based real-time operating system for embedded applications. In Proceedings of the International Symposium on Component-Based Software Engineering. Springer, 2009, pp. 209–226. [9]. K. Mallachiev, N. Pakulin, A. Khoroshilov, and D. Buzdalov. Using modularization in embedded OS. Trudy ISP RAN / Proc. ISP RAS, vol. 29, issue. 4, 2017, pp. 283–294. DOI: 10.15514/ISPRAS-2017-29(4)-19 148 DOI: 10.15514/ISPRAS-2018-30(3)-11 DOI: 10.15514/ISPRAS-2018-30(3)-11 For citation: Andryukhin A.A. Methods of protecting decentralized autonomous organizations from crashes and attacks. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 149-164. DOI: 10.15514/ISPRAS-2018-30(3)-11 Methods of protecting decentralized autonomous organizations from crashes and attacks A.A. Andryukhin <Alexandr@kcdigital.ru> KCD, office 3, 131, prospect Mira, Moscow, 129226, Russia Abstract. Field of study: Blockchain technology, decentralized autonomous organizations, smart contract and their resistance to attacks and failures. Theoretical and practical significance: Due to the fact that such a form of organization is experimental, participants often face problems of attacks on the organization, the consequences of incorrectly written rules and of fraud. The task of creating decentralized autonomous organizations that are resistant to failures and attacks, and research on the causes of such problems has become relevant for software developers and architects. Goals and objectives of work: Investigation of attack algorithms and development of methods for ensuring the sustainability of decentralized autonomous organizations for attacks on the basis of analysis of the subprocesses of border events and logs using the methods of Process Mining. The methods to be developed should promptly identify and prevent inconsistencies between the alleged and actual behavior of smart contracts that lead to such errors in the operation, such as the content of spam contracts, empty transactions, increased block processing time, etc. Keywords: blockchain; decentralized autonomous organizations; process mining; smart contract; security 1.1 Blockchain and crypto-currencies In the past few years, thanks to the popularization of blockchain technology, which represents a continuous series of blocks containing information, built according to certain rules, there were created many services and applications using various crypto-currencies [13]. Many crypto-currencies are inextricably linked with this technology for such reasons: decisions on the blockchain do not require trust between the participants, they are open and validated. Success of the Bitcoin, decentralized crypto-currency with the capitalization of more than $ 10 billion, is of 149 Andryukhin A.A. Methods of protecting decentralized autonomous organizations from crashes and attacks. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 149-164 genuine interest both in industry, government, and in science [2]. A whitepaper, written under the pseudonym Satoshi Nakamoto in 2008, is the basic document for any form of organization created on blocking technology. This document for the first time outlines the Bitcoin structure and introduces the concept of blockchain [30]. The theoretical basis used in the creation of decentralized autonomous organizations is based on the research of automatic verification systems [31, 32], cryptography [33, 34] and distributed databases [35, 36]. 1.2 Decentralized autonomous organizations The economic theory and research of organizations [19, 20], the theory of contracts [21, 22], auction mechanisms [23, 24], the theory of innovation [25, 26], as well as virtual organizations [27, 28] played an important role in the emergence of decentralized autonomous organizations [30]. Bitcoin can be called the very first decentralized autonomous organization created to carry out paid transactions [1]. The most famous decentralized autonomous organization, based on the Bitcoin code, was created in 2014 and was given the current name Dash in March 2015. Dash is currently experiencing a stage of rapid growth. In September 2017, the company's market capitalization was $ 2.5 billion. However, the most promising platforms for the development of decentralized autonomous organizations are platforms that use smart contracts and the Turing- complete programming language, such as Ethereum [2]. p p g g g g On April 30, 2016, the first decentralized crowd-funding project, known as The Dao (Decentralized Autonomous Organization), was launched on the Ethereum blockchain platform. The organization was established as a venture capital fund with transparent and democratic flows of project financing, in which each investor would have a voice whose weight is directly proportional to the funds invested. The technology of smart contracts was laid for the first time in the basis of the functioning of the organization The DAO. The Dao in record time attracted more than $ 168 million in investment almost immediately became a target of intruders and was repeatedly attacked to steal or freeze funds. As a result of one of the attacks, more than $ 50 million was stolen from the organization, and as a result of the other, more than $ 150 million was frozen [14, 16]. The imperfection in the code of smart contracts and the existing vulnerabilities, as well as the inability to change them lead to so-called softfork and hardfork. The Dao is not the only decentralized organization deployed on Ethereum. Fermat (www.fermat.org), Digix.global also operate on the Ethereum blockchain platform and are managed collectively by the participants who own the tokens by voting. 1.3 Smart contracts In 1994, cryptographer Nick Szabo proposed the use of cryptography and computer technology to automate the process of concluding, executing and auditing various contracts [29]. The development of this direction led to the creation of smart 150 Андрюхин А.А. Методы защиты децентрализованных автономных организаций от системных отказов и атак. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 149-164 contracts on the basis of the blockade - special electronic algorithms introduced into the block, where they are monitored by the decentralized computer network itself. This allows you to expand the capabilities of the block-up to a computing platform for centralized execution of common tasks [5]. contracts on the basis of the blockade - special electronic algorithms introduced into the block, where they are monitored by the decentralized computer network itself. This allows you to expand the capabilities of the block-up to a computing platform for centralized execution of common tasks [5]. Smart contracts allow you to exclude from the process of intermediaries because computer algorithm independently and automatically confirms the fulfillment of the terms of the contract and determines what to do with the asset for which the contract was created. Smart contracts are protected from uncoordinated changes in the terms of the transaction, allow you to automate the audit and make it in real time. The most famous framework for smart contracts is Ethereum, a decentralized virtual machine, where the Turing-complete programming language is used to create smart contracts. A distinctive feature of Ethereum is the ability to transfer ETH crypto currency between users and contracts. Users create transactions on the Ethereum network in order to create a new contract, call a contract, or transfer ETH to a contract or another user. Blockchain allows you to track the status of each contract and the balance of each user. Smart contracts are unchangeable: after they are deployed in the core network, updates and changes are not possible, they are publicly available. The main serious problem of creating smart contracts is their formal verification: for example, in the Etherium network, verifying the decentralized virtual machine (EVM) code is very difficult, so unverified smart contracts are often the subject of hacker attacks. Later in the article, known vulnerabilities and attacks will be examined using the example of the Ethereum network and the distributed decentralized autonomous organization The Dao. 1.3 Smart contracts In this article, special attention will be paid to the security of decentralized autonomous organizations, which are based on smart contracts, examines examples of existing attacks. The problem of attacks on DAO is currently relevant, although it is currently not very well covered [2, 4]. 2. Structure of the DAO based on smart-contracts Ideally, the entire mining network performs a function call and skips or does not miss the contract, depending on the consensus reached, based on the consensus protocol. The result of the calculation is replicated through the blockchain and provides a commission for the transaction for the miners in accordance with the established interest rates. In addition to being used as a reward, the service fee also protects against denial-of- service attacks when an attacker tries to slow down the entire network by requesting time-consuming calculations. Each operation consumes a certain level of gas, the upper consumption threshold and the unit price of which are indicated in the transaction. Unused gas comes back, and if during the calculation all gas was consumed, then the process stops and all gas is lost. EVM allows contract functions to have a local state, while contracts themselves can contain global variables stored in the blockchain. Contracts can also refer to other contracts via message calls. The output of these calls is part of the same transaction and is also returned during the runtime of the transaction. It is important to note that calls can also send the Ether to other contracts and non-contractual addresses. The balance of the contract can be read by any member of the blockchain, but it can be changed only by calls originating from other contracts or initiated from outside the transaction. Only contracts with white list addresses can receive funding from the organization, and track the addition of new contract addresses, the main purpose of which is financing, curators [9]. The main motive for the introduction of human control is the screening out of malicious addresses, through which the "51% attack" is carried out, the purpose of which is to transfer most of the company's funds to one block. After adding the contract address to the white list, further decisions on it are made by voting all the holders of the tokens. At the time of voting, the balance sheets of the voters are "frozen" to the voting results. The withdrawal of funds from the organization is possible only by creating a sub-organization, where the withdrawing funds is the sole curator. The decision on separation (creation of a new DAO) is also adopted by a general vote. The entire process of creating a new DAO takes a little more than 30 days [4, 10]. 2. Structure of the DAO based on smart-contracts A decentralized autonomous organization is a supposedly "democratic" organization operating in a distributed network through a combination of "smart" contracts and a rich scripting language. Technically, DAO is the implementation of a financial service by performing all necessary calculations directly in smart contracts when using the scripting language. A distributed ledger, for example, a host, provides a secure environment for computing and storing data across the entire network, and, as a consequence, eliminates the need for a central trusted party [1]. As an example of the structure of a decentralized autonomous organization, TheDAO can be considered, where the main smart contract is used, serving as a "factory" for sub-contracts, the number of which is already in the millions. Smart contracts in Ethereum run on Ethereum Virtual Machine (EVM), the predominant language of contracts is Solidity. A smart contract is an autonomous agent with its own software logic, an identification address in the network and the associated 151 Andryukhin A.A. Methods of protecting decentralized autonomous organizations from crashes and attacks. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 149-164 balance of the Ether. After the initialization, the contract code can no longer be changed, the contract can be called repeatedly and stored on the network forever, until it executes the bytecode of the suicide instruction, after which the contract is no longer subject to a call and is called dead [7,9]. Each contract call is carried out by sending a transaction to the address of the contract together with the input data and charges (the so-called gas). Ideally, the entire mining network performs a function call and skips or does not miss the contract, depending on the consensus reached, based on the consensus protocol. The result of the calculation is replicated through the blockchain and provides a commission for the transaction for the miners in accordance with the established interest rates. balance of the Ether. After the initialization, the contract code can no longer be changed, the contract can be called repeatedly and stored on the network forever, until it executes the bytecode of the suicide instruction, after which the contract is no longer subject to a call and is called dead [7,9]. Each contract call is carried out by sending a transaction to the address of the contract together with the input data and charges (the so-called gas). 3. Vulnerabilities of DAO Despite the huge amounts of 53% of DAO funds and curatorial verification of the addresses of financed contracts, there is a possibility of cartel collusion with the aim of raising funds for interrelated projects. Attack of parallel voting. For the voting period, the balances of the voters are blocked, which can be used for voting for a malicious contract with a smaller voting period. Attack of parallel voting. For the voting period, the balances of the voters are blocked, which can be used for voting for a malicious contract with a smaller voting period. Attack on reward. To reduce the payments to the separated participants of the system, the remaining participants can deliberately create overheads for maintenance by looping money in fake contracts. Logical vulnerability of voting. The nature of voting in the existing DAO does not allow to build a logical chain during voting. For example, (vote "yes" the proposal A if the proposal B is not funded). Because social processes are non-linear, it is impossible to foresee how competing or conflicting proposals run simultaneously. Logical vulnerability of voting. The nature of voting in the existing DAO does not allow to build a logical chain during voting. For example, (vote "yes" the proposal A if the proposal B is not funded). Because social processes are non-linear, it is impossible to foresee how competing or conflicting proposals run simultaneously. Attacks that exploit the behavioral features of the system, for the most part, require tremendous resources and considerable training, while attacks based on technical vulnerabilities and bugs can be carried out with minimal costs, thus such attacks are the most interesting and dangerous. Attacks that exploit the behavioral features of the system, for the most part, require tremendous resources and considerable training, while attacks based on technical vulnerabilities and bugs can be carried out with minimal costs, thus such attacks are the most interesting and dangerous. According to the studies [5], the Ethereum blockchain contains over 34,000 vulnerable smart contracts per 1 million researched contracts. Vulnerable contracts were divided into 3 conditional groups: suicidal contracts, prodigal contracts and greedy contracts - such contracts allow either to block funds for an indefinite period, or to destroy the contract after implementation, or allow leakage means of purse to arbitrary users. There are several types [2] of major vulnerabilities that make the contract dangerous for the system. Call to the unknown. 3. Vulnerabilities of DAO Attacks of the DAO system based both on the technical imperfection of the system and on the behavioral characteristics of the DAO participants [15, 16, 17]. The behavior of participants allowed the appearance of the following types of attacks, some of which are still used for malicious actions in the system [4]. 152 Андрюхин А.А. Методы защиты децентрализованных автономных организаций от системных отказов и атак. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 149-164 Андрюхин А.А. Методы защиты децентрализованных автономных организаций от системных отказов и атак. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 149-164 Stalker Attack. During the separation and creation of a subsidiary DAO in order to withdraw funds from the system (the withdrawal is possible only under this scheme), an attacker can seize tokens created by the DAO and have a negative impact on the withdrawal of funds. p Attack of the last moment. At the last moment of voting, a large investor is added with a hug number of tokens with which he votes "yes" and pushes an unprofitable or absurd project. Attack of the value of the token. Sowing panic among tokens holders, forcing to sell tokens, and not invest in system projects. There is a buying up of tokens at a low price and the acquisition of a larger stake in the DAO. Attack of the value of the token. Sowing panic among tokens holders, forcing to sell tokens, and not invest in system projects. There is a buying up of tokens at a low price and the acquisition of a larger stake in the DAO. Attack of extra-balance. The attacker provokes the separation of DAO to increase the book value of tokens. The more participants are separated from the DAO, the higher the value of the extra balance as a percentage. Attack of extra-balance. The attacker provokes the separation of DAO to increase the book value of tokens. The more participants are separated from the DAO, the higher the value of the extra balance as a percentage. Attack of 53%. Despite the huge amounts of 53% of DAO funds and curatorial verification of the addresses of financed contracts, there is a possibility of cartel collusion with the aim of raising funds for interrelated projects. Attack of 53%.  the gas has ended; 3. Vulnerabilities of DAO When the code is illiterate, the call, send, and delegate call primitives can result in sending to an unset address or returning a broadcast by calling a backup function. ception disorder. In Solidity, exceptions are used in the following cases: 153 Andryukhin A.A. Methods of protecting decentralized autonomous organizations from crashes and attacks. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 149-164  the call stack has reached the limit;  the call stack has reached the limit;  a command throw is executed [2].  a command throw is executed [2]. However, in some cases (often these are chains of nested calls), exceptions can lead to an unplanned cancellation of the actions performed, while gas consumption is not returned [11].  a command throw is executed [2]. However, in some cases (often these are chains of nested calls), exceptions can lead to an unplanned cancellation of the actions performed, while gas consumption is not returned [11]. Gasless send. The lack of gas in the transmission of Ether can cause unpredictable behavior. Type casts. Using the compiler does not guarantee the correct operation of the contract. Reentrancy. It can often be confusing to realize that if a function is not recursive, then it will not allow repeated repetitions. However, this misconception can lead to the fact that a non-recursive function starts a cycle of calls that ultimately consume all the gas [11]. Keeping secret. Fields in contracts can be both private and public for all users. However, declaring a field private does not guarantee its inaccessibility to others. This is due to the fact that to set the privacy of the field, the user must send the corresponding transaction to the miners who will then publish it in the blockchain. Since the blockchain itself is public, any user can check the contents of the transaction and make changes to the privacy of the field. In order to best protect the contract field, you need to use suitable cryptographic methods [12]. Immutable bugs. As already known, after the publication of the contract in the detachment, it is already impossible to change it, so contracts with errors can manifest themselves completely unpredictably. Sometimes, when the consequences of executing such contracts have an extremely negative impact on the entire detachment, the community comes to the decision to use softfork or hardfork. y f f f Ether lost in transfer. Some addresses in the blockchain are not associated with either specific users or contracts, so when sending airtime to these addresses, it is lost irrevocably. y Stack size limit. The stack size is limited to 1024 frames. Every time there is a call to another contract or even yourself, the stack size increases by 1.  the call stack has reached the limit; The vulnerability lies in the fact that an attacker can try to create his own block with the content controlled by him in order to evade the result of the generator and shift the probability of distribution of pseudo-random numbers. Time constraints. Time constraints are used to identify permitted or mandatory actions and contain a timestamp that is consistent with all participants in the process. Contracts can extract timestamps and set their own. Attackers can exploit this vulnerability to gain temporary advantages over other participants in the process. The Threat of Quantum Computing. One of the potential vulnerabilities is the instability of cryptography to quantum attacks. The most popular public-key encryption algorithms, for example, RSA in the near future can be destroyed with the help of a quantum computer.  the call stack has reached the limit; If the rules for rejecting a call when reaching a stack limit are incorrectly set, then the attacker has the opportunity to exploit the vulnerability. The vulnerability was closed in 2016 by limiting gas at a rate of 63/64 from the existing one. Since the current gas limit is limited to 4.7M units, the depth of the stack is always less than 1024. Unpredictable state. When sending a transaction to the network, the user can not always be sure of the status of the contract, which is determined by the cost of its fields and balance. This can happen because at the time of sending the contract status was changed by another transaction, or the contract contains dynamic variables associated with other contracts. Such vulnerability can be used by attackers to link the called contract to malicious components that allow stealing the broadcast. Stack size limit. The stack size is limited to 1024 frames. Every time there is a call to another contract or even yourself, the stack size increases by 1. If the rules for rejecting a call when reaching a stack limit are incorrectly set, then the attacker has the opportunity to exploit the vulnerability. The vulnerability was closed in 2016 by limiting gas at a rate of 63/64 from the existing one. Since the current gas limit is limited to 4.7M units, the depth of the stack is always less than 1024. Unpredictable state. When sending a transaction to the network, the user can not always be sure of the status of the contract, which is determined by the cost of its fields and balance. This can happen because at the time of sending the contract status was changed by another transaction, or the contract contains dynamic variables associated with other contracts. Such vulnerability can be used by attackers to link the called contract to malicious components that allow stealing the broadcast. Generating randomness. Due to the fact that execution of the bytecode on EVM is deterministic, i.e. all participants as a result of processing the transaction should 154 Андрюхин А.А. Методы защиты децентрализованных автономных организаций от системных отказов и атак. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 149-164 receive the same result, unless otherwise specified, to obtain non-deterministic results, some contracts (for example, games, lotteries) use pseudo-random number generators. Such blocks usually have timestamps. 4. Levels of attacks on smart-contracts In connection with the fact that the basis of any decentralized autonomous organization is the implementation of smart contracts, the main attacks are aimed at them. The existing vulnerabilities of smart contracts can be conditionally divided into three classes, depending on the level at which the vulnerability is detected (Solidity, EVM bytecode, blockchain). Each vulnerability class can spawn one or more known attack types [2, 16, 17] (fig. 1). In the study [2], the simplest test DAO was simulated, on which the following attacks, existing in real Ethereum, were made. on which the following attacks, existing in real Ethereum, were made. The DAO Attack. In the well-known attacks on the DAO, the purpose of which was to seize the organization's funds, the call to the unknown and reentrancy vulnerabilities were exploited, which could have a negative impact, because the broadcast was broadcast before the credit was reduced. Examples of contracts used in attacks: g , g , The DAO Attack. In the well-known attacks on the DAO, the purpose of which was to seize the organization's funds, the call to the unknown and reentrancy vulnerabilities were exploited, which could have a negative impact, because the broadcast was broadcast before the credit was reduced. Examples of contracts used in attacks: O Attack. In the well-known attacks on the DAO, the purpose of which w 155 Andryukhin A.A. Methods of protecting decentralized autonomous organizations from crashes and attacks. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 149-164 Andryukhin A.A. Methods of protecting decentralized autonomous organizations from crashes and attacks. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 149-164 Andryukhin A.A. Methods of protecting decentralized autonomous organizations from crashes and attacks. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 149-164 Andryukhin A.A. Methods of protecting decentralized autonomous organizations from crashes and attacks. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 149-164 Vulnerability Solidity EVM Blockchain Attack Contracts G Gasless send Contracts G P Exception disorders Contracts P Reentrancy Contracts G P S Immutable bugs Contracts P D Ether lost in transfer Contracts G P S Stack size limit Contracts G P S Unpredictable stack Contracts P G i Contracts G P S Time constraints The Dao attack (1) King of the ether throne (2) Multi-player games (3) Rubixi (4) Goverrnmental (5) Dynamic libraries (6) G Greedy Contracts G P S Contracts G P S Type casts Keeping secrets Contracts P Call to the unknown Vulnerability Solidity EVM Blockchain Attack Contracts G Gasless send Contracts G P Exception disorders Contracts P Reentrancy Contracts G P S Immutable bugs Contracts P D Ether lost in transfer Contracts G P S Stack size limit Contracts G P S Unpredictable stack Contracts P Generating randomness Contracts G P S Time constraints The Dao attack (1) King of the ether throne (2) Multi-player games (3) Rubixi (4) Goverrnmental (5) Dynamic libraries (6) G - Greedy P - Prodigal S - Suicidal D - Dead Contracts G P S Contracts G P S Type casts Keeping secrets Contracts P Call to the unknown Fig. 1. Vulnerabilities of smart contracts Fig. 1. Vulnerabilities of smart contracts 156 Андрюхин А.А. Методы защиты децентрализованных автономных организаций от системных отказов и атак. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 149-164 Attack in the game King of Ether throne. The game is represented by simplified contracts with vulnerabilities. At first glance, contracts seem fair, but the lack of a send return check (1) and intentional call exceptions (2) can result both in unfair winnings and in theft of contract funds after the game is over. Attack in games with multiple players. In such games, hidden fields are often used, which are unknown during the game, but can be opened at the time of joining the game (vulnerability keeping secrets). Andryukhin A.A. Methods of protecting decentralized autonomous organizations from crashes and attacks. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 149-164 An example of a similar game with existing vulnerabilities is represented by a contract Using data from private fields, an attacker can lead a strategy of permanent winnings. Using data from private fields, an attacker can lead a strategy of permanent winnings. Attack of Rubixy. Was implemented in contracts that use the Ponzi scheme (financial pyramid). Attack was possible because the developers renamed the contract with DynamicPyramid Rubixy, forgetting to change the name of the constructor, which then became a function that everyone can call. Instead of a single use of DynamicPyramid when setting the owner's address, which is allowed to take profit, this function was used by intruders to set their addresses as owner addresses. 157 Andryukhin A.A. Methods of protecting decentralized autonomous organizations from crashes and attacks. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 149-164 Andryukhin A.A. Methods of protecting decentralized autonomous organizations from crashes and attacks. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 149-164 Attack GovernMental. As well as above, the contract is implemented by the Ponzi scheme. Money receives the final invested after 12 hours except for the fees of the organizers. After that, the array is cleared with the data of the participants. At some point, the list became so huge that it took much more gas to clean up the arrays than the maximum allowed for a single transaction. A simplified version of the game with all the existing vulnerabilities looks This scheme was also subjected to attacks using the exception disorder and stack size limit vulnerabilities. Thanks to these vulnerabilities, it became possible not to pay the winners to win by launching a new round of the game. Also, dishonest miners used the possibility of not adding new blocks to be the last ones invested, or adding a timestamp to the block in such a way that the block would become the last one each time. This scheme was also subjected to attacks using the exception disorder and stack size limit vulnerabilities. Thanks to these vulnerabilities, it became possible not to pay the winners to win by launching a new round of the game. Also, dishonest miners used the possibility of not adding new blocks to be the last ones invested, or adding a timestamp to the block in such a way that the block would become the last one each time. Andryukhin A.A. Methods of protecting decentralized autonomous organizations from crashes and attacks. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 149-164 An attack using dynamic libraries. Such attacks use the Unpredictable state vulnerability, since it is possible to update the library with malicious content after the publication of the contract. 5. Potential mitigations and solutions transparent) block is a simple protocol that relies solely on hashes and is better protected of quantum computers, because it does not use elliptic curves;  evidence with zero disclosure – the preservation of confidential information and at the same time the certification of its availability; it can be achieved by authentication of the "call-response" to verify the transaction, using the zkSNARK (zk-zero-knowledge, SNARK- Succinct Non-interactive Adaptive Argument of Knowledge) module for verification; it will make certain contract variables private, ensuring that they are stored out of the blockchain by users who using the SNARK protocol to prove that they adhere to its rules (requires a prior "trust"); the use of the zkSTARK (T-transparent, i.e. transparent) block is a simple protocol that relies solely on hashes and is better protected of quantum computers, because it does not use elliptic curves;  use of obfuscation (code entanglement);  use of oracles - parties transferring information between smart contracts and external data sources;  use of oracles - parties transferring information between smart contracts and external data sources;  use of the trusted environment for program execution.  use of the trusted environment for program execution. Verification of smart contracts. The development of tools for verification, the introduction of verification formats will make sure that the smart contract behaves the way it was intended. The complexity of verification of smart contracts lies in the complexity of verification of the EVM bytecode. Verification of smart contracts will also reduce the risk of virus infection and hacker attacks. Verification guarantees greater accuracy, than traditional approaches, for example, testing [8]. g g y pp p g Perfection of intra-organizational processes. Improving the voting processes by introducing a grace period that allows the movement of non-voting funds, adding the function of the voting office, prolonging the voting time for a statistical release, attracting more users to the process, developing secure withdrawal methods will prevent a number of attacks and increase trust in the system. Perfection of intra-organizational processes. Improving the voting processes by introducing a grace period that allows the movement of non-voting funds, adding the function of the voting office, prolonging the voting time for a statistical release, attracting more users to the process, developing secure withdrawal methods will prevent a number of attacks and increase trust in the system. Improving the mechanism for achieving consensus. 5. Potential mitigations and solutions Having considered the above vulnerabilities and attacks based on them, it is possible to draw conclusions and understand the need for steps to be taken in the field of DAO security. Confidentiality. Many mistakenly accept conditional anonymity of transactions in the blockchain for real: transactions are recorded and stored in the public registry and are linked to the address of the account that does not contain information about the real person behind this account. However, identifying information can be 158 Андрюхин А.А. Методы защиты децентрализованных автономных организаций от системных отказов и атак. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 149-164 obtained through web trackers and cookies. In addition, the required data is often contained directly in smart contracts [17]. Lack of confidentiality is a serious threat when it comes to medical records, identity documents, credential management, and a number of closed financial documents. Strengthen confidentiality in several ways:  addresses on the Diffie-Hellman-Merkle protocol on elliptical curves (ECDHM) will allow the use of the secret key by the two sides of the process;  creation of a decentralized mix-protocol for joining a group of payments into one pool, with the possibility of tracking amounts in a private registry, without a third party;  evidence with zero disclosure – the preservation of confidential information and at the same time the certification of its availability; it can be achieved by authentication of the "call-response" to verify the transaction, using the zkSNARK (zk-zero-knowledge, SNARK- Succinct Non-interactive Adaptive Argument of Knowledge) module for verification; it will make certain contract variables private, ensuring that they are stored out of the blockchain by users who using the SNARK protocol to prove that they adhere to its rules (requires a prior "trust"); the use of the zkSTARK (T-transparent, i.e. 5. Potential mitigations and solutions The use of the existing PoW (Proof-of-Work) protocol, which depends on computing power, jeopardizes the decentralization of the system and makes it possible for a cartel plot. Because Major mining pools have a great advantage over private miners in the extraction of blocks and profit distribution, centralization occurs in the blockage and several large mining pools own more than 70% of the hash. A more advanced PoS protocol 159 Andryukhin A.A. Methods of protecting decentralized autonomous organizations from crashes and attacks. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 149-164 Andryukhin A.A. Methods of protecting decentralized autonomous organizations from crashes and attacks. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 149-164 (Proof-of-Stake) practically excludes the possibility of bundle aggregations in terms of computing power, but at the same time it requires solutions to problems such as «Nothing at stake», when forming forecaster, the miners vote simultaneously for several different blocks on one altitude, or start fork from any place, getting validators of previous participants and creating a million blocks in a new blockchain, preventing users from understanding which of the blockchain is «correct». Creation of the necessary tools for development. Creation of the necessary tools for development. At the moment, in the ecosystem of the toolkit of the developer of smart contracts, the weak points are: Creation of the necessary tools for development. At the moment, in the ecosystem of the toolkit of the developer of smart contracts, the weak points are:  Integrated Development Environment (IDE);  the code assembly system and compiler program;  deployment tools;  storage medium;  debugging and logging tools;  security audit;  Integrated Development Environment (IDE);  Integrated Development Environment (IDE);  the code assembly system and compiler program;  deployment tools;  storage medium; d b i d l i l  the code assembly system and compiler program;  debugging and logging tools;  security audit;  analytical tools. Improving the development toolkit will have a positive impact on the functioning of the entire DAO. Improving the development toolkit will have a positive impact on the functioning of the entire DAO. References [1]. Williams J. The Seconomics (Security-Economics) Vulnerabilities of Decentralized Autonomous Organizations. Lecture Notes in Computer Science, vol. 10476, 2017, pp. 171-179. [1]. Williams J. The Seconomics (Security-Economics) Vulnerabilities of Decentralized Autonomous Organizations. Lecture Notes in Computer Science, vol. 10476, 2017, pp. 171-179. [2]. Atzei N., Bartoletti M., Cimoli T. A survey of attacks on Ethereum smart contracts (SoK). In Proc. of the International Conference on Principles of Security and Trust, 2017, pp. 164-186. [3]. Mehar M. et al. Understanding a Revolutionary and Flawed Grand Experiment in Blockchain: The DAO Attack. Available at SSRN: https://ssrn.com/abstract=3014782, accessed 29.05.2018. [4]. DuPont Q. Experiments in algorithmic governance: A history and ethnography of “The DAO,” a failed decentralized autonomous organization. In Bitcoin and Beyond: Cryptocurrencies, Blockchains and Global Governance, Routledge, 2017, 212 p. [5]. Nikolic I. et al. Finding The Greedy, Prodigal, and Suicidal Contracts at Scale. arXiv preprint arXiv:1802.06038, 2018. [6]. Grossman S. et al. Online detection of effectively callback free objects with applications to smart contracts. Proceedings of the ACM on Programming Languages, vol. 2, issue POPL, article 48, 2017, 20 p. [7]. Gurfinkel A. et al. The SeaHorn verification framework. In Proc. of the International Conference on Computer Aided Verification, 2015, pp. 343-361. [8]. Bhargavan K. et al. Formal verification of smart contracts. In Proc. of the ACM Workshop on Programming Languages and Analysis for Security, 2016, pp. 91-96. [9]. Delmolino K. et al. Step by step towards creating a safe smart contract: Lessons and insights from a cryptocurrency lab. In Proc. of the International Conference on Financial Cryptography and Data Security, 2016, pp. 79-94. 160 Андрюхин А.А. Методы защиты децентрализованных автономных организаций от системных отказов и атак. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 149-164 [10]. Wüst K., Gervais A. Ethereum Eclipse Attacks. Report, ETH Zurich Research Collection, 2016, 7 p. [11]. Chen T. et al. An Adaptive Gas Cost Mechanism for Ethereum to Defend Against Under-Priced DoS Attacks. In Proc. of the International Conference on Information Security Practice and Experience, 2017, pp. 3-24. [12]. Luu L. et al. Making smart contracts smarter. In Proc. of the ACM SIGSAC Conference on Computer and Communications Security, 2016, pp. 254-269. [13]. Dhillon V., Metcalf D., Hooper M. The DAO Hacked. In Blockchain Enabled Applications, Apress. Berkeley, CA, 2017, pp. 67-78. [14]. Mayer H. ECDSA security in bitcoin and ethereum: a research survey. CoinFabrik, 2016. Available at https://blog.coinfabrik.com/wp-content/uploads/2016/06/ECDSA- Security-in-Bitcoin-and-Ethereum-a-Research-Survey.pdf, accessed 29.05.2018. References [15]. Marcus Y., Heilman E., Goldberg S. Low-Resource Eclipse Attacks on Ethereum’s Peer-to-Peer Network. IACR Cryptology ePrint Archive, Available at https://eprint.iacr.org/2018/236.pdf, accessed 29.05.2018. [16]. Dika A. Ethereum Smart Contracts: Security Vulnerabilities and Security Tools, Master’s thesis, NTNU, 2017. [17]. Wöhrer M., Zdun U. Smart Contracts: Security Patterns in the Ethereum Ecosystem and Solidity. In Proc. of the International Workshop on Blockchain Oriented Software Engineering (IWBOSE), 2018, 8 p. [18]. Biryukov A., Khovratovich D., Tikhomirov S. Findel: Secure Derivative Contracts for Ethereum. In Proc. of the International Conference on Financial Cryptography and Data Security, 2017, pp. 453-467. [19]. Ross S. A. The economic theory of agency: The principal's problem. The American Economic Review, vol. 63, №. 2, 1973, pp. 134-139. [20]. Eisenhardt K. M. Agency theory: An assessment and review. Academy of management review, vol. 14, № 1, 1989, pp, 57-74. [21]. Gale D., Hellwig M. Incentive-compatible debt contracts: The one-period problem. The Review of Economic Studies, vol. 52, №. 4, 1985, pp. 647-663. [22]. Bolton P., Dewatripont M. Contract theory. MIT press, 2005, 744 p. [23]. Edelman B., Ostrovsky M., Schwarz M. Internet advertising and the generalized second- price auction: Selling billions of dollars’ worth of keywords. American economic review, vol. 97, №. 1, 2007, pp. 242-259. [24]. Roth A. E., Ockenfels A. Last-minute bidding and the rules for ending second-price auctions: Evidence from eBay and Amazon auctions on the Internet. American economic review, vol. 92, №. 4, 2002, pp. 1093-1103. [25]. Greenstein S. How the internet became commercial: Innovation, privatization, and the birth of a new network. Princeton University Press, 2015, 488 p. [26]. Moeen M., Agarwal R. Incubation of an industry: Heterogeneous knowledge bases and modes of value capture. Strategic Management Journal, vol. 38, №. 3, 2017, pp. 566- 587. [27]. Handy C. Trust and the virtual organization. Harvard business review, vol. 73, №. 3, 1995, pp. 40-51. [28]. Markus M. L., Agres B. M. C. E. What makes a virtual organization work? MIT Sloan Management Review, vol. 42, №. 1. 2000, 16 p. [29]. Szabo N. The idea of smart contracts. Nick Szabo’s Papers and Concise Tutorials. Available at 161 http://www.fon.hum.uva.nl/rob/Courses/InformationInSpeech/CDROM/Literature/LOT winterschool2006/szabo.best.vwh.net/smart_contracts_idea.html, accessed 29.05.2018. http://www.fon.hum.uva.nl/rob/Courses/InformationInSpeech/CDROM/Literature/LOT winterschool2006/szabo.best.vwh.net/smart_contracts_idea.html, accessed 29.05.2018. [30]. Nakamoto S. Bitcoin: A peer-to-peer electronic cash system. 2008. Available at https://bitcoin.org/bitcoin.pdf, accessed 29.05.2018. [31]. Haber S., Stornetta W. S. How to time-stamp a digital document. In Proc. Методы защиты децентрализованных автономных организаций от системных отказов и атак А.А. Андрюхин <Alexandr@kcdigital.ru> А.А. Андрюхин <Alexandr@kcdigital.ru> ООО "КЕЙСИДИ", 129226, г Москва, проспект Мира, 131, офис 3 А.А. Андрюхин <Alexandr@kcdigital.ru> ООО "КЕЙСИДИ", 129226, г Москва, проспект Мира, 131, офис 3 Аннотация. В статье обсуждаются технология блокчейнов, децентрализованные автономные организации, смарт-контракты и их устойчивость к атакам и сбоям. Из-за того, что такая форма организаций является экспериментальной, их участники часто сталкиваются с проблемами атак на организацию, последствиями неправильно написанных правил и мошенничества. Задача создания децентрализованных автономных организаций, которые устойчивы к отказам и атакам, и исследование причин этих проблем стало актуальным для проектировщиков и разработчиков программного обеспечения. В статье исследуются алгоритмы атак и предлагаются методы обеспечения устойчивости децентрализованных автономных организаций для атак на основе анализа подпроцессов пограничных событий и журналов с использованием методов Process Mining. Методы, которые необходимо разработать, должны оперативно выявлять и предотвращать несоответствия между предполагаемым и фактическим поведением смарт-контрактов, которые приводят к таким ошибкам в функционировании, как пустые транзакции, увеличенное время обработки блоков и т. д. Ключевые слова: блокчейн; децентрализованные автономные организации; анализ процессов; смарт-контракт; security DOI: 10.15514/ISPRAS-2018-30(3)-11 References of the Conference on the Theory and Application of Cryptography, 1990, pp. 437-455. [32]. Massias H., Avila X. S., Quisquater J. J. Design of a secure timestamping service with minimal trust requirement. In Proc. of the 20th Symposium on Information Theory in the Benelux, 1999, pp. 79-86. [33]. Merkle R. C. Protocols for public key cryptosystems. In Proc. of the IEEE Symposium on Security and Privacy, 1980, pp. 122-122. y y pp [34]. Katz J. et al. Handbook of applied cryptography. CRC press, 1996, 810 p. Ö [35]. Özsu M. T., Valduriez P. Principles of distributed database systems. Springer Science & Business Media, 2011, 846 p. [36]. Bernstein P. A., Hadzilacos V., Goodman N. Concurrency control and recovery in database systems. 1987. Available at https://www.microsoft.com/en-us/research/wp- content/uploads/2016/05/ccontrol.zip, accessed 29.05.2018. Для цитирования: Андрюхин А.А. Методы защиты децентрализованных автономных организаций от системных отказов и атак. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 149-164 (на английском языке). DOI: 10.15514/ISPRAS-2018-30(3)-11 DOI: 10.15514/ISPRAS-2018-30(3)-11 Для цитирования: Андрюхин А.А. Методы защиты децентрализованных автономных организаций от системных отказов и атак. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 149-164 (на английском языке). DOI: 10.15514/ISPRAS-2018-30(3)-11 162 Андрюхин А.А. Методы защиты децентрализованных автономных организаций от системных отказов и атак. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 149-164 Список литературы [1]. Williams J. The Seconomics (Security-Economics) Vulnerabilities of Decentralized Autonomous Organizations. Lecture Notes in Computer Science, vol. 10476, 2017, pp. 171-179. [2]. Atzei N., Bartoletti M., Cimoli T. A survey of attacks on Ethereum smart contracts (SoK). In Proc. of the International Conference on Principles of Security and Trust, 2017, pp. 164-186. [3]. Mehar M. et al. Understanding a Revolutionary and Flawed Grand Experiment in Blockchain: The DAO Attack. Available at SSRN: https://ssrn.com/abstract=3014782, accessed 29.05.2018. [4]. DuPont Q. Experiments in algorithmic governance: A history and ethnography of “The DAO,” a failed decentralized autonomous organization. In Bitcoin and Beyond: Cryptocurrencies, Blockchains and Global Governance, Routledge, 2017, 212 p. [5]. Nikolic I. et al. Finding The Greedy, Prodigal, and Suicidal Contracts at Scale. arXiv preprint arXiv:1802.06038, 2018. [6]. Grossman S. et al. Online detection of effectively callback free objects with applications to smart contracts. Proceedings of the ACM on Programming Languages, vol. 2, issue POPL, article 48, 2017, 20 p. p [7]. Gurfinkel A. et al. The SeaHorn verification framework. In Proc. of the International Conference on Computer Aided Verification, 2015, pp. 343-361. [8]. Bhargavan K. et al. Formal verification of smart contracts. In Proc. of the ACM Workshop on Programming Languages and Analysis for Security, 2016, pp. 91-96. [9]. Delmolino K. et al. Step by step towards creating a safe smart contract: Lessons and insights from a cryptocurrency lab. In Proc. of the International Conference on Financial Cryptography and Data Security, 2016, pp. 79-94. yp g p y y pp [10]. Wüst K., Gervais A. Ethereum Eclipse Attacks. Report, ETH Zurich Research Collection, 2016, 7 p. [11]. Chen T. et al. An Adaptive Gas Cost Mechanism for Ethereum to Defend Against Under-Priced DoS Attacks. In Proc. of the International Conference on Information Security Practice and Experience, 2017, pp. 3-24. [12]. Luu L. et al. Making smart contracts smarter. In Proc. of the ACM SIGSAC Conference on Computer and Communications Security, 2016, pp. 254-269. [13]. Dhillon V., Metcalf D., Hooper M. The DAO Hacked. In Blockchain Enabled Applications, Apress. Berkeley, CA, 2017, pp. 67-78. [14]. Mayer H. ECDSA security in bitcoin and ethereum: a research survey. CoinFabrik, 2016. Режим доступа: https://blog.coinfabrik.com/wp- content/uploads/2016/06/ECDSA-Security-in-Bitcoin-and-Ethereum-a-Research- Survey.pdf, дата обращения 29.05.2018. [15]. Marcus Y., Heilman E., Goldberg S. Low-Resource Eclipse Attacks on Ethereum’s Peer-to-Peer Network. IACR Cryptology ePrint Archive, Режим доступа https://eprint.iacr.org/2018/236.pdf, дата обращения 29.05.2018. p p g p [16]. Dika A. Список литературы Ethereum Smart Contracts: Security Vulnerabilities and Security Tools, Master’s thesis, NTNU, 2017. [17]. Wöhrer M., Zdun U. Smart Contracts: Security Patterns in the Ethereum Ecosystem and Solidity. In Proc. of the International Workshop on Blockchain Oriented Software Engineering (IWBOSE), 2018, 8 p. 163 Andryukhin A.A. Methods of protecting decentralized autonomous organizations from crashes and attacks. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 149-164 Andryukhin A.A. Methods of protecting decentralized autonomous organizations from crashes and attacks. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 149-164 [18]. Biryukov A., Khovratovich D., Tikhomirov S. Findel: Secure Derivative Contracts for Ethereum. In Proc. of the International Conference on Financial Cryptography and Data Security, 2017, pp. 453-467. y pp [19]. Ross S. A. The economic theory of agency: The principal's problem. The American Economic Review, vol. 63, №. 2, 1973, pp. 134-139. [20]. Eisenhardt K. M. Agency theory: An assessment and review. Academy of management review, vol. 14, № 1, 1989, pp, 57-74. [21]. Gale D., Hellwig M. Incentive-compatible debt contracts: The one-period problem. The Review of Economic Studies, vol. 52, №. 4, 1985, pp. 647-663. [22]. Bolton P., Dewatripont M. Contract theory. MIT press, 2005, 744 p. [23]. Edelman B., Ostrovsky M., Schwarz M. Internet advertising and the generalized second- price auction: Selling billions of dollars’ worth of keywords. American economic review, vol. 97, №. 1, 2007, pp. 242-259. [24]. Roth A. E., Ockenfels A. Last-minute bidding and the rules for ending second-price auctions: Evidence from eBay and Amazon auctions on the Internet. American economic review, vol. 92, №. 4, 2002, pp. 1093-1103. pp [25]. Greenstein S. How the internet became commercial: Innovation, privatization, and the birth of a new network. Princeton University Press, 2015, 488 p. [26]. Moeen M., Agarwal R. Incubation of an industry: Heterogeneous knowledge bases and modes of value capture. Strategic Management Journal, vol. 38, №. 3, 2017, pp. 566- 587. [27]. Handy C. Trust and the virtual organization. Harvard business review, vol. 73, №. 3, 1995, pp. 40-51. [28]. Markus M. L., Agres B. M. C. E. What makes a virtual organization work? MIT Sloan Management Review, vol. 42, №. 1. 2000, 16 p. [29]. Szabo N. The idea of smart contracts. Nick Szabo’s Papers and Concise Tutorials. Режим доступа: http://www.fon.hum.uva.nl/rob/Courses/InformationInSpeech/CDROM/Literature/LOT winterschool2006/szabo.best.vwh.net/smart_contracts_idea.html, дата обращения 29.05.2018. [30]. Nakamoto S. Bitcoin: A peer-to-peer electronic cash system. 2008. Режим доступа https://bitcoin.org/bitcoin.pdf, дата обращения 29.05.2018. [31]. DOI: 10.15514/ISPRAS-2018-30(3)-12 For citation: Prokopev S.E. Cryptographic Stack Machine Notation One. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 165-182. DOI: 10.15514/ISPRAS-2018- 30(3)-12 For citation: Prokopev S.E. Cryptographic Stack Machine Notation One. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 165-182. DOI: 10.15514/ISPRAS-2018- 30(3)-12 Cryptographic Stack Machine Notation One S.E. Prokopev <s.e.pr@mail.ru> Independent researcher Moscow, Russia Abstract. A worthy cryptographic protocol specification has to be human-readable (declara- tive and concise), executable and formally verified in a sound model. Keeping in mind these requirements, we present a protocol message definition notation named CMN.1, which is based on an abstraction named cryptographic stack machine. The paper presents the syntax and semantics of CMN.1 and the principles of implementation of the CMN.1-based executa- ble protocol specification language. The core language library (the engine) performs all the message processing, whereas a specification should only provide the declarative definitions of the messages. If an outcoming message must be formed, the engine takes the CMN.1 defi- nition as input and produces the binary data in consistency with it. When an incoming mes- sage is received, the engine verifies the binary data with respect to the given CMN.1 defini- tion memorizing all the information needed in the further actions. The verification is com- plete: the engine decrypts the ciphertexts, checks the message authentication codes and signa- tures, etc. Currently, the author's proof-of-concept implementation of the language (embedded in Haskell) can translate a CMN.1-based specifications both to the interoperable implementa- tions and to the programs for the ProVerif protocol analyzer. The excerpts from the CMN.1-based TLS protocol specification and corresponding automatically generated ProVer- if program are provided as an illustration. Keywords: cryptographic stack machine; cryptographic protocol message notation; executable cryptographic protocol specification languages; embedded domain-specific languages; Haskell; ProVerif; TLS. Список литературы Haber S., Stornetta W. S. How to time-stamp a digital document. In Proc. of the Conference on the Theory and Application of Cryptography, 1990, pp. 437-455. [32]. Massias H., Avila X. S., Quisquater J. J. Design of a secure timestamping service with minimal trust requirement. In Proc. of the 20th Symposium on Information Theory in the Benelux, 1999, pp. 79-86. [33]. Merkle R. C. Protocols for public key cryptosystems. In Proc. of the IEEE Symposium on Security and Privacy, 1980, pp. 122-122. y y pp [34]. Katz J. et al. Handbook of applied cryptography. CRC press, 1996, 810 p. Ö [35]. Özsu M. T., Valduriez P. Principles of distributed database systems. Springer Science & Business Media, 2011, 846 p. [36]. Bernstein P. A., Hadzilacos V., Goodman N. Concurrency control and recovery in database systems. 1987. Режим доступа https://www.microsoft.com/en-us/research/wp- content/uploads/2016/05/ccontrol.zip, дата обращения 29.05.2018. 164 2. Related work There exist many formal notations for data structures: ASN.1, JSON, etc. These notations are often provided with the engines, which can automatically generate the binary data using the provided data structure definition and, in the opposite direction, automatically unpack the binary data in accordance with the definition. Such projects as CSN.1 [12], TSN.1 [13], BinPAC [14], NetPDL [15] are targeted specifically at the network protocols. While the readability of some of these notations can be suitable, their expressiveness (in the domain of cryptographic protocols) does not. We need to have behind the notation not simply a message generator/parser waiting to be embedded to some bigger program, but a generic cryptographic protocol implementation waiting for (semi-)declarative specification to adjust to specific case. Therefore, the primary challenge is to find such powerful underlying abstraction, whereas the notation would have to be naturally emerged from it. 1. Introduction The establishment of good soundness relations between cryptographic protocol implementations and their formal models is a popular research area. The existing approaches differ by the starting point of development (implementation first [1-6] or formal model first [7-9]), by the degree of cryptographic soundness of the models (symbolic [10] or computational [9]), by the presence of the formal proof of the soundness of the model-to-implementation (or vice verse) translation procedure, by implementation usability area and by other aspects. 165 Prokopev S.E. Cryptographic Stack Machine Notation One. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 165-182 E. Cryptographic Stack Machine Notation One. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp Prokopev S.E. Cryptographic Stack Machine Notation One. Trudy ISP RAN/Proc. ISP RAS, vol. 30, iss 165-182 Our aim is to soundly tie not two (implementation and formal model) but three elements of the protocol development process: implementation, formal model and specification. By the latter, we mean a human-readable protocol description that is usually placed in RFC. The models' languages, which are based on logics or special versions of general-purpose programming languages, are not quite suitable for this task: they are either not convenient for capturing the low-level details or are firmly imperative. Therefore, our goal is a declarative specification language that could be directly used in the RFCs to considerably enhance the degree of formalization of these documents. Yet, the specification must be automatically translatable both to the interoperable implementation and to the programs for the state-of-the-art protocol model analyzers such as ProVerif [10] and Tamarin [11]. 3 Cryptographic Stack Machine Notation One We propose an abstraction named cryptographic stack machine (abbreviated as CSM), which is a stack machine specifically tailored to the needs of cryptographic protocols. Within the proposed approach, the message definition is in fact a sequence of the CSM instructions. The instructions set is divided into "bare-metal" and "sugared" parts. The "sugared" instructions make the message definitions (which in their essence are imperative) looking declarative. The instructions set may be expanded if needed. To reflect the fact that the declarative style of the protocol message definitions is one of the main targets, we name our notation «Cryptographic Stack Machine Notation One» (abbreviated as CMN.1) adopting the naming style of the ASN.1, CSN.1 and TSN.1 notations. 166 Прокопьев С.Е. Нотация криптографической стековой машины версии один. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 165-182 Е. Нотация криптографической стековой машины версии один. Труды ИСП РАН, том 30, вып. 3, 5 182 Прокопьев С.Е. Нотация криптографической стековой машины версии один. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 165-182 3.1 CMN.1 syntax Prog ::= "["{Instr,}[Instr]"]" Instr ::= BareMetal \| Sugared Prog ::= "["{Instr,}[Instr]"]" Instr ::= BareMetal \| Sugared Instr ::= BareMetal \| Sugared BareMetal ::= Const Word8List \| Var VarName Role VarType \| V VarName \| SEnc' SEncAlg \| Enco' EncoAlg \| Xor' Int \| ModAdd' \| ModMult' \| ModInv' \| Add' Integer \| Rev RFun \| Hash' HashAlg \| Pad' Int Word8List \| Mod' \| ModExp' \| Take' IntList \| Split' IntList \| SplitE' Int \| ECMult' \| ECAdd' \| C' \| CE' \| Len' LenHdr \| InsertTo Int \| PickFrom Int \| Dup Int \| Free Int \| Elem Int Prog \| Map' Prog Int Int \| Sort' Int Int \| SA' Int Int Prog \| Select' CaseList \| M Prog \| L Int Inst Instr ::= BareMetal \| Sugared BareMetal ::= Const Word8List \| Var VarName Role VarType \| V VarName \| SEnc' SEncAlg \| Enco' EncoAlg \| Xor' Int \| ModAdd' \| ModMult' \| ModInv' \| Add' Integer \| Rev RFun \| Hash' HashAlg \| Pad' Int Word8List \| Mod' \| ModExp' \| Take' IntList \| Split' IntList \| SplitE' Int \| ECMult' \| ECAdd' \| C' \| CE' \| Len' LenHdr \| InsertTo Int \| PickFrom Int \| Dup Int \| Free Int \| Elem Int Prog \| Map' Prog Int Int \| Sort' Int Int \| SA' Int Int Prog \| Select' CaseList \| M Prog \| L Int Inst BareMetal ::= Const Word8List \| Var VarName Role VarType \| V VarName \| SEnc' SEncAlg \| Enco' EncoAlg \| Xor' Int \| ModAdd' \| ModMult' \| ModInv' \| Add' Integer \| Rev RFun \| Hash' HashAlg \| Pad' Int Word8List \| Mod' \| ModExp' \| Take' IntList \| Split' IntList \| SplitE' Int \| ECMult' \| ECAdd' \| C' \| CE' \| Len' LenHdr \| InsertTo Int \| PickFrom Int \| Dup Int \| Free Int \| Elem Int Prog \| Map' Prog Int Int \| Sort' Int Int \| SA' Int Int Prog \| Select' CaseList \| M Prog \| L Int Inst Sugared ::= C Prog \| CE Prog \| Hash HashAlg Prog \| SEnc SEncAlg Prog \| Enco EncoAlg Prog \| Mod Prog \| ModAdd Prog \| ModMult Prog \| ModExp Prog \| ModInv Prog \| ECMult Prog \| ECAdd Prog \| Len LenHdr Prog \| Xor Prog \| Add Integer Prog \| Take IntList Prog \| Split IntList Prog \| SplitE Int Prog \| Pad Int Word8List Prog \| Map Prog Int Prog \| Sort Int Prog \| Select Inst CaseList \| SA Prog \| WithLen LenHdr Prog \| VarL Int VarName Role VarType \| VL Int VarName \| SelectV VarName CaseList Sugared ::= C Prog \| CE Prog \| Hash HashAlg Prog \| SEnc SEncAlg Prog \| Enco EncoAlg Prog \| Mod Prog \| ModAdd Prog \| ModMult Prog \| ModExp Prog \| ModInv Prog \| ECMult Prog \| ECAdd Prog \| Len LenHdr Prog \| Xor Prog \| Add Integer Prog \| Take IntList Prog \| Split IntList Prog \| SplitE Int Prog \| Pad Int Word8List Prog \| Map Prog Int Prog \| Sort Int Prog \| Select Inst CaseList \| SA Prog \| WithLen LenHdr Prog \| VarL Int VarName Role VarType \| VL Int VarName \| SelectV VarName CaseList VarName ::= "["{String,}String"]" VarType ::= Plain Int \| Primary Int \| Modulo Inst \| UTC \| ECx Inst \| Sublist Prog \| Choice Prog \| Subset Prog \| Is Prog Word8List ::= "["{Word8,}[Word8]"]" IntList ::= "["{Int,}Int"]" IntegerList ::= "["{Integer,}Integer"]" SEncAlg ::= AES128CBC \| AES256CBC … HashAlg ::= SHA1 \| SHA256 … EncoAlg ::= SSLPad Int \| B2DERInt \| B2DERBits … LenHdr ::= BE Int \| LE Int \| DER CaseTy ::= Case Word8List Prog \| Cases "["{Word8List,}Word8List "]" Prog \| Case' Condition Prog \| Otherwise Prog \| CaseUndef Prog Case' Condition Prog \| Otherwise Prog \| CaseUndef Prog CaseList ::= "["{CaseTy,}CaseTy"]" Condition ::= Bytes Word8List \| Equal Integer \| Less Integer \| More Integer \| LessOrEq Integer \| MoreOrEq Integer \| OneOf IntegerList \|Otherwise' Condition ::= Bytes Word8List \| Equal Integer \| Less Integer \| More Integer \| LessOrEq Integer \| MoreOrEq Integer \| OneOf IntegerList \|Otherwise' Condition ::= Bytes Word8List \| Equal Integer \| Less Integer \| More Integer \| LessOrEq Integer \| MoreOrEq Integer \| OneOf IntegerList \|Otherwise' Role ::= Clnt \| Serv \| A \| B \| S \| CA \| RA \| TTP … 3.1 CMN.1 syntax Below, the terms 'String', 'Integer', 'Int', 'Word8' denote the sets of strings, unlimited integers, integers ranged from 0 to 232-1 and integers ranged from 0 to 28-1, respectively. The curled brackets mean repetition, the square ones – optionality. The symbol ',' means comma itself, not concatenation. 3.2 CMN.1 semantics CSM has one main stack and varying number of temporary stacks, random-number generator, real-time clock, the storage s_var containing the values of the protocol variables (actually they don’t vary in CSM) and the register s_rol containing the identifier of the protocol role (fig. 1). 167 Cryptographic Stack Machine Notation One. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 165-182 Fig. 1. Cryptographic stack machine The language of the CSM instructions extends the line of the stack-oriented languages. It supports branching but doesn't support looping or recursing (table 1). Table 1. CSM instructions semantics Instruction CSM actions "Bare-metal" instructions Const bs CSM pushes the byte string bs onto the stack. Var s r t If the storage s_var contains the variable named s, then CSM pushes this variable value onto the stack. Otherwise, if r /= s_rol, CSM returns an error. Otherwise, it generates a new element of type t, stores its value under the name s in the s_var storage and puts this value in the stack. The currently defined variable types: Plain n – random n bytes; Primary n – random primary integer of n-bit length; Modulo is – random integer modulo n, where n is the big-endian value of the result of the instruction is execution; ECx is – random point on the curve curve_id, where curve_id is the value of the result of the instruction is execution; UTC – the time and date in standard UNIX 32-bit format; Sublist p (Choice p, Subset p) – random sublist (element, subset) of the list comprised of resulting elements of the program p execution; Is p – equivalent to Choice [C p]. V s If the storage s_var contains the variable with name s, then CSM pushes the value of this variable onto the stack. Otherwise, it returns an error. SEnc' alg CSM takes the top 3 elements of the stack as arguments: a, b, c. CSM encrypts a with b as initial vector and c as the key using symmetric encryption algorithm alg. Here and after: 1) if the stack is underflowed, CSM returns an error; 2) the l i h li i l d h f h k 3) h Fig. 1. Cryptographic stack machine Fig. 1. Cryptographic stack machine Fig. 1. Cryptographic stack machine The language of the CSM instructions extends the line of the stack-oriented languages. 3.2 CMN.1 semantics It supports branching but doesn't support looping or recursing (table 1). Table 1. CSM instructions semantics Instruction CSM actions "Bare-metal" instructions Const bs CSM pushes the byte string bs onto the stack. Var s r t If the storage s_var contains the variable named s, then CSM pushes this variable value onto the stack. Otherwise, if r /= s_rol, CSM returns an error. Otherwise, it generates a new element of type t, stores its value under the name s in the s_var storage and puts this value in the stack. The currently defined variable types: Plain n – random n bytes; Primary n – random primary integer of n-bit length; Modulo is – random integer modulo n, where n is the big-endian value of the result of the instruction is execution; ECx is – random point on the curve curve_id, where curve_id is the value of the result of the instruction is execution; UTC – the time and date in standard UNIX 32-bit format; Sublist p (Choice p, Subset p) – random sublist (element, subset) of the list comprised of resulting elements of the program p execution; Is p – equivalent to Choice [C p]. V s If the storage s_var contains the variable with name s, then CSM pushes the value of this variable onto the stack. Otherwise, it returns an error. SEnc' alg CSM takes the top 3 elements of the stack as arguments: a, b, c. CSM encrypts a with b as initial vector and c as the key using symmetric encryption algorithm alg. Here and after: 1) if the stack is underflowed, CSM returns an error; 2) the last argument in the argument list is located at the top of the stack; 3) the arguments of the function are removed from the stack; 4) the result is pushed to the stack. Enco' alg Encoding of a using algorithm alg. List of arguments: a. The language of the CSM instructions extends the line of the stack-oriented languages. It supports branching but doesn't support looping or recursing (table 1). 168 Прокопьев С.Е. Нотация криптографической стековой машины версии один. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 165-182 169 Xor' n Exclusive OR. Arguments: the top n elements of the stack. ModAdd', ModMult' Addition (multiplication) of a and b modulo m. List of arguments: a, b, m. 3.2 CMN.1 semantics Len' e The length of the top element of the stack written in e format, where e can be 169 Cryptographic Stack Machine Notation One. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. Prokopev S.E. Cryptographic Stack Machine Notation One. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 165-182 one of: BE n (packing into n big-endian bytes), LE n (packing into n little- endian bytes), DER (packing using ASN.1 DER format). Insert i CSM moves the top element of the stack to the i-th position. Pick i, Dup i CSM moves (for Pick) or copies (for Dup) the i-th element of the stack to the top position. Free i CSM removes the i-th element from the stack. Elem i p CSM executes the program p using temporary empty stack and then puts in the current working stack the i-th element of temporary stack. SA' n k p CSM copies n elements from the current working stack to temporary stack, executes the program p using a new temporary stack and then inserts the resulting elements between the (k+1)-th and k-th elements of the current working stack. Map' p i n The stack must contain at least in elements. CSM executes the program p n times using at each iteration a new temporary stack to which the next i elements from the current working stack are moved (beginning from the depths of the stack). At each iteration the elements containing in temporary stack after execution of p are moved to the current working stack. Sort' i n CSM considers the top in elements of the stack as a list of n elements, where each element, in turn, is a list of i elements. CSM sorts this list of n elements comparing their first (from the depths of the stack) elements. Select' cs CSM converts the list of the cases cs into the form: [Case' c1 p1 , … , Case' cn pn]. If CSM finds in the list cs (from left to right) the condition ci to which the top element of the stack satisfies, then it removes the top element from the stack and executes the program pi. Otherwise, it returns an error. M p Macro instruction: CSM simply executes the program p. L n p Macro instruction supplemented by the total length of the resulting elements of p execution (parameter n). 3.2 CMN.1 semantics Here and after: the byte strings are interpreted as integers basing on the `big endian` agreement. ModInv' Inverse of a under modulo m. List of arguments: a, m. Add' n Let a is the top element of the stack. CSM adds n to a modulo 2^(8k), where k is the length of a in bytes. Rev fun The function that is reverse to the function fun, where fun must be one of: Enco' alg, SEnc' alg, Xor' n, ModMult', ModAdd', ModInv', Add' n. Mod' Modulo operation. List of arguments: a. ModExp' Modular exponentiation: a^b mod m. List of arguments: a,b,m. Hash' alg CSM calculates the hash of a using algorithm alg. List of arguments: a. Pad' n ws Padding of a using the bytes ws until the length of the result reaches n (n must be equal or greater than length of a). List of arguments: a. Take' ns Here ns is the list of numbers. If the length of the top element of the stack is less than the sum of the elements of ns, then CSM returns the specification error. Otherwise, CSM cuts the top element of the stack into n parts considering the numbers from the ns list as lengths of elements and pushes (from left to right) the resulting n elements onto the stack, where n is the length of the ns list. The remainder of the top element is dropped (if any). Split' ns The same as the instruction Take' ns, except that the length of the top element of the stack must be exactly equal to the sum of the numbers from the ns list. SplitE' n Is equivalent to the instruction Split' [k,k...k], where k = len / n, where len is the length of the top element of the stack (len must be dividable by n). ECMult' Elliptic curve scalar multiplication. List of arguments: curve_id (curve identifier), x (x-coordinate), y (y-coordinate), k (the scalar). Instruction produces 2 elements of the stack: x-coordinate and y-coordinate. ECAdd' Elliptic curve addition of points (x1,y1) and (x2,y2). List of arguments: curve_id (curve identifier), x1, y1, x2, y2. Instruction produces 2 elements of the stack: x-coordinate and y-coordinate. C' n Concatenation. Arguments: the top n elements of the stack. CE' n Concatenation of the equal-sized arguments. 3.2 CMN.1 semantics "Sugared" instructions C p, CE p, Xor p, SEnc al p, Mod p, ModMult p, ModAdd p, ModExp p, ModInv p, ECMult p, ECAdd p CSM executes the program p using temporary empty stack and copies the resulting m elements onto the current working stack. Then it executes the "bare-metal" counterpart of the "sugared" instruction: C' m, CE' m, Xor' m, SEnc' al, Mod', ModMult', ModAdd', ModExp', ModInv', ECMult' or ECAdd'. In the end, CSM moves the resulting elements (two elements in the case of the ECMult' or ECAdd' instruction and one element in the other cases) to the current working stack. 170 Е. Нотация криптографической стековой машины версии один. Труды ИСП РАН, том 30, вып. 3, 5 182 Прокопьев С.Е. Нотация криптографической стековой машины версии один. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 165-182 Map q n p Sort n p CSM executes the program p using temporary empty stack. If m mod n /= 0, CSM returns an error (where m is the number of elements of temporary stack after execution of p). Otherwise, it copies the resulting m elements onto the current working stack executes the "bare-metal" counterpart: Map' q i n or Sort' i n, where i = m / n. Hash al p, Enco al p, Add n p, Pad n bs p, Len e p, Take lst p, Split lst p, SplitE n p CSM executes the program [C p] using temporary empty stack and copies the resulting element onto the current working stack. After that, CSM executes the "bare-metal" counterpart of the "sugared" instruction: Hash' al, Enco' al, Add' n, Pad' n bs, Len' e, Take' ls, Split' ls or SplitE' n. Select is cs CSM tries to execute the program [C [is]] using temporary empty stack. If the program was successfully executed, CSM copies the resulting element onto the current working stack and executes the instruction Select' cs. If the execution failed (due to unknown variable), CSM checks if the list cs does contain the element CaseUnkno p. If so, CSM executes the program p, otherwise it returns an error. 3.2 CMN.1 semantics VarL n s r t Is equivalent to: L n (Var s r t) VL n s Is equivalent to: L n (V s) SA p Is equivalent to: SA' 1 0 p WithLen e p Is equivalent to: M [C p, SA' 1 1 [Len' e]] SelectV s cs Is equivalent to: Select (V s) cs 4. Simple CMN.1-based specification language 165-182 s_var belonging to the CSM instance of the role r, where val is concatenation of the resulting elements of the execution of is. select r is acs This action provides a branching support in the same manner as the CSM instruction Select is cs does. The difference between the lists cs and acs is that cs consists of elements Case value p, where p is a CSM program, whereas acs consist of elements Case value a, where a is a sequence of protocol actions. trusted r id p This action takes from a trusted storage the binary data stored under the name id and processes these data using CMN.1 definition p and the CSM instance of the role r. connect r port addr If this action is present, the specification turns into the client implementation acting as the protocol role role. The action carries out the connection to a third- party server implementation listening on the port port of the IP-address addr. accept role port The specification turns into the server implementation acting as role and listening on the port port. printPV printPV' Both actions generate the ProVerif program corresponding to the protocol events that took place at the time of the call. The first action generates a full program, the second one ignores the lengths fields of messages and related events as non-essential in order to make this program more concise and productive. Bearing in mind the elegant and concise syntax of the Haskell language and advantages of embedded domain-specific languages, we integrate our CMN.1-based specification language in Haskell. As an illustration, we present an excerpt from the CMN.1-based specification of the TLS protocol (fig. 2; note that the order of declarations can be arbitrary in the Haskell language). A specification, which serves as source for this excerpt, comprises about 500 lines (the total for client and server) covering substantial part of the TLS v.1.2 protocol including four ciphersuites and X.509 certificates support and excluding extensions and renegotiations. The specification turned into the implementation (see the actions connect and accept in the table 2) was successfully tested for interoperability with the OpenSSL v.1.0.2o tool (both in the client and server roles). Bearing in mind the elegant and concise syntax of the Haskell language and advantages of embedded domain-specific languages, we integrate our CMN.1-based specification language in Haskell. 4. Simple CMN.1-based specification language The language presented below is simple in the sense that it doesn't capture the protocol automata in full. A specification consists of the CMN.1-based message definitions and a sequence of protocol actions with simple branching support (table 2). Table 2. Protocol actions Action Description roles rlist The action sets the roles participating in the protocol. Each role runs its own CSM instance. msg src dst p The message with the CMN.1 definition p is transferred from the role src to the role dst. set r vvlist Here vvlist is the list of pairs of type (V name, is). For each pair, the action executes the CSM instruction is and includes the pair (name, val) in a storage 171 Prokopev S.E. Cryptographic Stack Machine Notation One. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 165-182 s_var belonging to the CSM instance of the role r, where val is concatenation of the resulting elements of the execution of is. select r is acs This action provides a branching support in the same manner as the CSM instruction Select is cs does. The difference between the lists cs and acs is that cs consists of elements Case value p, where p is a CSM program, whereas acs consist of elements Case value a, where a is a sequence of protocol actions. trusted r id p This action takes from a trusted storage the binary data stored under the name id and processes these data using CMN.1 definition p and the CSM instance of the role r. connect r port addr If this action is present, the specification turns into the client implementation acting as the protocol role role. The action carries out the connection to a third- party server implementation listening on the port port of the IP-address addr. accept role port The specification turns into the server implementation acting as role and listening on the port port. printPV printPV' Both actions generate the ProVerif program corresponding to the protocol events that took place at the time of the call. The first action generates a full program, the second one ignores the lengths fields of messages and related events as non-essential in order to make this program more concise and productive. Cryptographic Stack Machine Notation One. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. Prokopev S.E. Cryptographic Stack Machine Notation One. Trudy ISP RAN/Proc. ISP RAS, vol. 4. Simple CMN.1-based specification language 30, issue 3, 2018, pp. 165-182 s_var belonging to the CSM instance of the role r, where val is concatenation of the resulting elements of the execution of is. select r is acs This action provides a branching support in the same manner as the CSM instruction Select is cs does. The difference between the lists cs and acs is that cs consists of elements Case value p, where p is a CSM program, whereas acs consist of elements Case value a, where a is a sequence of protocol actions. trusted r id p This action takes from a trusted storage the binary data stored under the name id and processes these data using CMN.1 definition p and the CSM instance of the role r. connect r port addr If this action is present, the specification turns into the client implementation acting as the protocol role role. The action carries out the connection to a third- party server implementation listening on the port port of the IP-address addr. accept role port The specification turns into the server implementation acting as role and listening on the port port. printPV printPV' Both actions generate the ProVerif program corresponding to the protocol events that took place at the time of the call. The first action generates a full program, the second one ignores the lengths fields of messages and related events as non-essential in order to make this program more concise and productive. Bearing in mind the elegant and concise syntax of the Haskell language and advantages of embedded domain-specific languages, we integrate our CMN.1-based specification language in Haskell. As an illustration, we present an excerpt from the CMN.1-based specification of the TLS protocol (fig. 2; note that the order of declarations can be arbitrary in the Haskell language). A specification, which serves as source for this excerpt, comprises about 500 lines (the total for client and server) covering substantial part of the TLS v.1.2 protocol including four ciphersuites and X.509 certificates support and excluding extensions and renegotiations. The specification turned into the implementation (see the actions connect and accept in the table 2) was successfully tested for interoperability with the OpenSSL v.1.0.2o tool (both in the client and server roles). Prokopev S.E. Cryptographic Stack Machine Notation One. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 4. Simple CMN.1-based specification language As an illustration, we present an excerpt from the CMN.1-based specification of the TLS protocol (fig. 2; note that the order of declarations can be arbitrary in the Haskell language). A specification, which serves as source for this excerpt, comprises about 500 lines (the total for client and server) covering substantial part of the TLS v.1.2 protocol including four ciphersuites and X.509 certificates support and excluding extensions and renegotiations. The specification turned into the implementation (see the actions connect and accept in the table 2) was successfully tested for interoperability with the OpenSSL v.1.0.2o tool (both in the client and server roles). 172 172 Е. Нотация криптографической стековой машины версии один. Труды ИСП РАН, том 30, вып. 3, 5 182 2018 г., стр. 165-182 173 173 E. Cryptographic Stack Machine Notation One. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp p yp g p y , , , , pp 165-182 165-182 174 Прокопьев С.Е. Нотация криптографической стековой машины версии один. Труды ИСП РАН, том 2018 г., стр. 165-182 Прокопьев С.Е. Нотация криптографической стековой машины версии один. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 165-182 . Нотация криптографической стековой машины версии один. Труды ИСП РАН, том 30, вып. 3, 5 182 5. Translation to the ProVerif program The ProVerif program presented in the fig. 3 was generated automatically from the above specification (it is a console output of the call printPV'; see the line 115 in the fig. 2). This program corresponds to the protocol trace based on the ciphersuite TLS-DHE-DSS-WITH-AES-256-CBC-SHA. The program passed the ProVerif compiler checks without warnings. The events and queries of interest have to be inserted manually because CMN.1-based specifications do not contain such information. 175 Prokopev S.E. Cryptographic Stack Machine Notation One. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 165-182 176 Прокопьев С.Е. Нотация криптографической стековой машины версии один. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 165-182 Fig. 3. The corresponding ProVerif program (an excerpt) копьев С.Е. Нотация криптографической стековой машины версии один. Труды ИСП РАН, том 30 г., стр. 165-182 С.Е. Нотация криптографической стековой машины версии один. Труды ИСП РАН, том 30, вып. . 165-182 Fig. 3. The corresponding ProVerif program (an excerpt) Fig. 3. The corresponding ProVerif program (an excerpt) 6. Engine implementation details The engine implements the functionality that is significantly more powerful than the CSM machine presented in the section 3. The engine does not execute the CMN.1- notated programs as straightforward as CSM does. It executes the programs symbolically: the elements of the stack are not byte strings but symbolic expressions. This well-known technique allows the engine to fully take over the task of verification of the incoming messages using the same CMN.1-definitions that are used in the direct task of message generation. The verification is complete: the engine decrypts the ciphertexts, checks MACs and signatures, etc. Throughout a protocol execution, the engine accumulates the generated symbolic expressions, their values, lengths and types. It uses this information to generate or verify the 177 Prokopev S.E. Cryptographic Stack Machine Notation One. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 165-182 Prokopev S.E. Cryptographic Stack Machine Notation One. Trudy ISP RAN/Proc. ISP RAS, vol. 30, iss 165-182 E. Cryptographic Stack Machine Notation One. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp protocol messages in the future. In addition, the engine logs such events as calculations of the values of the symbolic expressions and applications of the rewriting rules. This information can be used by the engine's environment to extract symbolic traces and convert them to the programs for symbolic verifiers, e.g. ProVerif (as was presented in the previous section). The scheme of the verification is as follows. Let the byte string bs is considered by the engine as a protocol message with the CMN.1 definition p. Let EQ is a set variable containing equations, i.e. pairs of type (symbolic expression, byte string). The engine implements the verification procedure as follows. Step 1. The engine executes the program p symbolically resulting the symbolic expression exp. EQ is initialized with the equation (exp,bs). Step 1. The engine executes the program p symbolically resulting the symbolic expression exp. EQ is initialized with the equation (exp,bs). Step 2. For every new equation (exp,bs) from EQ, until neither of Step 2.1 or Step 2.2 can be applied anymore: Step 2.1. The engine tries to apply a rewriting rule to this equation. This rule can be a simple inversion (for Enco, SEnc, Xor, ModMult, ModAdd, ModInv or Add functions) or be a complex group operation taking into account other equations from EQ (e.g. for Split). 7. Conclusion We presented cryptographic protocol message notation (named CMN.1) based on the instruction set of a stack machine specifically tailored to the needs of cryptographic protocols (named cryptographic stack machine, or CSM). The principles of implementation of the protocol specification language based on this notation also presented. Within such an approach, specifications are executable and also translatable to the programs for symbolic verifiers, such as ProVerif. The readability of CMN.1-notated specifications is brought in the court of public opinion. In addition, the validation of the proposed notation on a wider spectrum of cryptographic protocols is needed. The validation will certainly cause minor additions to the notation (at least regarding cryptographic key types) without affecting currently defined CSM instructions. Taking into account the fact that the author's proof-of-concept implementation of the core language library (the engine) comprises only 700 lines of the Haskell code (excluding cryptographic primitives), it seems logical to provide in the future a formal description of the engine's algorithm and, basing on it, a proof of the soundness of the ProVerif-translation procedure. 6. Engine implementation details cEvent ev – The engine logs the event ev (i.e. the environment can insert additional events into the engine log). 7. cGetLog – The engine returns content of its log. 7. cGetLog – The engine returns content of its log. 6. Engine implementation details The application of the rule produces one or several new equations, which are inserted in EQ. If some rule was applied, the engine returns to the beginning of the Step 2. Otherwise, it goes to the Step 2.2. Step 2.2. If the values of all the arguments of the top operation of the symbolic expression exp are known, the engine calculates the value of exp. If this value is equal to bs, the engine removes the equation from EQ. Otherwise, it returns the message verification error. Step 2.2. If the values of all the arguments of the top operation of the symbolic expression exp are known, the engine calculates the value of exp. If this value is equal to bs, the engine removes the equation from EQ. Otherwise, it returns the message verification error. The engine knows about the equality (ab)c = (ac)b and analogous equality for the elliptic curve scalar multiplication, so Diffie-Hellman key exchange and ElGamal asymmetric encryption do not ask for special treatment. Yet the engine uses specific rewrites for expressions relevant to the DSA and ECDSA algorithms or to their relatives. he calls exported by the engine are presented below. The calls exported by the engine are presented below. 1. cSymExec p – The engine executes the program p symbolically and returns the descriptor of the generated symbolic expression. 2. cCalc d – The engine calculates the value of the symbolic expression with descriptor d. 3. cGetVal d – The engine returns the value of the symbolic expression with descriptor d. 4. cSetVal d bs – The engine assigns the value bs to the symbolic expression with descriptor d. 5. cVerify d bs – The engine verifies the byte string bs with respect to the symbolic expression with descriptor d. If verification has failed, it returns 8 178 Прокопьев С.Е. Нотация криптографической стековой машины версии один. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 165-182 . Нотация криптографической стековой машины версии один. Труды ИСП РАН, том 30, вып. 3, 5-182 Прокопьев С.Е. Нотация криптографической стековой машины версии один. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 165-182 an error, otherwise, it returns the superfluous remainder of the byte string bs (if present). 6. cEvent ev – The engine logs the event ev (i.e. the environment can insert additional events into the engine log). 6. References [1]. S. Chaki and A. Datta. Aspier: An automated framework for verifying security protocol implementations. In Proceedings of the Computer Security Foundations Workshop, 2009, pp. 172–185. [2]. J. Goubault-Larrecq and F. Parrennes. Cryptographic protocol analysis on real C code. In Proceedings of the 6th International Conference on Verification, Model Checking and Abstract Interpretation (VMCAI’05), Lecture Notes in Computer Science, vol. 3385, 2005, pp. 363–379. [3]. Mihhail Aizatulin, Andrew D. Gordon, and Jan Jurjens. Extracting and verifying cryptographic models from C protocol code by symbolic execution. In Proc. of the 18th ACM Conference on Computer and Communications Security (CCS’11), 2011, pp. 331– 340. [4]. Nicholas O’Shea. Using Elyjah to analyse Java implementations of cryptographic protocols. In Proc. of the Joint Workshop on Foundations of Computer Security, Automated Reasoning for Security Protocol Analysis and Issues in the Theory of Security (FCS-ARSPA-WITS’08), 2008. y ( ) [5]. Karthikeyan Bhargavan, Cedric Fournet, Andrew Gordon, and Stephen Tse. Verified interoperable implementations of security protocols. ACM Transactions on Programming Languages and Systems (TOPLAS), vol. 31, no. 1, 2008. 179 Prokopev S.E. Cryptographic Stack Machine Notation One. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 165-182 Prokopev S.E. Cryptographic Stack Machine Notation One. Trudy ISP RAN/Proc. ISP RAS, vol. 30, iss 165-182 E. Cryptographic Stack Machine Notation One. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp Prokopev S.E. Cryptographic Stack Machine Notation One. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 165-182 [6]. Karthikeyan Bhargavan, Bruno Blanchet, and Nadim Kobeissi. Verified Models and Reference Implementations for the TLS 1.3 Standard Candidate. In Proc. of the IEEE Symposium on Security and Privacy, 2017. [6]. Karthikeyan Bhargavan, Bruno Blanchet, and Nadim Kobeissi. Verified Models and Reference Implementations for the TLS 1.3 Standard Candidate. In Proc. of the IEEE Symposium on Security and Privacy, 2017. [7]. Matteo Avalle, Alfredo Pironti, Riccardo Sisto, and Davide Pozza. The JavaSPI framework for security protocol implementation. In Proc. of the 6th International Conference on Availability, Reliability and Security (ARES’11), 2011, pp. 746–751. [8]. David Cade, Bruno Blanchet. From Computationally-Proved Protocol Specifications to Implementations and Application to SSH. Available at: http://prosecco.gforge.inria.fr/personal/dcade/CadeBlanchetJoWUA13.pdf, accessed 10.06.2018. [9]. A. Delignat-Lavaud et al. Implementing and Proving the TLS 1.3 Record Layer. In Proc. of the 2017 IEEE Symposium on Security and Privacy (SP), 2017, pp. 463-482. [10]. Bruno Blanchet. Automatic Verification of Security Protocols in the Symbolic Model: the Verifier ProVerif. References In Foundations of Security Analysis and Design VII, FOSAD Tutorial Lectures, Lecture Notes in Computer Science, vol. 8604, 2014, pp. 54-87. [11]. Cas Cremers, Marko Horvat, Jonathan Hoyland, Sam Scott, and Thyla van der Merwe. 2017. Source files and annotated RFC for TLS 1.3 analysis. (2017). Available at: https://tls13tamarin.github.io/TLS13Tamarin/, accessed 10.06.2018. [12]. Concrete Syntax Notation One (CSN.1). Available at: http://csn1.info, accepted 10.06.2018. [13]. Transfer Syntax Notation One (TSN.1). Available at: http://www.protomatics.com/tsn1. Html, accessed 10.06.2018.. [14]. The BinPAC language. Available at: https://www.bro.org/sphinx/components/binpac/ README.html, accessed 10.06.2018.. [15]. Mario Baldi, Fulvio Risso. NetPDL: An Extensible XML-Based Language for Packet Header Description. Computer Networks, vol, 50, issue 5, 2006, pp. 688-706. DOI: 10.15514/ISPRAS-2018-30(3)-12 DOI: 10.15514/ISPRAS-2018-30(3)-12 Для цитирования: Прокопьев С.Е. Нотация криптографической стековой машины версии один. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 165-182 (на английском языке). DOI: 10.15514/ISPRAS-2018-30(3)-12 Для цитирования: Прокопьев С.Е. Нотация криптографической стековой машины версии один. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 165-182 (на английском языке). DOI: 10.15514/ISPRAS-2018-30(3)-12 С.Е. Прокопьев <s.e.pr@mail.ru> г. Москва С.Е. Прокопьев <s.e.pr@mail.ru> г. Москва 180 Аннотация. Хорошая спецификация криптографического протокола должна легко восприниматься человеком (быть декларативной и лаконичной), быть исполнимой и пройти процедуру формальной верификации в некоторой адекватной модели. Нацеливаясь на эти требования, в статье предложена нотация CMN.1, предназначенная для описания сообщений криптографических протоколов и основанная на вычислительной абстракции под названием криптографическая стековая машина (CSM). Статья описывает синтаксис и семантику CMN.1, а также представляет результаты разработки языка спецификаций криптографических протоколов, построенного на основе данной нотации и встроенного в язык Haskell. В авторской реализации вся обработка сообщений инкапсулирована внутри базового библиотечного модуля, в то время как спецификация должна лишь дать декларативные определения этих сообщений. При формировании исходящего сообщения протокола базовый модуль берет описание данного сообщения в нотации CMN.1 и возвращает фрагмент данных, сгенерированный по этому описанию. При обработке входящего сообщения базовый модуль берет поступивший фрагмент данных и описание ожидаемого сообщения в 180 Е. Нотация криптографической стековой машины версии один. Труды ИСП РАН, том 30, вып. 3, 5 182 нотации CMN.1 и возвращает вердикт об их соответствии друг другу, извлекая и запоминая при этом все содержащиеся в сообщении данные, необходимые для формирования или верификации следующих сообщений протокола. Процесс верификации является полным: базовый модуль осуществляет расшифрование, проверку кодов аутентификации сообщений и значений цифровой подписи и т.д. Текущая реализация языка включает функции трансляции спецификаций в исполняемый код, совместимый с существующими программными реализациями протоколов, а также функции конвертации спецификаций в программы на входном языке анализатора протоколов ProVerif. В качестве иллюстрации приводятся выдержки из CMN.1-спецификации протокола TLS и соответствующей ей автоматически сгенерированной программы для ProVerif. Ключевые слова: язык спецификаций криптографических протоколов; нотация сообщений криптографических протоколов; криптографическая стековая машина; встроенные предметно-ориентированные языки; Haskell; ProVerif; TLS Список литературы [1]. S. Chaki and A. Datta. Aspier: An automated framework for verifying security protocol implementations. In Proceedings of the Computer Security Foundations Workshop, 2009, pp. 172–185. [1]. S. Chaki and A. Datta. Aspier: An automated framework for verifying security protocol implementations. In Proceedings of the Computer Security Foundations Workshop, 2009, pp. 172–185. [2]. J. Goubault-Larrecq and F. Parrennes. Cryptographic protocol analysis on real C code. In Proceedings of the 6th International Conference on Verification, Model Checking and Abstract Interpretation (VMCAI’05), Lecture Notes in Computer Science, vol. 3385, 2005, pp. 363–379. [3]. Mihhail Aizatulin, Andrew D. Gordon, and Jan Jurjens. Extracting and verifying cryptographic models from C protocol code by symbolic execution. In Proc. of the 18th ACM Conference on Computer and Communications Security (CCS’11), 2011, pp. 331– 340. [4]. Nicholas O’Shea. Using Elyjah to analyse Java implementations of cryptographic protocols. In Proc. of the Joint Workshop on Foundations of Computer Security, Automated Reasoning for Security Protocol Analysis and Issues in the Theory of Security (FCS-ARSPA-WITS’08), 2008. [5]. Karthikeyan Bhargavan, Cedric Fournet, Andrew Gordon, and Stephen Tse. Verified interoperable implementations of security protocols. ACM Transactions on Programming Languages and Systems (TOPLAS), vol. 31, no. 1, 2008. [6]. Karthikeyan Bhargavan, Bruno Blanchet, and Nadim Kobeissi. Verified Models and Reference Implementations for the TLS 1.3 Standard Candidate. In Proc. of the IEEE Symposium on Security and Privacy, 2017. y p y y [7]. Matteo Avalle, Alfredo Pironti, Riccardo Sisto, and Davide Pozza. The JavaSPI framework for security protocol implementation. In Proc. of the 6th International Conference on Availability, Reliability and Security (ARES’11), 2011, pp. 746–751. 181 Prokopev S.E. Cryptographic Stack Machine Notation One. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 165-182 E. Cryptographic Stack Machine Notation One. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp [8]. David Cade, Bruno Blanchet. From Computationally-Proved Protocol Specifications to Implementations and Application to SSH. Режим доступа: http://prosecco.gforge.inria.fr/personal/dcade/CadeBlanchetJoWUA13.pdf, дата обращения 10.06.2018. [9]. A. Delignat-Lavaud et al. Implementing and Proving the TLS 1.3 Record Layer. In Proc. of the 2017 IEEE Symposium on Security and Privacy (SP), 2017, pp. 463-482. [10]. Bruno Blanchet. Automatic Verification of Security Protocols in the Symbolic Model: the Verifier ProVerif. In Foundations of Security Analysis and Design VII, FOSAD Tutorial Lectures, Lecture Notes in Computer Science, vol. 8604, 2014, pp. 54-87. [11]. Cas Cremers, Marko Horvat, Jonathan Hoyland, Sam Scott, and Thyla van der Merwe. 2017. [15]. Mario Baldi, Fulvio Risso. NetPDL: An Extensible XML-Based Language for Packet Header Description. Computer Networks, vol, 50, issue 5, 2006, pp. 688-706. Список литературы Source files and annotated RFC for TLS 1.3 analysis. (2017). Режим доступа: https://tls13tamarin.github.io/TLS13Tamarin/, дата обращения 10.06.2018. [12]. Concrete Syntax Notation One (CSN.1). Режим доступа: http://csn1.info, дата обращения 10.06.2018. [13]. Transfer Syntax Notation One (TSN.1). Режим доступа: http://www.protomatics.com/tsn1. Html, дата обращения 10.06.2018. Режим [14]. The BinPAC language. Режим доступа: https://www.bro.org/sphinx/components/binpac/ README.html, дата обращения 10.06.2018.. [15]. Mario Baldi, Fulvio Risso. NetPDL: An Extensible XML-Based Language for Packet Header Description. Computer Networks, vol, 50, issue 5, 2006, pp. 688-706. 182 Construction of validation modules based on reference functional models in a standalone verification of communication subsystem D.A. Lebedev <lebedev_d@mcst.ru> I.A. Stotland <stotl_i@ mcst.ru > MCST, 24 Vavilov st., Moscow, 119334, Russia D.A. Lebedev <lebedev_d@mcst.ru> I.A. Stotland <stotl_i@ mcst.ru > MCST, 24 Vavilov st., Moscow, 119334, Russia Abstract. The paper proposes some approaches to functional verification of microprocessor communication controllers based on developing layered UVM (Universal Verification Methodology) test systems. In modern microprocessor systems there are a lots of controllers operating with their own data types. Communication controllers support transferring and transformation data between microprocessor units. Such transformation must be carried out quickly and without data corruption for the correct functioning of the whole system. Communication controllers could carry additional functions such transmission values of copies of the system registers, address translation and others. Brief overview of verification tools and benefits of application standalone simulation based verification for checking the correctness of communication subsystems are marked out in the paper. We present the approaches of construction a standalone UVM-based verification environment with checking module implemented in external functional reference model. We also propose some techniques for checking the correctness of communication subsystems: checking multiple- clock controllers using parametrized clock generator, supporting of credit exchange mechanisms. Presented approaches were used to verify the communication subsystem — Host-Bridge — of Sparc V9 eight-core microprocessor developed by MCST. The difficulties discovered in the process of test system developing and its resolutions are described in the paper. The results of using presented solutions for verification of communicating subsystem controllers and further plan of the test system enhancement are considered. Keywords: test system; communication controller; functional verification; Universal Verification Methodology (UVM); reference model. DOI: 10.15514/ISPRAS-2018-30(3)-13 DOI: 10.15514/ISPRAS-2018-30(3)-13 For citation: Lebedev D.A., Stotland I.A. Construction of validation modules based on reference functional models in a standalone verification of communication subsystem. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 183-194. DOI: 10.15514/ISPRAS-2018- 30(3)-13 183 Lebedev D.A., Stotland I.A. Construction of validation modules based on reference functional models in a standalone verification of communication subsystem. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 183-194 1 I t d ti Lebedev D.A., Stotland I.A. Construction of validation modules based on reference functional models in a standalone verification of communication subsystem. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 183-194 Lebedev D.A., Stotland I.A. Construction of validation modules based on reference functional models in a standalone verification of communication subsystem. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 183-194 Lebedev D.A., Stotland I.A. Construction of validation modules based on reference functional models in a standalone verification of communication subsystem. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 183-194 1. Introduction The state of the art in microprocessors composition includes a variety of hardware controllers, which differ in complexity, speed rate, volume and types of data transmitted over them. The characteristics of data are continuously increasing. At the same time, verification costs are increasing, because the possibilities of verification methods are significantly lagging behind the development of microprocessor systems and, accordingly, the correctness checking requires more resources [1]. Each peripheral controller in the system could have its own data format. Converting data format is one of the functions of the interface communication controllers. The communication controllers can be a part of communication subsystem also known as northbridge. The northbridge typically handles communications among the CPU, I/O and in some cases RAM. Therefore, these transformations must be carried out quickly and without data loss. For this reason, the verification of communication controllers is an important step in the development of the microprocessor system. The rest of the paper is organized as follows. Section 2 reviews the existing techniques for verifying communication controllers. Section 3 suggests an approach to the problem of developing test system. Section 4 describes a case study and the suggested approaches. Section 5 reveals results and Section 6 concludes the paper. Each peripheral controller in the system could have its own data format. Converting data format is one of the functions of the interface communication controllers. The communication controllers can be a part of communication subsystem also known as northbridge. The northbridge typically handles communications among the CPU, I/O and in some cases RAM. Therefore, these transformations must be carried out quickly and without data loss. For this reason, the verification of communication controllers is an important step in the development of the microprocessor system. The rest of the paper is organized as follows. Section 2 reviews the existing techniques for verifying communication controllers. Section 3 suggests an approach to the problem of developing test system. Section 4 describes a case study and the suggested approaches. Section 5 reveals results and Section 6 concludes the paper. . Functional verification of the communication controllers It is necessary to simulate the operation of entire environment while providing standalone verification of a controller. This requires a test system, the development of which could be started at the earliest stages of whole microprocessor development, as soon as module specification and the RTL-model becomes available. Standalone verification allows detecting errors in the early stages of project. In addition, it helps to create complicated, critical and incorrect situations for the verified module. The achivement of such situations using system verification of the whole microprocessor model takes lot of resources. It is also important to note that the localization of the error is faster, what reduces the debugging time of the controller. Due to its location between the CPU and the peripheral interface controller, the communication controller, in addition to its basic data format transformation function, could include copies of registers, buffers, FIFOs, parts of distributed control systems, and perform other additional functions. A number of these features should be taken into account in the standalone verification of communication controllers. It is essential that, according to the classification proposed in [2], the properties of communication controllers include the absence of a pipeline, the absence of strict time (in the system clock frequency) restrictions on transaction processing and tagging of transmitted data. Accordingly, when the devices of this type are verified it is possible to use event-checking modules. 184 Лебедев Д.А., Стотланд И.А. Построение модулей проверки на основе эталонных функциональных моделей при автономной верификации подсистемы связи. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 183-194 Лебедев Д.А., Стотланд И.А. Построение модулей проверки на основе эталонных функциональных моделей при автономной верификации подсистемы связи. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 183-194 There are a number of methods to build a test system and implement a standalone verification of microprocessor controllers. Among them there is a tool created in the "MCST" named Alone-env, the development of the ISP RAS named C++TESKHW and methodology UVM [3]. The Alone-env tool simplifies implementation of standalone Verilog tests by creating test sequences in C++. Its library provides a wrapper-class over Verilog description of the verified module. Despite the relative simplicity of using Alone-env tool, there are some disadvantages: the lack of collecting coverage means, high requirements for the testing reference model and the inability to reuse the test system. . Functional verification of the communication controllers One of the C++TESKHW tool features is availability of test generation based on the device state graph traversal. However, sometimes it is very hard to define all of the states of device and it needs high accuracy of documentation and checking reference model. UVM is the most widespread verification methodology developed by Accellera Systems [4]. UVM is a library with well-described tools for building portable and reusable testbenches and their components. The test system based on UVM can generate pseudo-random constrained input requests to cover all the possible states of the verified device. Most of well-known simulation tools (like Incisive, VCS, etc.) support the methodology. Moreover, most of VIP (Verification IP) support UVM-based interfaces. We also have a number of test systems and libraries already written and debugged. Therefore, we choose to use UVM for verification of RTL implemented modules of microprocessor systems. Alone-env and C++TESKHW do not support UVM and we cannot use these tools for UVM-based test system development. UVM provides the universal approach for all types of devices to develop test systems. In this way, test systems are becoming more complex and worse in debugging. UVM also has no additional approaches for construction of validation modules based on reference functional models. Therefore, the main purpose of our investigation is to develop and extend the methods of standalone verification of communication controllers using UVM and program reference models. 3. Principles for testing communication controllers Standalone verification of communication controllers can be carried out using simulation reference models that are part of the test systems - specially implemented software environment for the verified device. Test system functions includes:  generating of input requests;  monitoring of reactions from the verified device and the reference mo  checking of reactions;  forming a conclusion about the completeness of testing. Uvm_sequence_item and uvm_sequence extension classes are defined to generate pseudorandom constrained impacts. The first one defines a set of variables that are required for serialization of set of impacts into a serial bit format. The second performs a single or multiply generating of a set of variables to transmit a request. 185 Lebedev D.A., Stotland I.A. Construction of validation modules based on reference functional models in a standalone verification of communication subsystem. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 183-194 Lebedev D.A., Stotland I.A. Construction of validation modules based on reference functional models in a standalone verification of communication subsystem. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 183-194 Lebedev D.A., Stotland I.A. Construction of validation modules based on reference functional models in a standalone verification of communication subsystem. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 183-194 The request generated by the uvm_sequence_item object is processed by a special uvm_sequencer class and passed to the uvm_driver class. Uvm_driver produces a transformation of generated random requests into sequential bit-vectors in accordance with the interface exchange protocol. Uvm_monitor class is passive. It tracks changes in the interface of the verified device, indicating the appearance of input or output data, then packages the serial bit signals into the uvm_sequence_item format and transmits for further analysis to the checking blocks. To simplify the structure of the test environment perception, the uvm_driver, uvm_monitor, and uvm_sequencer are combined in the uvm_agent class, shown on fig.1. ITEM SEQUENCER DRIVER MONITOR AGENT ITEM ITEM DUV INTERFACE Fig 1. An example of the uvm_agent structure Fig 1. An example of the uvm_agent structure Checking the reactions of the verified device can be carried out by internal means of the UVM library, however, if the verified device has a complex structure and many states, the checking module is based on the external to the controller environment reference model usually written in C++. A typical reference model-based test system is shown in Fig. 2. In Fig. 3. Principles for testing communication controllers 2 DUV (Design Under Verification) is RTL-model of the verified device, ENV (Environment) - test environment. The number of agents are determined by the number of interface groups of the verified device (tracking the reactions uvm_monitor object can be taken outside from the agents). The reference model generates reference responses when impacts from the test environment are implied. Uvm_scoreboard is a checking module compares the response from the verification device and the reference model and makes a conclusion about the correctness of the operating. Using the DPI (Directed Programming Interface) of System Verilog is necessary to reconcile the types and classes of the test system written in SystemVerilog hardware description language with the C++ language. in which the reference model is developed. 186 Лебедев Д.А., Стотланд И.А. Построение модулей проверки на основе эталонных функциональных моделей при автономной верификации подсистемы связи. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 183-194 Лебедев Д.А., Стотланд И.А. Построение модулей проверки на основе эталонных функциональных моделей при автономной верификации подсистемы связи. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 183-194 у D U V M O D E L ENVIROMENT DRIVER AGENT MONITOR S C O R E B O A R D AGENT AGENT DPI Fig 2. A typical structure of a test system for testing communications controllers. Fig 2. A typical structure of a test system for testing communications controllers. The reference models could be divided into three types: cycle-accurate, discrete- event with time accounting and event models [5]. The choice of model type depends on the type of verified device, its architecture, and the complexity of developing a test environment. As stated earlier, the use of event models is justified for communication controllers, as they require less time to develop, maintain changes and can fully simulate the operation of such a controllers. 4.2 Support of credit exchange mechanisms To control the flow of requests and free positions in the transaction buffers, Host- bridge supports a credit mechanism, which is a one-bit signal transmission that informs about the availability of free space in the buffers of connected devices. The management of this mechanism allows creating tests with full filling of all positions in the device buffers and needing to wait the vacated space to handle new requests, or on the other hand, tests, when the release of positions provides very fast, and the requests are executed almost instantly. As a result, it is possible to create test scenarios that are difficult to implement during system testing. 4.1 Several synchro signals parameters randomization The main task of communication controllers is to coordinate the requests and data of several devices of the microprocessor system operating at different frequencies of the synchro signal. The parts of the communication controller in which several synchro signals interact should be checked carefully. Generating of random periods of synchro signals and their shifts that are relative to each other can be used for this purpose. The controller specification defines the operating ranges of each synchro signal. At the beginning of each test the frequency and start time of random synchro signal generators are pre-calculated. Thus, it is possible to detect errors in synchronization of internal units: the detection of metastability, desynchronization of releasing requests and data, sticking data in some positions of the buffers. 4. Functional verification of communication controller – Host- Bridge Host-Bridge is a part of northbridge of microprocessor MCST-R2000 CPU with Sparc V9 architecture developing by MCST. The Host-bridge interface communication controller connects the system with external devices, accepting requests from the system and the I/O space while maintaining the transaction formats accepted in the system and I/O space. The Host-bridge receives requests from the System Commutator, communicate with two I/O channel controllers (IO- links) and provide translation of the virtual address to physical addresses. In addition, controller provides access to the system registers, registers of inter- processor links, memory controller’s registers, transferring the new values of the registers to the local copies of them, transmitting interrupts, status signals and collecting snoop-responses. Each type of the registers has its own interface for communication with the destination device. All these features should be taken into account when verifying Host-bridge. Some approaches for standalone verification using UVM and reference models were observed in [2, 6-8]. In [2] authors used buffers between testbench and reference model for checking marked transactions. In [6, 7] assertions and checking reference model were applied. In [8] there was reference model with very complicated algorithms. In our work, we used model buffers and assertions for checking correctness of transactions. In addition, we present a number of new solutions of the standalone verification process, which was applied for verification of Host-Bridge. 187 Lebedev D.A., Stotland I.A. Construction of validation modules based on reference functional models in a standalone verification of communication subsystem. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 183-194 Lebedev D.A., Stotland I.A. Construction of validation modules based on reference functional models in a standalone verification of communication subsystem. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 183-194 4.3 Verification of address translation controller One of the components of the Host-Bridge is the address translation controller IOMMU. It translates a virtual address received from the I/O subsystem requests to physical addresses. The controller sends a request for information about the physical address to a special memory area for providing translation. Virtual address mapping to physical addresses is stored in a special controller buffer – IOTLB (Input-Output Translation Lookaside Buffer). If the buffer is full, the oldest element is displaced. Translation Lookaside Buffer). If the buffer is full, the oldest element is displaced. The algorithm of the translation could be represented in the form of several consecutive steps: 1) receiving DMA request p ← start(x), 2) analyzing of the input request then matching in the cache IOMMU with the following scenarios: 2) analyzing of the input request then matching in the cache IOMMU with the following scenarios: a) match is found (hit IOMMU) – a request with the translated address x'' is executed; b) match is not found (miss IOMMU) – a request for a physical address x' is executed , then waiting for a response p.receive(y) with the data, and only after that translation of the address is done. b) match is not found (miss IOMMU) – a request for a physical address x' is executed , then waiting for a response p.receive(y) with the data, and only after that translation of the address is done. 188 Under dynamic test conditions, there may be situations when a lane in the IOMMU cache is not yet displaced in the RTL-model and the address can be translated without additional request, but it is not present in the reference model. In this case, Лебедев Д.А., Стотланд И.А. Построение модулей проверки на основе эталонных функциональных моделей при автономной верификации подсистемы связи. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 183-194 the reference model will give unnecessary requests to the test system. A global transaction counter was introduces in the reference model to solve this problem. The task of this counter was to identify the source of the requests. In addition, the responses generated by the test system are analyzed too. 4.3 Verification of address translation controller In the case when the request is successfully translated on the RTL-model side, and the reference model has already given an extra request for a physical address, the test environment generates a response that is marked with a special identifier and sent it to the reference model. When processing a response, the model concludes that the translation was not performed p.model_check(y), calculates the desired transaction identifier y.id and sends it to a special buffer of canceled requests 𝑄𝑖𝑑 for the physical address. When checking the interchange buffers with reference model after the test completes, the identifier values in the buffer of canceled requests 𝑄𝑟𝑒𝑞 are compared with the identifiers of the remaining unprocessed requests for physical addresses from the reference model. Such remaining requests are not treated as erroneous and are deleted delete(req.id). The pseudo-code of the algorithm is presented below. DMA handling: while true do wait p ← start(x) if 𝑥′ then p.ncheck(𝑥′) else if 𝑥′′ then begin wait p.receive(y) p.model_check(y) 𝑄𝑖𝑑← y.id p.finish() end end After test checking: for i ∈ 𝑄𝑟𝑒𝑞 do if c.check(req.id, 𝑄𝑖𝑑) then delete(req.id) else report(req.id) end DMA handling: while true do wait p ← start(x) if 𝑥′ then p.ncheck(𝑥′) else if 𝑥′′ then begin wait p.receive(y) p.model_check(y) 𝑄𝑖𝑑← y.id p.finish() end end After test checking: for i ∈ 𝑄𝑟𝑒𝑞 do if c.check(req.id, 𝑄𝑖𝑑) then delete(req.id) else report(req.id) end 4.4 The correct organization of the exchange in terms of the uncertainty of the issuance of queries In high-load dynamic tests with many input requests and responses, labeling requests and responses with tags that correspond to positions in the controller’s buffers may differ from the values of tags in the reference model. This happens because of the inability to predict time of release of the buffer's position in the 189 Lebedev D.A., Stotland I.A. Construction of validation modules based on reference functional models in a standalone verification of communication subsystem. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 183-194 event-driven models. Therefore, it is important to match the input and output requests of the model and the verified device. Each request, whether it is an I/O request or a PIO request, has several stages of execution. To ensure the correct functioning of the test scenarios we need to use an associative memory (mappers) for matching. The function of this memory is to store the matching of RTL request tags with reference model tags, when the remaining request data field, such as an address, destination device ID, processor number, and others are compared. Later, when you receive responses to the request, you have to pass to the model the same tag, which was allocated by the model at the stage of forming the request. event-driven models. Therefore, it is important to match the input and output requests of the model and the verified device. Each request, whether it is an I/O request or a PIO request, has several stages of execution. To ensure the correct functioning of the test scenarios we need to use an associative memory (mappers) for matching. The function of this memory is to store the matching of RTL request tags with reference model tags, when the remaining request data field, such as an address, destination device ID, processor number, and others are compared. Later, when you receive responses to the request, you have to pass to the model the same tag, which was allocated by the model at the stage of forming the request. Communication controllers in multi-core systems can participate in the coherence protocol and accept snoop responses. Depending on the mode of operation and the content of the fields of the first received for the request response could come as several responses or only one. 4.5 After test checking The correct behavior of the communication controller is determined in providing a certain number of responses to requests and receiving the exact number of responses to them. Incorrect operation can be identified by counting the number of received requests, converted to another format requests, and accepted responses. For this purpose, the test system includes transaction counters. They capture all kinds of transactions while the test is running. At the end of the test, special algorithm checks values of these counters and make a conclusion about correctness. After the test scenario is complete, it also needs to verify absence of unanswered requests in the buffers that link the reference model to the test environment. The presence of such requests signals about error of either the verified device or the reference model. 4.4 The correct organization of the exchange in terms of the uncertainty of the issuance of queries To complete the request check in coherent mode, it needed to pass all the responses with the correct tags to the model Communication controllers in multi-core systems can participate in the coherence protocol and accept snoop responses. Depending on the mode of operation and the content of the fields of the first received for the request response could come as several responses or only one. To complete the request check in coherent mode, it needed to pass all the responses with the correct tags to the model. 6. Conclusion Communication controllers are among the important parts of multi-core microprocessor systems have to be thoroughly tested. The principles described in the article do not depend mainly on the implementation of these controllers and allow their full standalone verification. The article suggests the ways to organize the interaction of the test system and the event reference model, as well as ways to resolve the difficulties encountered in the development of the test system. The proposed approaches have been applied in the verification of the controller Host-bridge as a part of eight-core microprocessor, developed by "MCST". The developed test system and tests made it possible to detect and correct a number of logical errors that were not detected by other test methods. In the future, it is planned to expand the test environment by adding a part of the interrupt system transmitting signals directly to the core. Fig. 3 shows a generalized diagram of a Host¬bridge, and the dotted lines illustrate such extension. HOST BRIDGE CPU I/O links registers JTAG status interrupts aerrs interfaces signals apic apic Fig 3. A typical structure of a Host Bridge controller connected with interrupt controllers I/O links Fig 3. A typical structure of a Host Bridge controller connected with interrupt controllers 5. Results The approaches described above were applied to standalone verification of the Host- Bridge of microprocessor MCST «R2000». Parametrization of synchro signals allowed finding metastability in the controller interfaces. Using different types of credit exchange rate helped to locate deadlocks and livelocks in the controller. Based on one test system the built-in IOMMU controller was also verified. Different configuration of answers with physical address were verified, which helps finding errors in displacement algorithms. After test, checking of model buffers and requests counters provide finding of not released responses to the system. For the Host-Bridge controller, due to its functional and structural features that belong to the class of communication controllers, a test environment is developed with a checker based on reference event-model. Due to standalone verification of the device 67 errors that have not been found by other means of verification were found and corrected. Code and functional coverage was carried out and 94% The approaches described above were applied to standalone verification of the Host- Bridge of microprocessor MCST «R2000». Parametrization of synchro signals allowed finding metastability in the controller interfaces. Using different types of credit exchange rate helped to locate deadlocks and livelocks in the controller. Based on one test system the built-in IOMMU controller was also verified. Different configuration of answers with physical address were verified, which helps finding errors in displacement algorithms. After test, checking of model buffers and requests counters provide finding of not released responses to the system. For the Host-Bridge controller, due to its functional and structural features that belong to the class of communication controllers, a test environment is developed with a checker based on reference event-model. Due to standalone verification of the device 67 errors that have not been found by other means of verification were found and corrected. Code and functional coverage was carried out and 94% 190 Лебедев Д.А., Стотланд И.А. Построение модулей проверки на основе эталонных функциональных моделей при автономной верификации подсистемы связи. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 183-194 coverage was extracted. Gaps in coverage will be eliminated with the further expanding of the test environment with external parts of interrupt system. Total result indicates about effectiveness of standalone verification of communication controller. [1]. Kamkin A.M., Kotsynyak S.A, Smolov A.A.. Sortov A.D., Tatarnikov, Chupilko M.M. Tools for functional verification of microprocessors. Trudy ISP RAN/Proc. ISP RAS, vol. 26, issue 1, 2014, pp. 149-200 (in Russian). References [1]. Kamkin A.M., Kotsynyak S.A, Smolov A.A.. Sortov A.D., Tatarnikov, Chupilko M.M. Tools for functional verification of microprocessors. Trudy ISP RAN/Proc. ISP RAS, vol. 26, issue 1, 2014, pp. 149-200 (in Russian). 191 191 Lebedev D.A., Stotland I.A. Construction of validation modules based on reference functional models in a standalone verification of communication subsystem. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 183-194 Lebedev D.A., Stotland I.A. Construction of validation modules based on reference functional models in a standalone verification of communication subsystem. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 183-194 [2]. Shmelev V.A., Stotland I.A. Standalone verification of microprocessors using reference models with various levels of abstraction. Problemy razrabotki perspektivnyh mikro- i nanojelektronnyh sistem [Problems of development of promising micro- and nanoelectronic systems], no 1, 2012, pp. 435-440 (in Russian).. [3]. A.N. Meshkov, M.P. Ryzhov, V.A. Shmelev. The developement of the verification tools of the Elbrus-2S microprocessor. Voprosy radioelektroniki [Issues of radio electronics], ser. EVT, 2014, no. 3, pp. 5-17 (in Russian). [4]. Standard Universal Verification Methodology http://accellera.org/downloads/standards/uvm (1 5]. Kelton W., Law A. Simulation modeling. Classics of CS. 3rd ed. SPb.: Piter, 200 [6]. Li-Bo Cheng, Francis Anghinolfi, Ke Wang, Hong-Bo Zhu, Wei-Guo Lu, Zhen-An Liu. A UVM Based Testbench Research for ABCStar. In Proc. of the IEEE-NPSS Real Time Conference (RT), 2016. https://indico.cern.ch/event/390748/contributions/1825090/attachments/1280814/190641 3/CR_PosterSession2_268.pdf (12.06.2018). [7]. Abhineet Bhojak, Tejbal Prasad. A UVM Based Methodology for Processor Verification. Proc. of the Design and Verification Conference and Exhibition (DVCON), 2015. https://dvcon-india.org/sites/dvcon- india.org/files/archive/2015/proceedings/6_UVM_Based_Processor_Verification_paper. pdf (12.06.2018). [8]. Kamkin A., Petrochenkov M. A Model-Based Approach to Design Test Oracles for Memory Subsystems of Multicore Microprocessors. Trudy ISP RAN/Proc. ISP RAS, vol. 27, issue 3, 2015, pp. 149-160. DOI: 10.15514/ISPRAS-2015-27(3)-11. DOI: 10.15514/ISPRAS-2018-30(3)-13 Для цитирования: Лебедев Д.А., Стотланд И.А. Построение модулей проверки на основе эталонных функциональных моделей при автономной верификации подсистемы связи. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 183-194 (на английском языке). DOI: 10.15514/ISPRAS-2018-30(3)-13 Построение модулей проверки на основе эталонных функциональных моделей при автономной верификации подсистемы связи Дмитрий Лебедев <lebedev_d@mcst.ru> Ирина Стотланд <stotl_i@ mcst.ru > АО «МЦСТ», 119334, Москва, Россия, ул. Вавилова, д. 24 Аннотация. В статье предложены подходы к функциональной верификации контроллеров сопряжения интерфейсов в составе микропроцессоров на основе разработки многоуровневых тестовых-систем по методологии UVM. В современных микропроцессорных системах существует множество контроллеров, работающих с собственными типами данных. Контроллеры сопряжения интерфейсов учувствуют в передаче и преобразовании данных между блоками микропроцессора. Такое преобразование должно осуществляться быстро и без повреждения данных для корректного функционирования всей системы. Контроллеры сопряжения интерфейсов могут выполнять дополнительные функции, такие как передача значений копий системных регистров, преобразование адресов и другие. В статье дан краткий обзор средств верификации и преимуществ применения автономной имитационной верификации для проверки корректности контроллеров сопряжения интерфейсов в составе подсистем связи. Представлены подходы к построению автономной 192 Лебедев Д.А., Стотланд И.А. Построение модулей проверки на основе эталонных функциональных моделей при автономной верификации подсистемы связи. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 183-194 Лебедев Д.А., Стотланд И.А. Построение модулей проверки на основе эталонных функциональных моделей при автономной верификации подсистемы связи. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 183-194 Лебедев Д.А., Стотланд И.А. Построение модулей проверки на основе эталонных функциональных моделей при автономной верификации подсистемы связи. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 183-194 верификационной тестовой системы на основе методологии UVM с модулем проверки, реализованным во внешней функциональной эталонной модели. Так же предложены методы проверки корректности подсистем связи: проверка контроллеров, работающих с несколькими синхросигналами при помощи параметризованного генератора синхросигналов, поддержка механизмов обмена кредитами. Представленные подходы использованы для верификации подсистемы связи - Host-Bridge - восьмиядерного микропроцессора с архитектурой Sparc V9, разработанного компанией АО «МЦСТ». В статье описаны проблемы, обнаруженные в процессе разработки тестовой системы и способы их разрешения. Представлены результаты использования рассмотренных решений для верификации контроллеров подсистем связи и дальнейший план совершенствования тестовой системы. верификационной тестовой системы на основе методологии UVM с модулем проверки, реализованным во внешней функциональной эталонной модели. Так же предложены методы проверки корректности подсистем связи: проверка контроллеров, работающих с несколькими синхросигналами при помощи параметризованного генератора синхросигналов, поддержка механизмов обмена кредитами. Представленные подходы использованы для верификации подсистемы связи - Host-Bridge - восьмиядерного микропроцессора с архитектурой Sparc V9, разработанного компанией АО «МЦСТ». В статье описаны проблемы, обнаруженные в процессе разработки тестовой системы и способы их разрешения. Представлены результаты использования рассмотренных решений для верификации контроллеров подсистем связи и дальнейший план совершенствования тестовой системы. Ключевые слова: тестовая система; контроллер сопряжения интерфейсов; функциональная верификация; Universal Verification Methodology (UVM); эталонная модель. [8]. Kamkin A., Petrochenkov M. A Model-Based Approach to Design Test Oracles for Memory Subsystems of Multicore Microprocessors. Trudy ISP RAN/Proc. ISP RAS, vol. 27, issue 3, 2015, pp. 149-160. DOI: 10.15514/ISPRAS-2015-27(3)-11. Lebedev D.A., Stotland I.A. Construction of validation modules based on reference functional models in a standalone verification of communication subsystem. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 183-194 india.org/files/archive/2015/proceedings/6_UVM_Based_Processor_Verification_paper. pdf (12.06.2018). Lebedev D.A., Stotland I.A. Construction of validation modules based on reference functional models in a standalone verification of communication subsystem. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 183-194 Lebedev D.A., Stotland I.A. Construction of validation modules based on reference functional models in a standalone verification of communication subsystem. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 183-194 india.org/files/archive/2015/proceedings/6_UVM_Based_Processor_Verification_paper. pdf (12.06.2018). [8]. Kamkin A., Petrochenkov M. A Model-Based Approach to Design Test Oracles for Memory Subsystems of Multicore Microprocessors. Trudy ISP RAN/Proc. ISP RAS, vol. 27, issue 3, 2015, pp. 149-160. DOI: 10.15514/ISPRAS-2015-27(3)-11. verification of communication subsystem. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 183-194 india.org/files/archive/2015/proceedings/6_UVM_Based_Processor_Verification_paper. pdf (12.06.2018). Список литературы [1]. А.С. Камкин, А.М. Коцыняк, С.А. Смолов, А.А. Сортов, А.Д. Татарников, М.М. Чупилко. Средства функциональной верификации микропроцессоров. Труды ИСП РАН, том 26, вып. 1, 2014, стр. 149-199. [2]. Шмелев В.А., Стотланд И.А. Автономная верификация микропроцессоров на основе эталонных моделей разного уровня абстракции. Проблемы разработки перспективных микро- и наноэлектронных систем (МЭС), No. 1, 2012, стр. 435- 440. [3]. А.Н. Мешков, М.П. Рыжов, В.А. Шмелев. Развитие средств верификации микропроцессора «Эльбрус-2S». Вопросы радиоэлектроники, сер. ЭВТ, вып.3, 2014, стр. С. 5-17. [4]. Standard Universal Verification Methodology http://accellera.org/downloads/standards/uvm (12.06.2018). [4]. Standard Universal Verification Methodology http://accellera.org/downloads/standards/uvm (12.06.2018). [5]. Кельтон В., Лоу А. Имитационное моделирование. Классика CS. 3-е изд. СПб.: Питер, 2004. [6]. Li-Bo Cheng, Francis Anghinolfi, Ke Wang, Hong-Bo Zhu, Wei-Guo Lu, Zhen-An Liu. A UVM Based Testbench Research for ABCStar. In Proc. of the IEEE-NPSS Real Time Conference (RT), 2016. https://indico.cern.ch/event/390748/contributions/1825090/attachments/1280814/190641 3/CR_PosterSession2_268.pdf (12.06.2018). [7]. Abhineet Bhojak, Tejbal Prasad. A UVM Based Methodology for Processor Verification. Proc. of the Design and Verification Conference and Exhibition (DVCON), 2015. https://dvcon-india.org/sites/dvcon- 193 194 DOI: 10.15514/ISPRAS-2018-30(3)-14 For citation: Petrochenkov M.V., Mushtakov R.E., Shpagilev D.I. Verification of System on Chip Integrated Communication Controllers. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 195-206. DOI: 10.15514/ISPRAS-2018-30(3)-14 For citation: Petrochenkov M.V., Mushtakov R.E., Shpagilev D.I. Verification of System on Chip Integrated Communication Controllers. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 195-206. DOI: 10.15514/ISPRAS-2018-30(3)-14 Verification of System on Chip Integrated Communication Controllers M.V. Petrochenkov <petroch_m@mcst.ru> R.E. Mushtakov <mushtakov_r@mcst.ru> D.I. Shpagilev <shpagilev_d@mcst.ru> MCST, 1 Nagatinskaya st., Moscow, 117105, Russia Abstract. This article presents an approach used to verify communication controllers developed for Systems on Chip (SOC) at MCST. We provide a list of communication controllers developed in MCST and present their characteristics. We describe principles of communication controller’s operation on transaction, data link and physical layers and highlight their similarities. Then we describe a common method of device verification: principles of test system design, constrained random test stimuli generation and checking of device behavior. Based on common features of the controllers, we provide the general design of their test system. It includes components to work with transaction level interface (system agent of system on chip communication protocol) and physical interface (physical agent of protocol for SOC communication on a single board), configuration agent that determines device mode of operation and a scoreboard. Because controllers only execute transformation of transactions between different representation, scoreboard checks accordance of in and outgoing transactions. In addition, we describe specific features of devices that require the adjustments to the common approach. We describe how verification of those features affected the design of different test systems. We explain how a replacement of a physical agent with a second communication controller allows to speed up the development of test systems. We explain challenges of link training and status state machine (LTSSM) verification. We provide a way to work with devices with direct memory access (DMA) in a system agent. In conclusion, we present a list of found errors and directions of further research. Keywords: Elbrus; system on chip; communication controller; Ethernet; DDR4; PCI Express; UVM; stand-alone verification DOI: 10.15514/ISPRAS-2018-30(3)-14 1. Introduction Modern systems on chip (SOC) may include multiple microprocessor cores, complex hierarchy of caches, peripheral controllers and other types of data processing modules. The task of interconnection between different systems on chip is solved by communication controller (CC) modules. Those modules solve the 195 Petrochenkov M.V., Mushtakov R.E., Shpagilev D.I. Verification of System on Chip Integrated Communication Controllers. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 195-206 problem of interprocessor communications, communication between CPU and random access memory (RAM), CPU and peripheral devices, network interfaces, etc. Performance and reliability of communication controllers is crucial for the quality of the whole system. To ensure that communication controllers satisfy all requirements, they must be thoroughly verified. Verification of complex communication controllers is a time-consuming task [1]. One of the widely used approaches to verification of SOC is system verification - execution of test programs (implemented in assembly language) on the model of microprocessor. Another approach is stand-alone verification of SOC components. In this approach, model of the device under verification (DUT) is included in a special program – a test system, which goal is to ensure that DUT satisfies all requirements. This article describes a problem of stand-alone verification of communication controllers with physical media access interfaces in the industrial setting. The rest of the paper is organized as follows. Section 2 describes communication controllers for physical media access interfaces developed by MCST company. Section 3 presents a common approach to the design a test system and describes its components. In section 4 we provide a case study for suggested approach applied to specific devices, and adjustments to the approach that were implemented to verify specific features of those devices. In conclusion, we present of verification and provide a direction of further research. 2. Overview of communication controllers in «Elbrus-16C» microprocessor “Elbrus-16C” System on Chip includes many communication controllers. In the following list we will describe ones that require the stand-alone verification: the most complex ones and the ones which reliability is crucial for the functionality of the system. 1. DDR4 Memory Controller is a digital circuit that manages the flow of data going to and from the computer's main memory. The controller contains the logical circuits necessary to perform read and write operations in DRAM, with all necessary delays (for example, between reading and writing). The flow of incoming requests is converted into sequences of DRAM commands, while monitoring various conflicts on banks, buses and channels. To increase the effective bandwidth of the memory channel, incoming requests can be buffered and reordered. The reordering mechanism is implemented on the basis of a sequential combination filter system. 2. PCI Express Root Complex (RC) Controller transforms packets from in- house protocol to standard PCI Express transaction level packets and implements RC configuration space for communication with peripheral devices. The controller is connected directly to on-chip network to improve throughput and reduce delays. The controller supports up to 16 lanes with speed up to 8 GT/s [2]. 2. PCI Express Root Complex (RC) Controller transforms packets from in- house protocol to standard PCI Express transaction level packets and implements RC configuration space for communication with peripheral devices. The controller is connected directly to on-chip network to improve throughput and reduce delays. The controller supports up to 16 lanes with speed up to 8 GT/s [2]. 2. PCI Express Root Complex (RC) Controller transforms packets from in- house protocol to standard PCI Express transaction level packets and implements RC configuration space for communication with peripheral devices. The controller is connected directly to on-chip network to improve throughput and reduce delays. The controller supports up to 16 lanes with speed up to 8 GT/s [2]. 196 Петроченков М.В., Муштаков Р.Е. Шпагилев Д.И. Верификация контроллеров связи в системах на кристалле. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 195-206 Петроченков М.В., Муштаков Р.Е. Шпагилев Д.И. Верификация контроллеров связи в системах на кристалле. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 195-206 Петроченков М.В., Муштаков Р.Е. Шпагилев Д.И. Верификация контроллеров связи в системах на кристалле. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 195-206 3. 2. Overview of communication controllers in «Elbrus-16C» microprocessor Inter-Processor Communication Controller (IPCC) is designed to solve problems of organization of multiprocessor architectures with shared memory [3]. IPCC functions are logically divided into two levels: the link layer (DLL - Data Link Layer) and the physical layer (PHL - Physical Layer). Exchange by link is carried out by transport packages (containers) of fixed size. Packages contain information about the type of the channel, data, as well as the CRC checksum. Packages are formed into containers according to special rules in order to ensure the priority and maximize the bandwidth of the link. The protocol packets are distributed among several virtual channels (VC) or streams with different priorities. To ensure the integrity of the data during the transmission over the link, the mechanism of sequential container numbering and CRC encoding are used. 3. Inter-Processor Communication Controller (IPCC) is designed to solve problems of organization of multiprocessor architectures with shared memory [3]. IPCC functions are logically divided into two levels: the link layer (DLL - Data Link Layer) and the physical layer (PHL - Physical Layer). Exchange by link is carried out by transport packages (containers) of fixed size. Packages contain information about the type of the channel, data, as well as the CRC checksum. Packages are formed into containers according to special rules in order to ensure the priority and maximize the bandwidth of the link. The protocol packets are distributed among several virtual channels (VC) or streams with different priorities. To ensure the integrity of the data during the transmission over the link, the mechanism of sequential container numbering and CRC encoding are used. 4. Wide Link Communication Controller (WLCC) is used to connect south bridge controller to SOC using a protocol similar to PCI Express 2.0 but with reduced overhead. Controller supports memory and configuration space access operations. Supported link width is up to 16 lanes with speed 2.5 or 5 GT/s for each lane. To ensure channel reliability transmitted packets are protected by 16 bit CRC. After transmission, packets are stored in replay buffer waiting for receive confirmation. If negative packet acknowledge is received or time-out is reached, packets are retransmitted. Controller supports up to 8 virtual channels. 4. Wide Link Communication Controller (WLCC) is used to connect south bridge controller to SOC using a protocol similar to PCI Express 2.0 but with reduced overhead. Controller supports memory and configuration space access operations. 2. Overview of communication controllers in «Elbrus-16C» microprocessor ISP RAS, vol. 30, issue 3, 2018, pp. 195-206 Petrochenkov M.V., Mushtakov R.E., Shpagilev D.I. Verification of System on Chip Integrated Communication Controllers. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 195-206  Controllers don’t possess complex internal state and don’t implement complex data processing or caching mechanisms. They transform packets between different representations: system level communication protocol packets (used for on-chip communications) and physical interface signals (used for communication on distances beyond the single chip).  Controllers implement data link layer (DLL) that performs error detection and/or correction using such mechanisms as Cyclic Redundancy Checks (CRC) or forward error correction (FEC).  Controllers implement data link layer (DLL) that performs error detection and/or correction using such mechanisms as Cyclic Redundancy Checks (CRC) or forward error correction (FEC).  Controllers expose the physical interface and implement logical and electrical parts of physical layer for communication with other components on a board. All aforementioned controllers communicate using low voltage differential signaling (LVDS). To ensure clock recovery and dc balancing devices use physical encoding schemes (for example 8b/10b, 64b/66b, 128b/130b) and signal scrambling. 2. Overview of communication controllers in «Elbrus-16C» microprocessor Supported link width is up to 16 lanes with speed 2.5 or 5 GT/s for each lane. To ensure channel reliability transmitted packets are protected by 16 bit CRC. After transmission, packets are stored in replay buffer waiting for receive confirmation. If negative packet acknowledge is received or time-out is reached, packets are retransmitted. Controller supports up to 8 virtual channels. 5. 10 Gigabit Ethernet Controller uses 10GBASE-KR interface [4]. It sends and receives Ethernet frames over backplane electrical interface. On a physical layer, it supports procedures of Clause 73 Auto-negotiation and Clause 72 Auto-adaptation. This device supports hardware calculation and checking of Ethernet CRC, IPv4, TCP and UDP checksums, various filtering mechanisms based on MAC addresses and VLAN tags and automatic handling of pause frames. 5. 5. 10 Gigabit Ethernet Controller uses 10GBASE-KR interface [4]. It sends and receives Ethernet frames over backplane electrical interface. On a physical layer, it supports procedures of Clause 73 Auto-negotiation and Clause 72 Auto-adaptation. This device supports hardware calculation and checking of Ethernet CRC, IPv4, TCP and UDP checksums, various filtering mechanisms based on MAC addresses and VLAN tags and automatic handling of pause frames. 6. Gigabit Ethernet Controller uses 1000BASE-KX interface [4]. Ethernet frames are sent using backplane electrical interface. It supports calculation and checking of Ethernet frame CRC, calculation and checking of IPv4, TCP and UDP checksums, filtering based on mac and IP addresses and automatic handling of pause frames. Despite the fact that those devices implement sufficiently different protocols, they nonetheless solve a lot of similar problems and implement similar features. Common features of controllers are: Despite the fact that those devices implement sufficiently different protocols, they onetheless solve a lot of similar problems and implement similar features. Common features of controllers are:  Register transfer level (RTL) models of this devices are implemented using Verilog and SystemVerilog [5] hardware description languages. Register transfer level (RTL) models of this devices are implemented usin Verilog and SystemVerilog [5] hardware description languages.  Controllers communicate with other components on chip using the system interface that implements on-chip communication protocol, and represents transaction layer of the device. 197 Petrochenkov M.V., Mushtakov R.E., Shpagilev D.I. Verification of System on Chip Integrated Communication Controllers. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 195-206 Petrochenkov M.V., Mushtakov R.E., Shpagilev D.I. Verification of System on Chip Integrated Communication Controllers. Trudy ISP RAN/Proc. 3. Test system structure Test systems are usually implemented using either general purpose programming languages (C++), hardware description languages (VHDL, Verilog) or dedicated verification languages (SystemVerilog, e, OpenVera). In our company we use SystemVerilog [5] with Universal Verification Methodology [6] (UVM). Use of this language allows for an easy interface with Verilog and SystemVerilog devices, and UVM describes a general test system structure and provides a library of basic verification components. p Fig 1. Structure of test system of communication controllers Fig 1. Structure of test system of communication controllers Fig 1. Structure of test system of communication controllers 198 Петроченков М.В., Муштаков Р.Е. Шпагилев Д.И. Верификация контроллеров связи в системах на кристалле. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 195-206 Петроченков М.В., Муштаков Р.Е. Шпагилев Д.И. Верификация контроллеров связи в системах на кристалле. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 195-206 Common principles of controller behaviour determine the general structure of the test system. All test systems include a set of basic components. Test system structure is presented in fig.1. Common principles of controller behaviour determine the general structure of the test system. All test systems include a set of basic components. Test system structure is presented in fig.1. A. Test stimuli generators are based on constrained randomization. In our case, stimuli generators communicate with system and physical interfaces of DUT. Transactions are described in terms of their attributes and constraints. To specify some test scenario, one must define specific constraints for transactions that will be issued by request generators. SystemVerilog offers a native support for constrained randomization constructs. In addition to transaction transmission and reception. physical agent is able to model some “non-standard” types of behavior: injection of corrupted or non-standard compliant transactions, or handling of received transactions in user-specified way (for example, send negative acknowledge for non-corrupted packet, drop the response to request from DUT, etc..). B. Test system scoreboard implements a correctness checks. Devices under verification do not possess complex data processing logic and simply perform transformation of transactions between different representations. Scoreboard receives transactions from system and physical interface monitors and performs comparison between ingress and egress transaction. If discrepancy between expected (transmitted) and received packets is detected, module reports an error in the test system. B. C. In addition to global test system scoreboard, test system contains local (system and physical) interface protocol checkers. 3. Test system structure Their goal is to check that interface rules and invariants are not violated and otherwise report an error. D. Configuration agent is used to access a set of memory-mapped configuration registers in the controllers. Those registers are accessed using separate configuration interface. Initial phase of a test is writing desired values to this registers. D. Configuration agent is used to access a set of memory-mapped configuration registers in the controllers. Those registers are accessed using separate configuration interface. Initial phase of a test is writing desired values to this registers. D. Configuration agent is used to access a set of memory-mapped configuration registers in the controllers. Those registers are accessed using separate configuration interface. Initial phase of a test is writing desired values to this registers. 4. Case study This chapter describes the adjustment and highlights specific implementation details of different test systems. 4.1 Verification of Link Training and Status State Machine One of the features of PCI Express, WLCC and IPCC links is a complex procedure of link initialization and training. During the initialization procedure device sends data patterns containing device capabilities and its current state across the link. Those data patterns are called a training sequence (TS). At the same time, using information from received training sequences, the controller detects the presence of a link partner, determines its active lanes and abilities. Based on this information, 199 Petrochenkov M.V., Mushtakov R.E., Shpagilev D.I. Verification of System on Chip Integrated Communication Controllers. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 195-206 Petrochenkov M.V., Mushtakov R.E., Shpagilev D.I. Verification of System on Chip Integrated Communication Controllers. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 195-206 pair of devices establishes common mode of operation for transaction transfer. In addition, training sequences are used to change the state of the link (for example, from active to low power mode or to the disabled state). pair of devices establishes common mode of operation for transaction transfer. In addition, training sequences are used to change the state of the link (for example, from active to low power mode or to the disabled state). Presence of the LTSSM provides several additional challenges for the device verification.  To send the transactions across the link, the active link must be established first. Thus, first action that the controller and its physical link agent partner performs is a link training sequence.  One must test ability of the device to change its state and check that it reacts correctly to the state change of the link partner.  One must test ability of the device to change its state and check that it reacts correctly to the state change of the link partner.  In addition to “main” device states there are several “transient” states that the device passes when switching from one main state to another. Depending on training sequences received from link partner in transient states, link training procedure either continues successfully or terminates while reporting the error status. It should be said that, despite the internal complexity of LTSSM protocols, they are almost invisible to the transaction layer. Only information available to transaction layer is whenever link is currently active or not. 4.2 Test systems based on a pair of controllers To verify implementations of in-house communication protocols (IPCC and WLCC) additional type of test system was used [7]. It is based on the pair of RTL-models of communication controllers. In these test systems two controllers are connected using their corresponding physical interfaces. Errors are injected by manipulating the signals of physical interface. The structure of the test system is presented in fig.2. Fig 2. Structure of test system based on a pair of controllers Fig 2. Structure of test system based on a pair of controllers Advantages of the approach are as following. Advantages of the approach are as following.  Simulation of device behaviour in realistic scenarios. Those devices (IPCC and WLCC) use our company’s proprietary protocols to connect identical 200 Петроченков М.В., Муштаков Р.Е. Шпагилев Д.И. Верификация контроллеров связи в системах на кристалле. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 195-206 Петроченков М.В., Муштаков Р.Е. Шпагилев Д.И. Верификация контроллеров связи в системах на кристалле. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 195-206 devices, developed in-house. Thus, test system of this kind represents a realistic use-case of the device. devices, developed in-house. Thus, test system of this kind represents a realistic use-case of the device. devices, developed in-house. Thus, test system of this kind represents a realistic use-case of the device.  Simplicity of implementation. The development of physical level agent is a labor-intensive and time-consuming, and its development cannot be avoided by purchasing a third party Verification IP (VIP). In this approach, the development of only a system agent is necessary, and verification can start earlier. Disadvantages are as following.  Lower simulation performance is caused by the need to simulate two identical controllers. This doubles the required computational resources.  Lower simulation performance is caused by the need to simulate two identical controllers. This doubles the required computational resources.  More difficult state and error injection control. To inject errors into sent and received transactions one must either directly manipulate external signals of the controller or use hierarchical access to modify the behaviour of the controllers.  Inability to detect “self-correcting” bugs (for example, incorrect CRC polynomial). This disadvantage is mitigated by the fact those bugs will also self-correct in “real” device.  Absence of checks on lower protocol levels. The main way to detect an error is to receive an unexpected packet on system interfaces. 4.2 Test systems based on a pair of controllers This may cause difficulties in bug detection and localization in many cases. For example, an error that causes an incorrect request to repeat a transaction can be detected only by performance degradation. One can reduce the disadvantages while keeping most of some of the advantages of the approach by adding physical monitor on a link between devices. 4.3 Complex system agent in the Ethernet test systems Distinctive feature of Ethernet test systems (both 10 Gigabit and Gigabit) is a complex system agent [8]. To reduce CPU usage and increase device efficiency controllers implement Direct Memory Access (DMA). Instead of sending Ethernet frames directly to device interfaces, frames are stored in system memory and the device reads the memory when it is ready for frame transmission. In a same way, the system must prepare a memory space for device to store received frames The device will write the data to this location after the frame reception. Ethernet controllers are managed using a set of memory-mapped registers. The most important ones are descriptor pointer registers (head and tail). Descriptors contain an Ethernet frame metadata (size of frame, memory location address, higher-level protocol information, etc...). The head register points to the first descriptor available to the controller, and the tail points to the last processed by it. Using those registers the controller reads and writes transaction descriptors and a frame memory. The structure of Ethernet agents is presented in fig. 3. 201 Petrochenkov M.V., Mushtakov R.E., Shpagilev D.I. Verification of System on Chip Integrated Communication Controllers. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 195-206 .V., Mushtakov R.E., Shpagilev D.I. Verification of System on Chip Integrated Communication dy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 195-206 Controllers. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 195 206 Fig 3. Structure of Ethernet controller test system Fig 3. Structure of Ethernet controller test system 4.4 DDR4 Memory Controller protocol checks A system agent in the memory controller test system consists of a set of two modules: the management agent of the information written into the memory and the agent for transferring requests from the system to the controller. The test system requires more sophisticated physical protocol checkers. For this purpose, two modules are used: the DFI protocol verification module and the DDR protocol verification module. Before active work with the memory is started, the controller performs programming of the operating modes of the DRAM memory modules, conducts its initialization and training. To verify these processes, the DDR Protocol Checker is used. In addition to the fact that the module monitors the initialization and training of the memory, it also controls the execution of all the time constraints imposed to the controller when it issues commands to the memory. Another important function of the memory controller is to periodically update the data stored in the DRAM using a refresh command. Without periodic updates, DRAM memory chips would gradually lose information, as capacitors storing bits are discharged by leakage currents. DDR protocol checker is used to analyze transactions on physical interface and to check if Refresh commands are issued within specified timing constraints. In addition, the memory state is checked before executing the Refresh command. The memory must be in the IDLE state. The controller has built-in noise immunity mechanisms that allow to check the integrity 202 Петроченков М.В., Муштаков Р.Е. Шпагилев Д.И. Верификация контроллеров связи в системах на кристалле. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 195-206 Петроченков М.В., Муштаков Р.Е. Шпагилев Д.И. Верификация контроллеров связи в системах на кристалле. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 195-206 of the data, and to correct it if necessary. Such mechanisms include: rectification of parity errors of the DDR bus, calculation of checksums, correction of CRC errors on the data bus of the DFI interface while writing, and correction of ECC errors on the DFI data bus during reading. Verification of noise immunity of transmitted data is provided by the DFI Protocol Checker. In addition, checker provides a way to verify the process of switching to and from power saving modes of memory chips by checking their timing parameters. Table 1. Results of stand-alone verification Table 1. Results of stand-alone verification Table 1. Results of stand-alone verification Device Number of bugs DDR4 MC 32 PCI Express RC 48 IPCC 13 WLCC 2 10 Gigabit Ethernet 51 Gigabit Ethernet 22 Verification of those devices is still ongoing. Our future work is aimed at improving those test systems, developing additional test scenarios and using the approach to verify other devices. 5. Conclusion Methods described in the paper were used to verify components of “Elbrus-16C” microprocessor. Errors found in the controllers as a result of stand-alone verification are presented in table 1. References [1] Stotland I., Shpagilev D., Starikovskaya N. UVM based approaches to functional verification of communication controllers of microprocessor systems. In Proc. of the 2016 IEEE East-West Design & Test Symposium (EWDTS). 2] PCI Express Base Specification Revision 3.0, http://pcisig.com/specifications [3] Belyanin I, Petrakov P., Feldman V. Functional organization and hardware means of network interconnection of modules in computer cluster on «Elbrus» microproseccors. Voprosy radioelektroniki [Issues of radio electronics], ser. EVT, no. 3, 2015, pp. 7–20 (in Russian). [4] IEEE Standard for Ethernet. IEEE Std 802.3-2012. 1983 p. [5] IEEE Standard for SystemVerilog — Unified Hardware Design, Specification, and Verification Language. IEEE Std 1800-2012 [6] 1800.2-2017 - IEEE Standard for Universal Verification Methodology Language Reference Manual [7] Stotland I., Shpagilev D., Petrochenkov M. Features of High Speed Communication Controllers Standalone Verification of «Elbrus» Microprocessor Systems. Voprosy radioelektroniki [Issues of radio electronics], ser. EVT, 3, 2017, pp. 69-75 (in Russian). [8] S. Chitti, P. Chandrasekhar, M. Asha Rani. Gigabit Ethernet Verification using Efficient Verification Methodology. In Proc. of the International Conference on Industrial Instruments and Control (ICIC), 2015, pp.1231-1235. 203 Petrochenkov M.V., Mushtakov R.E., Shpagilev D.I. Verification of System on Chip Integrated Communication Controllers. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 195-206 Для цитирования: Петроченков М.В., Муштаков Р.Е., Шпагилев Д.И. Верификация контроллеров связи в системах на кристалле. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 195-206 (на английском языке). DOI: 10.15514/ISPRAS-2018-30(3)-14 Верификация контроллеров связи в системах на кристалле Аннотация. В статье описаны подходы, которые использовались для верификации контроллеров связи в системах на кристалле, разрабатываемых в МЦСТ. Представлен список контроллеров связи, а также их характеристики. Приведены принципы работы контроллеров на уровне транзакций, канальном и физическом, и отмечен их общий функционал. Затем описан общий подход к верификации устройств: принцип проектирования тестовой системы, генерации случайных тестовых воздействий и проверки поведения устройства. Представлена общая структура тестовой системы, основанная на общих свойствах устройств. Она включает компоненты для работы с интерфейсом уровня транзакций (системный агент, реализующий коммуникационный протокол системы на кристалле), интерфейсом физического уровня (физический агент, реализующий коммуникационный протокол между различными системами на кристалле на одной плате), модуль конфигурационного интерфейса, определяющего режим работы устройства, а также модуль проверки. Отмечено, что поскольку устройства исполняют только преобразования транзакций между различными представлениями, заключение о корректности поведения осуществляется на основании простой проверки совпадения входящих и исходящих транзакций. Кроме того, приведены особенности функционала устройств, которые требуют адаптации общего подхода. Объяснено, как верификация данных особенностей работы устройств определила детали структуры тестовых систем. Описано, как замена физического агента на второй контроллер связи позволяет ускорить разработку тестовой системы. Представлены методы и сложности верификации конечного автомата тренировки и состояния линка (LTSSM). Описана структура и принцип работы системных агентов, поддерживающих прямой доступ к памяти (DMA). В заключение приведен список найденных ошибок и направления дальнейшей работы. Ключевые слова: Эльбрус; система на кристалле; контроллер связи; Ethernet; DDR4; PCI Express; UVM; автономная верификация DOI: 10.15514/ISPRAS-2018-30(3)-14 Для цитирования: Петроченков М.В., Муштаков Р.Е., Шпагилев Д.И. Верификация контроллеров связи в системах на кристалле. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 195-206 (на английском языке). DOI: 10.15514/ISPRAS-2018-30(3)-14 Для цитирования: Петроченков М.В., Муштаков Р.Е., Шпагилев Д.И. Верификация контроллеров связи в системах на кристалле. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 195-206 (на английском языке). DOI: 10.15514/ISPRAS-2018-30(3)-14 204 Петроченков М.В., Муштаков Р.Е. Шпагилев Д.И. Верификация контроллеров связи в системах на кристалле. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 195-206 Петроченков М.В., Муштаков Р.Е. Шпагилев Д.И. Верификация контроллеров связи в системах на кристалле. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 195-206 Petrochenkov M.V., Mushtakov R.E., Shpagilev D.I. Verification of System on Chip Integrated Communication Controllers. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 195-206 Список литературы [1] Stotland I., Shpagilev D., Starikovskaya N. UVM based approaches to functional verification of communication controllers of microprocessor systems. In Proc. of the 2016 IEEE East-West Design & Test Symposium (EWDTS). 2] PCI Express Base Specification Revision 3.0, http://pcisig.com/specifications [3] Белянин И., Петраков П., Фельдман В. Функциональная организация и аппаратура сетевого взаимодействий модулей в вычислительном кластере на базе микропроцессоров с архитектурой «Эльбрус». Вопросы радиоэлектроники, серия ЭВТ, вып. 3, 2015 г., стр. 7-20. [4] IEEE Standard for Ethernet. IEEE Std 802.3-2012. 1983 p. [5] IEEE Standard for SystemVerilog — Unified Hardware Design, Specification, and Verification Language. IEEE Std 1800-2012 [6] 1800.2-2017 - IEEE Standard for Universal Verification Methodology Language Reference Manual [7] Стотланд И., Шпагилев Д., Петроченков М. Особенности функциональной веривикации контроллеров высокоскоростных каналов обмена микропроцессорных систем семейства «Эльбрус». Вопросы радиоэлектроники, серия ЭВТ, вып. 3, 2017, стр. 69-75. [8] S. Chitti, P. Chandrasekhar, M. Asha Rani. Gigabit Ethernet Verification using Efficient Verification Methodology. In Proc. of the International Conference on Industrial Instruments and Control (ICIC), 2015, pp.1231-1235. 205 Petrochenkov M.V., Mushtakov R.E., Shpagilev D.I. Verification of System on Chip Integrated Communication Controllers. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 195-206 206 Ключевые слова: интерфейс; адаптивность; индуктивное моделирование; веб- интерфейс Ключевые слова: интерфейс; адаптивность; индуктивное моделирование; веб- интерфейс DOI: 10.15514/ISPRAS-2018-30(3)-15 Для цитирования: Зосимов В.В., Христодоров А.В., Булгакова А.С. Программные решения для динамического изменения пользовательского интерфейса на основе автоматически собранной информации о пользователе. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 207-220. DOI: 10.15514/ISPRAS-2018-30(3)-15 Для цитирования: Зосимов В.В., Христодоров А.В., Булгакова А.С. Программные решения для динамического изменения пользовательского интерфейса на основе автоматически собранной информации о пользователе. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 207-220. DOI: 10.15514/ISPRAS-2018-30(3)-15 Программные решения для динамического изменения пользовательского интерфейса на основе автоматически собранной информации о пользователе В.В. Зосимов <zosimovvv@gmail.com> А.В. Христодоров <belfegor26@gmail.com> А.С. Булгакова <sashabulgakova2@gmail.com> Николаевский национальный университет им. В.А. Сухомлинского Украина, 54030, г. Николаев, ул. Никольская, 24 В.В. Зосимов <zosimovvv@gmail.com> А.В. Христодоров <belfegor26@gmail.com> А.С. Булгакова <sashabulgakova2@gmail.com> Николаевский национальный университет им. В.А. Сухомлинского Украина, 54030, г. Николаев, ул. Никольская, 24 Аннотация. В статье описываются функциональные возможности и структура программного модуля для автоматизированной адаптации интерфейсов веб- приложений. Особое внимание уделяется идентификации и различению псевдоанонимных пользователей веб-приложений для адаптации интерфейса под конкретного пользователя. Разработанный подход обеспечивает возможность псевдоидентификации киберсущностей в контексте поведения пользователей и автоматизированную адаптацию интерфейсов под особенности выделенного пользователя в зависимости от поставленных задач. 1. Введение Интерфейсы – это неотъемлемая составляющая звеньев восприятия программных продуктов и управления ими. Именно они обеспечивают управление программными продуктами и связь пользователей с программой. Неустанная интеграция IT-технологий в среду существования современного человека порождает все больший спрос на разработки в области адаптации 207 Zosimov V.V., Khrystodorov O.V., Bulgakova O.S. Dynamically changing user interfaces: software solutions based on automatically collected user information. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 207-220 Zosimov V.V., Khrystodorov O.V., Bulgakova O.S. Dynamically changing user interfaces: software solutions based on automatically collected user information. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 207-220 интерфейсов программных продуктов, направленных на удовлетворение потребностей пользователей. Программные продуктов, ориентированные на повышение удобства пользования интерфейсами, обладают следующими достоинствами: Программные продуктов, ориентированные на повышение удобства пользования интерфейсами, обладают следующими достоинствами:  упрощение восприятия бизнес-логики программного продукта конкретным пользователем;  гибкость модели представления информации;  повышение производительности работы с интерфейсом. Под динамическим изменением интерфейса следует понимать его адаптацию в результате выполнения некоторого сценария на основе поведенческого портрета пользователя. Наиболее полный обзор содержания адаптивного поведения человеко- машинного интерфейса (ЧМИ) содержится в [1]. Авторы предлагают три параметра интерфейса, которые могут меняться: Наиболее полный обзор содержания адаптивного поведения человеко- машинного интерфейса (ЧМИ) содержится в [1]. Авторы предлагают три параметра интерфейса, которые могут меняться:  содержание представляемой информации;  форма представления информации и ведения диалога;  распределение задач между человеком и машиной (уровень автоматизации). В других работах декларируется, что адаптивность интерфейса проявляется:  в настройке уровня детализации диалога с пользователем – от подробного диалога, «ведущего» пользователя к цели шаг за шагом через иерархию меню, к короткому, с использованием сокращенных команд и макросов в режиме «вопрос-ответ» [2];  в подсказках, ограничении доступа к приложениям, регулировании интенсивности информационного обмена и изменению внешнего вида интерфейса [3,4];  в фильтрации и расстановке приоритетов контента, предложенного пользователю; это может происходить не только в соответствии с собственными предпочтениями, но и под влиянием внешних факторов и контекста (например, пользователю предъявляется продукт, который выгодно купить именно сейчас [5]);  в изменении темпа подачи информации [6];  в настройке параметров изображения (толщины линий, размера шрифта, яркости и др.).  в настройке параметров изображения (толщины линий, размера шрифта, яркости и др.). В условиях конкуренции современных разработок в области разработки и применения интерактивных интерфейсов повышается актуальность исследований в направлении автоматизированной адаптации пользовательских интерфейсов. Под автоматизированной адаптацией понимается динамическое изменение пользовательских интерфейсов на основе автоматически собранной информации о пользователе. 1. Введение В условиях конкуренции современных разработок в области разработки и применения интерактивных интерфейсов повышается актуальность исследований в направлении автоматизированной адаптации пользовательских интерфейсов. Под автоматизированной адаптацией понимается динамическое изменение пользовательских интерфейсов на основе автоматически собранной информации о пользователе. 208 Зосимов В.В., Христодоров А.В., Булгакова А.С. Программные решения для динамического изменения пользовательского интерфейса на основе автоматически собранной информации о пользователе. Труды ИСП РАН, том 30, вып. 2, 2018 г., стр. 207-220 Зосимов В.В., Христодоров А.В., Булгакова А.С. Программные решения для динамического изменения пользовательского интерфейса на основе автоматически собранной информации о пользователе. Труды ИСП РАН, том 30, вып. 2, 2018 г., стр. 207-220 Значительную роль в исследовании моделей и методов адаптации интерфейсов играют достижения Питера Экерсли [7]. Отдельные вопросы проектирования и анализа использования возможностей псевдоидентификации рассматриваются в трудах исследователей Г. Кришна, К. Докинз, Дж. Гилмор [8, 9]. Некоторые вопросы использования технологий сбора данных в автономных киберфизических системах рассматриваются в работах отечественных ученых А. Бочкарева, В. Голембо [10]. Значительную роль в исследовании моделей и методов адаптации интерфейсов играют достижения Питера Экерсли [7]. Отдельные вопросы проектирования и анализа использования возможностей псевдоидентификации рассматриваются в трудах исследователей Г. Кришна, К. Докинз, Дж. Гилмор [8, 9]. Некоторые вопросы использования технологий сбора данных в автономных киберфизических системах рассматриваются в работах отечественных ученых А. Бочкарева, В. Голембо [10]. В [8-10] рассмотрены фундаментальные принципы и концепции, которые применяются при реализации программных продуктов, представлены различные модели и методы, направленные на улучшение восприятия и взаимодействия с интерфейсами конечных программных продуктов. Но все эти аспекты рассматривались как отдельные задачи, и отсутствовало их объединение в единую адаптивную систему. Анализ результатов предыдущих исследований позволил прийти к выводу, что на сегодняшний день неуклонно растет спрос на системы адаптации интерфейсов во всех сферах их использования, а работоспособные системы отсутствуют. В данной статье описывается разработанный авторами подход, доведенный до программной реализации. 1. Индуктивный подход к построению адаптивных интерфейсов Пусть CR = {cr1, … cr\|CR\|} ‒ множество критериев, которые будут изменятся, то есть адаптироваться под пользователя, Fact = {fact1, … fact\|Fact\|} ‒ множество факторов, которые будут влиять на выбор того или иного критерия. На основе выделенных заранее факторов, которые могут влиять на интерфейс, будут меняться критерии построения интерфейса. Индуктивность процесса заключается в том, что адаптация интерфейса происходит от конкретных данных наблюдений, то есть факторов – к общей модели [11], которая включает в себя множество критериев, которые будут меняться (например, при известной информации о возрасте пользователя (фактор –возраст), увеличить размер шрифта (для пожилых людей) – критерий. Следовательно, исходя из вышесказанного, можно построить функцию Φ, отражающую процесс адаптации, где θ – параметры модели: Общая функциональная схема механизма адаптации интерфейса представлена на рис.1. Механизм состоит из нескольких блоков: блок определения Общая функциональная схема механизма адаптации интерфейса представлена на рис.1. Механизм состоит из нескольких блоков: блок определения 209 Zosimov V.V., Khrystodorov O.V., Bulgakova O.S. Dynamically changing user interfaces: software solutions based on automatically collected user information. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 207-220 факторов, где формируются/хранятся факторы пользователя, которые могут влиять на критерии изменения интерфейса, то есть адаптации; блок формирования критериев (количество критериев для каждого фактора может быть разной); блок обработки данных, в котором программным образом будет обрабатываться содержание критериев, приводит к изменению интерфейса [12]. Рис.1. Составные части механизма адаптации интерфейса Fig.1. Components of the interface adaptation mechanism Рис.1. Составные части механизма адаптации интерфейса Fig.1. Components of the interface adaptation mechanism Рис.1. Составные части механизма адаптации интерфейса Fig.1. Components of the interface adaptation mechanism 2. Требования к программному продукту Программный продукт должен обеспечивать следующие функциональные возможности:  сбор и хранение информации о пользователях соответствующего веб- приложения;  сбор и хранение информации о пользователях соответствующего веб- приложения;  псевдоидентификацию пользователей на основании собранных данных;  псевдоидентификацию пользователей на основании собранных данных;  автоматизированную  автоматизированную адаптацию псевдоидентифицированных пользователей. Программный продукт должен соответствовать следующим требованиям: 1. контролируемый сбор информации о пользователях веб приложения; 2. создание и поддержка базы данных полученной информации; 3. псевдоидентификация пользователей веб-приложения на основе собранной информации; 210 Зосимов В.В., Христодоров А.В., Булгакова А.С. Программные решения для динамического изменения пользовательского интерфейса на основе автоматически собранной информации о пользователе. Труды ИСП РАН, том 30, вып. 2, 2018 г., стр. 207-220 Зосимов В.В., Христодоров А.В., Булгакова А.С. Программные решения для динамического изменения пользовательского интерфейса на основе автоматически собранной информации о пользователе. Труды ИСП РАН, том 30, вып. 2, 2018 г., стр. 207-220 4. автоматизированная адаптация пользовательского интерфейса веб- приложения на основании собранных и обработанных программным продуктом данных. 4. автоматизированная адаптация пользовательского интерфейса веб- приложения на основании собранных и обработанных программным продуктом данных. программным продуктом данных. Ниже проведен сравнение предъявленных требований с возможностями существующих аналогов. За последние годы сформировалось и продолжает развиваться направление корректировки контента, структуры сайта, контроля внутренних и внешних факторов пребывания и взаимодействия пользователя с интерфейсом. Это неотъемлемая составляющая успеха маркетинговых и информационных сайтов. Существует достаточно большое количество систем сбора информации о пользователях, например, Amplitude [13], Mixpanel [14] и др. Эти программные продукты удовлетворяют требованиям 1 и 2, но не удовлетворяют требованиям 3 и 4. Практически отсутствуют системы, которые удовлетворяют всем четырем перечисленным требованиям. 3. Структурная схема программного продукта Программные решения для динамического изменения пользовательского интерфейса на основе автоматически собранной информации о пользователе. Труды ИСП РАН, том 30, вып. 2, 2018 г., стр. 207-220 В блоке «Set token» – программная система «AAUI» генерирует идентификатор Генерация идентификатора происходит параллельно со Зосимов В.В., Христодоров А.В., Булгакова А.С. Программные решения для динамического изменения пользовательского интерфейса на основе автоматически собранной информации о пользователе. Труды ИСП РАН, том 30, вып. 2, 2018 г., стр. 207-220 пользователя, рис. 4. Рис 4. Реализация блока генерации идентификатора Рис 4. Реализация блока генерации идентификатора Fig. 4. Implementation of the ID generation unit «Checking visit» – проверка на возможного предыдущего посещения сервиса пользователем. «Checking visit» – проверка на возможного предыдущего посещения сервиса пользователем.  Если пользователь идентифицируется («User identified») происходит анализ поведенческих характеристик «Analysis rules», собранных на основе объектов интерфейса программного продукта, ориентированных на отслеживание изменений; применение персональных настроек «Involvement rules» и дальнейшее наблюдение за изменением потенциально изменяемых настроек интерфейса T ki h В б ф й  Если пользователь идентифицируется («User identified») происходит анализ поведенческих характеристик «Analysis rules», собранных на основе объектов интерфейса программного продукта, ориентированных на отслеживание изменений; применение персональных настроек «Involvement rules» и дальнейшее наблюдение за изменением потенциально изменяемых настроек интерфейса «Tracking status changes». Все изменения настроек веб-интерфейса, которые отслеживаются, сохраняются в базе данных каждый раз, когда они происходят – «Saving the latest user settings»; за изменением потенциально изменяемых настроек интерфейса «Tracking status changes». Все изменения настроек веб-интерфейса, которые отслеживаются, сохраняются в базе данных каждый раз, когда они происходят – «Saving the latest user settings»;  В случае, если пользователь впервые посещает веб-сервис, т.е. «User not identified», происходит отслеживание изменений, задаваемых пользователем, и их в базе данных сохранение «Saving changes made by the user». 3. Структурная схема программного продукта На рис. 2 представлена схема взаимодействия программной системы «AAUI - Automatic Adaptation of User Interfaces». Рис 2. Схема взаимодействия программной системы «AAUI» Fig. 2. Diagram of interaction of the "AAUI" software system Блок «AAUI» является начальным модулем, начиная с которого происходит взаимодействие с компонентами программного продукта. «Identification» реализует формирование сущности на основании собранной информации об объекте, который проявил активность. «Operating system» обеспечивает сбор Рис 2. Схема взаимодействия программной системы «AAUI» Рис 2. Схема взаимодействия программной системы «AAUI» Fig. 2. Diagram of interaction of the "AAUI" software system Блок «AAUI» является начальным модулем, начиная с которого происходит взаимодействие с компонентами программного продукта. «Identification» реализует формирование сущности на основании собранной информации об объекте, который проявил активность. «Operating system» обеспечивает сбор 211 Zosimov V.V., Khrystodorov O.V., Bulgakova O.S. Dynamically changing user interfaces: software solutions based on automatically collected user information. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 207-220 Zosimov V.V., Khrystodorov O.V., Bulgakova O.S. Dynamically changing user interfaces: software solutions based on automatically collected user information. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 207-220 Zosimov V.V., Khrystodorov O.V., Bulgakova O.S. Dynamically changing user interfaces: software solutions based on automatically collected user information. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 207-220 информации о программных средствах активного объекта посещения. «Browser identification» отвечает за сбор информации о программном обеспечении, которое задействовано во взаимодействии с конечным программным продуктом. «Human behavior» обеспечивает проверку возможного прошлого присутствия псевдодеанимизированного объекта. «Adaptation» отвечает за программную адаптацию на основе созданного отпечатка псевдодеанимизированного объекта взаимодействия с конечным продуктом. «Database» отвечает за сохранность проанализированной информации, токенов псевдодеанимизации и правил адаптации интерфейсов. «Process identifier» отвечает за согласование идентификации существующих объектов взаимодействия. На рис. 3 представлена блок-схема использования программного проду дставлена блок-схема использования программного про Рис 3. Схема использования программного продукта Fig. 3. Scheme of using the software Рис 3. Схема использования программного продукта Fig. 3. Scheme of using the software 212 Зосимов В.В., Христодоров А.В., Булгакова А.С. Программные решения для динамического изменения пользовательского интерфейса на основе автоматически собранной информации о пользователе. Труды ИСП РАН, том 30, вып. 2, 2018 г., стр. 207-220 В блоке «Set token» – программная система «AAUI» генерирует идентификатор. Генерация идентификатора происходит параллельно со сбором дополнительных данных для более точной идентификации пользователя, рис. 4. Рис 4. Реализация блока генерации идентификатора Fig. 4. Implementation of the ID generation unit Зосимов В.В., Христодоров А.В., Булгакова А.С. 4. Программная реализация Программные решения для динамического изменения пользовательского интерфейса на основе автоматически собранной информации о пользователе. Труды ИСП РАН, том 30, вып. 2, 2018 г., стр. 207-220 Для визуального представления схемы базы данных была использована методология IDEF1X, направленная на моделирование реляционных баз данных (рис.6). Указанный стандарт входит в семейство методологий IDEF позволяющее исследовать структуру, параметры и характеристики производственно-технических и организационно-экономических систем. Методология IDEF1X адаптирована для совместного использования с IDEF0 в рамках единой технологии моделирования. На основе IDEF0 детализируются функциональные блоки, а IDEF1X позволяет детализировать «стрелки». Разработка базы данных производилась с помощью инструмента для проектирования баз данных MySQL Workbench – инструмента, интегрирующего проектирование, моделирование, создание и эксплуатацию БД в единое окружение для системы баз данных MySQL с использованием, в частности, нотации IDEF1X. Рис 6. Фрагмент ER-диаграммы базы данных в нотации IDEF1X Рис 6. Фрагмент ER-диаграммы базы данных в нотации IDEF1X Fig. 6. Fragment of the ER-diagram of the database in IDEF1X notation Интересной в данном продукте является реализация системы псевдоидентификации пользователей. Взаимодействие частей программного продукта демонстрируют части кода, представленные на рис.7 и 8. На рис.7. представлен блок получения информации об установленных пользователем языков и часового пояса. Функция автоматической адаптации локализации веб-приложения на основе собранных данных представлена на рис.8. 4. Программная реализация Основным преимуществом программной системы является примененная модель идентификации анонимных пользователей конечных программных продуктов с дальнейшим использованием динамического идентификатора для автоматической адаптации интерфейса под идентифицированного пользователя. 213 Zosimov V.V., Khrystodorov O.V., Bulgakova O.S. Dynamically changing user interfaces: software solutions based on automatically collected user information. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 207-220 Zosimov V.V., Khrystodorov O.V., Bulgakova O.S. Dynamically changing user interfaces: software solutions based on automatically collected user information. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 207-220 «AAUI» является серверным программным продуктом, написанным на скриптовом языке программирования PHP. Программный продукт обеспечивает: «AAUI» является серверным программным продуктом, написанным на скриптовом языке программирования PHP. Программный продукт обеспечивает:  администрирование программного продукта (управление списком задействованных фильтров идентификации, изменение, редактирование, создание правил адаптации); редактирование, создание правил адаптации);  хранение и обработка информации о пользователях;  обеспечение хранения информации о сессиях. В «AAUI» предусмотрен механизм настройки псевдодеанонимизации пользователей, и их группировки по заданным фильтрам. Функции групп пользователей: 1. группировка списков по странам посещения; 2. группировка списков пользователей по языку пользователей; 3. группировка по конфигурации конечного устройства посетител 4. комбинация групп для выделения целевой аудитории. Надежность и устойчивое функционирование «AAUI» достигается совокупностью следующих организационно-технических мероприятий: Надежность и устойчивое функционирование «AAUI» достигается совокупностью следующих организационно-технических мероприятий: 1. организация надежной защиты специалистами по кибербезопасности веб-приложений; 2. организации регулярного и качественного технического обслуживания серверной части; 3. своевременное обслуживание базы данных. 3. своевременное обслуживание базы данных. Серверное приложение «AAUI» взаимодействует с посетителем веб- приложения с начала его взаимодействия с интерфейсом веб-приложения (рис. 5). Уровень «поглощения» собранной информации о посетителе устанавливает администратор программной системы. Рис 5. Схема модели взаимодействия Fig. 5. Diagram of interaction model Пользователь AAUI Веб- приложение Веб- приложение AAUI Рис 5. Схема модели взаимодействия Fig. 5. Diagram of interaction model Основным источником хранения данных является база данных. В ней хранится информация о полученных сущностях псевдодеанонимизированных пользователях. Для хранения информации используется свободная система управления реляционными базами данных MySQL. Структура базы данных разработана на основе разработанной методологии идентификации пользователей. 214 Зосимов В.В., Христодоров А.В., Булгакова А.С. Программные решения для динамического изменения пользовательского интерфейса на основе автоматически собранной информации о пользователе. Труды ИСП РАН, том 30, вып. 2, 2018 г., стр. 207-220 Зосимов В.В., Христодоров А.В., Булгакова А.С. Программные решения для динамического изменения пользовательского интерфейса на основе автоматически собранной информации о пользователе. Труды ИСП РАН, том 30, вып. 2, 2018 г., стр. 207-220 Зосимов В.В., Христодоров А.В., Булгакова А.С. 5. Экспериментальное использование технологии После интеграции «AAUI» с Интернет магазином было проведено исследование, направленное на выявление процентного соотношения прироста количества новых посетителей. В течение тридцати дней система работала в режиме сбора поведенческих сущностей, псевдодеанимизации и формирования базы посетителей. 215 Zosimov V.V., Khrystodorov O.V., Bulgakova O.S. Dynamically changing user interfaces: software solutions based on automatically collected user information. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 207-220 Zosimov V.V., Khrystodorov O.V., Bulgakova O.S. Dynamically changing user interfaces: software solutions based on automatically collected user information. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 207-220 Zosimov V.V., Khrystodorov O.V., Bulgakova O.S. Dynamically changing user interfaces: software solutions based on automatically collected user information. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 207-220 Была получена выборка из 2000 уникальных поведенческих портретов. Из них 30% возвращались на сайт Интернет магазина в дальнейшем. На основании анализа собранных данных была запрограммирована функция адаптация, которая была более всего востребована – сортировка товаров по цене (от минимальной до максимальной). Р 7 Б б ф 6. Заключение В работе представлено описание программного обеспечения для автоматизированной адаптации пользовательских интерфейсов. Проведенный анализ систем, направленных на сбор информации о пользователях, таких как Amplitude и Mixpanel, позволил сделать вывод, что указанные программные системы обеспечивают только сбор профилированной информации о входящих пользователях и не обеспечивают автоматизированной адаптации интерфейсов под нужды пользователей. Разработанная система «AAUI» направлена на автоматизированную адаптацию интерфейсов под нужды пользователей. Система обеспечивает псевдоидентификацию пользователей (построение базы анонимных пользователей и правил на основе их пребывания в веб-приложениях). Оригинальной особенностью системы является использованная модель идентификации анонимных пользователей конечных программных продуктов с дальнейшим использованием динамического идентификатора для автоматической адаптации интерфейса под идентифицированного пользователя. В открытом доступе отсутствует аналогичное программное обеспечение. К перспективам дальнейшей разработки «AAUI» можно отнести увеличение количества маркеров идентификации для повышения достоверности идентификации пользователей, внедрение в системы управления контентом и оптимизацию работы с данными. Представлено экспериментальное применение технологии на примере Интернет-магазина, в котором в течение 30 дней было зафиксировано 2000 уникальных поведенческих портретов. В результате эксперимента процент посетителей веб-приложения, которые снова обратились в Интернет-магазин, вырос с 30% до 50%. [1] Rothrock L., Koubek R., Fuchs F. et al. Review and reappraisal of adaptive interfaces: toward biologically inspired paradigms. Theoretical Issues in Ergonomics Science, vol. 3, No. 1, 2002, рр. 47-84. Рис 7. Блок сбора информации Fig. 7. Information collection block Рис 8. Функция автоматической адаптации Fig. 8. Automatic adaptation function 216 Зосимов В.В., Христодоров А.В., Булгакова А.С. Программные решения для динамического изменения пользовательского интерфейса на основе автоматически собранной информации о пользователе. Труды ИСП РАН, том 30, вып. 2, 2018 г., стр. 207-220 Зосимов В.В., Христодоров А.В., Булгакова А.С. Программные решения для динамического изменения пользовательского интерфейса на основе автоматически собранной информации о пользователе. Труды ИСП РАН, том 30, вып. 2, 2018 г., стр. 207-220 Программная система «AAUI», в течение тридцати дней генерировала выборку поведенческих портретов. При повторном посещении псевдодеанимизированного пользователя система автоматически адаптировала запрограммированную на отслеживание функцию в зависимости от полученного состояния сортировки. Программная система «AAUI», в течение тридцати дней генерировала выборку поведенческих портретов. При повторном посещении псевдодеанимизированного пользователя система автоматически адаптировала запрограммированную на отслеживание функцию в зависимости от полученного состояния сортировки. у В результате использования «AAUI» процент посетителей веб-приложения, которые снова обратились в Интернет-магазин, вырос с 30% до 50%. В дальнейшем будет проведено более детальное экспериментальное исследование возможностей предлагаемой технологии. В результате использования «AAUI» процент посетителей веб-приложения, которые снова обратились в Интернет-магазин, вырос с 30% до 50%. В дальнейшем будет проведено более детальное экспериментальное исследование возможностей предлагаемой технологии. В дальнейшем будет проведено более детальное экспериментальное исследование возможностей предлагаемой технологии. Список литературы [1] Rothrock L., Koubek R., Fuchs F. et al. Review and reappraisal of adaptive interfaces: toward biologically inspired paradigms. Theoretical Issues in Ergonomics Science, vol. 3, No. 1, 2002, рр. 47-84. [1] Rothrock L., Koubek R., Fuchs F. et al. Review and reappraisal of adaptive interfaces: toward biologically inspired paradigms. Theoretical Issues in Ergonomics Science, vol. 3, No. 1, 2002, рр. 47-84. 217 Zosimov V.V., Khrystodorov O.V., Bulgakova O.S. Dynamically changing user interfaces: software solutions based on automatically collected user information. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 207-220 Zosimov V.V., Khrystodorov O.V., Bulgakova O.S. Dynamically changing user interfaces: software solutions based on automatically collected user information. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 207-220 [2] Ходаков В.Е., Ходаков Д.В. Адаптивный пользовательский интерфейс: проблемы построения. Автоматика. Автоматизация. Электротехнические комплексы и системы, № 1 (11), 2002, стр. 45-57. [3] Курзанцева Л.И. Методика комплексного исследования адаптивного человеко- машинного интерфейса. Математические машины и системы, № 4, 2011, стр. 69- 77. [4] Курзанцева Л.И. Об адаптивном интеллектуальном интерфейсе «пользователь – система массового применения». Компьютерные средства, сети и системы, №7, 2008, стр. 110-116. [5] Langley P. User modeling in adaptive interfaces. In Proceedings of the Seventh International Conference on User Modeling, 1999, pp. 357-370. [6] Karwowski W. A review of human factors challenges of complex adaptive systems: discovering and understanding chaos in human performance. Human Factors, vol. 54, No. 6, 2012, pp. 983-995. [7] Peter Eckersley, How Unique Is Your Web Browser? Electronic Frontier Foundation Режим доступа: https://www.eff.org/, дата обращения 02.05.2017. [8] Gilmore J. Easy Laravel 5. Leanpub, 2016, 235 р. [9] Dockins K. Design Patterns in PHP and Laravel. Appers, 2017, 45 р. [10] О. Бочкарьов, В. Голембо. Використання інтелектуальних технологій збору даних у автономних кіберфізичних системах. Lviv Polytechnic National University Institutional Repository. Сборник научных трудов, № 830, 2015, стр. 7–11 (на украинском). [11] Stepashko V., Bulgakova O., Zosimov V. Construction and research of the generalized iterative GMDH algorithm with active neurons. Advances in Intelligent Systems and Computing, vol. 689, 2018, рр. 492-510. [12] Булгакова А.С. Концепция построения адаптивного интерфейса с использованием индуктивного подхода. Индуктивное моделирование сложных систем. Сборник трудов, выпуск 8. Киев: МННЦ ІТС, 2016, стр. 73-78 (на украинском) [13] Amplitude. Analytics for modern product teams. Режим доступа: https://amplitude.com/, дата обращения 02.05.2017. [14] Mixpanel. Product and User Analytics for Mobile, Web, and Beyond. Режим доступа: https://mixpanel.com/, дата обращения 02.05.2017. DOI: 10.15514/ISPRAS-2018-30(3)-15 For citation: Zosimov V.V., Khrystodorov O.V., Bulgakova O.S. Dynamically changing user interfaces: software solutions based on automatically collected user information. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 207-220 (in Russian). DOI: 10.15514/ISPRAS-2018-30(3)-15 Dynamically changing user interfaces: software solutions based on automatically collected user information V.V. Zosimov <zosimovvv@gmail.com> O.V. Khrystodorov <belfegor26@gmail.com> O.S. Bulgakova <sashabulgakova2@gmail.com> V.O. Sukhomlynsky Mykolaiv National University, 24 Nikolska St, Mykolayiv, 54030, Ukraine Abstract. The developed system "AAUI" is aimed at the automated adaptation of interfaces to the needs of users. The system provides pseudo-identification of users (building anonymous users and rules database based on their stay in web applications). The original feature of the system is the used model of identifying anonymous users of the end products 218 Зосимов В.В., Христодоров А.В., Булгакова А.С. Программные решения для динамического изменения пользовательского интерфейса на основе автоматически собранной информации о пользователе. Труды ИСП РАН, том 30, вып. 2, 2018 г., стр. 207-220 Зосимов В.В., Христодоров А.В., Булгакова А.С. Программные решения для динамического изменения пользовательского интерфейса на основе автоматически собранной информации о пользователе. Труды ИСП РАН, том 30, вып. 2, 2018 г., стр. 207-220 Зосимов В.В., Христодоров А.В., Булгакова А.С. Программные решения для динамического изменения пользовательского интерфейса на основе автоматически собранной информации о пользователе. Труды ИСП РАН, том 30, вып. 2, 2018 г., стр. 207-220 with the further use of the dynamic identifier to automatically adapt the interface to the identified user's need. The analysis of the systems aimed at collecting information about users made it possible to conclude that these software systems provide only the collection of profiled information about incoming users and do not provide for automated adaptation of interfaces to the needs of users. There is no similar software in the open access. The prospects for the further development of AAUI include an increase in the number of identification markers to improve the authenticity of user identification, the integratioin with content management systems and the optimization of data management. with the further use of the dynamic identifier to automatically adapt the interface to the identified user's need. The analysis of the systems aimed at collecting information about users made it possible to conclude that these software systems provide only the collection of profiled information about incoming users and do not provide for automated adaptation of interfaces to the needs of users. There is no similar software in the open access. The prospects for the further development of AAUI include an increase in the number of identification markers to improve the authenticity of user identification, the integratioin with content management systems and the optimization of data management. Keywords: interface, adaptiveness, inductive modelling, web-interface [12] Bulgakova O. The concept of constructing an adaptive interface using an inductive approach. Induktivnoe modelirovanie slozhnyh sistem [Inductive modeling of complex systems], pp. 73-78 (in Ukrain) Zosimov V.V., Khrystodorov O.V., Bulgakova O.S. Dynamically changing user interfaces: software solutions based on automatically collected user information. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 207-220 [13] Amplitude. Analytics for modern product teams. Available at: https://amplitude.com/, accessed 02.05.2017. [14] Mixpanel. Product and User Analytics for Mobile, Web, and Beyond. Available at: https://mixpanel.com/, accessed 02.05.2017. References [1] Rothrock L., Koubek R., Fuchs F. et al. Review and reappraisal of adaptive interfaces: toward biologically inspired paradigms. Theoretical Issues in Ergonomics Science, vol. 3, No. 1, 2002, рр. 47-84. [1] Rothrock L., Koubek R., Fuchs F. et al. Review and reappraisal of adaptive interfaces: toward biologically inspired paradigms. Theoretical Issues in Ergonomics Science, vol. 3, No. 1, 2002, рр. 47-84. [2] Khodakov V.E., Khodakov D.V. Adaptive user interface: problems of building. Avtomatika. Avtomatizacija. Jelektrotehnicheskie kompleksy i sistemy [Automation. Automatization. Electrotechnical complexes and systems], № 1 (11), 2002, pp. 45-57 (in Russian). [3] Kurgantseva L.I. Methodology of integrated studies of the adaptive human-machine interface. Matematicheskie mashiny i sistemy [Mathematical Machines and Systems], № 4, 2011, pp. 69-77 (in Russian). [4] Kurgantseva L.I. About the adaptive intellectual interface “The user – system of mass application”. Komp'juternye sredstva, seti i sistemy [Computer means, networks and systems], №7, 2008, pp. 110-116 (in Russian). [5] Langley P. User modeling in adaptive interfaces. In Proceedings of the Seventh International Conference on User Modeling, 1999, pp. 357-370. [6] Karwowski W. A review of human factors challenges of complex adaptive systems: discovering and understanding chaos in human performance. Human Factors, vol. 54, No. 6, 2012, pp. 983-995. pp [7] Peter Eckersley, How Unique Is Your Web Browser? Electronic Frontier Foundation Режим доступа: https://www.eff.org/, дата обращения 02.05.2017. [8] Gilmore J. Easy Laravel 5. Leanpub, 2016, 235 р. [8] Gilmore J. Easy Laravel 5. Leanpub, 2016, 235 р. [9] Dockins K. Design Patterns in PHP and Laravel. Appers, 2017, 45 р. [10] Botchkaryov A., Golembo V. Applying intelligent technologies of data acquisition to autonomous cyber-physical systems. Lviv Polytechnic National University Institutional Repository, № 830, 2015, стр. 7–11 (in Ukrain). [11] Stepashko V., Bulgakova O., Zosimov V. Construction and research of the generalized iterative GMDH algorithm with active neurons. Advances in Intelligent Systems and Computing, vol. 689, 2018, рр. 492-510. 219 [13] Amplitude. Analytics for modern product teams. Available at: https://amplitude.com/, accessed 02.05.2017. [14] Mixpanel. Product and User Analytics for Mobile, Web, and Beyond. Available at: https://mixpanel.com/, accessed 02.05.2017. 220 DOI: 10.15514/ISPRAS-2018-30(3)-16 For citation: Gordenko M.K., Avdoshin S.M. The Variants of Chinese Postman Problems and Way of Solving through Transformation into Vehicle Routing Problems. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 221-232. DOI: 10.15514/ISPRAS-2018-30(3)-16 M.K. Gordenko <mgordenko@hse.ru> S.M. Avdoshin <savdoshin@hse.ru> Department of Software Engineering, National Research University Higher School of Economics, 20, Myasnitskaya st., Moscow, 101000 Russia M.K. Gordenko <mgordenko@hse.ru> S.M. Avdoshin <savdoshin@hse.ru> Department of Software Engineering, National Research University Higher School of Economics, 20, Myasnitskaya st., Moscow, 101000 Russia Abstract. In this article, the routing problems are described. It is shown, that almost all routing problem can be transformed into each other. An example of the Mixed Chinese Postman problem is discussed. The article gives an overview of various variants of Chinese Postman Problem. For all problems the mathematical formulation is given. Moreover, the useful real- life application is presented, too. Then, the article provides a table of possible Chinese Postman problems and identifies parameters that can be varied for obtaining new problems. Five parameters have been identified, such as: presence of set of edges; presence of set of arcs; presence of edges with cost, depending on traversing; the presence of set of required edges; the presence of set of required arcs. It was shown that by varying these parameters one can obtain tasks that were not described earlier but can be used in real life. Four new tasks were identified. Then it is shown that the Chinese Postman problem can be solved as another routing tasks through graph transformations. The method for transforming Chinese Postman problem into the Generalized Travelling Salesman problem is given. Then the results of solving the above problem are presented by simple algorithms, and their effectiveness is shown. The research is not over yet. The testing of other algorithms is planned. Keywords: Generalized Routing Problem; Arc Routing Problem; Chinese Postman Problem; Generalized Travelling Salesman Problem 2. The Variations of Chinese Postman Problem There are a lot of variations of CPP. Below, some of them are described. 1. Introduction The General Routing Problem (GRP) is a routing problem defined on a graph where a minimum cost tour is to be found and where the route must include visiting certain required vertices and traversing certain required edges [1]. The routing problems are 221 Gordenko M.K., Avdoshin S.M. The Variants of Chinese Postman Problems and Way of Solving through Transformation into Vehicle Routing Problems. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 221-232 Gordenko M.K., Avdoshin S.M. The Variants of Chinese Postman Problems and Way of Solving through Transformation into Vehicle Routing Problems. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 221-232 closely related to the logistic and transportation management. From the theoretical point of view, routing problems are mainly related to determining the optimal set of routes in a graph. In practice, the routing problems are not only the tasks of determining optimal set of routes, they are also the tasks of testing robots, the correctness of links in the application menu and operating systems, interactivity usability of web-sites [2]. The Travelling Salesman Problem (TSP) is one of the routing problem consisting in finding a minimal length closed tour that visits each city once. The TSP is one of the most well-known routing problem. Another practical, but less well-known problem is the Chinese Postman Problem (CPP). The CPP is finding a shortest closed path that visits every edge or arc of a graph. The CPP has a simple formulation and a lot of potentially useful applications, but today is poorly understood. The article gives an overview of various CPPs, provides mathematical formulations of problems, and describes the scope of the problem. In addition, the article cites references to the literature, in which the various ways of transforming different types of ARP to VRP is described. Also, the results of the current research of various algorithms for solving the problem of a Generalized Traveling Salesman Problem (GTSP) are presented. 𝑣𝑖, 𝑣𝑗) ∈𝐴 ordered pair of vertices, meaning the traversing an arc (𝑣𝑖,𝑗) 𝑣𝑖to 𝑣𝑗vertex. 𝑗 𝑗 𝑗 Let arc be (𝑣𝑖, 𝑣𝑗) ∈𝐴 ordered pair of vertices, meaning the traversing an arc (𝑣𝑖,𝑗) from vertex 𝑣𝑖 to 𝑣𝑗 vertex. 2.2 The Undirected Rural Chinese Postman problem The Undirected Rural Chinese Postman Problem (URCPP) is a particular WRCPP which consists of determining a minimum cost circuit on a graph so that it is possible to traverse a given subset of required edges. The Undirected Rural Chinese Postman Problem (URCPP) is a particular WRCPP which consists of determining a minimum cost circuit on a graph so that it is possible to traverse a given subset of required edges. DCPP is a special case of WRCPP, where 𝐴 = ∅, and there is not edges, which satisfy (1). So, ∀{𝑣𝑖,𝑣𝑗} ∈ 𝐸, 𝐶(𝑣𝑖,𝑣𝑗) = 𝐶(𝑣𝑗,𝑣𝑖). DCPP is a special case of WRCPP, where 𝐴 = ∅, and there is not edges, which satisfy (1). So, ∀{𝑣𝑖,𝑣𝑗} ∈ 𝐸, 𝐶(𝑣𝑖,𝑣𝑗) = 𝐶(𝑣𝑗,𝑣𝑖). The URCPP is known to be an NP-hard problem and it has some interesting real-life applications. The URCPP is known to be an NP-hard problem and it has some interesting real-life applications. 2.1 The Windy Rural Chinese Postman problem We give a formal formulation of the WRCPP problem, extending it to the case of a mixed multigraph. Let 𝐼= {1,2,. . . , \|𝐸𝑅+ 𝐴𝑅\|}, 𝐿= {1,2,. . . , \|𝑉\|}. On the set of vertices 𝑉 of 𝐺 define indexation 𝑖𝑛𝑣 = 𝑉 → 𝐿, (∀𝑣𝑖 ∈𝑉)(∀𝑣𝑗 ∈𝑉)(𝑣𝑖 ≠𝑣𝑗→𝑖 ≠𝑗),𝑖= 𝑖𝑛𝑣(𝑣𝑖). On the set 𝐸𝑅∪𝐴𝑅 of 𝐺 define indexation 𝑖𝑛𝑒𝑎= 𝐸𝑅∪𝐴𝑅 →𝐼, (∀𝑢𝑖 ∈(𝐸𝑅 ∪𝐴𝑅))(∀𝑢𝑗 ∈(𝐸𝑅 ∪𝐴𝑅)) (𝑢𝑖 ≠𝑢𝑗→𝑖≠𝑗), 𝑖= 𝑖𝑛𝑢(𝑢𝑖). The solution of WRCPP is the route 𝜇 = (𝑣𝑙1, 𝑢𝑝1, 𝑣𝑙2, 𝑢𝑝2, . . . , 𝑣𝑙𝑘, 𝑢𝑝𝑘), which satisfy for the following [11]: 𝑗 𝑗 The solution of WRCPP is the route 𝜇 = (𝑣𝑙1, 𝑢𝑝1, 𝑣𝑙2, 𝑢𝑝2, . . . , 𝑣𝑙𝑘, 𝑢𝑝𝑘), which satisfy for the following [11]: 𝑢𝑝𝑖= { (𝑣𝑙𝑖, 𝑣𝑙𝑖+1),(𝑣𝑙𝑖, 𝑣𝑙𝑖+1) ∈𝐸 (𝑣𝑙𝑖,𝑣𝑙𝑖+1), (𝑣𝑙𝑖,𝑣𝑙𝑖+1) ∈𝐴𝑖= 1, 2 …, 𝑘−1 𝑢𝑝𝑘= { (𝑣𝑙𝑘,𝑣𝑙1),(𝑣𝑙𝑘, 𝑣𝑙1) ∈𝐸 (𝑣𝑙𝑘,𝑣𝑙1), (𝑣𝑙𝑘, 𝑣𝑙1) ∈𝐴𝑖= 1,2 … , 𝑘−1 𝐸𝑅∪𝐴𝑅{𝑢𝑝1, 𝑢𝑝2, … , 𝑢𝑝𝑘} = ∅ 𝑘 We denote by 𝐶(𝜇) = ∑ 𝑘 𝑖= 1 𝐶(𝑢𝑝𝑖) the cost of traversing the route. 𝑖 1 Let 𝑀 is a set of WRCPP routes. It is needed to find 𝜇0 ∈ 𝑀, where (∀𝜇 ∈ 𝑀) (𝐶(𝜇0) ≤ С(𝜇)). 𝑖 1 Let 𝑀 is a set of WRCPP routes. It is needed to find 𝜇0 ∈ 𝑀, where (∀𝜇 ∈ 𝑀) (𝐶(𝜇0) ≤ С(𝜇)). ( 𝜇 ) ( (𝜇) (𝜇)) A lot of theoretical and computational works is devoted to WRCPP. WRCPP cannot be solved for polynomial time. In general, the problem of WRCPP is NP-hard [12]. A lot of theoretical and computational works is devoted to WRCPP. WRCPP cannot be solved for polynomial time. In general, the problem of WRCPP is NP-hard [12]. 2.1 The Windy Rural Chinese Postman problem The Windy Rural Chinese Postman Problem (WRCPP) is a special case of ARP, in which 𝐴𝑅 ⊆ 𝐴, 𝐸𝑅 ⊆ 𝐸, and the cost of traversing the edges is depended from the direction of traversing. The Windy Rural Chinese Postman Problem (WRCPP) is a special case of ARP, in which 𝐴𝑅 ⊆ 𝐴, 𝐸𝑅 ⊆ 𝐸, and the cost of traversing the edges is depended from the direction of traversing. WRCPP is a generalization of the CPP in a mixed multigraph. In original CPP problem, it is necessary to find a closed route of minimum length that contains all edges and arcs of the original multigraph at least once. In the real world, it is not always necessary to traverse absolutely all edges and arcs, it is enough to traverse only a certain set of them. Besides, the cost of traversing the edges depend from direction of traversing. The problem of this type is known as the Windy Rural Chinese Postman Problem, which is finding a closed route of minimum length that contains all required edges or arcs of the original multigraph at least once and can contains non-required edges or arcs, so, that the cost of traversing edges depends on traversing direction [5, 6]. g p yp y Postman Problem, which is finding a closed route of minimum length that contains all required edges or arcs of the original multigraph at least once and can contains non-required edges or arcs, so, that the cost of traversing edges depends on traversing direction [5, 6]. Fix the edge {𝑣𝑖, 𝑣𝑗} (non-oriented pair of vertex) from 𝐸. Define (𝑣𝑖, 𝑣𝑗) as ordered pair of vertices, meaning the traversing an edge {𝑣𝑖 , 𝑣𝑗 } from vertex 𝑣𝑖 to 𝑣𝑗 vertex. Note, that (∃{𝑣𝑖, 𝑣𝑗} ∈ 𝐸)(𝐶(𝑣𝑖, 𝑣𝑗) ≠ 𝐶(𝑣𝑖, 𝑣𝑗)) (1) (1) Let arc be (𝑣𝑖, 𝑣𝑗) ∈𝐴 ordered pair of vertices, meaning the traversing an arc (𝑣𝑖,𝑗) from vertex 𝑣𝑖 to 𝑣𝑗 vertex. Let arc be (𝑣𝑖, 𝑣𝑗) ∈𝐴 ordered pair of vertices, meaning the traversing an arc (𝑣𝑖,𝑗) from vertex 𝑣𝑖 to 𝑣𝑗 vertex. 222 Горденко М.К., Авдошин С.М. Варианты задач китайского почтальона и их решения через преобразование в задачи маршрутизации. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 221-232 We give a formal formulation of the WRCPP problem, extending it to the case of a mixed multigraph. 2.4 The Directed Rural Chinese Postman problem The Directed Rural Chinese Postman Problem (DRCPP) is a special case of the WRCPP where a subset of the set of arcs of a given directed graph is required to be traversed at minimum cost [2, 8]. DRCPP is a special case of WRCPP, where 𝐸 = ∅. DRCPP is a special case of WRCPP, where 𝐸 = ∅. In general, the DRCPP is NP-hard for directed multigraphs. This problem also known as the Selecting Chinese Postman problem [8]. This problem also known as the Selecting Chinese Postman problem [8]. 2.3 The Undirected Chinese Postman problem The Chinese Postman problem in the undirected graph (Undirected Chinese Postman Problem, UCPP) is the original statement of the CPP problem, which was firstly introduced by the mathematician Kwang-Mei-Ko in 1960 [2]. UCPP is a special case of WRCPP, where 𝐴 = ∅, 𝐸𝑅= 𝐸 and there is not edges, which introduced by the mathematician Kwang Mei Ko in 1960 [2]. UCPP is a special case of WRCPP, where 𝐴 = ∅, 𝐸𝑅 = 𝐸 and there is not edges, which satisfy (1). So, ∀{𝑣𝑖,𝑣𝑗} ∈ 𝐸, 𝐶(𝑣𝑖,𝑣𝑗) = 𝐶(𝑣𝑗,𝑣𝑖) UCPP is a special case of WRCPP, where 𝐴 = ∅, 𝐸𝑅 = 𝐸 and there is not edges, which satisfy (1). So, ∀{𝑣𝑖,𝑣𝑗} ∈ 𝐸, 𝐶(𝑣𝑖,𝑣𝑗) = 𝐶(𝑣𝑗,𝑣𝑖) 𝑗 𝑗 𝑗 If multigraph has Eulerian circuit then this cycle is a solution of UCPP. The algorithm for constructing the Eulerian circuit has 𝑂(\|𝐸\|) time complexity [5]. If multigraph has Eulerian circuit then this cycle is a solution of UCPP. The algorithm for constructing the Eulerian circuit has 𝑂(\|𝐸\|) time complexity [5]. The Eulerian circuit is existing in an undirected multigraph if multigraph is connected and every vertex has an even degree. A multigraph satisfying the conditions for the existence the Eulerian circuit is called Eulerian multigraph. If the original multigraph 223 Gordenko M.K., Avdoshin S.M. The Variants of Chinese Postman Problems and Way of Solving through Transformation into Vehicle Routing Problems. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 221-232 is not Eulerian, then for UCPP solution some edges must be traversed more than once. In other words, the multigraph should be supplemented with copies of some the edges to the Eulerian multigraph, so that the cost of the added copies of the edges is minimal [4]. 2.5 The Directed Chinese Postman problem The Chinese Postman problem in the directed graph (Directed Chinese Postman Problem, DCPP) is a special case of the WRCPP problem, in which defined on directed graph and all arcs should be traversed. In some articles DCPP also called New York Street Sweeper Problem [8]. DCPP is a special case of WRCPP, where 𝐸 = ∅, 𝐴𝑅 = 𝐴. DCPP is a special case of WRCPP, where 𝐸 = ∅, 𝐴𝑅 = 𝐴. If multigraph has Eulerian trail, then this trail is a solution of DCPP. The algorithm for constructing the Eulerian trail has 𝑂(\|𝐴\|) time complexity [9]. If multigraph has Eulerian trail, then this trail is a solution of DCPP. The algorithm for constructing the Eulerian trail has 𝑂(\|𝐴\|) time complexity [9]. The Eulerian trail is existing in a directed multigraph if multigraph is strongly connected and outdegree of each vertex is equal to indegree. A multigraph satisfying the conditions for the existence the Eulerian trail is called Eulerian multigraph. If the original multigraph is not Eulerian, then for DCPP solution some edges must be traversed more than once. In other words, the multigraph should be supplemented with copies of some the arcs to the Eulerian multigraph, so that the cost of the added copies of the arcs is minimal [2]. 2.6 The Undirected Windy Rural Chinese Postman problem The Undirected Windy Rural Chinese Postman Problem (UWRCPP) is an important ARP which generalizes most of the single-vehicle ARP and can be defined as follows [2, 9]. UWRCPP is a special case of WRCPP, where 𝐴 = ∅ and there is edges, which satisfy (1). UWRCPP is a special case of WRCPP, where 𝐴 = ∅ and there is edges, which satisfy (1). 2.7 The Undirected Windy Chinese Postman problem 2.7 The Undirected Windy Chinese Postman problem The Undirected Windy Chinese Postman problem is the NP- hard problem of finding the minimum cost of a tour traversing all edges of an undirected multigraph, where the cost of traversal of an edge depends on the direction [10]. UWCPP is a special case of WRCPP, where 𝐴 = ∅ and there is not edges, which satisfy (1). So, ∀{𝑣𝑖,𝑣𝑗} ∈ 𝐸, 𝐶(𝑣𝑖,𝑣𝑗) = 𝐶(𝑣𝑗,𝑣𝑖). The Undirected Windy Chinese Postman problem is the NP- hard problem of finding the minimum cost of a tour traversing all edges of an undirected multigraph, where the cost of traversal of an edge depends on the direction [10]. UWCPP is a special case of WRCPP, where 𝐴 = ∅ and there is not edges, which satisfy (1). So, ∀{𝑣𝑖,𝑣𝑗} ∈ 𝐸, 𝐶(𝑣𝑖,𝑣𝑗) = 𝐶(𝑣𝑗,𝑣𝑖). 224 Горденко М.К., Авдошин С.М. Варианты задач китайского почтальона и их решения через преобразование в задачи маршрутизации. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 221-232 If multigraph has Eulerian circuit then this cycle is a solution of WCPP. The algorithm for constructing the Eulerian circuit has 𝑂(\|𝐸\|) time complexity [5]. If the original multigraph is not Eulerian, then some should be traversed more than once. In other words, the multigraph should be supplemented with copies of the edges to the Eulerian multigraph so that the cost of the added copies of the edges is minimal. The solution of the complement problem for a graph that does not satisfy properties (9) and (10) is an NP-hard problem. Thus, WCPP belongs to the class of NP- hard that cannot be solved in polynomial time [13]. 2.9 The Mixed Windy Chinese Postman Problem The Mixed Windy Chinese Postman Problem (MWCCP, also called WCPP) is a special case of WRCPP. In MWCCP the cost of traversing the edges is depended from the direction of traversing. g UWRCPP is a special case of WRCPP, where there are edges, which satisfy (1). In many theoretical works it was shown that problem is NP- hard. UWRCPP is a special case of WRCPP, where there are edges, which satisfy (1). In many theoretical works it was shown that problem is NP- hard. 2.8 The Mixed Chinese Postman problem Mixed Chinese Postman Problem (MCPP) it is a version of WRCPP, where multigraph consists from edges and arcs, simultaneously, and all of them should be traversed [11, 12]. MCPP is a special case of WRCPP, where 𝐴𝑅 = 𝐴, 𝐸𝑅 = 𝐸. and there is not edges, which satisfy (1). So, ∀{𝑣𝑖, 𝑣𝑗} ∈ 𝐸, 𝐶(𝑣𝑖, 𝑣𝑗) = 𝐶(𝑣𝑗, 𝑣𝑖). In 1962, Ford and Fulkerson proposed necessary and sufficient conditions for a mixed graph to be Eulerian. It is necessary and sufficient that in a strongly connected multigraph, the degrees of all vertices are even, and the divergence of each vertex is zero. If a mixed multigraph does not satisfy these conditions, then it must be supplemented by copies of arcs and edges to the Eulerian multigraph, so that the cost of the added copies of the arcs and edges is minimal. The addition of a mixed multigraph to Eulerian is an NP-difficult problem [13]. • direction (C), • the presence of set of arcs (D), • the presence of set of arcs (D), • the presence of set of required arcs (E). The results are shown in “Table 1”. As we can see, there are four problems, which today are not existing (yellow cells in table), but also can have real-world applications. Table 1. The Classification of CPP UCPP URCPP UWCPP UWRCPP DCPP DRCPP MCPP DRUCPP MWCPP DURWCPP URDCPP MRCPP DRUWCPP WRMCPP A - - - - + + + + + + + + + + B - - - - - + - - - - + + + + C - + - + - - - + - + - + - + D + + + + - - + + + + + + + + E - - + + - - - - + + - - + + . Solving the Variations of Chinese Postman Problem 2.10 The Mixed Windy Chinese Postman Problem The Mixed Rural Chinese Postman Problem (MRCCP) is a special case of WRCPP. In MRCCP not all edges and arcs should be traversed. There is a set of arcs and edges, which must appear in solution, other arcs and edges may appear in solution or may not. MRCPP is a special case of WRCPP, where there are not edges, which satisfy (1). In many theoretical works it was shown that problem is NP- hard [14]. y p We tried to build a classification of different CPP. Combine the existing CPP in a table containing the following criteria: We tried to build a classification of different CPP. Combine the existing CPP in a table containing the following criteria: We tried to build a classification of different CPP. Combine the existing CPP in a table containing the following criteria: • the presence of set of edges (A), • the presence of set of required edges (B), • the presence of edges with cost, depending on traversing 225 Gordenko M.K., Avdoshin S.M. The Variants of Chinese Postman Problems and Way of Solving through Transformation into Vehicle Routing Problems. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 221-232 Gordenko M.K., Avdoshin S.M. The Variants of Chinese Postman Problems and Way of Solving through Transformation into Vehicle Routing Problems. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp • direction (C), Горденко М.К., Авдошин С.М. Варианты задач китайского почтальона и их решения через преобразование в задачи маршрутизации. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 221-232 Горденко М.К., Авдошин С.М. Варианты задач китайского почтальона и их решения через преобразование в задачи маршрутизации. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 221-232 3. Solving the Variations of Chinese Postman Problem g In many sources was shown that almost all ARP problems can be transformed into VRP problems, predominantly in generalized travelling salesman problems (GTSP) [13, 15, 16, 17]. For example, in [16] paper is described how the Capacitated Arc Routing Problem can be formulated as a standard vehicle routing problem. This allows us to transform arc routing into node routing problems and, therefore, establishes the equivalence of these two classes of problems. A well-known transformation by Pearn, Assad and Golden [16] reduces arc routing problem (ARP) into an equivalent vehicle routing problem (VRP). However, that transformation is regarded as unpractical, since an original instance with n required edges is turned into a VRP over a complete graph with 3n+1 vertices. In [15] article was proposed a similar transformation that reduces this graph to 2n+1 vertices, with the additional restriction that a previously known set of n pairwise disconnected edges must belong to every solution. Thus, one can move from less studied problems ARP to well-known problems VRP, such as TSP and GTSP, which have a lot of different approximation algorithms for solving. In the next sections, we try to compare the simplest algorithms for solving the GTSP. Generalized travelling salesman problem (GTSP) is an expansion of well-known TSP (Travelling Salesman Problem). In GTSP all vertices of graph are grouped in separate clusters. The solution of GTSP is a minimum-cost route, which traverse each cluster exactly once. 226 4. Methods for Solving the Generalized Travelling Salesman Problem Now, we investigate the following simple approximate algorithms for solving GTSP: • Nearest Neighbor Heuristic (NN) [19]; • Repetitive Nearest Neighbor Heuristic (RNN) [20]; • Improved Nearest Neighbor Heuristic (INN) [21];  Repetitive Improved Nearest Neighbor Heuristic (RINN) [22]; • Loneliest Neighbor Heuristic (NLN) [23]. To evaluate the developed algorithms, the source code was written in the C++language. Experiments was conducted on Apple Macbook Pro 13 a1502. Measurements were made of the executing time of the algorithm and the error rate of the solution. The results is presented in Table 1, 2, 3, 4, 5 and 6. Min(T), max(T) and M(T) means minimum, maximum and average time of algorithm working. Min(C), max(C) and M(C) means minimum, maximum and average error rate of algorithms. Table 2. The measurements of NN algorithms Table 2. The measurements of NN algorithms Table 2. The measurements of NN algorithms Table 2. The measurements of NN algorithms Table 3. The measurements of RNN algorithms Table 2. The measurements of NN algorithms Table 3. The measurements of RNN algorithms Table 3. The measurements of RNN algorithms 227 Gordenko M.K., Avdoshin S.M. The Variants of Chinese Postman Problems and Way of Solving through Transformation into Vehicle Routing Problems. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp Table 4. The measurements of INN algorithms Gordenko M.K., Avdoshin S.M. The Variants of Chinese Postman Problems and Way of Solving through Transformation into Vehicle Routing Problems. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 221-232 Table 4. The measurements of INN algorithms Gordenko M.K., Avdoshin S.M. The Variants of Chinese Postman Problems and Way of Solving through Transformation into Vehicle Routing Problems. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 221-232 Table 4. The measurements of INN algorithms Gordenko M.K., Avdoshin S.M. The Variants of Chinese Postman Problems and Way of Solving through Transformation into Vehicle Routing Problems. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 221-232 Table 5. The measurements of RINN algorithms Table 6. The measurements of NLN algorithms Table 6. The measurements of NLN algorithms References [1]. Eglese R., Letchford A., General Routing Problem. In Encyclopedia of Optimization. Springer, Boston, MA. 2008. [2]. Thimbleby, H. The directed chinese postman problem. Software: Practice and Experience, 33(11), 2003, pp. 1081-1096. [3]. Toth P., Vigo D. (ed.). The vehicle routing problem. – Society for Industrial and Applied Mathematics, 2002. [4]. Hertz A., Laporte G., Mittaz M. A tabu search heuristic for the capacitated arc routing problem. Operations research, vol. 48, no. 1, 2000, pp. 129-135. [5]. Zerbino D. R., Birney E. Velvet: algorithms for de novo short read assembly using de Bruijn graphs. Genome research, vol. 18, no. 5, 2008, pp. 821-829. [6]. Edmonds J., Johnson E. L. Matching, Euler tours and the Chinese postman. Mathematical programming, vol. 5, no. 1, 1973, pp. 88- 124. [7]. Kolmogorov V. Blossom V: a new implementation of a minimum cost perfect matching algorithm. Mathematical Programming Computation, vol. 1, no. 1, 2009, pp. 43-67. [8]. Robinson H. Graph theory techniques in model-based testing. In Proc. of the International Conference on Testing Computer Software, 1999. g [9]. Wilson R. J. An eulerian trail through Königsberg. Journal of graph theory, vol., 10, no. 3, 1986, pp. 265-275. [10]. Ababei C., Kavasseri R. Efficient network reconfiguration using minimum cost maximum flow-based branch exchanges and random walks-based loss estimations, IEEE Transactions on Power Systems, vol. 26, no. 1, 2010, pp.. 30-37. [11]. Chen W. H. Test sequence generation from the protocol data portion based on the Selecting Chinese Postman algorithm. Information Processing Letters, vol. 65, no. 5, pp. 261-268. [12]. Aho A.V. et al. An optimization technique for protocol conformance test generation based on UIO sequences and rural Chinese postman tours. IEEE transactions on communications, vol. 39, no. 11, 1991, pp, 1604-1615. pp [13]. Dror M. (ed.). Arc routing: theory, solutions and applications. Springer Science & Business Media, 2012. [14]. Ghiani G., Improta G. An algorithm for the hierarchical Chinese postman problem. Operations Research Letters, vol. 26, no. 1, 2000, pp. 27- 32. [15]. Longo H., De Aragão M. P., Uchoa E. Solving capacitated arc routing problems using a transformation to the CVRP. Computers & Operations Research, vol. 33, no. 6, pp. 1823- 1837. [16]. Pearn W. L., Assad A., Golden B. L. Transforming arc routing into node routing problems, Computers & operations research, vol. 14, no. 4, 1987, pp. 285-288. [17]. Laporte G. Modeling and solving several classes of arc routing problems as traveling salesman problems. 5. Summary This article provides an overview of the known CPP. An attempt to systematize and classify these problems has been made. Mathematical formulations of new types of CPP was founded. The paper also shows that almost all problems of the ARP can be transformed to VRP. In addition, for solving the Chinese Postman problems the way of transformation it into VRP (mainly in GTSP) has been chosen. Горденко М.К., Авдошин С.М. Варианты задач китайского почтальона и их решения через преобразование в задачи маршрутизации. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 221-232 Горденко М.К., Авдошин С.М. Варианты задач китайского почтальона и их решения через преобразование в задачи маршрутизации. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 221-232 At this stage, the research is not complete. It is necessary to investigate the various ways of transformation ARP is into VRP. In addition, it is necessary to investigate the various ways of solving the GTSP. And the key idea of future research is the use of transformation algorithms and algorithms for solving the GTSP for solving the different modifications of CPP. At this stage, the research is not complete. It is necessary to investigate the various ways of transformation ARP is into VRP. In addition, it is necessary to investigate the various ways of solving the GTSP. And the key idea of future research is the use of transformation algorithms and algorithms for solving the GTSP for solving the different modifications of CPP. References Computers & operations research, vol. 24, no. 11, 1997, pp. 1057- 1061. 229 Gordenko M.K., Avdoshin S.M. The Variants of Chinese Postman Problems and Way of Solving through Gordenko M.K., Avdoshin S.M. The Variants of Chinese Postman Problems and Way of Solving through Transformation into Vehicle Routing Problems. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 221-232 Transformation into Vehicle Routing Problems. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp [18]. Fischetti M., Salazar González J. J., Toth P. A branch-and-cut algorithm for the symmetric generalized traveling salesman problem. Operations Research, vol. 45, no. 3. 1997. pp. 378-394. [19]. Solomon M. M. Algorithms for the vehicle routing and scheduling problems with time window constraints, Operations research, vol. 35, no. 2, 1987, pp. 254-265. [20]. Modares A., Somhom S., Enkawa T. A self-organizing neural network approach for multiple traveling salesman and vehicle routing problems. International Transactions in Operational Research, vol. 6, no. 6, 1999, pp. 591-606. p pp [21]. Cheung K.L., Fu A.W.C. Enhanced nearest neighbor search on the R-tree. ACM SIGMOD Record, vol. 27, no. 3, 1998. pp. 16-21. [22]. Tao Y., Papadias D., Shen Q. Continuous nearest neighbor search. In Proceedings of the 28th International Conference on Very Large Databases, 2002, pp. 287-298. [23]. Pimentel F. G. S. L. Double-ended nearest and loneliest neighbour: a nearest neighbour heuristic variation for the travelling salesman problem. Revista de Ciências da Computação, vol. 6, issue 6, 2011. DOI: 10.15514/ISPRAS-2018-30(3)-16 DOI: 10.15514/ISPRAS-2018-30(3)-16 Для цитирования: Горденко М.К., Авдошин С.М. Варианты задач китайского почтальона и их решения через преобразование в задачи маршрутизации. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 221-232 (на английском языке). DOI: 10.15514/ISPRAS- 2018-30(3)-16 Варианты задач китайского почтальона и их решения через преобразование в задачи маршрутизации М.К. Горденко <mgordenko@hse.ru> С.М. Авдошин <savdoshin@hse.ru> Департамент программной инженерии, Национальный исследовательский университет “Высшая школа экономики”, 101000, Россия, г. Москва, ул. Мясницкая, д. 20 Abstract. В статье описаны проблемы маршрутизации. Показано, что почти все проблемы маршрутизации дуг могут быть преобразованы в другие проблемы маршрутизации. Это продемонстрировано на примере задачи китайского почтальона в смешанном мультиграфе. Также в статье приведен обзор различных задач китайского почтальона (в зависимости от типа графа, функции стоимости и обязательности прохождения элементов графа). Для каждой проблемы дана математическая постановка. Кроме того, описаны примеры потенциально полезных приложений, где задачи могут быть применены. Приведена таблица различных вариантов задачи китайского почтальона и выбраны параметры для идентификации различных типов задач. Выделено пять параметров: наличие дуг, наличие ребер, наличие обязательных дуг, наличие обязательных ребер, наличие ребер со стоимостью, зависящей от прохода. Показано, что, варьируя эти параметры, можно получить новые задачи, которые могут быть полезны в реальной жизни, однако еще не описаны. Выявлены четыре таких задачи. Показано, что задача китайского почтальона может быть решена путем преобразования в другие задачи маршрутизации. Приведен метод, позволяющий преобразовать задачу в обобщенную задачу коммивояжера. Показаны результаты применения простейших алгоритмов для решения преобразованного варианта задачи (результаты приведены для алгоритмов ближайшего соседа и их модификаций). Исследование еще не завершено, планируется продолжать тестировать алгоритмы решения смежных задач маршрутизации и алгоритмы для преобразований задач в эквивалентные. 230 Горденко М.К., Авдошин С.М. Варианты задач китайского почтальона и их решения через преобразование в задачи маршрутизации. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 221-232 р ру ру р Keywords: обобщенная задача коммивояжера; задача маршрутизации дуг; задача маршрутизации; обобщенная задача маршрутизации; задача китайского почтальона DOI: 10.15514/ISPRAS-2018-30(3)-16 Keywords: обобщенная задача коммивояжера; задача маршрутизации дуг; задача маршрутизации; обобщенная задача маршрутизации; задача китайского почтальона Список литературы: [1]. Eglese R., Letchford A., General Routing Problem. In Encyclopedia of Optimization. Springer, Boston, MA. 2008. [2]. Thimbleby, H. The directed chinese postman problem. Software: Practice and Experience, 33(11), 2003, pp. 1081-1096. [3]. Toth P., Vigo D. (ed.). The vehicle routing problem. – Society for Industrial and Applied Mathematics, 2002. [4]. Hertz A., Laporte G., Mittaz M. A tabu search heuristic for the capacitated arc routing problem. Operations research, vol. 48, no. 1, 2000, pp. 129-135. [5]. Zerbino D. R., Birney E. Velvet: algorithms for de novo short read assembly using de Bruijn graphs. Genome research, vol. 18, no. 5, 2008, pp. 821-829. [6]. Edmonds J., Johnson E. L. Matching, Euler tours and the Chinese postman. Mathematical programming, vol. 5, no. 1, 1973, pp. 88- 124. [7]. Kolmogorov V. Blossom V: a new implementation of a minimum cost perfect matching algorithm. Mathematical Programming Computation, vol. 1, no. 1, 2009, pp. 43-67. [8]. Robinson H. Graph theory techniques in model-based testing. In Proc. of the International Conference on Testing Computer Software, 1999. [9]. Wilson R. J. An eulerian trail through Königsberg. Journal of graph theory, vol., 10, no. 3, 1986, pp. 265-275. [10]. Ababei C., Kavasseri R. Efficient network reconfiguration using minimum cost maximum flow-based branch exchanges and random walks-based loss estimations, IEEE Transactions on Power Systems, vol. 26, no. 1, 2010, pp.. 30-37. [11]. Chen W. H. Test sequence generation from the protocol data portion based on the Selecting Chinese Postman algorithm. Information Processing Letters, vol. 65, no. 5, pp. 261-268. [12]. Aho A.V. et al. An optimization technique for protocol conformance test generation based on UIO sequences and rural Chinese postman tours. IEEE transactions on communications, vol. 39, no. 11, 1991, pp, 1604-1615. [13]. Dror M. (ed.). Arc routing: theory, solutions and applications. Springer Science & Business Media, 2012. [14]. Ghiani G., Improta G. An algorithm for the hierarchical Chinese postman problem. Operations Research Letters, vol. 26, no. 1, 2000, pp. 27- 32. [15]. Longo H., De Aragão M. P., Uchoa E. Solving capacitated arc routing problems using a transformation to the CVRP. Computers & Operations Research, vol. 33, no. 6, pp. 1823- 1837. [16]. Pearn W. L., Assad A., Golden B. L. Transforming arc routing into node routing problems, Computers & operations research, vol. 14, no. 4, 1987, pp. 285-288. 231 ordenko M.K., Avdoshin S.M. DOI: 10.15514/ISPRAS-2018-30(3)-17 For citation: Beresneva E., Avdoshin S. Analysis of Mathematical Formulations of Capacitated Vehicle Routing Problem and Methods for their Solution. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 233-250. DOI: 10.15514/ISPRAS-2018-30(3)-17 For citation: Beresneva E., Avdoshin S. Analysis of Mathematical Formulations of Capacitated Vehicle Routing Problem and Methods for their Solution. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 233-250. DOI: 10.15514/ISPRAS-2018-30(3)-17 Analysis of Mathematical Formulations of Capacitated Vehicle Routing Problem and Methods for their Solution E. Beresneva <eberesneva@hse.ru> S. Avdoshin <savdoshin@hse.ru> Department of Software Engineering, National Research University Higher School of Economics, 20, Myasnitskaya st., Moscow, 101000 Russia E. Beresneva <eberesneva@hse.ru> S. Avdoshin <savdoshin@hse.ru> Department of Software Engineering, National Research University Higher School of Economics, 20, Myasnitskaya st., Moscow, 101000 Russia Abstract. Vehicle Routing Problem (VRP) is one of the most widely known questions in a class of combinatorial optimization problems. It is concerned with the optimal design of routes to be used by a fleet of vehicles to serve a set of customers. In this study we analyze Capacitated Vehicle Routing Problem (CVRP) – a subcase of VRP, where the vehicles have a limited capacity. CVRP is mostly aimed at savings in the global transportation costs. The problem is NP-hard, therefore heuristic algorithms which provide near-optimal polynomial-time solutions will be considered instead of the exact ones. The aim of this article is to make a survey on mathematical formulations of CVRP and on methods for solving each type of this problem. The first part presents a general information about the problem and restrictions of this work. In the second part, the classical mathematical formulations of CVRP are described. In the third part, a classification of most popular subcases of CVRP is given, including description of additional constraints with their math formulations. This section also includes most perspective methods that can be applied for solving special types of CVRP. The forth part contains an important note about the most powerful algorithm LKH-3. Finally, the fourth part consists of table with solving techniques for each subproblem and of scheme with basic problems of the CVRP class and their interconnections. Keywords: capacitated vehicle routing problem; mathematical formulation; metaheuristics; classification of cvrp Список литературы: The Variants of Chinese Postman Problems and Way of Solving through Gordenko M.K., Avdoshin S.M. The Variants of Chinese Postman Problems and Way of Solving through Transformation into Vehicle Routing Problems. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 221-232 Gordenko M.K., Avdoshin S.M. The Variants of Chinese Postman Problems and Way of Solving through Transformation into Vehicle Routing Problems. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp [17]. Laporte G. Modeling and solving several classes of arc routing problems as traveling salesman problems. Computers & operations research, vol. 24, no. 11, 1997, pp. 1057- 1061. [18]. Fischetti M., Salazar González J. J., Toth P. A branch-and-cut algorithm for the symmetric generalized traveling salesman problem. Operations Research, vol. 45, no. 3. 1997. pp. 378-394. [19]. Solomon M. M. Algorithms for the vehicle routing and scheduling problems with time window constraints, Operations research, vol. 35, no. 2, 1987, pp. 254-265. [20]. Modares A., Somhom S., Enkawa T. A self-organizing neural network approach for multiple traveling salesman and vehicle routing problems. International Transactions in Operational Research, vol. 6, no. 6, 1999, pp. 591-606. p pp [21]. Cheung K.L., Fu A.W.C. Enhanced nearest neighbor search on the R-tree. ACM SIGMOD Record, vol. 27, no. 3, 1998. pp. 16-21. [22]. Tao Y., Papadias D., Shen Q. Continuous nearest neighbor search. In Proceedings of the 28th International Conference on Very Large Databases, 2002, pp. 287-298. [23]. Pimentel F. G. S. L. Double-ended nearest and loneliest neighbour: a nearest neighbour heuristic variation for the travelling salesman problem. Revista de Ciências da Computação, vol. 6, issue 6, 2011. 232 1. Introduction The Vehicle Routing Problem (VRP) is one of the most widely known questions in a class of combinatorial optimization problems. VRP is directly related to Logistics 233 Beresneva E., Avdoshin S. Analysis of Mathematical Formulations of Capacitated Vehicle Routing Problem and Methods for their Solution. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp, 233-250 Beresneva E., Avdoshin S. Analysis of Mathematical Formulations of Capacitated Vehicle Routing Problem and Methods for their Solution. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp, 233-250 transportation problem and it is meant to be a generalization of the Travelling Salesman Problem (TSP). In contrast to TSP, VRP produces solutions containing some (usually, more than one) looped cycles, which are started and finished at the same point called “depot”. The objective is to minimize the cost (time or distance) for all tours. For the identical type of input data, VRP has higher solving complexity than TSP. Both problems belong to the class of NP-hard tasks. Specialized algorithms are able to find optimal solutions for cases with up to about 50 customers; larger problems have been solved to optimality in some cases, but often at the expense of considerable computing time. Thus, actuality of research and development of heuristics algorithms for solving VRP is on its top, because such approximate algorithms can produce near- optimal solutions in a polynomial time. It is especially important in real-based tasks when there are more than one hundred clients in a delivery net. Real world applications may be mail delivery, solid waste collection, street cleaning, distribution of commodities, design telecommunication, transportation networks, school bus routing, dial–a–ride systems, transportation of handicapped persons, and routing of sales people and maintenance units. A survey of real–world applications is in [1]. This work is aimed at analysis of VRP subcase, which is called Capacitated Vehicle Routing Problem (Capacitated VRP, CVRP), where the vehicles have a limited capacity. It means that there is a physical restriction on transportation more than determined amount of weight for each machine. Capacitated vehicle routing problems CVRP form the core of logistics planning and are hence of great practical and theoretical interest. Nowadays, there is a great range of different variations of both classical mathematical model of CVRP and its subcases. It can be too difficult to understand all the details for newcomers in this field of study. 2. CVRP mathematical model In this paper, mathematical formulation of Asymmetrical CVRP (ACVRP) proposed by original authors [2] is adopted in a new way as follows. This new variant of math model is created because only Integer Programming models were found in other articles. ACVRP is chosen for basic formulation instead of Symmetrical CVRP (SCVRP) because the first one is a general variant of the second problem. In the paper we will use CVRP abbreviation having in mind the next formulation. Given a complete weighted oriented graph 𝐺= (𝑉, 𝐴). Let 𝐼= {0, 1, … , 𝑁}, where 𝑁= \|𝑉\|. Graph vertices are indexed as = 𝑉→𝐼, (∀𝑣∈𝑉)(∀𝑤∈𝑉) 𝑣≠𝑤⟹ 𝑖𝑛𝑑(𝑣) ≠𝑖𝑛𝑑(𝑤). Thus, 𝑉= {𝑣0, 𝑣1,… , 𝑣𝑁} is set of vertices, here 𝑖= 𝑖𝑛𝑑(𝑣𝑖), and 𝐴 is set of arcs. Let 𝑣0 be a depot, where vehicles are located, and 𝑣𝑖 be the destination points of a delivery, 𝑖 ≠0. The distance between two vertices 𝑣𝑖 and 𝑣𝑗 is calculated using a distance function 𝑐(𝑣𝑖, 𝑣𝑗). Here a real-valued function 𝑐(∙,∙) on 𝑉× 𝑉 satisfies [3]: 𝑐(𝑣𝑖, 𝑣𝑗). Here a real-valued function 𝑐(∙,∙) on 𝑉× 𝑉 satisfies [3]: ⎼ 𝑐(𝑣𝑖, 𝑣𝑗) ≥0 (non-negativity axiom) ⎼ 𝑐(𝑣𝑖, 𝑣𝑗) ≥0 (non-negativity axiom) ⎼ 𝑐(𝑣𝑖, 𝑣𝑗) = 0 if and only if 𝑣𝑖= 𝑣𝑗 (identity axiom) Each destination point 𝑣𝑖, 𝑖= 0. . 𝑁 ̅̅̅̅̅̅, is associated with a known nonnegative demand, 𝑑𝑖, to be delivered, and the depot has a fictitious demand 𝑑0 = 0. The total demand of the set 𝑉′ ⊆𝑉 is calculated as 𝑑(𝑉′) = ∑ 𝑑𝑖 𝑖∈\|𝑉′\| . Let 𝐾 be a number of available vehicles at the depot 𝑣0. Each vehicle has the same capacity – 𝐶. Let us assume that every vehicle may perform at most one route and 𝐾≥𝐾𝑚𝑖𝑛, where 𝐾𝑚𝑖𝑛 is a minimal number of vehicles needed to serve all the customers due to restriction on 𝐶. Clearly, next condition must be fulfilled – (∀𝑣𝑖 ∈𝑉) 𝑑𝑖≤𝐶, which prohibits goods transportation that exceed maximum vehicle capacity. Let introduce 𝑉0 = {𝑣0}, where 𝑣0 ∈𝑉. We divide 𝑉 in 𝐾+ 1 sets: 𝑆= {𝑉0, 𝑉1, … , 𝑉𝐾}, each subset, except for 𝑉0, represent a set of customers to be served for one vehicle. 𝑆𝑎𝑙𝑙= {𝑆} is a set of all possible partitions of 𝑉. Let 𝐽= {0, 1, … , 𝐾} be a set that keeps indexes. Then (∀𝑖∈𝐽) \|𝑉𝑖\| ≥1. Береснева Е.Н., Авдошин С.М. Анализ математических постановок задачи маршрутизации с ограничением по грузоподъемности и методов их решения. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 233-250 Береснева Е.Н., Авдошин С.М. Анализ математических постановок задачи маршрутизации с ограничением по грузоподъемности и методов их решения. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 233-250 1. Introduction It is important to have an ability not to waste personal time doing observation but to quickly get the best solution methods for the current problem. Unfortunately, there are no articles concerned with CVRP, which have both a full classification of the subcases and a list of the solving algorithms. So, the purpose of this study is to make a survey on subcases of CVRP and on state-of- the-art heuristic methods for solving each extension of this problem. Also, it was decided to provide a new variant of mathematical model differed from Integer Linear Programming models. Clearly, a study of this type is inevitably restricted by various constraints, in this research only CVRP subcases with static and deterministic input are considered instead of the dynamic and stochastic ones. Another condition is that classification is based according to various types of constraints. The paper is structured as follows. In the second part, the classical mathematical formulations of CVRP are described. In the third part, a classification of most popular subcases of CVRP is given, including description of additional constraints with their math formulations. This section also includes most perspective methods that can be applied for solving special types of CVRP. Finally, the fourth part consists of scheme with basic problems of the CVRP class and their interconnections and of conclusion. 234 2. CVRP mathematical model According to [1], another variant of mathematical formulation of CVRP allows to leave some vehicles unused, it means that at most 𝐾 circuits must be determined. Of course, the number of 𝐾𝑚𝑖𝑛 must be less or equal than 𝐾. In this case in the basic formulation described above we should subsequently divide 𝑉 in 𝐾′ + 1 sets: 𝑆= {𝑉0, 𝑉1, … , 𝑉𝐾}, where 𝐾′ ∈[𝐾𝑚𝑖𝑛;𝐾]. Alternative variant takes its place from real-based situations where available vehicles have their own capacity 𝐶𝑖, 𝑖= 1. . 𝐾 ̅̅̅̅̅̅. Due to this fact, next restriction appears: (𝑉0𝑖) ≤𝐶𝑖, 𝑖= 1. . 𝐾 ̅̅̅̅̅̅. Alternative variant takes its place from real-based situations where available vehicles have their own capacity 𝐶𝑖, 𝑖= 1. . 𝐾 ̅̅̅̅̅̅. Due to this fact, next restriction appears: (𝑉0𝑖) ≤𝐶𝑖, 𝑖= 1. . 𝐾 ̅̅̅̅̅̅. However, most researches put this alternative to another class of problems not connected with CVRP which is known as the Mixed Fleet VRP or as the Heterogeneous Fleet VRP. Thus, this variant will not be taken into consideration in this paper. Among the best-known heuristic algorithms are those proposed by Pisinger and Ropke (2007) [4], Nagata and Braysy (2009) [5], and Vidal et al. (2012) [6]. 2. CVRP mathematical model There should be no duplicates in any of subsets from 𝑆: (∀𝑖∈𝐽)(∀𝑗∈𝐽) (𝑖≠𝑗⇒𝑉𝑖∩𝑉𝑗= ∅). Also, all subsets from 𝑆 must form set 𝑉. Thus, 𝑉= ⋃ 𝑉𝑖 𝐾 𝑖=0 . In this notation, we should make 𝑉0𝑖= 𝑉0 ∪𝑉𝑖, 𝑖= 1. . 𝐾 ̅̅̅̅̅̅. It is obvious that 𝑑(𝑉𝑜𝑖) ≤𝐶, 𝑖= 1. . 𝐾 ̅̅̅̅̅̅. Let introduce 𝑀𝑖= {1, … , 𝑁𝑖}, 𝑁𝑖= \|𝑉𝑖\|, ∑ 𝑁𝑖 𝐾 𝑖=1 = 𝑁.So, 𝑀0𝑖= {0} ∪𝑀𝑖. Let 𝐼𝑖= {𝑖\|𝑖= 𝑖𝑛𝑑(𝑣), 𝑣∈𝑉𝑖} be a set of vertex indices from 𝑉𝑖. Then 𝐼0𝑖= {0} ∪𝐼𝑖. Let 𝐻𝑖= {𝑝𝑖: 𝑀0𝑖→𝐼0𝑖\| 𝑝𝑖(0) = 0 & (∀𝑥∈𝑀0𝑖)(∀𝑦∈𝑀0𝑖) 𝑥≠𝑦 ⟹𝑝𝑖(𝑥) ≠𝑝𝑖(𝑦)} be a set of codes of all Hamiltonian cycles ℎ𝑖= (𝑣𝑝𝑖(0), 𝑣𝑝𝑖(1), … , 𝑣𝑝𝑖(𝑁𝑖)) of 𝑉0𝑖. Weight of a Hamiltonian cycle ℎ𝑖∈𝐻𝑖 can be Let introduce 𝑀𝑖= {1, … , 𝑁𝑖}, 𝑁𝑖= \|𝑉𝑖\|, ∑ 𝑁𝑖 𝐾 𝑖=1 = 𝑁.So, 𝑀0𝑖= {0} ∪𝑀𝑖. Let 𝐼𝑖= {𝑖\|𝑖= 𝑖𝑛𝑑(𝑣), 𝑣∈𝑉𝑖} be a set of vertex indices from 𝑉𝑖. Then 𝐼0𝑖= {0} ∪𝐼𝑖. Let 𝐻𝑖= {𝑝𝑖: 𝑀0𝑖→𝐼0𝑖\| 𝑝𝑖(0) = 0 & (∀𝑥∈𝑀0𝑖)(∀𝑦∈𝑀0𝑖) 𝑥≠𝑦 ⟹𝑝𝑖(𝑥) ≠𝑝𝑖(𝑦)} be a set of codes of all Hamiltonian cycles ℎ𝑖= (𝑣𝑝𝑖(0), 𝑣𝑝𝑖(1), … , 𝑣𝑝𝑖(𝑁𝑖)) of 𝑉0𝑖. Weight of a Hamiltonian cycle ℎ𝑖∈𝐻𝑖 can be 235 Beresneva E., Avdoshin S. Analysis of Mathematical Formulations of Capacitated Vehicle Routing Problem and Methods for their Solution. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp, 233-250 Beresneva E., Avdoshin S. Analysis of Mathematical Formulations of Capacitated Vehicle Routing Problem and Methods for their Solution. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp, 233-250 found as 𝑓(ℎ𝑖) = 𝑐(𝑣𝑝𝑖(0), 𝑣𝑝𝑖(𝑁𝑖)) + ∑ 𝑐(𝑣𝑝𝑖(𝑗), 𝑣𝑝𝑖(𝑗+1)) 𝑁𝑖−1 𝑗=0 . Let 𝑆′ be a set of {𝑉01, 𝑉02, … , 𝑉0𝐾}. In this notation the weight of 𝑆′ is calculated as 𝐹(𝑆′) = ∑ 𝑓(ℎ𝑖) 𝑖=1..𝐾 ̅̅̅̅̅ . 0 0 𝐹(𝑆) ∑ 𝑓(ℎ) 𝑖=1..𝐾 . Overall, the formulation of CVRP is to find such 𝑆0: 𝐹(𝑆0) = min 𝑆∈𝑆𝑎𝑙𝑙𝐹(𝑆). If 𝑐(𝑣𝑖,𝑣𝑗) = 𝑐(𝑣𝑗, 𝑣𝑖) for ∀𝑣𝑖∈𝑉 ∀𝑣𝑗∈𝑉 then the problem is symmetrical (SCVRP) and triangle inequality axiom must be hold 𝑐(𝑣𝑖,𝑣𝑘) ≤ 𝑐(𝑣𝑖,𝑣𝑗) + 𝑐(𝑣𝑗, 𝑣𝑘). If 𝑐(𝑣𝑖,𝑣𝑗) = 𝑐(𝑣𝑗, 𝑣𝑖) for ∀𝑣𝑖∈𝑉 ∀𝑣𝑗∈𝑉 then the problem is symmetrical (SCVRP) and triangle inequality axiom must be hold 𝑐(𝑣𝑖,𝑣𝑘) ≤ 𝑐(𝑣𝑖,𝑣𝑗) + 𝑐(𝑣𝑗, 𝑣𝑘). 3.2. Distance-Constrained CVRP (DCVRP) The next extension of CVRP to be considered is Distance-Constrained CVRP [14]. It suggests introducing the maximum length or time constraint for each route. It means that the total travelled distance by each vehicle in the solution is less than or equal to the maximum possible travelled distance 𝑇. Thus, new function 𝑡(𝑣𝑖,𝑣𝑗), returning travel time between 𝑣𝑖 and 𝑣𝑗, appears. Function 𝑡(∙,∙) on 𝑉× 𝑉 satisfies the same axioms as 𝑐(∙,∙). 𝑓𝑇(ℎ𝑖) = 𝑡(𝑣𝑝𝑖(0), 𝑣𝑝𝑖(𝑁𝑖)) + ∑𝑡(𝑣𝑝𝑖(𝑗), 𝑣𝑝𝑖(𝑗+1)) 𝑁𝑖−1 𝑗=0 𝐹𝑇(𝑆′) = ∑ 𝑓𝑇(ℎ𝑖) 𝑖=1..𝐾 ̅̅̅̅̅ (∀𝑖= 1. . 𝐾 ̅̅̅̅̅̅)(𝐹𝑇(ℎ𝑖) ≤𝑇𝑀𝑎𝑥) Most heuristics applied to simple CVRP can be easily converted for solving DCVRP cases. However, one heuristic proposed by Li et al. stands out from them [15]. It transforms the DCVRP into a multiple traveling salesman problem with time windows. Береснева Е.Н., Авдошин С.М. Анализ математических постановок задачи маршрутизации с ограничением по грузоподъемности и методов их решения. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 233-250 Береснева Е.Н., Авдошин С.М. Анализ математических постановок задачи маршрутизации с ограничением по грузоподъемности и методов их решения. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 233-250 3.1. Open VRP (OVRP) The OVRP is a variant of the CVRP where the vehicles need not return to the depot after visiting the last customer of a given route. Any OVRP instance can be converted to an ACVRP instance by simply setting 𝑐(𝑣𝑖, 𝑣0) = 0. There is only one heuristic algorithm for solving OVRP proposed by Salari et al. (2010) [9]. Their method is based on Integer Linear Programming Improvement Procedure. There is a good variety of metaheuristics. Most known and important are following algorithms: Hybrid evolution strategy algorithm by Repoussis et al. (2010) [10], variant of Variable Neighborhood Search (VNS) algorithm for OVRP by Fleszar et al. (2009) [11], method based on Tabu Search (TS) with route-evaluations memories by Zachariadis and Kiranoudis (2010) [12], Yu et al. (2011) Genetic algorithm and the last one is Particle swarm optimization metaheuristic proposed by MirHassani and Abolghasemi (2011) [13]. 236 Береснева Е.Н., Авдошин С.М. Анализ математических постановок задачи маршрутизации с ограничением по грузоподъемности и методов их решения. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 233-250 Береснева Е.Н., Авдошин С.М. Анализ математических постановок задачи маршрутизации с ограничением по грузоподъемности и методов их решения. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 233-250 3.3. VRP with Time Windows (VRPTW) In VRPTW there is a constraint on time interval [𝑎𝑖; 𝑏𝑖] associated with each 𝑣𝑖, called time window. It means that service of each customer must start only after the time 𝑎𝑖 comes and this service must end before the time 𝑏𝑖. Obviously, 𝑎0 = 0 and 𝑏0 = ∞ for 𝑣0. Let us assume that if 𝑡𝑐𝑢𝑟 is a current time, then all vehicles leave 𝑣0 when 𝑡𝑐𝑢𝑟= 0. If a vehicle arrives to 𝑣𝑖 at the moment when 𝑡𝑐𝑢𝑟< 𝑎𝑖, then it is obliged to wait until 𝑡𝑐𝑢𝑟= 𝑎𝑖 and to start serving only after that moment. New function 𝑡(𝑣𝑖,𝑣𝑗), returning travel time between 𝑣𝑖 and 𝑣𝑗, appears. Also, a variable 𝑠𝑟𝑣𝑖 keeping serving time of 𝑣𝑖 is introduced. It is clear, that the problem can be solved if {(∀𝑣𝑖∈𝑉)(∃𝑣𝑗∈𝑉) \| 𝑎𝑖+ 𝑠𝑟𝑣𝑖+ 𝑡(𝑣𝑖, 𝑣𝑗) + 𝑠𝑟𝑣𝑗≤𝑏𝑗}. There are a lot of metaheuristics for solving VRPTW, but the most actual and state- of-the-art ones are given. The guided Evolutionary algorithm of Repoussis et al. (2009) [16] combines evolution, ruin-and-recreate mutations and guided local search. Prescott-Gagnon et al. (2009) [17] suggests a Large Neighborhood search (LNS) combined with branch-and-price for solution reconstruction. The method proposed by Nagata et al. (2010) [18] uses an interesting relaxation scheme with penalized returns in time. Another algorithm (Vidal et al. (2013)) [19] also applies time- constraint relaxations during the search to benefit from infeasible solutions in the search space. There are a lot of metaheuristics for solving VRPTW, but the most actual and state- of-the-art ones are given. The guided Evolutionary algorithm of Repoussis et al. (2009) [16] combines evolution, ruin-and-recreate mutations and guided local search. P tt G t l (2009) [17] t L N i hb h d h (LNS) 237 Beresneva E., Avdoshin S. Analysis of Mathematical Formulations of Capacitated Vehicle Routing Problem and Methods for their Solution. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp, 233-250 3.4. VRP with Backhauls (VRPB) Beresneva E., Avdoshin S. Analysis of Mathematical Formulations of Capacitated Vehicle Routing Problem and Methods for their Solution. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp, 233-250 Beresneva E., Avdoshin S. Analysis of Mathematical Formulations of Capacitated Vehicle Routing Problem and Methods for their Solution. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp, 233-250 3.5. VRP with Backhauls and Time Windows (VRPBTW) Like in VRPB, VRPBTW suggests having linehaul and backhaul customers. In addition, with every location 𝑣𝑖 there is a service time 𝑠𝑟𝑣𝑖 associated for loading/unloading and a time window [𝑎𝑖; 𝑏𝑖], which specifies the time in which this service has to be provided. In the same way as for VRPTW, when arriving too early at a location 𝑣𝑖, i.e., before 𝑎𝑖, the vehicle is allowed to wait until 𝑎𝑖 to start the service. Also, the linehaul customers must be served before any backhaul customers. Thus, mathematical formulation of VRPBTW is a combination of both formulations of VRPTW and VRP. The most powerful algorithms for solving VRPBTW are those which are proposed by Thangiah et al. (1996) [24] and by Kucukoglu et al (2015) [25]. The first method is based on insertion procedure with improving through the application of λ- interchange and 2-opt exchange procedures. The second one includes combination of TS and SA. 3.4. VRP with Backhauls (VRPB) VRPB is another extension to CVRP. To define VRPB we need to divide the set of customers 𝑉𝑜𝑖 into two subsets: the first set contains customers who require the product to be delivered, these customers are called linehaul customers 𝐿𝑖⊂𝑉𝑜𝑖. The other set contains customers who require the product to picked up, they are called backhaul customers 𝐵𝑖⊂𝑉𝑜𝑖. 𝐿𝑖∩ 𝐵𝑖= ∅, 𝐿𝑖∪𝐵𝑖= 𝑉𝑜𝑖. Also, neither all deliveries nor all pick-ups should exceed vehicle capacity: 𝑑(𝐿𝑖) ≤𝐶 & 𝑑(𝐵𝑖) ≤𝐶. If the tour contains customers from both sets, the linehaul customers must serve before any backhaul customers. Note that tours with backhaul customers only are not allowed in some formulations [1]. In basic formulation 𝐻𝑖 shoud be changed as follows: 𝐻𝑖= {𝑝𝑖: 𝑀𝑜𝑖→𝐼0𝑖\| 𝑝𝑖(0) = 0 & (∀𝑥∈ 𝑀𝑜𝑖)(∀𝑦∈ 𝑀𝑜𝑖) ⟹𝑝𝑖(𝑥) ≠𝑝𝑖(𝑦) & ((𝑥< 𝑦) ⟹(𝑣𝑝𝑖(𝑥) ∉𝐵𝑖) 𝑜𝑟 (𝑣𝑝𝑖(𝑦) ∉ 𝐿𝑖))} The best metaheuristics, according to [20], include the Adaptive LNS (ALNS) of Ropke and Pisinger (2006) [21], the Tabu Search (TS) of Zachariadis and Kiranoudis (2012) [22] which uses long-term memories to direct the search toward inadequately exploited characteristics; and finally multi-ant colony system algorithm by Gajpal and Abad (2009) [23], which suggests two multi-route local search schemes. The best metaheuristics, according to [20], include the Adaptive LNS (ALNS) of Ropke and Pisinger (2006) [21], the Tabu Search (TS) of Zachariadis and Kiranoudis (2012) [22] which uses long-term memories to direct the search toward inadequately exploited characteristics; and finally multi-ant colony system algorithm by Gajpal and Abad (2009) [23], which suggests two multi-route local search schemes. 3.7. VRP with Simultaneous Pickup and Delivery (VRPSPD VRPSPD is a subcase of VRPPD where each customer is a linehaul as well as a backhaul customer. In VRPSPD each customer not only requires a given quantity of products to be delivered but also requires a given quantity of products to be picked up. A complete service (i.e., delivery and pickup) to the customer is provided by a vehicle in a single visit. Thus, there is no need to explicitely indicate both variables 𝑂𝑖 and 𝐷𝑖 as in VRPPD. It is found in the literature that the heuristics of Subramanian et al. (2010) [27], Zachariadis and Kiranoudis (2011) [28] and Souza et al. (2010) [29] together produce the best known results. 3.6. VRP with Pickup and Delivery (VRPPD) In the basic version of VRPPD, each customer 𝑣𝑖 requests either two demands, 𝑑𝑖 to be delivered and 𝑝𝑖 to be picked up, or only 𝑑𝑖= 𝑑𝑖−𝑝𝑖, that represents the difference between two demands. In addition, we need to add for each customer 𝑣𝑖 two new variables: 𝑂𝑖 which denotes the vertex where the source of delivery 238 Береснева Е.Н., Авдошин С.М. Анализ математических постановок задачи маршрутизации с ограничением по грузоподъемности и методов их решения. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 233-250 originates and 𝐷𝑖 which denotes the customer where the destination of the pick up exists. It should be noted that for each customer the delivery must be implemented before the pick up. Let define 𝑑(𝑉𝑝𝑎𝑟𝑡 0𝑖) −𝑝(𝑉𝑝𝑎𝑟𝑡 0𝑖) ≤𝐶, 𝑝𝑎𝑟𝑡= 0. . 𝑁𝑖 ̅̅̅̅̅̅̅, as the weight of the current load of the vehicle after visiting 𝑣𝑝𝑖(𝑝𝑎𝑟𝑡), where 𝑑(𝑉𝑝𝑎𝑟𝑡 0𝑖) = ∑ 𝑑𝑗 𝑗∈\|0..𝑝𝑎𝑟𝑡\| , 𝑝𝑎𝑟𝑡≤𝑁𝑖 and 𝑝(𝑉𝑝𝑎𝑟𝑡 0𝑖) = ∑ 𝑝𝑗 𝑗∈\|0..𝑝𝑎𝑟𝑡\| . In basic formulation 𝐻𝑖 shoud be changed as follows: 𝐻𝑖= {𝑝𝑖: 𝑀𝑜𝑖→𝐼0𝑖\| 𝑝𝑖(0) = 0 & (∀𝑥∈ 𝑀0𝑖)(∀𝑦∈ 𝑀𝑜𝑖) (𝑥≠𝑦 ⟹𝑝𝑖(𝑥) ≠𝑝𝑖(𝑦)) & (∀𝑥 ∈ 𝑀𝑜𝑖)(∀𝑦 ∈𝑀𝑜𝑖) ((𝑣𝑝𝑖(𝑦) = 𝐷𝑝𝑖(𝑥)) ⟹ (𝑥< 𝑦)) & (∀𝑥∈ 𝑀𝑜𝑖)(∀𝑦∈ 𝑀𝑜𝑖) ((𝑣𝑝𝑖(𝑥) = 𝑂𝑝𝑖(𝑦)) ⟹ (𝑥> 𝑦))} A great number of heuristics and metaheuristics are presented in [26]. 3.8. VRP with Mixed Pickup and Delivery (VRPMPD) VRPMPD is also a subcase of VRPPD where each customer has either a delivery demand or pickup. Therefore, there is 𝑑𝑖> 0 and 𝑝𝑖= 0 in the first case and 𝑝𝑖> 0 and 𝑑𝑖= 0. Nevertheless, in basic formulation 𝐻𝑖 shoud be changed the same way as it was shown for VRPPD. The best known heuristics are those of Subramanian (2013) which is based on Iterative Local Search (ILS) idea [30] and hybrid algorithm proposed by Subramanian, Uchoa and Ochi (2013) [31]. 3.9. VRP with Pickup and Delivery and Time Windows (VRPPDTW) The VRPPDTW in this paper contains all constraints in the VRPPD plus added constraints in which both pickup and delivery have given time windows. With every location 𝑣𝑖 there is a service time 𝑠𝑟𝑣𝑖 associated for loading/unloading and a time window [𝑎𝑖;𝑏𝑖], which specifies the time in which this service has to be provided. In 239 neva E., Avdoshin S. Analysis of Mathematical Formulations of Capacitated Vehicle Routing Problem and ods for their Solution. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp, 233-250 Beresneva E., Avdoshin S. Analysis of Mathematical Formulations of Capacitated Vehicle Routing Problem and Methods for their Solution. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp, 233-250 the same way as for VRPTW, when arriving too early at a location 𝑣𝑖, i.e., before 𝑎𝑖, the vehicle is allowed to wait until 𝑎𝑖 to start the service. Also, for each customer the delivery must be implemented before the pick up. Efficient neighborhood-centered metaheuristics have been proposed, including the ALNS of Ropke and Pisinger (2006) [32] and the two-phase method of Bent, and Van Hentenryck (2006) [33], which combines SA to reduce the number of routes with LNS to optimize the distance. However, these methods were recently outperformed by the memetic algorithm of Nagata and Kobayashi (2011) [34], which exploits a well-designed crossover focused on transmitting parent characteristics without introducing too many new arcs in the offspring. 3.10. Multi-depot VRP (MDVRP) The MDVRP is a generalization of the CVRP where more than one depot may be considered. Obviously, the vehicle must start and end at the same depot. So, part of basic formulation should be changed as follows: Let 𝐺 be a number of depots. Let introduce 𝑉0 = {𝑣0 1, 𝑣0 2, … , 𝑣0 𝐺}, where 𝑣0 𝑖∈𝑉. In this case, we should make 𝑉0𝑖= {𝑣0 𝑗∈𝑉0} ∪𝑉𝑖, 𝑖= 1. . 𝐾 ̅̅̅̅̅̅, 𝑗= 1. . 𝐺 ̅̅̅̅̅̅. The best heuristic approaches for the MDVRP are considered to be developed by Pisinger and Ropke (2006) [35] and Vidal et al. (2012) [6]. Overall, the formulation of VRPM is to find: 𝑆0: 𝐹(𝑆0) = min 𝑆∈ 𝑆𝑎𝑙𝑙𝐹(𝑆) 𝑆0: 𝐹(𝑆0) = min 𝑆∈ 𝑆𝑎𝑙𝑙𝐹(𝑆) Metaheuristic inspired by ideas of TS and adaptive memory-based search (AMS) (Taillard (1993) [36]) still shows good results. In addition, another variant of AMS by Olivera and Viera (2007) [37] is considered to be competitive. 3.11. VRP with Multiple Use of Vehicles (VRPM) or Multi-Trip VRP (MTVRP) 3, 2018 г., стр. 233-250 Let 𝐻𝑚𝑡 𝑖 = {𝑝𝑚𝑡 𝑖: 𝑀𝑚𝑡 0𝑖→ 𝐼𝑚𝑡 0𝑖\| 𝑝𝑚𝑡 𝑖(0) = 0 & (∀𝑥∈𝑀𝑚𝑡 0𝑖)(∀𝑦∈𝑀𝑚𝑡 0𝑖) 𝑥≠𝑦 ⟹ 𝑝𝑚𝑡 𝑖(𝑥) ≠𝑝𝑚𝑡 𝑖(𝑦)} be a set of codes of all Hamiltonian cycles ℎ𝑚𝑡 𝑖 = (𝑣𝑝𝑚𝑡 𝑖 (0), 𝑣𝑝𝑚𝑡 𝑖 (1), … , 𝑣𝑝𝑚𝑡 𝑖 (𝑁𝑖)) of 𝑉𝑚𝑡 0𝑖. Let 𝐻𝑚𝑡 𝑖 = {𝑝𝑚𝑡 𝑖: 𝑀𝑚𝑡 0𝑖→ 𝐼𝑚𝑡 0𝑖\| 𝑝𝑚𝑡 𝑖(0) = 0 & (∀𝑥∈𝑀𝑚𝑡 0𝑖)(∀𝑦∈𝑀𝑚𝑡 0𝑖) 𝑥≠𝑦 ⟹ 𝑝𝑚𝑡 𝑖(𝑥) ≠𝑝𝑚𝑡 𝑖(𝑦)} be a set of codes of all Hamiltonian cycles ℎ𝑚𝑡 𝑖 = (𝑣𝑝𝑚𝑡 𝑖 (0), 𝑣𝑝𝑚𝑡 𝑖 (1), … , 𝑣𝑝𝑚𝑡 𝑖 (𝑁𝑖)) of 𝑉𝑚𝑡 0𝑖. Let 𝐻𝑚𝑡 𝑖 = {𝑝𝑚𝑡 𝑖: 𝑀𝑚𝑡 0𝑖→ 𝐼𝑚𝑡 0𝑖\| 𝑝𝑚𝑡 𝑖(0) = 0 & (∀𝑥∈𝑀𝑚𝑡 0𝑖)(∀𝑦∈𝑀𝑚𝑡 0𝑖) 𝑥≠𝑦 ⟹ 𝑝𝑚𝑡 𝑖(𝑥) ≠𝑝𝑚𝑡 𝑖(𝑦)} be a set of codes of all Hamiltonian cycles ℎ𝑚𝑡 𝑖 = (𝑣𝑝𝑚𝑡 𝑖 (0), 𝑣𝑝𝑚𝑡 𝑖 (1), … , 𝑣𝑝𝑚𝑡 𝑖 (𝑁𝑖)) of 𝑉𝑚𝑡 0𝑖. Weight of a Hamiltonian cycle ℎ𝑚𝑡 𝑖 ∈𝐻𝑚𝑡 𝑖 can be found according to the formula: 𝑓(ℎ𝑚𝑡 𝑖) = 𝑐(𝑣𝑝𝑚𝑡 𝑖 (0), 𝑣𝑝𝑚𝑡 𝑖 (𝑁𝑚𝑡 𝑖 )) + ∑𝑐(𝑣𝑝𝑚𝑡 𝑖 (𝑗), 𝑣𝑝𝑚𝑡 𝑖 (𝑗+1)) 𝑁𝑚𝑡 𝑖 −1 𝑗=0 Let 𝑆′ be a set of {𝑉𝑚𝑡 01, 𝑉𝑚𝑡 02, … , 𝑉𝑚𝑡 0𝐾}, 𝑚𝑡= 1. . 𝑀𝑇𝑖 ̅̅̅̅̅̅̅̅̅. In this notation the weight of 𝑆′ is calculated as 𝐹(𝑆′) = ∑ ∑ 𝑓(ℎ𝑚𝑡 𝑖) 𝑚𝑡=1..𝑀𝑇𝑖 ̅̅̅̅̅̅̅̅ 𝑖=1..𝐾 ̅̅̅̅̅ . Function 𝑡(∙,∙) on 𝑉× 𝑉 satisfies the same axioms as 𝑐(∙,∙). And it is defined as 𝑓𝑇(ℎ𝑚𝑡 𝑖) = 𝑡(𝑣𝑝𝑚𝑡 𝑖 (0), 𝑣𝑝𝑚𝑡 𝑖 (𝑁𝑚𝑡 𝑖 )) + ∑ 𝑡(𝑣𝑝𝑚𝑡 𝑖 (𝑗),𝑣𝑝𝑚𝑡 𝑖 (𝑗+1)) 𝑁𝑚𝑡 𝑖 −1 𝑗=0 . Function 𝑡(∙,∙) on 𝑉× 𝑉 satisfies the same axioms as 𝑐(∙,∙). And 𝑓𝑇(ℎ𝑚𝑡 𝑖) = 𝑡(𝑣𝑝𝑚𝑡 𝑖 (0), 𝑣𝑝𝑚𝑡 𝑖 (𝑁𝑚𝑡 𝑖 )) + ∑ 𝑡(𝑣𝑝𝑚𝑡 𝑖 (𝑗),𝑣𝑝𝑚𝑡 𝑖 (𝑗+1)) 𝑁𝑚𝑡 𝑖 −1 𝑗=0 . The most important thing here is the next 𝑀𝑇 The most important thing here is the next constraint: (∀𝑖= 1. . 𝐾 ̅̅̅̅̅̅) ∑𝑓𝑇(ℎ𝑚𝑡 𝑖) 𝑀𝑇𝑖 𝑚𝑡=1 ≤𝑇𝑀𝑎𝑥 (∀𝑖= 1. . 𝐾 ̅̅̅̅̅̅) ∑𝑓𝑇(ℎ𝑚𝑡 𝑖) 𝑚𝑡=1 ≤𝑇𝑀𝑎𝑥 Overall, the formulation of VRPM is to find: 3.11. VRP with Multiple Use of Vehicles (VRPM) or Multi-Trip VRP (MTVRP) VRPM or MTVRP is a variant of standard CVRP in which the same vehicle can be assigned to several routes during a given planning period. Not only this constraint is introduced but also the sum of the durations of the trips assigned to the same vehicle must not exceed 𝑇𝑀𝑎𝑥. 𝑇𝑀𝑎𝑥 is a trip duration being the sum of the travel times on arcs used in the route. Thus, new function 𝑡(𝑣𝑖,𝑣𝑗), returning travel time between 𝑣𝑖 and 𝑣𝑗, appears. In this variant it is possible if 𝑑(𝑉𝑜𝑖) > 𝐶, 𝑖= 1. . 𝐾 ̅̅̅̅̅̅. We additionally divide 𝑉𝑖 in 𝑀𝑇𝑖 sets: 𝑉𝑖= {𝑉1 𝑖, 𝑉2 𝑖, … , 𝑉𝑀𝑇𝑖 𝑖 }, where 𝑀𝑇𝑖∈[1; \|𝑉𝑖\|]. Let 𝐽= {0, 1, … , 𝐾} be a set that keeps indexes. Then (∀𝑖∈ 𝐽) (∀𝑚𝑡∈ 1. . 𝑀𝑇𝑖 ̅̅̅̅̅̅̅̅̅) \|𝑉𝑚𝑡 𝑖\| ≥ 1. There should be no duplicates in any of subsets from 𝑉𝑖: (∀𝑖∈𝐽) (∀𝑚𝑡∈1 𝑀𝑇 ̅̅̅̅̅̅̅̅̅) (∀𝑚𝑡∈1 𝑀𝑇 ̅̅̅̅̅̅̅̅̅) (𝑚𝑡≠𝑚𝑡⇒𝑉𝑖 ∩𝑉𝑗 ∅) In this variant it is possible if 𝑑(𝑉𝑜𝑖) > 𝐶, 𝑖= 1. . 𝐾 ̅̅̅̅̅̅. We additionally divide 𝑉𝑖 in 𝑀𝑇𝑖 sets: 𝑉𝑖= {𝑉1 𝑖, 𝑉2 𝑖, … , 𝑉𝑀𝑇𝑖 𝑖 }, where 𝑀𝑇𝑖∈[1; \|𝑉𝑖\|]. Let 𝐽= {0, 1, … , 𝐾} be a set that keeps indexes. Then (∀𝑖∈ 𝐽) (∀𝑚𝑡∈ 1. . 𝑀𝑇𝑖 ̅̅̅̅̅̅̅̅̅) \|𝑉𝑚𝑡 𝑖\| ≥ 1. There should be no duplicates in any of subsets from 𝑉𝑖: (∀𝑖∈ 𝐽) (∀𝑚𝑡1 ∈ 1. . 𝑀𝑇𝑖 ̅̅̅̅̅̅̅̅̅) (∀𝑚𝑡2 ∈ 1. . 𝑀𝑇𝑖 ̅̅̅̅̅̅̅̅̅) (𝑚𝑡1 ≠𝑚𝑡2 ⇒𝑉𝑚𝑡1 𝑖 ∩𝑉𝑚𝑡2 𝑗 = ∅). Also 𝑉𝑖= ⋃ 𝑉𝑡 𝑖 𝑀𝑇𝑖 In this notation we should make (∀𝑚𝑡∈1 𝑀𝑇 ̅̅̅̅̅̅̅̅̅)𝑉𝑡 0𝑖= Let introduce 𝑀𝑚𝑡 𝑖 = {1, … , 𝑁𝑚𝑡 𝑖},𝑁𝑚𝑡 𝑖 = \|𝑉𝑚𝑡 𝑖\|, ∑ 𝑁𝑚𝑡 𝑖 𝑀𝑇𝑖 𝑚𝑡=1 = 𝑁𝑖. Then 𝑀𝑚𝑡 0𝑖= {0} ∪𝑀𝑚𝑡 𝑖. Let 𝐼𝑚𝑡 𝑖 = {𝑖 \| 𝑖= 𝑖𝑛𝑑(𝑣), 𝑣∈𝑉𝑚𝑡 𝑖} be a set of vertex indices from 𝑉𝑚𝑡 𝑖. Then 𝐼𝑚𝑡 0𝑖= {0} ∪𝐼𝑚𝑡 𝑖. Let introduce 𝑀𝑚𝑡 𝑖 = {1, … , 𝑁𝑚𝑡 𝑖},𝑁𝑚𝑡 𝑖 = \|𝑉𝑚𝑡 𝑖\|, ∑ 𝑁𝑚𝑡 𝑖 𝑀𝑇𝑖 𝑚𝑡=1 = 𝑁𝑖. Then 𝑀𝑚𝑡 0𝑖= {0} ∪𝑀𝑚𝑡 𝑖. Let 𝐼𝑚𝑡 𝑖 = {𝑖 \| 𝑖= 𝑖𝑛𝑑(𝑣), 𝑣∈𝑉𝑚𝑡 𝑖} be a set of vertex indices from 𝑉𝑚𝑡 𝑖. Then 𝐼𝑚𝑡 0𝑖= {0} ∪𝐼𝑚𝑡 𝑖. 240 Береснева Е.Н., Авдошин С.М. Анализ математических постановок задачи маршрутизации с ограничением по грузоподъемности и методов их решения. Труды ИСП РАН, том 30, вып. 3.13. Split Delivery VRP (SDVRP) In the SDVRP-MDA, more than one vehicle can service a customer, so that a customer’s demand can be split among several vehicles on different routes. The most important here is that split deliveries are allowed only if at least a minimum fraction of a customer’s demand is delivered by each vehicle visiting the customer. The first metaheuristic for a SDVRP is proposed in Chen et al. (2007) [41]. The idea of the approach is based on combination of the classical Clarke and Wright algorithm, the Mixed-Integer Linear Programming (MILP) model and variable length record-to- record travel methods. A similar procedure is applied in Gulczynski et al. (2010) [42] to the SDVRP with minimum delivery amounts, that is a SDVRP where each delivery to a customer should consist of at least a minimum amount of goods. Another metaheuristic which contains TS approach is proposed in 2008 by Archetti et al. [43] The main thing here is to obtain a reduced graph by removing some arcs and to apply a set covering MILP formulation for the best routes. And in Jin et al. (2008) [44] a set covering formulation is proposed and the problem is solved through column generation. 3.12. Periodic VRP (PVRP) The Periodic VRP (PVRP) is used when planning is done over a certain period and deliveries to the customer can be done in different days. For the PVRP, customers can be visited more than once, though often with limited frequency. Efficient algorithm for solving PVRP is parallel extension of UTS with neighborhood-centered search (Cordeau and Maischberger, 2012 [38]). Also, the VNS of Hemmelmayr et al. (2009) [39], and the hybrid record-to-record and integer programming metaheuristic of Gulczynski et al. (2011) [40] can be successfully applied. In addition, one more metaheuristic is one that proposed by Vidal et al. (2012) [6]. It produces the current best solutions by combining the GA search breadth with efficient LS, relaxations schemes, and diversity management procedures. 241 Beresneva E., Avdoshin S. Analysis of Mathematical Formulations of Capacitated Vehicle Routing Problem and Methods for their Solution. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp, 233-250 Beresneva E., Avdoshin S. Analysis of Mathematical Formulations of Capacitated Vehicle Routing Problem and Methods for their Solution. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp, 233-250 4. An important note There is an important note about recent state-of-the-art algorithm proposed by K. Helsgaun [48]. In the end of 2017 this author released an extension of the Lin- Kernighan-Helsgaun TSP Solver for Vehicle Routing Problems, called LKH-3. In his technical report it is said that his algorithm can often obtain best known solutions for benchmark instances, and even new best solutions were found. Unfortunately, his algorithm cannot solve all subcases of CVRP. MDVRP, VRPM, PVRP, SDVRP and CCVRP can be solved using other metaheuristics but not using LKH-3. 3.14. Cumulative CVRP (CCVRP) CCVRP minimizes the sum of the arrival times at the customers instead of minimizing the total distance (or travel time) as an objective. For the CCVRP, Ngueveu et al. (2010) [45] and Ribeiro and Laporte (2012) [46] modified the hybrid GA. Also, two-phase metaheuristic proposed by Ke and Feng in 2013 [47] is considered to be successful enough. Fig. 1 sums up relations between classes of the CVRP and forms the classification of its subtypes. In our future work, we are going to extend current survey adding dynamic and stochastic subcases of CVRP. 5. Conclusions Table 1 sums up abovementioned and shows a list of best metaheuristics for each defined subcase of CVRP. The presented study is undertaken in order to make a survey on CVRP subcases and on heuristic methods for solving each extension of this problem. In addition, author variants of mathematical formulations are given. 242 Береснева Е.Н., Авдошин С.М. Анализ математических постановок задачи маршрутизации с ограничением по грузоподъемности и методов их решения. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 233-250 Береснева Е.Н., Авдошин С.М. Анализ математических постановок задачи маршрутизации с ограничением по грузоподъемности и методов их решения. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 233-250 Table 1. Best metaheuristics for CVRP subcases # Problem Best metaheuristics 1 CVRP LKH-3, Tabu Search, Simulated Annealing, Ant Colony Optimization algorithm, Genetic algorithm, Variable Neighborhood Search 2 OVRP LKH-3, Evolution algorithm, Variable Neighborhood Search, Tabu Search, Genetic algorithm, Particle swarm optimization metaheuristic 3 DCVRP LKH-3, CVRP + transformation to mTSP with Time Windows 4 VRPTW LKH-3, Guided Evolutionary algorithm, Large Neighborhood search (LNS) 5 VRPB LKH-3, Adaptive LNS, Tabu Search, Multi-Ant Colony System algorithm 6 VRPBTW LKH-3, Tabu Search + Simulated Annealing, λ-interchange and 2-opt exchange procedures 7 VRPSPD LKH-3, Genetic algorithm, Guided Evolutionary algorithm, Iterated Local Search algorithm 8 VRPMPD LKH-3, Genetic algorithm, Guided Evolutionary algorithm, Iterated Local Search algorithm 9 VRPPDTW LKH-3, Adaptive LNS, Simulated Annealing + LNS 10 MDVRP Hybrid Genetic algorithm 11 VRPM or MTVRP Adaptive Memory-Based Search variants 12 PVRP Tabu Search, Variable Neighborhood Search, Genetic algorithm + LNS 13 SDVRP Clarke and Wright Savings + Mixed-Integer Linear Programming, Tabu Search 14 CCVRP Adaptive LNS, Variable Neighborhood Search Fig. 1 sums up relations between classes of the CVRP and forms the classification of its subtypes. In our future work, we are going to extend current survey adding dynamic and stochastic subcases of CVRP. 243 Beresneva E., Avdoshin S. Analysis of Mathematical Formulations of Capacitated Vehicle Routing Problem and Methods for their Solution. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp, 233-250 Fig. 1. The basic problems of the CVRP class and their interconnections neva E., Avdoshin S. Analysis of Mathematical Formulations of Capacitated Vehicle Routing Problem and ods for their Solution. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp, 233-250 Fig. 1. The basic problems of the CVRP class and their interconnections References [1] P. Toth and D. Vigo. An overview of vehicle routing problems, In The Vehicle Routing Problem, SIAM, 2002. [2] G. B. Dantzig and J. H. Ramser. The Truck Dispatching Problem. Management Science, vol. 6, no. 1, 1959, pp. 80-91. [3] M. Reed and B. Simon. Methods of Modern Mathematical Physics, London: Academic Press, 1972. [4] B. Golden, S. Raghavan and E. Wasil. The vehicle routing problem: Latest advances and new challenges, New York: Springer, 2008. [5] P. Pisinger and S. Ropke. A general heuristic for vehicle routing problems. Computers & Operations Research, vol. 34, no. 8, 2007, pp. 2403-2435. [6] Y. Nagata and O. Braysy. Edge assembly-based memetic algorithm for the capacitated vehicle routing problem. Networks, vol. 54, no. 4, 2009, pp. 205-215. [7] T. Vidal, T. Crainic, M. Gendreau, N. Lahrichi and W. Rei. A hybrid genetic algorithm for multi-depot and periodic vehicle routing problems. Operations Research, vol. 60, no. 3, 2012, pp. 611-624. [8] M. Salari, P. Toth and A. Tramontani. An ILP improvement procedure for the Open Vehicle Routing Problem. Computers & Operations Research, vol. 37, no. 12, 2010, pp. 2106-2120. [9] P. Repoussis, C. Tarantilis, O. Braysy and G. Ioannou. A hybrid evolution strategy for the open vehicle routing problem. Computers & Operations Research, vol. 37, no. 3, 2010, pp. 443-455. [10] K. Fleszar, I. Osman and K. Hindi. A variable neighbourhood search algorithm for the open vehicle routing problem. European Journal of Operational Research, vol. 195, no. 3, 2009, pp. 803-809. pp [11] E. Zachariadis and C. Kiranoudis. An open vehicle routing problem metaheuristic for examining wide solution neighborhoods. Computers & Operations Research, vol. 37, no. 4, 2010, pp. 712-723. 244 Береснева Е.Н., Авдошин С.М. Анализ математических постановок задачи маршрутизации с ограничением по грузоподъемности и методов их решения. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 233-250 [12] S. MirHassani and N. Abolghasem. A particle swarm optimization algorithm for open vehicle routing problem. Expert Systems with Applications, vol. 38, no. 9, 2011, pp. 11547-11551. [13] G. Laporte, Y. Nobert and M. Desrochers. Optimal routing under capacity and distance restrictions. Operations Research, vol. 33, no. 5, 1985, p. 1050–1073. [14] C. Li, D. Simchi-Levi and M. Desrochers. On the distance constrained vehicle routing problem. Operational Research, vol. 40, 1992, pp. 790-799. [15] P. Repoussis, C. Tarantilis and G. Ioannou. Arc-guided evolutionary algorithm for the vehicle routing problem with time windows. References IEEE Transactions on Evolutionary Computation, vol. 13, no. 3, 2009, pp. 624–647. [16] E. Prescott-Gagnon, G. Desaulniers and L. Rousseau. A branch-and-price-based large neighborhood search algorithm for the vehicle routing problem with time windows. Networks, vol. 54, no. 4, 2009, pp. 190–204. [17] Y. Nagata, O. Brä ysy and W. Dullaert. A penalty-based edge assembly memetic algorithm for the vehicle routing problem with time windows. Computers & Operations Research, vol. 37, no. 4, 2010, pp. 724–737. [18] T. Vidal, T. Crainic, M. Gendreau and C. Prins. A hybrid genetic algorithm with adaptive diversity management for a large class of vehicle routing problems with time-windows. Computers & Operations Research, vol. 40, no. 1, 2013, pp. 475–489. [19] K. Braekers, K. Ramaekers and I. Nieuwenhuyse. The vehicle routing problem: State of the art classification and review. Computers & Industrial Engineering, vol. 99, 2016, pp. 300-313. [20] S. Ropke and D. Pisinger. A unified heuristic for a large class of vehicle routing problems with backhauls. European Journal of Operational Research, vol. 171, no. 3, 2006, pp. 750– 775. [21] E. Zachariadis and C. Kiranoudis. An effective local search approach for the vehicle routing problem with backhauls. Expert Systems with Applications, vol. 39, no. 3, 2012, pp. 3174–3184. [22] Y. Gajpal and P. Abad. Multi-ant colony system (MACS) for a vehicle routing problem with backhauls. European Journal of Operational Research, vol. 196, no. 1, 2009, pp. 102– 117. [23] S. Thangiah, J.-Y. Potvin and T. Sun. Approaches to Vehicle Routing with Backhauls and Time. Windows International Journal of Computers and Operations Research, vol. 23, no. 11, 1996, pp. 1043-1057. [24] I. Küçükoğlu and N. Öztürk. An advanced hybrid meta-heuristic algorithm for the vehicle routing problem with backhauls and time windows. Computers and Industrial Engineering, vol. 86, no. 3, 2015, pp. 60-68. [25] S. Parragh, K. Doerner and R. Hartl. A survey on pickup and delivery problems. Journal fü r Betriebswirtschaft, vol. 58, 2008, pp. 81-117. [26] A. Subramanian, Drummond, C. Bentes, L. Ochi and R. Farias. A parallel heuristic for the vehicle routing problem with simultaneous pickup and delivery. Computers & Operations Research, vol. 37, no. 11, 2010, pp. 1899-1911. [27] E. Zachariadis and C. Kiranoudis. A local search metaheuristic algorithm for the vehicle routing problem with simultaneous pick-ups and deliveries. Expert Systems with Applications, vol. 38, no. 3, 2011, pp. 2717-2726. [28] M. Souza, M. Silva, M. Mine, L. Ochi and A. Subramanian. References A hybrid heuristic, based on iterated local search and GENIUS, for the vehicle routing problem with simultaneous 245 Beresneva E., Avdoshin S. Analysis of Mathematical Formulations of Capacitated Vehicle Routing Problem and Methods for their Solution. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp, 233-250 pickup and delivery. International Journal of Logistics Systems Management, vol. 10, no 2, 2010, pp. 142-156. pickup and delivery. International Journal of Logistics Systems Management, vol. 10, no. 2, 2010, pp. 142-156. [29] A. Subramanian and M. Battarra. An iterated local search algorithm for the travelling salesman problem with pickups and deliveries. Journal of the Operational Research Society, vol. 64, 2013, pp. 402-409. y pp [30] A. Subramanian, E. Uchoa and L. Ochi. A hybrid algorithm for a class of vehicle routing problems. Computers & Operations Research, vol. 40, no. 10, 2013, pp. 2519-2531. [31] S. Ropke and D. Pisinger. An adaptive large neighborhood search heuristic for the pickup and delivery problem with time windows. Transportation Science, vol. 40, no. 4, 2006, pp. 455–472. [32] R. Bent and P. Van Hentenryck. A two-stage hybrid algorithm for pickup and delivery vehicle routing problems with time windows. Computers & Operations Research, vol. 33, no. 4, 2006, pp. 875–893. [33] Y. Nagata and S. Kobayashi. A memetic algorithm for the pickup and delivery problem with time windows using selective route exchange crossover. In Proceedings of PPSN’11, vol. 6238, 2011, pp. 536– 545. [34] D. Pisinger and S. Ropke. A general heuristic for vehicle routing problems. Computers & Operations Research, vol. 34, no. 8, 2007, pp. 2403–2435. [35] E. Taillard, G. Laporte and M. Gendreau. Vehicle routeing with multiple use of vehicles. Journal of the Operational Research Society, vol. 47, no. 8, 1996, pp. 1065. [36] A. Olivera and O. Viera. Adaptive memory programming for the vehicle routing problem with multiple trips. Computers & Operations Research, vol. 34, no. 1, 2007, pp. 28–47. [37] J.-F. Cordeau and M. Maischberger. A Parallel Iterated Tabu Search Heuristic for Vehicle Routing Problems. Computers & Operations Research, vol. 39, no. 9, 2012, pp. 2033– 2050. [38] V. Hemmelmayr, K. Doerner and R. Hartl. A variable neighborhood search heuristic for periodic routing problems. European Journal of Operational Research, vol. 195, no. 3, 2009, pp. 791–802. [39] D. Gulczynski, B. Golden and E. Wasil. The period vehicle routing problem : New heuristics and real-world variants. neva E., Avdoshin S. Analysis of Mathematical Formulations of Capacitated Vehicle Routing Problem and ods for their Solution. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp, 233-250 References Transportation Research Part E: Logistics and Transportation Review, vol. 47, no. 5, 2011, pp. 648-668. [40] S. Chen, B. Golden and E. Wasil. The split delivery vehicle routing problem: Applications, algorithms, test problems, and computational results. Networks, vol. 49, 2007, pp. 318– 329. [41] D. Gulczynski, B. Golden and E. Wasil. The split delivery vehicle routing problem with minimum delivery amounts. Transportation Research Part E, vol. 46, 2010, pp. 612–626. [42] C. Archetti and M. Speranza. The split delivery vehicle routing problem: a survey. In The Vehicle Routing Problem Latest Advances and New Challenges, Operations Research, Computer Science Interfaces Series, 2008, pp. 103-122. [43] M. Jin, K. Liu and B. Eksioglu. A column generation approach for the split delivery vehicle routing problem. Operations Research Letters, vol. 36, 2008, pp. 265-270. [44] S. Ngueveu, C. Prins and C. Wolfler. An effective memetic algorithm for the cumulative capacitated vehicle routing problem. Computers & Operations Research , vol. 37, no. 11, 2010, pp. 1877–1885. [45] G. Ribeiro and G. Laporte. An adaptive large neighborhood search heuristic for the cumulative capacitated vehicle routing problem. Computers & Operations Research, vol. 39, no. 3, 2012, pp. 728–735. 246 Береснева Е.Н., Авдошин С.М. Анализ математических постановок задачи маршрутизации с ограничением по грузоподъемности и методов их решения. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 233-250 [46] L. Ke and Z. Feng. A two-phase metaheuristic for the cumulative capacitated vehicle routing problem. Computers & Operations Research, vol. 40, no. 2, 2013, pp. 633-638. [47] J. Sze, S. Salhi and N. Wassan. The cumulative capacitated vehicle routing problem with min-sum and min-max objectives: An effective hybridisation of adaptive variable neighbourhood search and large neighbourhood search. Transportation Research Part B: Methodological, vol. 101, 2017, pp. 162-184. [48] K. Helsgaun. An Extension of the Lin-Kernighan-Helsgaun TSP Solver for Constrained Traveling Salesman and Vehicle Routing Problems. Technical Report, Roskilde University, 2017. DOI: 10.15514/ISPRAS-2018-30(3)-17 DOI: 10.15514/ISPRAS-2018-30(3)-17 Для цитирования: Береснева Е.Н., Авдошин С.М. Анализ математических постановок задачи маршрутизации с ограничением по грузоподъемности и методов их решения. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 233-250 (на английском языке). DOI: 10.15514/ISPRAS-2018-30(3)-17 10.15514/ISPRAS-2018-30(3)-17 Список литературы [1] P. Toth and D. Vigo. An overview of vehicle routing problems, In The Vehicle Routing Problem, SIAM, 2002. [2] G. B. Dantzig and J. H. Ramser. The Truck Dispatching Problem. Management Science, vol. 6, no. 1, 1959, pp. 80-91. [3] M. Reed and B. Simon. Methods of Modern Mathematical Physics, London: Academic Press, 1972. [4] B. Golden, S. Raghavan and E. Wasil. The vehicle routing problem: Latest advances and new challenges, New York: Springer, 2008. [5] P. Pisinger and S. Ropke. A general heuristic for vehicle routing problems. Computers & Operations Research, vol. 34, no. 8, 2007, pp. 2403-2435. [6] Y. Nagata and O. Braysy. Edge assembly-based memetic algorithm for the capacitated vehicle routing problem. Networks, vol. 54, no. 4, 2009, pp. 205-215. [7] T. Vidal, T. Crainic, M. Gendreau, N. Lahrichi and W. Rei. A hybrid genetic algorithm for multi-depot and periodic vehicle routing problems. Operations Research, vol. 60, no. 3, 2012, pp. 611-624. [8] M. Salari, P. Toth and A. Tramontani. An ILP improvement procedure for the Open Vehicle Routing Problem. Computers & Operations Research, vol. 37, no. 12, 2010, pp. 2106-2120. [9] P. Repoussis, C. Tarantilis, O. Braysy and G. Ioannou. A hybrid evolution strategy for the open vehicle routing problem. Computers & Operations Research, vol. 37, no. 3, 2010, pp. 443-455. [10] K. Fleszar, I. Osman and K. Hindi. A variable neighbourhood search algorithm for the open vehicle routing problem. European Journal of Operational Research, vol. 195, no. 3, 2009, pp. 803-809. [11] E. Zachariadis and C. Kiranoudis. An open vehicle routing problem metaheuristic for examining wide solution neighborhoods. Computers & Operations Research, vol. 37, no. 4, 2010, pp. 712-723. [12] S. MirHassani and N. Abolghasem. A particle swarm optimization algorithm for open vehicle routing problem. Expert Systems with Applications, vol. 38, no. 9, 2011, pp. 11547-11551. [13] G. Laporte, Y. Nobert and M. Desrochers. Optimal routing under capacity and distance restrictions. Operations Research, vol. 33, no. 5, 1985, p. 1050–1073. [14] C. Li, D. Simchi-Levi and M. Desrochers. On the distance constrained vehicle routing problem. Operational Research, vol. 40, 1992, pp. 790-799. [15] P. Repoussis, C. Tarantilis and G. Ioannou. Arc-guided evolutionary algorithm for the vehicle routing problem with time windows. IEEE Transactions on Evolutionary Computation, vol. 13, no. 3, 2009, pp. 624–647. [16] E. Prescott-Gagnon, G. Desaulniers and L. Rousseau. Beresneva E., Avdoshin S. Analysis of Mathematical Formulations of Capacitated Vehicle Routing Problem and Methods for their Solution. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp, 233-250 Beresneva E., Avdoshin S. Analysis of Mathematical Formulations of Capacitated Vehicle Routing Problem and Methods for their Solution. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp, 233-250 Анализ математических постановок задачи маршрутизации с ограничением по грузоподъемности и методов их решения Е.Н. Береснева <eberesneva@hse.ru> С.М. Авдошин <savdoshin@hse.ru> Департамент программной инженерии, Национальный исследовательский университет “Высшая школа экономики”, 101000, Россия, г. Москва, ул. Мясницкая, д. 20 Е.Н. Береснева <eberesneva@hse.ru> С.М. Авдошин <savdoshin@hse.ru> Департамент программной инженерии, Национальный исследовательский университет “Высшая школа экономики”, 101000, Россия, г. Москва, ул. Мясницкая, д. 20 Аннотация. Задача маршрутизации является одной из важнейших NP-трудных задач комбинаторной оптимизации. Она заключается в нахождении множества оптимальных замкнутых маршрутов с целью развозки товаров клиентам, используя ограниченное количество транспортных средств. В данной работе анализируется особый вид задачи маршрутизации – задача маршрутизации с ограничением по грузоподъемности, в которой у каждого транспортного средства есть лимит на максимальный вес (объем) груза. Целью данной работы является составление классификации различных типов задачи маршрутизации с ограничением по грузоподъемности. В первой части работы дана общая информация о проблеме, указаны рамки, в которых проводилось исследование – не рассматривались динамические и стохастические подвиды задачи маршрутизации. Во второй части представлена впервые предложенная авторами работы математическая постановка трех классических вариантов задачи маршрутизации с ограничением по грузоподъемности. В третьей части работы представлен список подклассов рассматриваемой задачи, включающий описание, математические модели для некоторых задач, а также наиболее перспективные метаэвристики, с помощью которых можно решать поставленную задачу. В четвертой части приведено упоминание об алгоритме LKH-3, способном решать некоторые подклассы задач с меньшим отклонением от оптимального значения по сравнению с другими алгоритмами. В заключении, приведена таблица, объединяющая все методы, описанные ранее, и схема с взаимосвязями задачи маршрутизации с ограничением по грузоподъемности и её подклассами. В будущем авторы работы планируют расширить классификацию, включив в неё подклассы стохастических и динамических вариантов данной проблемы. Ключевые слова: задача маршрутизации с ограничением по грузоподъемности; математическая постановка; метаэвристики; классификация задач маршрутизации. 247 247 Beresneva E., Avdoshin S. Analysis of Mathematical Formulations of Capacitated Vehicle Routing Problem and Methods for their Solution. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp, 233-250 Список литературы A branch-and-price-based large neighborhood search algorithm for the vehicle routing problem with time windows. Networks, vol. 54, no. 4, 2009, pp. 190–204. 248 Береснева Е.Н., Авдошин С.М. Анализ математических постановок задачи маршрутизации с ограничением по грузоподъемности и методов их решения. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 233-250 [17] Y. Nagata, O. Brä ysy and W. Dullaert. A penalty-based edge assembly memetic algorithm for the vehicle routing problem with time windows. Computers & Operations Research, vol. 37, no. 4, 2010, pp. 724–737. [18] T. Vidal, T. Crainic, M. Gendreau and C. Prins. A hybrid genetic algorithm with adaptive diversity management for a large class of vehicle routing problems with time-windows. Computers & Operations Research, vol. 40, no. 1, 2013, pp. 475–489. p p pp [19] K. Braekers, K. Ramaekers and I. Nieuwenhuyse. The vehicle routing problem: State of the art classification and review. Computers & Industrial Engineering, vol. 99, 2016, pp. 300-313. [20] S. Ropke and D. Pisinger. A unified heuristic for a large class of vehicle routing problems with backhauls. European Journal of Operational Research, vol. 171, no. 3, 2006, pp. 750– 775. [21] E. Zachariadis and C. Kiranoudis. An effective local search approach for the vehicle routing problem with backhauls. Expert Systems with Applications, vol. 39, no. 3, 2012, pp. 3174–3184. [22] Y. Gajpal and P. Abad. Multi-ant colony system (MACS) for a vehicle routing problem with backhauls. European Journal of Operational Research, vol. 196, no. 1, 2009, pp. 102– 117. [23] S. Thangiah, J.-Y. Potvin and T. Sun. Approaches to Vehicle Routing with Backhauls and Time. Windows International Journal of Computers and Operations Research, vol. 23, no. 11, 1996, pp. 1043-1057. [24] I. Küçükoğlu and N. Öztürk. An advanced hybrid meta-heuristic algorithm for the vehicle routing problem with backhauls and time windows. Computers and Industrial Engineering, vol. 86, no. 3, 2015, pp. 60-68. g g pp [25] S. Parragh, K. Doerner and R. Hartl. A survey on pickup and delivery problems. Journal fü r Betriebswirtschaft, vol. 58, 2008, pp. 81-117. [26] A. Subramanian, Drummond, C. Bentes, L. Ochi and R. Farias. A parallel heuristic for the vehicle routing problem with simultaneous pickup and delivery. Computers & Operations Research, vol. 37, no. 11, 2010, pp. 1899-1911. [27] E. Zachariadis and C. Kiranoudis. A local search metaheuristic algorithm for the vehicle routing problem with simultaneous pick-ups and deliveries. Expert Systems with Applications, vol. 38, no. Список литературы 3, 2011, pp. 2717-2726. [28] M. Souza, M. Silva, M. Mine, L. Ochi and A. Subramanian. A hybrid heuristic, based on iterated local search and GENIUS, for the vehicle routing problem with simultaneous pickup and delivery. International Journal of Logistics Systems Management, vol. 10, no. 2, 2010, pp. 142-156. [29] A. Subramanian and M. Battarra. An iterated local search algorithm for the travelling salesman problem with pickups and deliveries. Journal of the Operational Research Society, vol. 64, 2013, pp. 402-409. [30] A. Subramanian, E. Uchoa and L. Ochi. A hybrid algorithm for a class of vehicle routing problems. Computers & Operations Research, vol. 40, no. 10, 2013, pp. 2519-2531. [31] S. Ropke and D. Pisinger. An adaptive large neighborhood search heuristic for the pickup and delivery problem with time windows. Transportation Science, vol. 40, no. 4, 2006, pp. 455–472. [32] R. Bent and P. Van Hentenryck. A two-stage hybrid algorithm for pickup and delivery vehicle routing problems with time windows. Computers & Operations Research, vol. 33, no. 4, 2006, pp. 875–893. 249 Beresneva E., Avdoshin S. Analysis of Mathematical Formulations of Capacitated Vehicle Routing Problem and Methods for their Solution. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp, 233-250 [33] Y. Nagata and S. Kobayashi. A memetic algorithm for the pickup and delivery problem with time windows using selective route exchange crossover. In Proceedings of PPSN’11, vol. 6238, 2011, pp. 536– 545. pp [34] D. Pisinger and S. Ropke. A general heuristic for vehicle routing problems. Computers & Operations Research, vol. 34, no. 8, 2007, pp. 2403–2435. [35] E. Taillard, G. Laporte and M. Gendreau. Vehicle routeing with multiple use of vehicles. Journal of the Operational Research Society, vol. 47, no. 8, 1996, pp. 1065. [36] A. Olivera and O. Viera. Adaptive memory programming for the vehicle routing problem with multiple trips. Computers & Operations Research, vol. 34, no. 1, 2007, pp. 28–47. [37] J.-F. Cordeau and M. Maischberger. A Parallel Iterated Tabu Search Heuristic for Vehicle Routing Problems. Computers & Operations Research, vol. 39, no. 9, 2012, pp. 2033– 2050. [38] V. Hemmelmayr, K. Doerner and R. Hartl. A variable neighborhood search heuristic for periodic routing problems. European Journal of Operational Research, vol. 195, no. 3, 2009, pp. 791–802. [39] D. Gulczynski, B. Golden and E. Wasil. The period vehicle routing problem : New heuristics and real-world variants. DOI: 10.15514/ISPRAS-2018-30(3)-18 For citation: Beresneva E., Gordenko M. Applying the methods of system analysis to teaching assistants’ evaluation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 251-270. DOI: 10.15514/ISPRAS-2018-30(3)-18 For citation: Beresneva E., Gordenko M. Applying the methods of system analysis to teaching assistants’ evaluation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 251-270. DOI: 10.15514/ISPRAS-2018-30(3)-18 Список литературы Transportation Research Part E: Logistics and Transportation Review, vol. 47, no. 5, 2011, pp. 648-668. [40] S. Chen, B. Golden and E. Wasil. The split delivery vehicle routing problem: Applications, algorithms, test problems, and computational results. Networks, vol. 49, 2007, pp. 318– 329. [41] D. Gulczynski, B. Golden and E. Wasil. The split delivery vehicle routing problem with minimum delivery amounts. Transportation Research Part E, vol. 46, 2010, pp. 612–626. [42] C. Archetti and M. Speranza. The split delivery vehicle routing problem: a survey. In The Vehicle Routing Problem Latest Advances and New Challenges, Operations Research, Computer Science Interfaces Series, 2008, pp. 103-122. p pp [43] M. Jin, K. Liu and B. Eksioglu. A column generation approach for the split delivery vehicle routing problem. Operations Research Letters, vol. 36, 2008, pp. 265-270. [44] S. Ngueveu, C. Prins and C. Wolfler. An effective memetic algorithm for the cumulative capacitated vehicle routing problem. Computers & Operations Research , vol. 37, no. 11, 2010, pp. 1877–1885. pp [45] G. Ribeiro and G. Laporte. An adaptive large neighborhood search heuristic for the cumulative capacitated vehicle routing problem. Computers & Operations Research, vol. 39, no. 3, 2012, pp. 728–735. [46] L. Ke and Z. Feng. A two-phase metaheuristic for the cumulative capacitated vehicle routing problem. Computers & Operations Research, vol. 40, no. 2, 2013, pp. 633-638. [47] J. Sze, S. Salhi and N. Wassan. The cumulative capacitated vehicle routing problem with min-sum and min-max objectives: An effective hybridisation of adaptive variable neighbourhood search and large neighbourhood search. Transportation Research Part B: Methodological, vol. 101, 2017, pp. 162-184. [48] K. Helsgaun. An Extension of the Lin-Kernighan-Helsgaun TSP Solver for Constrained Traveling Salesman and Vehicle Routing Problems. Technical Report, Roskilde University, 2017. 250 Applying the methods of system analysis to teaching assistants’ evaluation E. Beresneva <eberesneva@hse.ru> M. Gordenko <mgordenko@hse.ru> Department of Software Engineering, National Research University Higher School of Economics, 20, Myasnitskaya st., Moscow, 101000 Russia Abstract. This article presents the results of applying various methods of system analysis (CATWOE, Rich Picture, AHP, Fuzzy AHP) to evaluation of teaching assistants. The soft and hard methods were applied. Methods of system analysis are considered as an example at the Higher School of Economics (HSE) in program “Teaching assistant”. The article shows the process of interaction of teaching assistants with students and faculty in the form of Rich Picture. Selection and analysis of criteria for the evaluation of training assistants are carried out. Three groups of criteria were defined: professional skills, communicating skills, personal qualities. Each group has some subcriteria, which were defined in brainstorm. Its own method was determined, which immediately allow drop some assistants. In addition, the application of the methods AHP and Fuzzy AHP type-2 to evaluate teaching assistants is considered. The strengths and weaknesses of each method are revealed. It is also shown that, despite the power of the methods of system analysis, it is necessary to use common sense and logic. Do not rely only on the numbers obtained by the methods of system analysis. In the process of work, the best teaching assistant is selected, and the group of the best teaching assistants is defined. Keywords: system analysis; combination of soft and hard methods; multicriteria decision making (MCDM); AHP; type-2 fuzzy sets; Fuzzy AHP 1. Introduction At the Higher School of Economics (HSE) there is a program “Teaching assistant” which has been effective for sever-al years. Each teacher can invite an education assistant, who will take some of the routine tasks related to teaching the course (checking homework, developing test materials, etc.). 251 Beresneva E., Gordenko M. Applying the methods of system analysis to teaching assistants’ evaluation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 251-270 Beresneva E., Gordenko M. Applying the methods of system analysis to teaching assistants’ evaluation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 251-270 Beresneva E., Gordenko M. Applying the methods of system analysis to teaching assistants’ evaluation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 251-270 Beresneva E., Gordenko M. Applying the methods of system analysis to teaching assistants’ evaluation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 251-270 Every student or a graduate student of the HSE, who meets the criteria established by the faculty, can be a teaching assistant. The teacher (or group of teachers) formulates tasks for the teaching assistants and monitors the quality of their performance. The teacher is responsible for the results of the students' knowledge, the quality of materials prepared by the education assistant, methodical support of the teaching assistant’ work. At the moment, all faculties establish their own criteria for selecting teaching assistants independently. Now there is only one criterion for all disciplines: “A student must have a mark at least 8 on the course in which he/she is involved, or he/she must have a recommendation from the department, to which teaching of this discipline is fixed.” However, the practice shows that it is not enough to have only this criterion. There were no special studies about it before, but annual evidence showed that an excellent mark does not fully correlate with being a good teaching assistant. Recent year revealed that 60% of assistants were not able to cope with their work according to teachers. Most problems were connected with personal qualities, professional and communicative skills. For example, somebody did all the tasks slowly and did not do everything in time, or just did not have enough knowledge in the subject area. There were even some facts of disclosure of confidential information: one teaching assistant shared answers to the tests with students. 2. The Difference between Previous Works and Our Approach The literature review shows that there are a lot of researches that reveal a high success of applying the teaching assistant program in general. The most recent one is [3]. However, no one article is aimed neither at selection criteria for teaching assistants nor at searching methodology. The closest study to our problem is devoted to a proposed framework for evaluating student’s performance [4]. This work is based on the hard approach only. It uses the The closest study to our problem is devoted to a proposed framework for evaluating student’s performance [4]. This work is based on the hard approach only. It uses the variation of the most widely used approach for multi-criteria decision-making – Analytic Hierarchy Process that combines mathematics and expert judgment. Since Analytic Hierarchy Process suffers from the problem of imprecision and subjectivity, their paper proposes to use Fuzzy AHP instead of traditional method. student’s performance [4]. This work is based on the hard approach only. It uses the variation of the most widely used approach for multi-criteria decision-making – Analytic Hierarchy Process that combines mathematics and expert judgment. Since Analytic Hierarchy Process suffers from the problem of imprecision and subjectivity, their paper proposes to use Fuzzy AHP instead of traditional method. However, there is an opinion about useless of applying Fuzzy AHP method. In [3] it is said that “the numerical representation of judgments in the AHP is already fuzzy” and “making fuzzy judgments more fuzzy does not lead to a better more valid variation of the most widely used approach for multi criteria decision making Analytic Hierarchy Process that combines mathematics and expert judgment. Since Analytic Hierarchy Process suffers from the problem of imprecision and subjectivity, their paper proposes to use Fuzzy AHP instead of traditional method. y y p j g Analytic Hierarchy Process suffers from the problem of imprecision and subjectivity, their paper proposes to use Fuzzy AHP instead of traditional method. However, there is an opinion about useless of applying Fuzzy AHP method. In [3] it is said that “the numerical representation of judgments in the AHP is already fuzzy” and “making fuzzy judgments more fuzzy does not lead to a better more valid outcome and it often leads to a worse one.” Our article proves that Fuzzy AHP with type-2 modification can still be used in a decision making process. 2. The Difference between Previous Works and Our Approach Moreover, our study combines both hard and soft approaches be-cause this problem consists of not only main criteria but also it has a lot of additional ones. And these auxiliary factors can not be described using only formal algorithms. 1. Introduction Thus, there is a strong necessity to define a group of selective factors in a clever manner. Recently, the head of Computer Science faculty has ordered each teacher (or group of teachers) on all disciplines to choose the best teaching assistant to give him/her an incentive award. In addition, next year the number of students is reduced, and it is necessary to decrease the number of assistants. Now there is a tendency on «Discrete mathematics» course that the education assistants who come from year to year are the same. This situation prompted the idea that at the moment when assessing teaching assistants, it is worth using additional criteria that will allow the group of teachers to select the best assistant and choose the group of the most successful assistants. Thus, two tasks are faced – to choose the best assistant on «Discrete mathematics» course and to select the group of the most successful assistants, with whom it is possible to continue working on this course. The purpose of this work is the development of searching method, which will select the best assistant and select the group of the most successful ones according to the criteria set by the group of teachers. The rest of the paper is organized as follows. We discuss the problem specification in Section 2 and introduce our premises for model, which we use to illustrate our main results on Section 9. Sections 3, 4 and 6 present the different methods used for solution the problem. In sections 5 and 7 the derivations for the AHP and Fuzzy AHP are dis-cussed. Section 8 presents a sensitivity analyze. 252 3. Problem Definitions The problem of finding the best teaching assistant and the group of teaching assistants is closely related with searching the criteria by which the teaching assistants should be selected. To analyze the domain and determine its boundaries, the rich picture can be applied. Rich Picture is a collection of sketches, pictures, photos, symbols, signatures which represent a particular situation or a question of the real world from the point of view of the person or group of people who create it. Image components are people (stakeholders), systems, processes, inter-faces, data streams, information sources, infrastructure objects, attendant and impeding factors, emotions, points of view and attitude to them, etc. Rich Picture can reflect the interaction and connections of the system components (or the surrounding world), their influence, cause and effect. It can also represent such subjective elements as attitude (perception), point of view, prejudice [1]. It is used to explore and aggregate the physical, conceptual and emotional aspects of the actual situation (sys-tem/problem/need). Rich picture on subject «Teaching assistants» interactions in discipline «Discrete mathematic» is provided in Fig. 1. To analyze the subject area and project boundaries, the CATWOE technique is a good addition to Rich Pictures. 253 Beresneva E., Gordenko M. Applying the methods of system analysis to teaching assistants’ evaluation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 251-270 Beresneva E., Gordenko M. Applying the methods of system analysis to teaching assistants’ evaluation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 251-270 CATWOE is defined by Peter Checkland as a part of his Soft Systems Methodology (SSM). It is a simple checklist for thinking. CATWOE is an acronym, each letter stands for a specific word: Clients, Actors, Transformation, World view, Owner, Environmental constraints [2]. CATWOE is defined by Peter Checkland as a part of his Soft Systems Methodology (SSM). It is a simple checklist for thinking. CATWOE is an acronym, each letter stands for a specific word: Clients, Actors, Transformation, World view, Owner, Environmental constraints [2]. Fig. 1. Rich picture on subject “Teaching assistans’ interactions in “Discrete mathematics” Fig. 1. Rich picture on subject “Teaching assistans’ interactions in “Discrete mathematics” Table 1. The CATWOE analysis Role Description Clients Teachers who want to assess their teaching assistants. Students who need assistants’ help. Actors Groups of teachers who interested in evaluating of skills of teaching assistants and choosing the group of the best teaching assistants. 3. Problem Definitions The head of faculty who wants to encourage the best teaching assistant. Transformation Teaching assistant receives points for certain evaluation criteria. World View It is needed to define a group of the best teaching assistants and the best teaching assistant. The definition of a group of best teaching assistants is necessary in order to reduce the risks associated with incompetent and disinterested teaching assistants with the next year group of teaching assistants. Owner Teachers and the head of faculty. Environmental National educational and assessment standards. Table 1. The CATWOE analysis Description 254 Береснева Е.Н., Горденко М.К. Применение методов системного анализа к оцениванию работы учебных ассистентов. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 251-270 After analyzing the processes and interactions associated with the members of the system, a clear understanding of the subject area is emerged. There are three teachers: one lecturer (the leading teacher) and two seminarians at “Discrete mathematics” course. They compose a decision group for choosing best assistants. Fair and reliable evaluation results would be obtained by this group because its members have a strong relationship with teaching assistants during the course. In order to evaluate the assistants, it is decided to come up with evaluation criteria. After the first brainstorm, the list of criteria is similar to a chaotic list of records. Th t ti f th t h h th t f th it i id tifi d i th The next meeting of the teachers shows that some of the criteria identified in the first stage for assessing the assistants turned out to be duplicated or unnecessary. After long discussions and joint brainstorming, three main groups of criteria are identified: professional skills, communicating skills, personal qualities. The professional skills include the following sub-criteria:  active involvement in the process of forming the program of discipline;  initiative to compile new types of tasks for tests;  knowledge of the subject domain;  quality of homework checking;  speed of homework cheking;  experience of active use of the LMS; The communicating skills include the following sub-criteria: The communicating skills include the following sub-criteria:  pedagogical experience, the ability to correctly present information;  pedagogical experience, the ability to correctly present information;  openness to student issues (e g quick response to questions  pedagogical experience, the ability to correctly present information;  openness to student issues (e.g. 3. Problem Definitions quick response to questions, competent answers);  openness to student issues (e.g. quick response to questions, c answers);  participation in counseling sessions before the tests and examinations;  active communicating with teachers, participation in weekly meetings;  the ability to listen carefully. The personal qualities include the following sub-criteria: The personal qualities include the following sub-criteria:  ethical compliance;  punctuality;  self-motivation, the desire for development;  responsibility for work;  teamwork skills;  subordination;  striving to achieve common results;  resistance to conflict situations;  the ability to generate new and innovative ideas;  the ability to compromise;  benevolence. From the first group the next criteria are deleted: From the first group the next criteria are deleted: 255 Beresneva E., Gordenko M. Applying the methods of system analysis to teaching assistants’ evaluation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 251-270 Beresneva E., Gordenko M. Applying the methods of system analysis to teaching assistants’ evaluation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 251-270 active involvement in the process of forming the program of discipline. Th  active involvement in the process of forming the program of discipline. The teachers should do it, because drawing up a discipline program requires experience and entails a great responsibility; p g p g p teachers should do it, because drawing up a discipline program requires experience and entails a great responsibility; teachers should do it, because drawing up a discipline program requires experience and entails a great responsibility;  knowledge of the subject domain. Taking into account that each assistant is selected among the best students of the course, this requirement should be fulfilled by default.  knowledge of the subject domain. Taking into account that each assistant is selected among the best students of the course, this requirement should be fulfilled by default. And the next criteria are combined as they characterize the checking of homework and are closely interrelated: And the next criteria are combined as they characterize the checking of homework and are closely interrelated:  quality of homework checking;  quality of homework checking;  speed of homework checking.  speed of homework checking. From the second group the next criteria are deleted: From the second group the next criteria are deleted:  pedagogical experience, the ability to correctly present information. This ability can be learned. 3. Problem Definitions One of the goals of the "Teaching Assistant" program is the development of teaching skills;  pedagogical experience, the ability to correctly present information. This ability can be learned. One of the goals of the "Teaching Assistant" program is the development of teaching skills;  the ability to listen. In our opinion, this parameter is almost impossible to estimate. From the third group the next criteria are combined, because they are very interrelated and cannot be separated: From the third group the next criteria are combined, because they are very interrelated and cannot be separated:  self-motivation, the desire for development;  responsibility for work; And the next criteria are deleted:  teamwork skills. It is related with the responsibility of work criteria;  teamwork skills. It is related with the responsibility of work criteria;  ability to be subordinate. By default, the main person on the course is the teacher. This is necessary to understand at first;  striving to achieve common results. It is related with the responsibility of work criteria;  resistance to conflict situations. It is the responsibility of the teacher to resolve and prevent the emergence of conflict situations;  the ability to generate new and innovative ideas. This is not a paramount task of the teaching assistant. And the teaching assistant can work great, but do not come up with ideas, it's not scary;  the ability to compromise. The last word for the teacher;  benevolence. It is related with the ethical compliance and punctuality of work criteria. The final elected criteria and subcriteria are shown in Fig. 2. All the criteria and subcriteria have their own identification numbers. The final elected criteria and subcriteria are shown in Fig. 2. All the criteria and subcriteria have their own identification numbers. 256 снева Е.Н., Горденко М.К. Применение методов системного анализа к оцениванию работы учебных стентов. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 251-270 Береснева Е.Н., Горденко М.К. Применение методов системного анализа к оцениванию работы учебных ассистентов. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 251-270 Fig. 2. The final list of criteria. Fig. 2. The final list of criteria. 4. Exploring the alternatives There are ten teaching assistants A, B, C, D, E, F, G, H, I, J on the course. We can reduce the number of evaluating teaching assistants after assessing the involvement of teaching assistants in educational process. We have 3 groups of criteria, consisting of 9 sub-criteria. In order to assess the involvement of assistants in the educational process, we did not use the values of the last three subcriteria (3.1-3.3). These sub-criteria refer to a group of personal qualities and cannot be regarded as involvement in the educational process. Then the involvement of the teaching assistant in the educational process for each criterion is evaluated, based on expert judgment. The results are presented in Table 2. Table 2. The involvement in educational process A B C D E F G H I J 1.1. 5 5 5 5 4 5 3 2 1 1 1.2. 4 5 5 4 4 4 4 4 1 4 1.3. 1 2 1 4 5 1 1 1 1 1 2.1. 4 3 3 4 2 5 5 1 3 1 2.2. 5 4 4 4 3 5 2 1 2 2 2.3. 4 4 3 4 1 4 5 2 1 2 Table 2. The involvement in educational process 257 Beresneva E., Gordenko M. Applying the methods of system analysis to teaching assistants’ evaluation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 251-270 Beresneva E., Gordenko M. Applying the methods of system analysis to teaching assistants’ evaluation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 251-270 Let us understand which assistants are least involved in the work process, according to experts. Calculations of threshold equals to 3, 4 and 5 are shown in Tables 3, 4 and 5. Let us understand which assistants are least involved in the work process, according to experts. Calculations of threshold equals to 3, 4 and 5 are shown in Tables 3, 4 and 5. Table 3.Treshhold is equal to 3 Table 3.Treshhold is equal to 3 Table 3.Treshhold is equal to 3 A B C D E F G H I J 1.1. 1 1 1 1 1 1 1 0 0 0 1.2. 1 1 1 1 1 1 1 1 0 1 1.3. 0 0 0 1 1 0 0 0 0 0 2.1. 1 1 1 1 0 1 1 0 1 0 2.2. 4. Exploring the alternatives 1 1 1 1 1 1 0 0 0 0 2.3. 1 1 1 1 0 1 1 0 0 0 Table 3 and 4 allow to identify teaching assistants who are least involved in the educational process. educational process. The Table 5 with threshold equals to 5 shows that no one from H, I, J is not indispensable. The Table 5 with threshold equals to 5 shows that no one from H, I, J is not indispensable. Table 4.Treshhold is equal to 4  H was ill two month; Table 4.Treshhold is equal to 4 Table 4.Treshhold is equal to 4 A B C D E F G H I J 1.1. 1 1 1 1 1 1 0 0 0 0 1.2. 1 1 1 1 1 1 1 1 0 1 1.3. 0 0 0 1 1 0 0 0 0 0 2.1. 1 0 0 1 0 1 1 0 0 0 2.2. 1 1 1 1 0 1 0 0 0 0 2.3. 1 1 0 1 0 1 1 0 0 0 Table 5.Treshhold is equal to 5 A B C D E F G H I J 1.1. 1 1 1 1 0 1 0 0 0 0 1.2. 0 1 1 0 0 0 0 0 0 0 1.3. 0 0 0 0 1 0 0 0 0 0 2.1. 0 0 0 0 0 1 1 0 0 0 2.2. 1 0 0 0 0 1 0 0 0 0 2.3. 0 0 0 0 0 0 1 0 0 0 Thus, it is decided not to consider further the last three teaching assistants (H, I, J). However, little involvement in the educational process has its own explanations:  H was ill two month; A B C D E F G H I J 1.1. 1 1 1 1 1 1 0 0 0 0 1.2. 1 1 1 1 1 1 1 1 0 1 1.3. 0 0 0 1 1 0 0 0 0 0 2.1. 1 0 0 1 0 1 1 0 0 0 2.2. 1 1 1 1 0 1 0 0 0 0 2.3. 1 1 0 1 0 1 1 0 0 0 Table 5.Treshhold is equal to 5 Table 5.Treshhold is equal to 5 A B C D E F G H I J 1.1. 1 1 1 1 0 1 0 0 0 0 1.2. 0 1 1 0 0 0 0 0 0 0 1.3. 0 0 0 0 1 0 0 0 0 0 2.1. 0 0 0 0 0 1 1 0 0 0 2.2. 1 0 0 0 0 1 0 0 0 0 2.3. 0 0 0 0 0 0 1 0 0 0 Thus, it is decided not to consider further the last three teaching assistants (H, I, J). Table 4.Treshhold is equal to 4 However, little involvement in the educational process has its own explanations: Thus, it is decided not to consider further the last three teaching assistants (H, I, J). However, little involvement in the educational process has its own explanations:  H was ill two month; 258 Береснева Е.Н., Горденко М.К. Применение методов системного анализа к оцениванию работы учебных ассистентов. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 251-270  I was out of connection;  J decided to switch to another faculty. Preparation for the exams took all his spare time.  J decided to switch to another faculty. Preparation for the exams took all his spare time. Thus, seven candidates are remained. It is difficult to find the best one because each of them is successful in one or more criteria. Stakeholders are about to choose A as a winner because this assistant took part in all teacher meetings and he suggested new types of tasks for tests so regularly (approximately once every two weeks). Assistant A communicated with teachers a lot (flashed before their eyes), that is why they prefer him. However, this decision can be too unfair, that’s why multicriteria decision making (MCDM) prosess is decided to be applied. 5. Analytical Hierarchy Process Analytic Hierarchy Process (AHP) which is one of the most used MCDM approaches Указан недопустимый источник. is a structured multicriteria technique for organizing and analyzing complex decisions including many criteria. In this paper we use a classical AHP proposed by the author Указан недопустимый источник.. At the first step of AHP a model for the decision is developed. Experts break down the decision into a hierarchy of goals, criteria, sub-criteria and alternatives. After that, decisioners derive priorities (weights) for the criteria with respect to the desired goal. It is made in the form of pairwise comparisons using individual questionnaires. Since the evaluation criteria are subjective and qualitative in nature, it is very difficult for the decision maker to express the preferences using exact numerical values. That is why a special numerical scale Указан недопустимый источник. which consists of interpretation of linguistic terms is used (see Table 6). 259 Table 6. Saaty’s pairwise comparison scale Numeric value Linguistic terms 1 Equally important 3 Moderately more important 5 Strongly more important 7 Very Strongly more important 9 Extremely important 2, 4, 6, 8 The intermediate values that are used to address situations of uncertainty between the two adjacent judgments Results of comparisons of all experts are presented in the form of matrices (see Table 7). Table 7. Criteria pairwise comparisons obtained by experts Professional skills Communicative skills Personal qualities Exp.1 Exp.2 Exp.3 Exp.1 Exp.2 Exp.3 Exp.1 Exp.2 Exp.3 Table 6. Saaty’s pairwise comparison scale 259 Beresneva E., Gordenko M. Applying the methods of system analysis to teaching assistants’ evaluation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 251-270 Professional skills 1 1 1 5 5 4 3 2 1 Communicative skills 1/5 1/5 1/4 1 1 1 1/3 1/4 1/5 Personal qualities 1/3 1/2 1 3 4 5 1 1 1 Before calculating the weights, the consistency of the comparison matrix is checked. As a rule, only if consistency is less than 0.1, it considered as acceptable, otherwise the pair-wise comparisons should be revised. In this decision making process, all of them are less than 0.092 that shows answers are consistence. On the basis of Table 7 the final matrix is created by finding a mean between estimates of all experts (see Table 8). This metric is used because of solid decision to make all experts’ voices to be equal. Table 8. Table 10. Local and overall priorities of alternatives 5. Analytical Hierarchy Process Aggregate matrix with criteria pairwise comparisons Table 8. Aggregate matrix with criteria pairwise comparisons Professional skills Communicative skills Personal qualities Professional skills 1 4,66 2 Communicative skills 0,22 1 0,25 Personal qualities 0,6 4 1 The matrix from Table 8 is used in order to calculate criteria priority weights. The same way as it was earlier, a pairwise comparison of all the sub-criteria, with respect to each criterion, included in the decision-making model, is made. Obtained results are shown in Table 9. Table 9. Criteria and subcriteria priority weights 1. Professional skills 54,772% 1.1. New task types creation 25,232% 1.2. HW checking 26,068% 1.3. Experience in LMS 3,472% 2. Communicative skills 10,069% 2.1. Openness to students 2,946% 2.2. Communication with teachers 1,288% 2.3. Participation in consultations 5,834% 3. Personal qualities 35,159% 3.1. Punctuality 13,086% 3.2. Ethical compliance 10,062% 3.3. Self-motivation 12,011% Table 9. Criteria and subcriteria priority weights Table 9. Criteria and subcriteria priority weights Next step consists of deriving the relative priorities (preferences) of the alternatives with respect to each criterion. Overall priority weights of assistants are calculated by summing all local priorities. Final figures are shown in Table 10. Bar chart is built on the basis of overall preferences of the alternatives (see Fig. 3). Table 10. Local and overall priorities of alternatives 260 Береснева Е.Н., Горденко М.К. Применение методов системного анализа к оцениванию работы учебных ассистентов. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 251-270 A B C D E F G 1.1. 8,352% 6,711% 3,784% 2,728% 1,821% 1,214% 0,622% 1.2. 3,140% 6,060% 9,131% 3,836% 1,700% 0,966% 1,234% 1.3. 0,160% 0,249% 0,148% 1,001% 1,618% 0,148% 0,148% 2.1. 0,203% 0,154% 0,101% 0,441% 0,082% 0,803% 1,161% 2.2. 0,444% 0,089% 0,117% 0,062% 0,062% 0,444% 0,072% 2.3. 0,414% 0,387% 0,198% 0,833% 0,144% 1,628% 2,230% 3.1. 1,035% 2,677% 1,921% 3,670% 1,663% 1,041% 1,078% 3.2. 0,254% 1,928% 0,254% 3,822% 1,782% 0,993% 1,029% 3.3. 0,923% 2,040% 0,761% 4,201% 1,640% 1,108% 1,339% Totals 14,924% 20,296% 16,416% 20,595% 10,010% 8,846% 9,114% Fig. 3. Overall preferences of the alternatives Fig. 3. Overall preferences of the alternatives 6. A Discussion on AHP Results AHP analysis shows that the prompt decision of choosing A as the best assistant is totally unfair. Results reveal that experts did not take into account other important criteria that in general over weighted those, which were chosen at first. Another discovered problem of A is some of his/her estimates, which are the worst in comparison with others (for instance, criteria 3.1 and 3.2). This fact also decreases his/her chances to be a leader. The main interesting point of results are the highest figures which belong to both two assistants B and D. Let's describe each of them. Assistant B cannot be named as a brilliant employee. Nevertheless, he/she has showed good stable work without having bad results in any of the activities during the course. Despite not being the best in any of the criteria, B always was close to the leader. In the same manner as B, assistant D has shown quite strong results in technical and communicative estimates. In addition, D was on the top in the personal qualities 261 Beresneva E., Gordenko M. Applying the methods of system analysis to teaching assistants’ evaluation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 251-270 Beresneva E., Gordenko M. Applying the methods of system analysis to teaching assistants’ evaluation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 251-270 criteria. He/she produces an impression of too self-motivated person and D was never late on any events. Result of D exceeds B at an inconspicuous gap of 0,3. Since experts make an arrangement on having no less than 2% advantage taking by the leader, such difference is admitted being not crucial for them. In addition, there is a problematic situation with evaluation of the five best assistants. Four employees can be determined more or less clearly (A, B, C, and D). However, the difference between E and the closest competitor G is less than 1%, which is also insignificant. 7. Fuzzy Type-2 AHP Since experts want to be more confident in fairness of their choice, we decide to apply another MCDM approach for purpose of aiming our goal. It is called Fuzzy AHP. In classical AHP crisp numbers are used, for pairwise comparison evaluations. However, in Fuzzy AHP, the linguistic variables are represented as fuzzy numbers instead of crisp. In this case a fuzzy logic provides a mathematical strength to capture the uncertainties associated with human cognitive process. Many researchers Указан недопустимый источник., Указан недопустимый источник. who have studied the Fuzzy AHP have provided evidence that it shows relatively more sufficient description of decision making processes compared to the traditional AHP methods. According to Указан недопустимый источник., the membership functions of type-1 fuzzy sets have no uncertainty associated with it. Type-2 fuzzy sets generalize type-1 fuzzy sets and systems so that more uncertainty for defining membership functions can be handled. That’s why type-2 fuzzy logic is used. A type-2 fuzzy set 𝐴̃̃ in the universe of discourse X can be represented by a type-2 membership function 𝜇𝐴̃̃, shown as follows Указан недопустимый источник.: 𝐴̃̃ = {((𝑥, 𝑢), 𝜇𝐴̃̃(𝑥, 𝑢)) \|∀𝑥∈𝑋, ∀𝑢∈𝐽𝑥⊆[0, 1], 0 ≤𝜇𝐴̃̃(𝑥, 𝑢) ≤1}, where Jx denotes an interval [0, 1]. Ã̃ is called an interval type-2 fuzzy set if all μÃ̃ = 1 Указан недопустимый источник.. Interval type-2 fuzzy sets are the most commonly used type-2 fuzzy sets because of their simplicity and reduced computational effort with respect to general type-2 fuzzy sets. For this reason, we use interval type-2 fuzzy sets. A trapezoidal interval type-2 fuzzy set is illustrated as where 𝐴𝑖𝑈 ̃ and 𝐴𝑖𝐿 ̃ are type-1 fuzzy sets, 𝑎𝑖1 𝑈, 𝑎𝑖2 𝑈, … , 𝑎𝑖3 𝐿, 𝑎𝑖4 𝐿 are the references points of the interval type-2 fuzzy set 𝐴̃𝑖̃ ; 𝐻𝑗(𝐴𝑖𝑈 ̃) denotes the membership value of the element 𝑎𝑗(𝑗+1) 𝑈 in the upper trapezoidal membership function 𝐴𝑖𝑈 ̃ and 𝐻𝑗(𝐴𝑖𝐿 ̃) denotes the membership value of the element 𝑎𝑗(𝑗+1) 𝐿 in the lower trapezoidal membership function 𝐴𝑖𝐿 ̃, 𝑗= 1. .2 ̅̅̅̅̅ Указан недопустимый источник.. 7. Fuzzy Type-2 AHP where 𝐴𝑖𝑈 ̃ and 𝐴𝑖𝐿 ̃ are type-1 fuzzy sets, 𝑎𝑖1 𝑈, 𝑎𝑖2 𝑈, … , 𝑎𝑖3 𝐿, 𝑎𝑖4 𝐿 are the references points of the interval type-2 fuzzy set 𝐴̃𝑖̃ ; 𝐻𝑗(𝐴𝑖𝑈 ̃) denotes the membership value of the element 𝑎𝑗(𝑗+1) 𝑈 in the upper trapezoidal membership function 𝐴𝑖𝑈 ̃ and 𝐻𝑗(𝐴𝑖𝐿 ̃) denotes the membership value of the element 𝑎𝑗(𝑗+1) 𝐿 in the lower trapezoidal membership function 𝐴𝑖𝐿 ̃, 𝑗= 1. .2 ̅̅̅̅̅ Указан недопустимый источник.. 262 Береснева Е.Н., Горденко М.К. Применение методов системного анализа к оцениванию работы учебных ассистентов. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 251-270 Pairwise comparison matrices got from experts for AHP are directly applied for our needs in Fuzzy AHP. We introduce interval trapezoidal type-2 fuzzy scales of the linguistic variables (see Table 11). They represent a modified version of scales proposed by Указан недопустимый источник. and include intermediate values between the two adjacent judgments like in AHP. The priority weights of criteria (Table 12) and sub-criteria (Table 13) are demonstrated. Type 2 fuzzy and defuzzified overall weights of the alternatives are shown in Tables 14 and 15. For defuzzification the Defuzzified Trapezoidal Type-2 Fuzzy Set (DTraT) approach is used proposed by Указан недопустимый источник.. Bar chart is built on the basis of overall preferences of the alternatives (see Fig. 4). rt is built on the basis of overall preferences of the alternatives (see Fig. 4). Table 11. Trapezoidal interval type-2 fuzzy scales Numeric value from AHP Trapezoidal interval type-2 fuzzy scales 1 (1, 1, 1, 1; 1, 1) (1, 1, 1, 1; 1, 1) 2 (1, 1, 3, 4; 1, 1) (1.2, 1.2, 2.8, 3.8; 0.8, 0.8) 3 (1, 2, 4, 5; 1, 1) (1.2, 2.2, 3.8, 4.8; 0.8, 0.8) 4 (2, 3, 5, 6; 1, 1) (2.2, 3.2, 4.8, 5.8; 0.8, 0.8) 5 (3, 4, 6, 7; 1, 1) (3.2, 4.2, 5.8, 6.8; 0.8, 0.8) 6 (4, 5, 7, 8; 1, 1) (4.2, 5.2, 6.8, 7.8; 0.8, 0.8) 7 (5, 6, 8, 9; 1, 1) (5.2, 6.2, 7.8, 8.8; 0.8, 0.8) 8 (6, 7, 8.5, 9; 1, 1) (6.2, 7.2, 8.3, 8.8; 0.8, 0.8) 9 (7, 8, 9, 9; 1, 1) (7.2, 8.2, 8.8, 9; 0.8, 0.8) Table 12. Interval type-2 fuzzy weights of criteria Criteria Interval type-2 weights 1. Professional skills (0.275, 0.377, 0.754, 1.005; 1, 1) (0.304, 0.410, 0.706, 0.935; 0.8, 0.8) 2. 7. Fuzzy Type-2 AHP (0.004, 0.007, 0.031, 0.071; 1, 1) (0.004, 0.008, 0.025, 0.055; 0.8, 0.8) 2.3. (0.014, 0.029, 0.115, 0.24; 1, 1) (0.017, 0.034, 0.099, 0.199; 0.8, 0.8) 3.1. (0.055, 0.087, 0.212, 0.317; 1, 1) (0.062, 0.095, 0.191, 0.28; 0.8, 0.8) 3.2. (0.046, 0.069, 0.168, 0.265; 1, 1) (0.051, 0.075, 0.151, 0.23; 0.8, 0.8) 3.3. (0.046, 0.075, 0.196, 0.317; 1, 1) (0.052, 0.082, 0.175, 0.274; 0.8, 0.8) Fig. 4. Overall preferences of the alternatives Fig. 4. Overall preferences of the alternatives Table 14. Local and overall priorities of alternatives A, B, C, D Table 14. Local and overall priorities of alternatives A, B, C, D A B C D 1.1. (0.115, 0.212, 0.505, 0.806; 1, 1) (0.133, 0.234, 0.462, 0.7; 0.8, 0.8) (0.098, 0.174, 0.414, 0.692; 1, 1) (0.113, 0.191, 0.38, 0.612; 0.8, 0.8) (0.048, 0.091, 0.252, 0.484; 1, 1) (0.06, 0.1, 0.23, 0.4; 0.8, 0.8) (0.037, 0.063, 0.173, 0.342; 1, 1) (0.042, 0.07, 0.15, 0.3; 0.8, 0.8) 1.2. (0.04, 0.067, 0.158, 0.272; 1, 1) (0.045, 0.073, 0.14, 0.24; 0.8, 0.8) (0.083, 0.144, 0.314, 0.471; 1, 1) (0.095, 0.157, 0.3, 0.429; 0.8, 0.8) (0.186, 0.265, 0.477, 0.657; 1, 1) (0.202, 0.283, 0.45, 0.6; 0.8, 0.8) (0.067, 0.108, 0.26, 0.406; 1, 1) (0.077, 0.12, 0.24, 0.37; 0.8, 0.8) 1.3. (0.038, 0.042, 0.055, 0.067; 1, 1) (0.039, 0.043, 0.05, 0.064; 0.8, 0.8) (0.038, 0.051, 0.082, 0.106; 1, 1) (0.041, 0.054, 0.08, 0.1; 0.8, 0.8) (0.032, 0.037, 0.051, 0.067; 1, 1) (0.033, 0.04, 0.05, 0.063; 0.8, 0.8) (0.177, 0.229, 0.382, 0.536; 1, 1) (0.187, 0.24, 0.36, 0.5; 0.8, 0.8) 2.1. (0.029, 0.044, 0.086, 0.133; 1, 1) (0.032, 0.047, 0.08, 0.12; 0.8, 0.8) (0.027, 0.038, 0.072, 0.11; 1, 1) (0.029, 0.04, 0.07, 0.1; 0.8, 0.8) (0.021, 0.026, 0.046, 0.071; 1, 1) (0.022, 0.03, 0.043, 0.06; 0.8, 0.8) (0.064, 0.104, 0.237, 0.409; 1, 1) (0.072, 0.113, 0.22, 0.36; 0.8, 0.8) 2.2. (0.185, 0.262, 0.448, 0.587; 1, 1) (0.2, 0.278, 0.426, 0.554; 0.8, 0.8) (0.041, 0.05, 0.077, 0.106; 1, 1) (0.04, 0.05, 0.07, 0.098; 0.8, 0.8) (0.044, 0.055, 0.084, 0.111; 1, 1) (0.046, 0.057, 0.08, 0.1; 0.8, 0.8) (0.038, 0.043, 0.061, 0.076; 1, 1) (0.04, 0.05, 0.06, 0.07; 0.8, 0.8) 2.3. 7. Fuzzy Type-2 AHP Communicative skills (0.057, 0.073, 0.14, 0.211; 1, 1) (0.061, 0.078, 0.13, 0.19; 0.8; 0.8) 3. Personal qualities (0.188, 0.257, 0.519, 0.708; 1, 1) (0.203, 0.274, 0.477, 0.639; 0.8, 0.8) Table 13. Interval type-2 fuzzy weights of sub-criteria Sub-criteria Interval type-2 weights 1.1. (0.071, 0.134, 0.447, 0.811; 1, 1) (0.085, 0.154, 0.396, 0.703; 0.8, 0.8 1.2. (0.074, 0.138, 0.453, 0.811; 1, 1) (0.088, 0.158, 0.402, 0.705; 0.8, 0.8 1.3. (0.013, 0.022, 0.069, 0.124; 1, 1) (0.015, 0.025, 0.061, 0.108; 0.8, 0.8 2.1. (0.008, 0.014, 0.062, 0.149; 1, 1) (0.009, 0.016, 0.052, 0.118; 0.8, 0.8 Table 11. Trapezoidal interval type-2 fuzzy scales Numeric value from AHP Trapezoidal interval type-2 fuzzy scales 1 (1, 1, 1, 1; 1, 1) (1, 1, 1, 1; 1, 1) 2 (1, 1, 3, 4; 1, 1) (1.2, 1.2, 2.8, 3.8; 0.8, 0.8) 3 (1, 2, 4, 5; 1, 1) (1.2, 2.2, 3.8, 4.8; 0.8, 0.8) 4 (2, 3, 5, 6; 1, 1) (2.2, 3.2, 4.8, 5.8; 0.8, 0.8) 5 (3, 4, 6, 7; 1, 1) (3.2, 4.2, 5.8, 6.8; 0.8, 0.8) 6 (4, 5, 7, 8; 1, 1) (4.2, 5.2, 6.8, 7.8; 0.8, 0.8) 7 (5, 6, 8, 9; 1, 1) (5.2, 6.2, 7.8, 8.8; 0.8, 0.8) 8 (6, 7, 8.5, 9; 1, 1) (6.2, 7.2, 8.3, 8.8; 0.8, 0.8) 9 (7, 8, 9, 9; 1, 1) (7.2, 8.2, 8.8, 9; 0.8, 0.8) Table 11. Trapezoidal interval type-2 fuzzy scales 263 Beresneva E., Gordenko M. Applying the methods of system analysis to teaching assistants’ evaluation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 251-270 Beresneva E., Gordenko M. Applying the methods of system analysis to teaching assistants’ evaluation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 251-270 Beresneva E., Gordenko M. Applying the methods of system analysis to teaching assistants’ evaluation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 251-270 2.2. (0.004, 0.007, 0.031, 0.071; 1, 1) (0.004, 0.008, 0.025, 0.055; 0.8, 0.8) 2.3. (0.014, 0.029, 0.115, 0.24; 1, 1) (0.017, 0.034, 0.099, 0.199; 0.8, 0.8) 3.1. (0.055, 0.087, 0.212, 0.317; 1, 1) (0.062, 0.095, 0.191, 0.28; 0.8, 0.8) 3.2. (0.046, 0.069, 0.168, 0.265; 1, 1) (0.051, 0.075, 0.151, 0.23; 0.8, 0.8) 3.3. (0.046, 0.075, 0.196, 0.317; 1, 1) (0.052, 0.082, 0.175, 0.274; 0.8, 0.8) RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 251-270 2.2. Beresneva E., Gordenko M. Applying the methods of system analysis to teaching assistants’ evaluation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 251-270 2.2. (0.004, 0.007, 0.031, 0.071; 1, 1) (0.004, 0.008, 0.025, 0.055; 0.8, 0.8) 2.3. (0.014, 0.029, 0.115, 0.24; 1, 1) (0.017, 0.034, 0.099, 0.199; 0.8, 0.8) 3.1. (0.055, 0.087, 0.212, 0.317; 1, 1) (0.062, 0.095, 0.191, 0.28; 0.8, 0.8) 3.2. (0.046, 0.069, 0.168, 0.265; 1, 1) (0.051, 0.075, 0.151, 0.23; 0.8, 0.8) 3.3. (0.046, 0.075, 0.196, 0.317; 1, 1) (0.052, 0.082, 0.175, 0.274; 0.8, 0.8) 7. Fuzzy Type-2 AHP (0.033, 0.046, 0.091, 0.137; 1, 1) (0.035, 0.05, 0.08, 0.124; 0.8, 0.8) (0.034, 0.046, 0.083, 0.122; 1, 1) (0.036, 0.049, 0.08, 0.1; 0.8, 0.8) (0.019, 0.025, 0.046, 0.07; 1, 1) (0.02, 0.026, 0.04, 0.06; 0.8, 0.8) (0.065, 0.1, 0.22, 0.366; 1, 1) (0.072, 0.108, 0.2, 0.323; 0.8, 0.8) 3.1. (0.029, 0.042, 0.086, 0.13; 1, 1) (0.03, 0.045, 0.08, 0.116; 0.8, 0.8) (0.102, 0.161, 0.324, 0.462; 1, 1) (0.113, 0.174, 0.3, 0.425; 0.8, 0.8) (0.038, 0.052, 0.095, 0.138; 1, 1) (0.041, 0.055, 0.09, 0.126; 0.8, 0.8) (0.134, 0.215, 0.483, 0.71; 1, 1) (0.153, 0.24, 0.45, 0.649; 0.8, 0.8) 3.2. (0.015, 0.02, 0.035, 0.053; 1, 1) (0.016, 0.021, 0.03, 0.048; 0.8, 0.8) (0.068, 0.102, 0.198, 0.295; 1, 1) (0.075, 0.109, 0.2, 0.27; 0.8, 0.8) (0.015, 0.02, 0.035, 0.053; 1, 1) (0.02, 0.02, 0.03, 0.05; 0.8, 0.8) (0.163, 0.269, 0.543, 0.78; 1, 1) (0.184, 0.3, 0.51, 0.72; 0.8, 0.8) 3.3. (0.05, 0.059, 0.09, 0.115; 1, 1) (0.052, 0.061, 0.09, 0.108; 0.8, 0.8) (0.061, 0.089, 0.23, 0.325; 1, 1) (0.07, 0.099, 0.2, 0.294; 0.8, 0.8) (0.038, 0.047, 0.081, 0.115; 1, 1) (0.04, 0.05, 0.076, 0.105; 0.8, 0.8) (0.184, 0.262, 0.492, 0.632; 1, 1) (0.2, 0.28, 0.46, 0.59; 0.8, 0.8) Total fuzzy weight (0.018, 0.052, 0.372, 1.069; 1, 1) (0.023, 0.064, 0.3, 0.826; 0.8, 0.8) (0.026, 0.074, 0.496, 1.336; 1, 1) (0.034, 0.091, 0.4, 1.044; 0.8, 0.8) (0.023, 0.061, 0.384, 1.062; 1, 1) (0.03, 0.074, 0.32, 0.83; 0.8, 0.8) (0.035, 0.09, 0.552, 1.46; 1, 1) (0.045, 0.11, 0.454, 1.14; 0.8, 0.8) Total defuz. weight 0.331 0.426 0.337 0.471 264 Береснева Е.Н., Горденко М.К. Применение методов системного анализа к оцениванию работы уч ассистентов. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 251-270 Береснева Е.Н., Горденко М.К. Применение методов системного анализа к оцениванию работы учебных ассистентов. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 251-270 Total norm. defuz. weight 14.868% 19.119% 15.124% 21.128% Table 15. Local and overall priorities of alternatives E, F, G E F G 1.1. (0.025, 0.042, 0.123, 0.257; 1, 1) (0.029, 0.046, 0.109, 0.212; 0.8, 0.8) (0.019, 0.03, 0.082, 0.161; 1, 1) (0.021, 0.033, 0.073, 0.134; 0.8, 0.8) (0.013, 0.018, 0.04, 0.077; 1, 1) (0.014, 0.019, 0.036, 0.065; 0.8, 0.8) 1.2. (0.029, 0.046, 0.106, 0.192; 1, 1) (0.033, 0.05, 0.097, 0.165; 0.8, 0.8) (0.021, 0.028, 0.055, 0.091; 1, 1) (0.022, 0.03, 0.05, 0.08; 0.8, 0.8) (0.024, 0.033, 0.076, 0.131; 1, 1) (0.026, 0.036, 0.068, 0.111; 0.8, 0.8) 1.3. 7. Fuzzy Type-2 AHP (0.283, 0.378, 0.56, 0.674; 1, 1) (0.302, 0.4, 0.532, 0.648; 0.8, 0.8) (0.032, 0.037, 0.051, 0.067; 1, 1) (0.033, 0.038, 0.049, 0.063; 0.8, 0.8) (0.032, 0.037, 0.051, 0.067; 1, 1) (0.033, 0.038, 0.049, 0.063; 0.8, 0.8) 2.1. (0.015, 0.02, 0.037, 0.06; 1, 1) (0.016, 0.021, 0.034, 0.053; 0.8, 0.8) (0.128, 0.193, 0.392, 0.613; 1, 1) (0.14, 0.208, 0.363, 0.553; 0.8, 0.8) (0.177, 0.282, 0.546, 0.772; 1, 1) (0.198, 0.304, 0.509, 0.715; 0.8, 0.8) 2.2. (0.038, 0.043, 0.061, 0.076; 1, 1) (0.039, 0.045, 0.058, 0.072; 0.8, 0.8) (0.226, 0.292, 0.461, 0.589; 1, 1) (0.239, 0.306, 0.44, 0.559; 0.8, 0.8) (0.042, 0.049, 0.068, 0.086; 1, 1) (0.044, 0.05, 0.066, 0.081; 0.8, 0.8) 2.3. (0.014, 0.017, 0.031, 0.049; 1, 1) (0.014, 0.018, 0.029, 0.044; 0.8, 0.8) (0.135, 0.2, 0.409, 0.6; 1, 1) (0.148, 0.216, 0.377, 0.543; 0.8, 0.8) (0.179, 0.272, 0.537, 0.747; 1, 1) (0.2, 0.295, 0.5, 0.693; 0.8, 0.8) 3.1. (0.064, 0.094, 0.194, 0.316; 1, 1) (0.07, 0.1, 0.179, 0.28; 0.8, 0.8) (0.046, 0.064, 0.128, 0.209; 1, 1) (0.049, 0.068, 0.118, 0.184; 0.8, 0.8) (0.05, 0.067, 0.13, 0.197; 1, 1) (0.053, 0.071, 0.12, 0.175; 0.8, 0.8) 3.2. (0.088, 0.146, 0.327, 0.538; 1, 1) (0.099, 0.159, 0.3, 0.48; 0.8, 0.8) (0.047, 0.067, 0.145, 0.263; 1, 1) (0.051, 0.072, 0.132, 0.227; 0.8, 0.8) (0.056, 0.077, 0.145, 0.225; 1, 1) (0.06, 0.082, 0.135, 0.203; 0.8, 0.8) 3.3. (0.111, 0.129, 0.195, 0.233; 1, 1) (0.117, 0.135, 0.186, 0.222; 0.8, 0.8) (0.059, 0.074, 0.131, 0.175; 1, 1) (0.063, 0.078, 0.122, 0.159; 0.8, 0.8) (0.064, 0.078, 0.137, 0.167; 1, 1) (0.067, 0.082, 0.127, 0.154; 0.8, 0.8) Total weight (0.021, 0.049, 0.283, 0.789; 1, 1) (0.026, 0.059, 0.233, 0.604; 0.8, 0.8) (0.015, 0.035, 0.227, 0.681; 1, 1) (0.018, 0.042, 0.183, 0.509; 0.8, 0.8) (0.016, 0.037, 0.232, 0.652; 1, 1) (0.019, 0.044, 0.188, 0.496; 0.8, 0.8) Total defuzzy weight 0.251 0.208 0.205 Total norm. defuzzy weight 11.347% 9.334% 9.079% Table 15. Local and overall priorities of alternatives E, F, G 8. Discussion on Fuzzy-Type-2 AHP Results Now, we see that assistant D has higher priority weigth than B and difference between them (2%) is suitable for experts. In addition, it can be noticed that E should be in the top five group, for sure (difference is also about 2%). Thus, Fuzzy AHP does not change ranks of alternatives but makes it clearer. It means that more reliable results are maintained since interval type-2 fuzzy sets can better represent uncertainties. It is important to note that, contrary to the common belief, the system does not determine the decision we should make, rather, the results should be interpreted as a blueprint of preference and alternatives based on the level of importance obtained 265 Beresneva E., Gordenko M. Applying the methods of system analysis to teaching assistants’ evaluation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 251-270 Beresneva E., Gordenko M. Applying the methods of system analysis to teaching assistants’ evaluation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 251-270 for the different criteria taking into consideration our comparative judgments. In other words, the AHP methodology allows us to determine which alternative is the most consistent with our criteria and the level of importance that we give them. Taking this point into account, Sensitivity Analysis is used. It performs a “what-if” analysis to see how the final results would have changed if the weights of the criteria would have been different Указан недопустимый источник.. Let’s start with a goal of finding the best teaching assistant. The first criterion has the highest weight in our results (≈50%). If we decrease its weight and proportionally increase other weights then D will still be a leader. In this case D will have even more clear-cut victory. Otherwise, if we increase weight of this criterion up to 60% and more, then B will become a new leader. However, stakeholders come to one opinion that no one criterion should cost more than a half and they has highlighted that the first criterion (professional skills) should stay as the most important one. It means that weight of the first criterion should be in the next approximate range [33%; 50%]. Let’s now tune proportions of the second and the third criteria. Calculations show that D can stop be a winner only and only if the third criterion will cost more than the second. 9. Final Result and Conclusion Taking into consideration recommendations mentioned above, group of teachers has decided to follow first two instructions. They have selected D as the best teaching assistant on the course of «Discrete mathematics». Also, they have prolongated contracts with D and A, B, and C assistants. The main important step now is to choose the fifth assistant. Before making a choice, experts decide to use a retrospective and to look through all methods that were applied earlier. Lecture of the course noticed that since A, B, C, and D assistants are already confirmed it means that nobody will be responsible for communication with students (answering questions, having consultations) because assistant F did it before. However, now there is a choice between either E or G. And in this case G demonstrates a clear superiority compared with others as he/she is one of the top in this kind of work. Finally, G is chosen. At the very beginning teachers wanted to choose assistant A as the best teaching assistant. However, the soft methods of analysis helped us to choose another assistant. Also, neither AHP nor Fuzzy AHP chose G teaching assistant as the 5th best assistant in the group. Only a sound logic helped us to do this. The application of methods of system analysis can help to make a decision but it does not make a choice for us. We should look carefully at the results of system analysis methods, but make a balanced and considered decision. 8. Discussion on Fuzzy-Type-2 AHP Results Thus, this point was brought to expert discussion and they have unanimously decided that personal qualities (third criterion) should be appreciated higher than communicative ones. Another important note is change of proportions of subcriteria inside their criteria. There are no strong disputes about subcriteria weights, experts’ opinions differ no more than 10%. In this case change of subcriteria preferences in that range does not influence on the leader. It means that there is no opportunity to have another leader than D by introducing small changes in current proportions of criteria weights. At the same time, there is a complex situation with choosing top five assistants group. Analysis shows that four assistants are determined clearly. They are A, B, C, and D. The fifth assistant can be either E or G. Calculations reveal that position of assistant G is directly connected with the second criteria and if its weight is equal or more than 15% than G will be in top five group instead of E. However, now second criterion has only nearly 10%. Finally, after Sensitivity Analysis is done, next recommendations for the experts are given:  to choose assistant D as a winner;  to prolongate contracts with A, B, C and D;  to prolongate contract with E if experts think that personal qualities should be at least twice more important than communicating skills (finally, communicating skills should have a weight less than 15);  to prolongate contract with G, in other case. 266 Береснева Е.Н., Горденко М.К. Применение методов системного анализа к оцениванию работы учебных ассистентов. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 251-270 Применение методов системного анализа к Аннотация. В этой статье представлены результаты применения различных методов системного анализа (CATWOE, Rich Picture, AHP, Fuzzy AHP) для оценки учебных ассистентов для преподавателей. В статье рассмотрено применение soft- и hard- методов системного анализа. Методы системного анализа рассматриваются на примере реализации программы «Учебный ассистент» в Национальном Исследовательском Университете «Высшая школа экономики» (НИУ ВШЭ) на дисциплине «Дискретная математика». В статье показан процесс взаимодействия преподавателей с учениками и преподавателями в форме Rich Picture. Определены связующие мероприятия, встречи и даже отчеты, которые предоставляют ассистенты преподавателю. Затем описано каким образом были выбраны критерии для оценки ассистентов и оценена важность каждого критерия. Были определены три группы критериев: профессиональные навыки, навыки общения, личные качества. Каждая группа имеет некоторые подкритерий, которые были определены посредством уточняющих встреч и мозгового штурма. Также в работе был определен собственный метод оценки, который явился пререквизитом для AHP и позволивший сразу же отбросить наиболее неперспективных ассистентов. Кроме того, рассматривается применение методов AHP и Fuzzy AHP типа 2 для оценки учебных ассистентов. Выявлены сильные и слабые стороны каждого метода. Также показано, что, несмотря на мощь методов системного анализа, необходимо использовать здравый смысл и логику. Нельзя полагаться только на числа, полученные методами системного анализа, необходимо затем производить анализ результатов. В процессе работы выбирается лучший учебный ассистент, и определяется группа лучших учебных ассистентов. Ключевые слова: системный анализ; комбинация soft и hard методов; многокритериальное принятие решений; AHP; Fuzzy Type-2 AHP; нечеткие множества. Для цитирования: Береснева Е.Н., Горденко М.К. Применение методов системного анализа к оцениванию работы учебных ассистентов. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 251-270 (на английском языке). DOI: 10.15514/ISPRAS-2018-30(3)-18 References [1] J. Lopa. Using Undergraduate Students as Teaching Assistants. The Professional & Organizational Development Network in Higher Education, vol. 21, 2009, pp. 50-62. [2] M. Asad, S. Kermani and H. Hora. A Proposed Framework for Evaluating Student’s Performance and Selecting the Top Students in E-Learning System, Using Fuzzy AHP Method. In Proceedings of the International Conference on Management, Economics and Humanities, Istanbul-Turkey, 2015. [3] T. Saaty. There is no mathematical validity for using fuzzy number crunching in the analytic hierarchy process. Journal of Systems Science and Systems Engineering, vol. 15, no. 4, 2006, pp. 457-464. [4] Monk and S. Howard. The Rich Picture: A Tool for Reasoning about Work Context. [Online]. Available: http://www-users.york.ac.uk/~am1/RichPicture.pdf, accessed 12.05.2018. [5] Changing Minds. CATWOE. [Online]. Available: http://creatingminds.org/tools/catwoe, accessed 12.05.2018. [6] M. Velasquez and P. Hester. An Analysis of Multi-Criteria Decision Making Methods. International Journal of Operations Research, vol. 10, no. 2, 2013, pp. 56-66. [7] T. Saaty. The Analytic Hierarchy Process, New York: McGraw Hill, 1980. [8] C. Boender and J. De Graan. Multicriteria Decision Analysis with Fuzzy Pairwise Comparisons. Fuzzy Sets and Systems, vol. 29, 1989, pp. 133-143. [9] D. Chang. Applications of the Extent Analysis Method on Fuzzy-AHP. European Journal of Operational Research, vol. 95, 1996, pp. 649-655. 267 Beresneva E., Gordenko M. Applying the methods of system analysis to teaching assistants’ evaluation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 251-270 [10] J. Mendel and R. John. Type-2 fuzzy sets made simple. IEEE Transactions on Fuzzy Systems, vol. 10, no. 2, 2002. pp. 117-127. [11] C. Kahraman and B. Öztayşi. Fuzzy analytic hierarchy process with interval type-2 fuzzy sets. Knowledge-Based Systems, vol. 59, 2014, pp. 48-57. [12] E. Mu and M. Pereyra-Rojas. An Introduction to the Analytic Hierarchy Process (AHP). Using Super Decisions, vol. 2, New York: Springer, 2016. Применение методов системного анализа к оцениванию работы учебных ассистентов Е.Н. Береснева <eberesneva@hse.ru> М.К. Горденко <mgordenko@hse.ru> Департамент программной инженерии, Национальный исследовательский университет “Высшая школа экономики”, 101000, Россия, г. Москва, ул. Мясницкая, д. 20 Применение методов системного анализа к оцениванию работы учебных ассистентов Е.Н. Береснева <eberesneva@hse.ru> М.К. Горденко <mgordenko@hse.ru> Департамент программной инженерии, Национальный исследовательский университет “Высшая школа экономики”, 101000, Россия, г. Москва, ул. Мясницкая, д. 20 DOI: 10.15514/ISPRAS-2018-30(3)-18 Для цитирования: Береснева Е.Н., Горденко М.К. Применение методов системного анализа к оцениванию работы учебных ассистентов. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 251-270 (на английском языке). DOI: 10.15514/ISPRAS-2018-30(3)-18 268 Береснева Е.Н., Горденко М.К. Применение методов системного анализа к оцениванию работы учебных ассистентов. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 251-270 Beresneva E., Gordenko M. Applying the methods of system analysis to teaching assistants’ evaluation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 251-270 Список литературы [1] J. Lopa. Using Undergraduate Students as Teaching Assistants. The Professional & Organizational Development Network in Higher Education, vol. 21, 2009, pp. 50-62. [2] M. Asad, S. Kermani and H. Hora. A Proposed Framework for Evaluating Student’s Performance and Selecting the Top Students in E-Learning System, Using Fuzzy AHP Method. In Proceedings of the International Conference on Management, Economics and Humanities, Istanbul-Turkey, 2015. [3] T. Saaty. There is no mathematical validity for using fuzzy number crunching in the analytic hierarchy process. Journal of Systems Science and Systems Engineering, vol. 15, no. 4, 2006, pp. 457-464. [4] Monk and S. Howard. The Rich Picture: A Tool for Reasoning about Work Context. [Online]. Available: http://www-users.york.ac.uk/~am1/RichPicture.pdf, accessed 12.05.2018. [5] Changing Minds. CATWOE. [Online]. Available: http://creatingminds.org/tools/catwoe, accessed 12.05.2018. [6] M. Velasquez and P. Hester. An Analysis of Multi-Criteria Decision Making Methods. International Journal of Operations Research, vol. 10, no. 2, 2013, pp. 56-66. [7] T. Saaty. The Analytic Hierarchy Process, New York: McGraw Hill, 1980. [8] C. Boender and J. De Graan. Multicriteria Decision Analysis with Fuzzy Pairwise Comparisons. Fuzzy Sets and Systems, vol. 29, 1989, pp. 133-143. [9] D. Chang. Applications of the Extent Analysis Method on Fuzzy-AHP. European Journal of Operational Research, vol. 95, 1996, pp. 649-655. [10] J. Mendel and R. John. Type-2 fuzzy sets made simple. IEEE Transactions on Fuzzy Systems, vol. 10, no. 2, 2002. pp. 117-127. [11] C. Kahraman and B. Öztayşi. Fuzzy analytic hierarchy process with interval type-2 fuzzy sets. Knowledge-Based Systems, vol. 59, 2014, pp. 48-57. [12] E. Mu and M. Pereyra-Rojas. An Introduction to the Analytic Hierarchy Process (AHP). Using Super Decisions, vol. 2, New York: Springer, 2016. 269 Beresneva E., Gordenko M. Applying the methods of system analysis to teaching assistants’ evaluation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 251-270 Beresneva E., Gordenko M. Applying the methods of system analysis to teaching assistants’ evaluation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018. pp. 251-270 270 DOI: 10.15514/ISPRAS-2018-30(3)-19 For citation: Ilyin D.V., Fokina N.Yu., Semenov V.A. Static dependency analysis for semantic data validation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 271-284. DOI: 10.15514/ISPRAS-2018-30(3)-19 For citation: Ilyin D.V., Fokina N.Yu., Semenov V.A. Static dependency analysis for semantic data validation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 271-284. DOI: 10.15514/ISPRAS-2018-30(3)-19 Keywords: information systems; ACID; data consistency management; EXPRESS Keywords: information systems; ACID; data consistency management; EXPRESS Static dependency analysis for semantic data validation D.V. Ilyin <denis.ilyin@ispras.ru> N.Yu. Fokina <nfokina@ispras.ru> V.A. Semenov <sem@ispras.ru> Ivannikov Institute for Systems Programming of the RAS, 25, Alexander Solzhenitsyn Str., Moscow, 109004, Russia Abstract. Modern information systems manipulate data models containing millions of items, and the tendency is to make these models even more complex. One of the most crucial aspects of modern concurrent engineering environments is their reliability. The principles of ACID (atomicity, consistency, isolation, durability) are aimed at providing it, but directly following them leads to serious performance drawbacks on large-scale models, since it is necessary to control the correctness of every performed transaction. In the paper, a method for incremental validation of object-oriented data is presented. Assuming that a submitted transaction is applied to originally consistent data, it is guaranteed that the final data representation is also consistent if only the spot rules are satisfied. To identify data items subject to spot rule validation, a bipartite data-rule dependency graph is formed. To automatically build the dependency graph a static analysis of the model specifications is proposed to apply. In the case of complex object-oriented models defining hundreds and thousands of data types and semantic rules, the static analysis seems to be the only way to realize the incremental validation and to make possible to manage the data in accordance with the ACID principles. 1. Introduction Management of semantically complex data is one of the challenging problems tightly connected with emerging information systems such as concurrent engineering environments and product data management systems [1-4]. Although transactional guarantees ACID (Atomicity, Consistency, Isolation, and Durability) are widely recognized and recommended for any information system, it is difficult to maintain the consistency and integrity of data driven by complex object-oriented 271 Ilyin D.V., Fokina N.Yu., Semenov V.A. Static dependency analysis for semantic data validation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 271-284 Ilyin D.V., Fokina N.Yu., Semenov V.A. Static dependency analysis for semantic data validation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 271-284 models. Often such models are specified in EXPRESS language being part of the STEP standard on industrial automation systems and integration (ISO 10303). To be unambiguously interpretable by different systems the data must satisfy numerous semantic rules imposed by formal models. Maintaining data consistency and ensuring system interoperability become a serious computational problem. Full semantic validation requires extremely high costs, often exceeding the processing time of individual transactions. Periodic validation is possible, but at a high risk of violating rules and losing actual data. The paper presents an effective method for incremental validation of object-oriented data. An idea of incremental checks is well-understood and was successfully implemented for the validation of such specific data as UML charts, XML documents, deductive databases [5-7]. Unlike the aforementioned results, the proposed method can be applied to semantically complex data driven by arbitrary object-oriented models. Assuming that a submitted transaction is applied to originally consistent data, it is guaranteed that the final data representation is also consistent if only the spot rules are satisfied. To identify data items subject to spot rule validation, a bipartite data- rule dependency graph is formed. To automatically build the dependency graph a static analysis of the model specifications is proposed to apply. In the case of large- scale models defining hundreds and thousands of data types and semantic rules, static analysis seems to be the only way to realize the incremental validation and to make possible to effectively manage the data in accordance with the ACID principles. The structure of the paper is as follows. In section 2, we will shortly overview EXPRESS language with an emphasis on the data types and the rule categories admitted by the language. 1. Introduction Formal definitions of model-driven data, rules and transactions are also provided. In section 3, we will present a complete validation routine and then explain how an incremental validation can be arranged using the proposed dependency graph. This is accompanied by an example of the model specification. In conclusion, we summarise benefits of the proposed validation method and outdraw future efforts. 2.1 EXPRESS language Product data models and, particularly, semantic rules can be specified formally in EXPRESS (ISO 2004) language [8]. This object-oriented modeling language provides a wide range of declarative and imperative constructs to define both data types and constraints imposed upon them. The supported data types can be subdivided into the following groups: simple types (character, string, integer, float, double, Boolean, logical, binary), aggregate types (set, multi-set, sequence, array), selects, enumerations, and entity types. 272 Ильин Д.В., Фокина Н.Ю., Семенов В.А. Статический анализ зависимостей для семантической валидации данных. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 271-284 Ильин Д.В., Фокина Н.Ю., Семенов В.А. Статический анализ зависимостей для семантической валидации данных. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 271-284 Depending on the definition context, three basic sorts of constraints are distinguished in the modeling language: rules for simple user-defined data types, local rules for object types, and global rules for object type extents. Depending on the evaluation context these imply the following semantic checks: Depending on the definition context, three basic sorts of constraints are distinguished in the modeling language: rules for simple user-defined data types, local rules for object types, and global rules for object type extents. Depending on the evaluation context these imply the following semantic checks:  attribute type compliance (𝑅0);  limited widths of strings and binaries (𝑅1, 𝑅2);  size of aggregates (𝑅3);  multiplicity of direct and inverse associations in objects (𝑅4, 𝑅5);  uniqueness of elements in sets, unique lists and arrays (𝑅6);  mandatory attributes in objects (𝑅7);  mandatory elements in aggregates excluding sparse arrays (𝑅8);  value domains for primitive data types (𝑅9);  value domains restricting and interrelating the states of separate attributes within objects (𝑅10 or so-called local rules);  uniqueness of attribute values (optionally, their groups) on object type extents (𝑅11 or uniqueness rules);  value domains restricting and interrelating the states of whole object populations (𝑅12 or so-called global rules). Value domains can be specified in a general algebraic form by means of all the variety of imperative constructs available in the language (control statements, functions, procedures, etc.). Certainly, each product model defines own data types and rules. Therefore, semantic validation methods and tools should be developed in a model-driven paradigm allowing their application for any data whose model is formally specified in EXPRESS language. 2.1 EXPRESS language For a more detailed description refer to the mentioned above standard family which regulates the language. 2.2 Formalization of models, data and transactions 𝑎 if ∀𝑜∈{𝑜}, 𝑜. 𝑎⊂{𝑜∗}, ∀𝑜∗∈{𝑜∗} →∃𝑜∈{𝑜}: 𝑜∗∈𝑜. 𝑎. We will denote that as {𝑜} 𝑐.𝑎 → {𝑜∗}. Let a dataset 𝑥 is driven by the model 𝑀〈𝑇, ≺, 𝑅〉. An object set {𝑜∗} ⊂𝑥, 𝑡𝑦𝑝𝑒𝑜𝑓(𝑜∗) = 𝑐∗∈𝐶⊂𝑇 is said to be interlinked with the objects {𝑜} ⊂𝑥, 𝑡𝑦𝑝𝑒𝑜𝑓(𝑜) = 𝑐∈𝐶⊂𝑇 along the route {𝑐. 𝑎} if ∃ {𝑜′} ⊂𝑥, {𝑜′′} ⊂𝑥, …, so that {𝑜} 𝑐.𝑎 → {𝑜′} 𝑐′.𝑎′ → {𝑜′′} 𝑐′′.𝑎′′ → … →{𝑜∗}. A query returning the objects {𝑜∗} interlinked with a given set {𝑜} along the route {𝑐. 𝑎} is called the route query and is designated as 𝑄𝑟𝑜𝑢𝑡𝑒(𝑥, {𝑜}, {𝑐. 𝑎}). A query returning the objects {𝑜} by a given object set {𝑜∗} is called the reverse route query and is designated as 𝑄𝑟𝑜𝑢𝑡𝑒(𝑥, {𝑜∗}, 𝑟𝑒𝑣 {𝑐. 𝑎}). The object set 𝑥= {𝑜1, 𝑜2, … } driven by the model 𝑀〈𝑇, ≺, 𝑅〉 is called consistent if all the rules being instantiated and evaluated are satisfied on this data set: ∀ 𝑟∈ 𝑅→𝑟(𝑥) 𝑡𝑟𝑢𝑒 ∀𝑜∈{𝑜} →∃𝑜∈{𝑜}: 𝑜∈𝑜. 𝑎. We will denote that as {𝑜} {𝑜}. Let a dataset 𝑥 is driven by the model 𝑀〈𝑇, ≺, 𝑅〉. An object set {𝑜∗} ⊂𝑥, 𝑡𝑦𝑝𝑒𝑜𝑓(𝑜∗) = 𝑐∗∈𝐶⊂𝑇 is said to be interlinked with the objects {𝑜} ⊂𝑥, 𝑡𝑦𝑝𝑒𝑜𝑓(𝑜) = 𝑐∈𝐶⊂𝑇 along the route {𝑐. 𝑎} if ∃ {𝑜′} ⊂𝑥, {𝑜′′} ⊂𝑥, …, so that {𝑜} 𝑐.𝑎 → {𝑜′} 𝑐′.𝑎′ → {𝑜′′} 𝑐′′.𝑎′′ → … →{𝑜∗}. A query returning the objects {𝑜∗} interlinked with a given set {𝑜} along the route {𝑐. 𝑎} is called the route query and is designated as 𝑄𝑟𝑜𝑢𝑡𝑒(𝑥, {𝑜}, {𝑐. 𝑎}). A query returning the objects {𝑜} by a given object set {𝑜∗} is called the reverse route query and is designated as 𝑄𝑟𝑜𝑢𝑡𝑒(𝑥, {𝑜∗}, 𝑟𝑒𝑣 {𝑐. 𝑎}). The object set 𝑥= {𝑜1, 𝑜2, … } driven by the model 𝑀〈𝑇, ≺, 𝑅〉 is called consistent if all the rules being instantiated and evaluated are satisfied on this data set: ∀𝑟∈ The object set 𝑥= {𝑜1, 𝑜2, … } driven by the model 𝑀〈𝑇, ≺, 𝑅〉 is called consistent if all the rules being instantiated and evaluated are satisfied on this data set: ∀ 𝑟∈ 𝑅→𝑟(𝑥) = 𝑡𝑟𝑢𝑒. Finally, let us introduce the concept of the delta as a specific representation of transactions. 2.2 Formalization of models, data and transactions An object-oriented data model 𝑀 can be formally considered as a triple 𝑀= 〈𝑇∪ ≺ ∪𝑅〉, where the types 𝑇= {𝐶∪𝑆∪𝐴∪… } are classes 𝐶, simple types 𝑆, aggregates 𝐴 and other constructed structures allowed by EXPRESS. Generalization/specialization relations ≺ are defined among these types. Each class 𝑐∈𝐶 defines a set of attributes in the form 𝑐. 𝑎: 𝐶↦𝑇. The attributes 𝑐. 𝑎: 𝐶↦𝐶, 𝑐. 𝑎: 𝐶↦𝑎𝑔𝑔𝑟𝑒𝑔𝑎𝑡𝑒(𝐶) are single and multiple associations which play role of object references. The rules 𝑅= {𝑅0 ∪𝑅1 ∪𝑅2 ∪… ∪𝑅12} define the value domains of typed data in an algebraic way in accordance with EXPRESS. The rules are subdivided into 12 categories enumerated above. Let us define the key concepts that are used in further consideration. 273 Ilyin D.V., Fokina N.Yu., Semenov V.A. Static dependency analysis for semantic data validation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 271-284 Ilyin D.V., Fokina N.Yu., Semenov V.A. Static dependency analysis for semantic data validation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 271-284 N.Yu., Semenov V.A. Static dependency analysis for semantic data validation. Trudy ISP RAN/Proc. sue 3, 2018, pp. 271-284 An object-oriented dataset 𝑥= {𝑜1, 𝑜2, … } is said to be driven by the model 𝑀〈𝑇, ≺, 𝑅〉 if all the objects belong to its classes: ∀ 𝑜∈𝑥→𝑡𝑦𝑝𝑒𝑜𝑓(𝑜) ∈𝐶⊂𝑇. An object-oriented dataset 𝑥= {𝑜1, 𝑜2, … } is said to be driven by the model 𝑀〈𝑇, ≺, 𝑅〉 if all the objects belong to its classes: ∀ 𝑜∈𝑥→𝑡𝑦𝑝𝑒𝑜𝑓(𝑜) ∈𝐶⊂𝑇. Let a dataset 𝑥 is driven by the model 𝑀〈𝑇, ≺, 𝑅〉. All the objects {𝑜∗} ⊂𝑥 such that 𝑠𝑢𝑏𝑡𝑦𝑝𝑒𝑜𝑓(𝑜∗) = 𝑐∈𝐶⊂𝑇 are called an extent of the class 𝑐 on the dataset 𝑥. A query returning the class extent 𝑐 on the dataset 𝑥 is called the extent query and is designated as 𝑄𝑒𝑥𝑡𝑒𝑛𝑡(𝑥, 𝑐). Let a dataset 𝑥 is driven by the model 𝑀〈𝑇, ≺, 𝑅〉. An object set {𝑜∗} ⊂𝑥, 𝑡𝑦𝑝𝑒𝑜𝑓(𝑜∗) = 𝑐∗∈𝐶⊂𝑇 is said to be interlinked with the objects {𝑜} ⊂𝑥, 𝑡𝑦𝑝𝑒𝑜𝑓(𝑜) = 𝑐∈𝐶⊂𝑇 along the association 𝑐. 𝑎 if ∀𝑜∈{𝑜}, 𝑜. 𝑎⊂{𝑜∗}, ∀𝑜∗∈{𝑜∗} →∃𝑜∈{𝑜}: 𝑜∗∈𝑜. 𝑎. We will denote that as {𝑜} 𝑐.𝑎 → {𝑜∗}. Let a dataset 𝑥 is driven by the model 𝑀〈𝑇, ≺, 𝑅〉. An object set {𝑜∗} ⊂𝑥, 𝑡𝑦𝑝𝑒𝑜𝑓(𝑜∗) = 𝑐∗∈𝐶⊂𝑇 is said to be interlinked with the objects {𝑜} ⊂𝑥, 𝑡𝑦𝑝𝑒𝑜𝑓(𝑜) = 𝑐∈𝐶⊂𝑇 along the association 𝑐. Fig. 1. Complete validation routine As mentioned above, complete validation of semantically complex product data is a computationally costly task that can cause performance degradation when processing transactions. Incremental validation makes it possible to reduce the amount of checks to be performed. 2.2 Formalization of models, data and transactions Each delta Δ(𝑥′, 𝑥) aggregates the changes happened in the dataset 𝑥′ = {𝑜1 ′, 𝑜2 ′, … } compared with its original revision 𝑥= {𝑜1, 𝑜2, … }. It is assumed that both revisions are driven by the same model and the objects have unique identifiers that allows to uniquely map the objects and to compute delta in a formal way as Δ(𝑥′, 𝑥) = 𝑑𝑒𝑙𝑡𝑎(𝑥′, 𝑥). The delta can be arranged as bidirectional one and then any of the revisions can be restored by the given other: 𝑥′ = 𝑎𝑝𝑝𝑙𝑦(𝑥, Δ) and 𝑥= 𝑎𝑝𝑝𝑙𝑦(𝑥′, Δ−1). The delta is represented as a set of elementary and compound changes Δ = {𝛿}, where each change can be either the creation of an object, or its deletion or modification designated as 𝛿𝑛𝑒𝑤(𝑜), 𝛿𝑑𝑒𝑙(𝑜), 𝛿𝑚𝑜𝑑(𝑜) correspondingly. The modification, in turn, is represented as a change in the attributes 𝛿𝑚𝑜𝑑(𝑜) = {𝛿𝑚𝑜𝑑(𝑜.𝑎)} that in the case of aggregates is represented by the operations of insertion, removal and modification of the elements 𝛿𝑚𝑜𝑑(𝑜.𝑎) = {𝛿𝑖𝑛𝑠(𝑜.𝑎[]), 𝛿𝑟𝑒𝑚(𝑜.𝑎[]), 𝛿𝑚𝑜𝑑(𝑜.𝑎[])}. In what follows, we assume that each creation operation in the delta representation is complemented by the operations of initializing the attributes that are equivalent to the modification operations. Each deletion operation is supplemented by the operations of resetting the attributes to an undefined state, also representable by the modification operations. Regardless of the way, the delta is structured, only elementary operations are taken into account in the context of the studied validation problems. 274 Ильин Д.В., Фокина Н.Ю., Семенов В.А. Статический анализ зависимостей для семантической валидации данных. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 271-284 Ильин Д.В., Фокина Н.Ю., Семенов В.А. Статический анализ зависимостей для семантической валидации данных. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 271-284 3.1 Complete validation for each object o ∈ x in dataset for each attribute o.a in object for each attribute rule ∈ R0 U R1 U R2 U … U R9 defined for typeof( o.a ) check rule(o.a), log if violated for each local rule ∈ R10 defined for typeof( o ) check rule( o ), log if violated for each class c ∈ C defined in model for each uniqueness rule ∈ R11 defined for class c check rule( Q_extent( x, rule.c ) ), log if violated for each global rule ∈ R12 defined in model check rule( Q_extent( x, rule.c1 ), Q_extent( x, rule.c2 ),… ), log if violated for each object o ∈ x in dataset for each attribute o.a in object for each attribute rule ∈ R0 U R1 U R2 U … U R9 defined for typeof( o.a ) check rule(o.a), log if violated for each local rule ∈ R10 defined for typeof( o ) check rule( o ), log if violated for each class c ∈ C defined in model for each uniqueness rule ∈ R11 defined for class c check rule( Q_extent( x, rule.c ) ), log if violated for each global rule ∈ R12 defined in model check rule( Q_extent( x, rule.c1 ), Q_extent( x, rule.c2 ),… ), log if violated for each object o ∈ x in dataset for each attribute o.a in object for each attribute rule ∈ R0 U R1 U R2 U … U R9 defined for typeof( o.a ) check rule(o.a), log if violated for each local rule ∈ R10 defined for typeof( o ) check rule( o ), log if violated for each class c ∈ C defined in model for each uniqueness rule ∈ R11 defined for class c check rule( Q_extent( x, rule.c ) ), log if violated for each global rule ∈ R12 defined in model check rule( Q_extent( x, rule.c1 ), Q_extent( x, rule.c2 ),… ), log if violated 3.1 Complete validation The complete validation routine is provided below (see Figure 1). In a cycle on all objects their attributes are checked against the rules of the categories 𝑅1 ∪𝑅2 ∪… ∪ 𝑅9. The checks are performed individually for each attribute provided that the corresponding rules are imposed on their types. In case of detected violations, the error messages are logged. Rules 𝑅10 are evaluated for entire objects in the same loop. The second cycle is formed due to the need for checks of uniqueness rules 𝑅11. Since these rules are declared inside the class definitions, an additional cycle is arranged on the model classes. The rules are evaluated on the class extents. Finally, the third cycle allows to check global rules 𝑅12 which are defined directly in the model. Such checks are performed for the corresponding class extents. 3.2 Incremental validation The proposed incremental validation method is based on the idea of localizing spot rules that can be affected by a transaction and generating a set of semantic checks that is sufficient to detect all potential violations. For this purpose, the dependency graph is built by a given specification of the data model in EXPRESS language. For brevity, we just explain that this structure represents and omit the details of how it can be formed using static analysis of the specification. The dependency graph is a bipartite graph whose nodes represent the kinds of transaction operations and the categories of semantic rules both defined by the underlying model. An operation node is connected with the rule nodes by directed edges if only such operations can violate the rules being instantiated for particular data. Usually, the semantics of the operations imply what are the data it is applied to. Sometimes the inspected data are apriori unknown and have to be determined by 275 Ilyin D.V., Fokina N.Yu., Semenov V.A. Static dependency analysis for semantic data validation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 271-284 Ilyin D.V., Fokina N.Yu., Semenov V.A. Static dependency analysis for semantic data validation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 271-284 executing corresponding route queries. Therefore, each edge is formed by the dependency structure 𝜎 containing both a rule reference 𝜎. 𝑟𝑢𝑙𝑒 and an optional query route 𝜎. 𝑟𝑜𝑢𝑡𝑒. In some sense, the graph reflects the transaction structure as if it contains all possible kinds of changes and the data organisation as if all data types are present and all rules are potentially suffered to violations. As mentioned above, only elementary operations are involved in the dependency analysis. executing corresponding route queries. Therefore, each edge is formed by the dependency structure 𝜎 containing both a rule reference 𝜎. 𝑟𝑢𝑙𝑒 and an optional query route 𝜎. 𝑟𝑜𝑢𝑡𝑒. In some sense, the graph reflects the transaction structure as if it contains all possible kinds of changes and the data organisation as if all data types are present and all rules are potentially suffered to violations. As mentioned above, only elementary operations are involved in the dependency analysis. Thus, the dependency graph enables to determine spot rules that could be violated for particular data due to the accepted transaction. For example, if the node operation is a modification of the object attribute 𝑐. for each elementary operation δ(o),δ(o.a) ∈ delta { σ } = dependency_graph( kindof( δ ) ) for each dependency σ ∈ { σ } switch kindof( σ.rule ) case attribute_rule : check σ.rule( o.a ), log if violated case local_rule : Fig. 2. Incremental validation routine The validation routine presented in Figure 2 consists in the sequential traversing of delta operations, determining the nodes of the operation semantics, obtaining associated spot rule nodes, evaluating the rules directly or filling the checkset for the subsequent validation. The checkset is organized as an indexed set of records each of which stores references on the validated rule, query and factual data to perform the corresponding check. The use of the checkset is motivated by the fact that some operations lead to repeated checks of the same rules. Indexing of the checkset allows you to exclude repeated records and, thus, to avoid redundant computations. At the same time, the attribute rule checks are always produced once by the modification operations and, therefore, it is more expedient to execute them immediately, without overloading the checkset. Ильин Д.В., Фокина Н.Ю., Семенов В.А. Статический анализ зависимостей для семантической валидации данных. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 271-284 Ильин Д.В., Фокина Н.Ю., Семенов В.А. Статический анализ зависимостей для семантической валидации данных. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 271-284 данных. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 271-284 { o } = Query_route( x, o, rev (σ.route) ) for each o* ∈ { o* } checkset.put( σ.rule( o* ) ) case uniqueness_rule : checkset.put( σ ) case global_rule : checkset.put( σ ) for each check σ, σ(o) ∈ checkset switch kindof( σ.rule ) case local_rule : check σ.rule( o ), log if violated case uniqueness_rule : check σ.rule( Query_extent( x, σ.rule.c ) ), log if violated case global_rule : check σ.rule( Query_extent( x, σ.rule.c1 ), Query_extent( x, σ.rule.c2 ),… ), log if violated 3.2 Incremental validation 𝑎 and a rule 𝑟 ∈𝑅0 ∪ 𝑅1 ∪ 𝑅2 ∪… ∪𝑅9 is defined for its type, then the node 𝛿𝑚𝑜𝑑(𝑐.𝑎) is connected with the rule node 𝑟 by a corresponding edge. Having a specific operation of this kind 𝛿𝑚𝑜𝑑(𝑜.𝑎), 𝑡𝑦𝑝𝑒𝑜𝑓(𝑜) = 𝑐 in the delta representation the corresponding check 𝑟(𝑜. 𝑎) can be produced using the dependency edge. The method of the dependency graph construction is described in more detail in the next section. Still, here we will point out some of its important features. If the same attribute 𝑐. 𝑎 participates in an expression of the domain rule 𝑟∈𝑅10 for the class 𝑐, then the operation 𝛿𝑚𝑜𝑑(𝑜.𝑎), 𝑡𝑦𝑝𝑒𝑜𝑓(𝑜) = 𝑐 produces the check 𝑟(𝑜) for the object 𝑜. If the attribute 𝑐. 𝑎 participates in the uniqueness rule 𝑟∈𝑅11 defined for the class 𝑐, then another dependency edge must be associated with the operation node. In this case, the corresponding check 𝑟(𝑄𝑒𝑥𝑡𝑒𝑛𝑡(𝑥, 𝑐)) must be performed. There is a more difficult case when the attribute 𝑐. 𝑎 participates in an expression of the domain rule 𝑟∈𝑅10 defined for the other class 𝑐∗. The attribute 𝑐. 𝑎 is assumed to be accessed by traversing associated objects along the route {𝑐∗. 𝑎∗} from the objects 𝑜∗∈𝑐∗. Then the operation 𝛿𝑚𝑜𝑑(𝑜.𝑎), 𝑡𝑦𝑝𝑒𝑜𝑓(𝑜) = 𝑐 induces the checks 𝑟(𝑜∗) for all 𝑜∗∈𝑄𝑟𝑜𝑢𝑡𝑒(𝑥, 𝑜, 𝑟𝑒𝑣 {𝑐∗. 𝑎∗}). To identify and perform such checks the operation node must be connected with the evaluated rule node and a route {𝑐∗. 𝑎∗} must be prescribed to the edge. The dependency analysis of spot rules 𝑟∈𝑅12 is carried out in a similar way. Finally, we note that the operations of creating and deleting objects on the assumptions made above can only violate global rules and only in those cases if the cardinalities of class extents are computed. Considering object references as specific attribute types, it is possible to localize some spot rules more exactly. Differing operations on aggregates also leads to better localization of spot rules. For brevity we omit the details how the spot rules can be localized more carefully and provide an example in the next subsection. for each elementary operation δ(o),δ(o.a) ∈ delta { σ } = dependency_graph( kindof( δ ) ) for each dependency σ ∈ { σ } switch kindof( σ.rule ) case attribute_rule : check σ.rule( o.a ), log if violated case local_rule : 276 3.3 Dependency graph construction To construct the dependency graph, an abstract syntactic tree for the model is built. According to the retrieved data, for all attribute declarations operation nodes are built. Number and types of these nodes constructed for a single attribute depend on its type. For non-aggregate attributes 𝑐. 𝑎 only node 𝛿𝑚𝑜𝑑(𝑐. 𝑎), representing modification of the attribute, is built. For aggregate attributes 𝑐. 𝑎[] three nodes are created: (1) 𝛿𝑖𝑛𝑠(𝑐. 𝑎[]) – insertion of a new element; (2) 𝛿𝑚𝑜𝑑(𝑐. 𝑎[]) – modification of an element of the aggregate; (3) 𝛿𝑟𝑒𝑚(𝑐. 𝑎[]) – removal of an element. Construction of the dependency graph proceeds with generating rule nodes. We handle construction of nodes for rules R1-R9 and R10-R12 differently. Construction of the dependency graph proceeds with generating rule nodes. We handle construction of nodes for rules R1-R9 and R10-R12 differently. handle construction of nodes for rules R1-R9 and R10-R12 differently. For rules R1-R9 we take all explicit attributes and build rule nodes for each of them. The types of rule nodes depend on the type of the attribute in question. For instance, if it is a bounded string c.S, we generate a R1(c.S) (R1 – limited width of strings), connected with the node corresponding to the modification of S 𝛿𝑚𝑜𝑑(𝑐. 𝑆). Similarly, if an attribute is a bounded aggregate, we construct a node of type R4 and 277 Ilyin D.V., Fokina N.Yu., Semenov V.A. Static dependency analysis for semantic data validation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 271-284 Ilyin D.V., Fokina N.Yu., Semenov V.A. Static dependency analysis for semantic data validation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 271-284 connect it with the insertion 𝛿𝑖𝑛𝑠(𝑐. 𝑎[]) and/or removal 𝛿𝑟𝑒𝑚(𝑐. 𝑎[]) operation nodes of the attribute, depending on the side from which the aggregate is bounded – if it is bounded above, then only with insertion node, if below – with removal, if from both sides – with both of them. connect it with the insertion 𝛿𝑖𝑛𝑠(𝑐. 𝑎[]) and/or removal 𝛿𝑟𝑒𝑚(𝑐. 𝑎[]) operation nodes of the attribute, depending on the side from which the aggregate is bounded – if it is bounded above, then only with insertion node, if below – with removal, if from both sides – with both of them. The way of construction of rule nodes for R10-R12 is uniform. 3.3 Dependency graph construction We start with locating all local rules for R10, all uniqueness rules for R11 and all global rules for R12. For each of the rules, we find all attributes used in it. If an attribute is explicit, we only connect its modification with the rule node, and also with insertion and removal, if it is an aggregate used inside a SIZEOF operation. If an attribute is derived, we take its definition and find the attributes used in it; if inverse – we proceed with analyzing the attribute it references. For derived and explicit attributes, the analysis is performed recursively, until all the explicit attributes, directly and indirectly referenced by them, are located. Then all of them are connected with the rule node corresponding to the rule in question. If the during the analysis we find a node that is a function call, we substitute its formal parameters with actual and thus locate the attributes which are used in it; the analysis of a function body with the parameters substituted is completely identical to the analysis of a rule. An example illustrating the constructed graph is given in the next subsecti An example illustrating the constructed graph is given in the next subsection. FUNCTION TaskIsCyclic (T1 : Task, T2 : Task) : BOOLEAN; IF (SIZEOF(T1.Parent) = 0) THEN RETURN(FALSE); ELSE 3.4 Example of a dependency graph Let us consider a fragment of the EXPRESS specification of a project management system. Three classes depicted in Figure 3 – Task, Link and Calendar – are its core entities. The meaning of Task is self-evident; Link represents a connection defining a relation and execution order between two tasks. The fact that between two tasks might be only a single link of one type is reflected in uniqueness rule ur1. A Calendar defines a typical working pattern: working days, working times, holidays. The calendar can be assigned to specific tasks, and one calendar can be set as a default project calendar, that means that it will be used for tasks for which no task calendar is set. Besides that, it is possible to use an Elapsed calendar for a task implying that work will be performed 24/7. Global rule SingleProjectCalendar restricts the possible number of project calendars to no more than one. Moreover, local rule wr3 is used to check that if a task has got a task calendar, it the reference to it must be non-null. One more local rule, wr2, restricts the length of an EntityName to be between 1 and 32 characters. Let us consider a fragment of the EXPRESS specification of a project management system. Three classes depicted in Figure 3 – Task, Link and Calendar – are its core entities. The meaning of Task is self-evident; Link represents a connection defining a relation and execution order between two tasks. The fact that between two tasks might be only a single link of one type is reflected in uniqueness rule ur1. A Calendar defines a typical working pattern: working days, working times, holidays. TYPE LinkEnum = ENUMERATION OF TYPE LinkEnum = ENUMERATION OF (START_START, START_FINISH, FINISH_START, FINISH_FINISH); END_TYPE; TYPE LinkEnum = ENUMERATION OF (START_START, START_FINISH, FINISH_START, FINISH_FINISH); END_TYPE; TYPE CalendarRuleEnum = ENUMERATION OF (TASK, PROJECT, ELAPSED); END_TYPE; TYPE CalendarRuleEnum = ENUMERATION OF (TASK, PROJECT, ELAPSED); END_TYPE; FUNCTION TaskIsCyclic (T1 : Task, T2 : Task) : BOOLEAN; IF (SIZEOF(T1.Parent) = 0) THEN RETURN(FALSE); ELSE 278 Ильин Д.В., Фокина Н.Ю., Семенов В.А. Статический анализ зависимостей для семантической валидации данных. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 271-284 IF ((TaskIsCyclic(T1.Parent[1], T2) = TRUE) OR (T1 = T2)) THEN RETURN(TRUE) ENTITY Task; Name : EntityName; TaskCalendar : Calendar; CalendarRule : CalendarRuleEnum; Children : LIST [0:?] OF Task; DERIVE Children : LIST [0:?] OF Task; TaskDuration : Duration := ?; INVERSE TaskDuration : Duration := ?; INVERSE Parent : SET [0:1] OF Task FOR Children; DownstreamLinks : SET [0:?] OF Link FOR Predecessor; DownstreamLinks : SET [0:?] OF Link FOR Pr UpstreamLinks : SET [0:?] OF Link FOR Successor; WHERE wr3 : CalendarRule <> CalendarRuleEnum.TASK OR wr3 : CalendarRule <> CalendarRuleEnum.TASK OR EXISTS(TaskCalendar); wr3 : CalendarRule <> CalendarRuleEnum.TASK OR EXISTS(TaskCalendar); wr4 : (SIZEOF(Parent) = 0) OR (TaskIsCyclic(Parent[1 FALSE); UNIQUE ur1 : ID; END_ENTITY; LinkType : LinkEnum; Predecessor : Task; Successor : Task; UNIQUE Q ur2 : LinkType AND Predecessor.ID AND Successor.ID; 3 ID ur3 : ID; END_ENTITY; Ильин Д.В., Фокина Н.Ю., Семенов В.А. Статический анализ зависимостей для семантической валидации данных. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 271-284 RULE SingleProjectCalendar FOR (Calendar); WHERE wr1: SIZEOF(QUERY(Temp <* Calendar \| Temp.isProjectCalendar = TRUE)) <= 1; END_RULE; TYPE EntityName = STRING; WHERE wr2: (1 <= SELF) AND (SELF <= 32); END_TYPE; TYPE EntityName = STRING; wr2: (1 <= SELF) AND (SELF <= 32); END_TYPE; ENTITY Task; 4. Conclusion This paper presents the incremental method of model data validation. The method is applicable for semantically complex data driven by arbitrary object-oriented models. It allows to increase the performance of semantic validation and to effectively manage the data in accordance with the ACID principles. The planned work concerns basically the implementation of the method proposed and its evaluation for industry meaningful product data. The expected positive results will allow its wide introduction into new software engineering technologies and emerging information systems. This paper presents the incremental method of model data validation. The method is applicable for semantically complex data driven by arbitrary object-oriented models. It allows to increase the performance of semantic validation and to effectively manage the data in accordance with the ACID principles. The planned work concerns basically the implementation of the method proposed and its evaluation for industry meaningful product data. The expected positive results will allow its wide introduction into new software engineering technologies and emerging information systems. ENTITY Calendar; The corresponding edges are assigned by the routes by traversing of which the attributes could be accessed. The expression of the global rule SingleProjectCalendar references only one attribute IsProjectCalendar, so the appropriate graph nodes are connected by the edge as well. Modification of any attribute of the Link class can affect its uniqueness defined by ur2; hence the connections between LinkType, Predecessor and Successor and the uniqueness rule node. removed from Children. Therefore, the corresponding operation nodes should be connected with the aforementioned nodes of the rules that the operations may potentially violate. As the expression for the local rule wr3 includes the attributes CalendarRule and TaskCalendar, the nodes corresponding to the operations of modification of these attributes are connected with the wr3 rule node. For the rule wr2 defining the value range of the EntityName type, there is a connection between the EntityName modification node and the wr2 rule node. The corresponding edges are assigned by the routes by traversing of which the attributes could be accessed. The expression of the global rule SingleProjectCalendar references only one attribute IsProjectCalendar, so the appropriate graph nodes are connected by the edge as well. Modification of any attribute of the Link class can affect its uniqueness defined by ur2; hence the connections between LinkType, Predecessor and Successor and the uniqueness rule node. It is also possible that a change affects a constraint not directly but through an inverse association, or even a chain of them, where other classes can be involved. In this case, rules for all the chain of affected classes is added to the checkset. Furthermore, they can be affected not only by direct associations but also by the inverse. For instance, cardinality constraints on inverse aggregate attributes causes insertion of additional rule nodes to the graph. ENTITY Calendar; ENTITY Calendar; ENTITY Calendar; ID : INTEGER; Name : OPTIONAL EntityName; IsProjectCalendar : BOOLEAN; UNIQUE ur4 : ID; END_ENTITY; ID : INTEGER; Name : OPTIONAL EntityName; IsProjectCalendar : BOOLEAN; UNIQUE ur4 : ID; END_ENTITY; Fig. 3. An example of the model specification in EXPRESS language 279 Ilyin D.V., Fokina N.Yu., Semenov V.A. Static dependency analysis for semantic data validation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 271-284 Ilyin D.V., Fokina N.Yu., Semenov V.A. Static dependency analysis for semantic data validation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 271-284 cy graph for this fragment of the specification is shown in Figure 4. p y g p g p g δins(Children[]) δmod(Children[]) δmod(Name) δmod(LinkType) δmod(Predecessor) δmod(Successor) δmod(TaskCalendar) δmod(CalendarRule) δmod(IsProjectCalendar) R0(Children) R8(Children) R10(wr2) R12(SingleProjectCalendar) δrem(Children[]) R4(Children) R5(Parent) R11(ur2) R10(wr3) R0(Name) R7(Name) R0(TaskCalendar) R0(LinkType) R7(CalendarRule) R0(CalendarRule) R7(LinkType) R0(Predecessor) R7(Predecessor) R0(Successor) R7(Successor) R0(IsProjectCalendar) R7(IsProjectCalendar) R10(wr4) δmod(Task.ID) R11(Task.ID) δmod(Link.ID) R11(Link.ID) δmod(Calendar.ID) R11(Calendar.ID) Fig. 4. A fragment of the model dependency graph δmod(IsProjectCalendar) Fig. 4. A fragment of the model dependency graph 280 Each operation of attribute modification except for removal of elements from the list of task children is connected with the rules validating corresponding attribute type compliance R0 and availability of defined values for mandatory attributes R7. To avoid placement of null values to the list of mandatory elements the rule R8 should be validated as well after the operations have been performed. The insertion cannot violate multiplicity of the direct and inverse associations as their upper borders are unlimited, but checks R4, R5 should be performed when an element is Ильин Д.В., Фокина Н.Ю., Семенов В.А. Статический анализ зависимостей для семантической валидации данных. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 271-284 Ильин Д.В., Фокина Н.Ю., Семенов В.А. Статический анализ зависимостей для семантической валидации данных. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 271-284 removed from Children. Therefore, the corresponding operation nodes should be connected with the aforementioned nodes of the rules that the operations may potentially violate. As the expression for the local rule wr3 includes the attributes CalendarRule and TaskCalendar, the nodes corresponding to the operations of modification of these attributes are connected with the wr3 rule node. For the rule wr2 defining the value range of the EntityName type, there is a connection between the EntityName modification node and the wr2 rule node. Ключевые слова: информационные системы; ACID; управление целостностью данных; EXPRESS DOI: 10.15514/ISPRAS-2018-30(3)-19 References [1]. V.A. Semenov. Product Data Management with Solid Transactional Guarantees, In Transdisciplinary Engineering: A Paradigm Shift Series Advances in Transdisciplinary Engineering, IOS Press, 2017, pp. 592-599. [2]. L. Lämmer and M. Theiss. Product Lifecycle Management, In Concurrent Engineering in the 21st Century – Foundations, Developments and Challenges, Springer, 2015, pp. 455-490. [3]. J. Osborn. Survey of concurrent engineering environments and the application of best practices towards the development of a multiple industry, multiple domain environment. Clemson University, 2009. Accessed: 29/01/2018. Available: http://tigerprints.clemson.edu/all_theses/635/ [4]. M. Philpotts. An introduction to the concepts, benefits and terminology of product data management, Industrial Management & Data Systems, MCB University Press, vol. 96, no. 4, 1996, pp. 11–17. 281 Ilyin D.V., Fokina N.Yu., Semenov V.A. Static dependency analysis for semantic data validation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 271-284 Ilyin D.V., Fokina N.Yu., Semenov V.A. Static dependency analysis for semantic data validation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 271-284 [5]. X. Blanc, A. Mougenot, I. Mounier, T. Mens. Incremental Detection of Model Inconsistencies based on Model Operations. In Advanced Information Systems Engineering, CAiSE 2009, LNCS, vol. 5565, Springer, 2009, pp. 32-46. [6]. C. Xu, C.S. Cheung, W.K. Chan. Incremental Consistency Checking for Pervasive Context. In Proc. the 28th International Conference on Software Engineering, 2006, pp. 292-301. [7]. J. Harrison, S.W. Dietrich. Towards an Incremental Condition Evaluation Strategy for Active Deductive Databases. In Research and Practical Issues in Databases, World Scientific, 1992, pp. 81-95. pp [8]. ISO 10303-11: 2004. Industrial automation systems and integration – Product data representation and exchange – Part 11: Description methods: The EXPRESS language reference manual, ISO, 2004. Статический анализ зависимостей для семантической валидации данных Ильин Д.В. <denis.ilyin@ispras.ru> Фокина Н.Ю. <nfokina@ispras.ru> Семенов В.А. <sem@ispras.ru> Институт системного программирования им. В.П. Иванникова РАН, 109004, Россия, г. Москва, ул. А. Солженицына, д. 25 Статический анализ зависимостей для семантической валидации данных Семенов В.А. <sem@ispras.ru> Институт системного программирования им. В.П. Иванникова 109004, Россия, г. Москва, ул. А. Солженицына, д. 25 Аннотация. Современные информационные системы манипулируют моделями данных, содержащими миллионы объектов, и тенденция такова, что эти модели постоянно усложняются. Одним из важнейших аспектов современных параллельных инженерных сред является их надежность. Принципы ACID (атомарность, согласованность, изолированность, устойчивость) направлены на ее обеспечение, однако прямое следование им приводит к серьезному снижению производительности на крупномасштабных моделях, поскольку необходимо контролировать правильность каждой выполненной транзакции. В настоящей статье представлен метод инкрементальной валидации объектно-ориентированных данных. Предполагая, что транзакция применяется к первоначально согласованным данным, гарантируется, что окончательное представление данных также будет согласованным, если только будут выполнены локальные правила. Для определения объектов данных, подлежащих проверке, формируется двудольный граф зависимостей по данным. Для автоматического построения графа зависимостей предлагается применять статический анализ спецификаций модели. В случае сложных объектно-ориентированных моделей, включающих сотни и тысячи типов данных и семантических правил, статический анализ, по-видимому, является единственным способом реализации инкрементальной валидации и обеспечения возможности управления данными в соответствии с принципами ACID. Ключевые слова: информационные системы; ACID; управление целостностью данных; EXPRESS DOI: 10.15514/ISPRAS-2018-30(3)-19 282 Ильин Д.В., Фокина Н.Ю., Семенов В.А. Статический анализ зависимостей для семантической валидации данных. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 271-284 Для цитирования: Ильин Д.В., Фокина Н.Ю., Семенов В.А. Статический анализ зависимостей для семантической валидации данных. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 271-284 (на английском языке). DOI: 10.15514/ISPRAS-2018-30(3)-19 Ilyin D.V., Fokina N.Yu., Semenov V.A. Static dependency analysis for semantic data validation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 271-284 1This work is supported by the Basic Research Program at the National Research University Higher School of Economics and Russian Foundation for Basic Research, project No. 16-01-00546. DOI: 10.15514/ISPRAS-2018-30(3)-20 For citation: Nesterov R.A., Mitsyuk A.A., Lomazova I.A. Simulating Behavior of Multi- Agent Systems with Acyclic Interactions of Agents. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 285-302. DOI: 10.15514/ISPRAS-2018-30(3)-20 For citation: Nesterov R.A., Mitsyuk A.A., Lomazova I.A. Simulating Behavior of Multi- Agent Systems with Acyclic Interactions of Agents. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 285-302. DOI: 10.15514/ISPRAS-2018-30(3)-20 Simulating Behavior of Multi-Agent Systems with Acyclic Interactions of Agents1 1, 2 R.A. Nesterov <rnesterov@hse.ru, r.nesterov@campus.unimib.it> 1 A.A. Mitsyuk <amitsyuk@hse.ru> 1 I.A. Lomazova <ilomazova@hse.ru> 1 National Research University Higher School of Economics, 20, Myasnitskaya st., Moscow, 101000, Russia 2 Dipartimento di Informatica, Sistemistica e Communicazione, Università degli Studi di Milano-Bicocca, Viale Sarca 336 – Edificio U14, I-20126 Milano, Italia , 2 R.A. Nesterov <rnesterov@hse.ru, r.nesterov@campus.unimib.it> 1 A.A. Mitsyuk <amitsyuk@hse.ru> 1 I.A. Lomazova <ilomazova@hse.ru> 1 National Research University Higher School of Economics, 20, Myasnitskaya st., Moscow, 101000, Russia 2 Dipartimento di Informatica, Sistemistica e Communicazione, Università degli Studi di Milano-Bicocca, Viale Sarca 336 – Edificio U14, I-20126 Milano, Italia Abstract. In this paper, we present an approach to model and simulate models of multi-agent systems (MAS) using Petri nets. A MAS is modeled as a set of workflow nets. The agent-to- agent interactions are described by means of an interface. It is a logical formula over atomic interaction constraints specifying the order of inner agent actions. Our study considers positive and negative interaction rules. In this work, we study interfaces describing acyclic agent interactions. We propose an algorithm for simulating the MAS with respect to a given interface. The algorithm is implemented as a ProM 6 plug-in that allows one to generate a set of event logs. We suggest our approach to be used for evaluating process discovery techniques against the quality of obtained models since this research area is on the rise. The proposed approach can be used for process discovery algorithms concerning internal agent interactions of the MAS. Keywords: Petri nets; multi-agent systems; interaction; interface; simulation; eve Список литературы [1]. V.A. Semenov. Product Data Management with Solid Transactional Guarantees, In Transdisciplinary Engineering: A Paradigm Shift Series Advances in Transdisciplinary Engineering, IOS Press, 2017, pp. 592-599. [2]. L. Lämmer and M. Theiss. Product Lifecycle Management, In Concurrent Engineering in the 21st Century – Foundations, Developments and Challenges, Springer, 2015, pp. 455-490. [3]. J. Osborn. Survey of concurrent engineering environments and the application of best practices towards the development of a multiple industry, multiple domain environment. Clemson University, 2009. Дата обращения: 29/01/2018. Режим доступа: http://tigerprints.clemson.edu/all_theses/635/ [4]. M. Philpotts. An introduction to the concepts, benefits and terminology of product data management, Industrial Management & Data Systems, MCB University Press, vol. 96, no. 4, 1996, pp. 11–17. [5]. X. Blanc, A. Mougenot, I. Mounier, T. Mens. Incremental Detection of Model Inconsistencies based on Model Operations. In Advanced Information Systems Engineering, CAiSE 2009, LNCS, vol. 5565, Springer, 2009, pp. 32-46. [6]. C. Xu, C.S. Cheung, W.K. Chan. Incremental Consistency Checking for Pervasive Context. In Proc. the 28th International Conference on Software Engineering, 2006, pp. 292-301. [7]. J. Harrison, S.W. Dietrich. Towards an Incremental Condition Evaluation Strategy for Active Deductive Databases. In Research and Practical Issues in Databases, World Scientific, 1992, pp. 81-95. [8]. ISO 10303-11: 2004. Industrial automation systems and integration – Product data representation and exchange – Part 11: Description methods: The EXPRESS language reference manual, ISO, 2004. 283 Ilyin D.V., Fokina N.Yu., Semenov V.A. Static dependency analysis for semantic data validation. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 271-284 284 Simulating Behavior of Multi-Agent Systems with Acyclic Interactions of Agents1 1, 2 R.A. Nesterov <rnesterov@hse.ru, r.nesterov@campus.unimib.it> 1 A.A. Mitsyuk <amitsyuk@hse.ru> 1 I.A. Lomazova <ilomazova@hse.ru> 1 National Research University Higher School of Economics, 20, Myasnitskaya st., Moscow, 101000, Russia 2 Dipartimento di Informatica, Sistemistica e Communicazione, Università degli Studi di Milano-Bicocca, Viale Sarca 336 – Edificio U14, I-20126 Milano, Italia 1. Introduction We model individual agents using workflow nets, whereas interfaces are specified using special formulae. They are constructed using a declarative formalism that we introduce to describe basic asynchronous interactions between agents. Based on agent models and a declarative interface formula our generator derives the operational semantics that describes a MAS behavior. We show that both of MAS representations are equivalent, i.e. they have the same set of possible model runs. Thus, this semantics can be used to simulate the model and generate event logs. In this paper, we describe a new event log generator that aims at preparing artificial event logs for MAS’s. We model individual agents using workflow nets, whereas interfaces are specified using special formulae. They are constructed using a declarative formalism that we introduce to describe basic asynchronous interactions between agents. Based on agent models and a declarative interface formula our generator derives the operational semantics that describes a MAS behavior. We show that both of MAS representations are equivalent, i.e. they have the same set of possible model runs. Thus, this semantics can be used to simulate the model and generate event logs. The main contributions of this paper are:  a formalism for a declarative description of the requirements for agent interactions is defined;  the operational semantics representing the behavior of a multi-agent system with declarative requirements for interactions of agents is defined;  an algorithm for generating event logs from given agent models and declarative constraints on their interactions based on the operational semantics is developed;  the approach is implemented as a prototype software and evaluated.  the approach is implemented as a prototype software and evaluated. This paper is structured as follows. The next section gives an overview of existing approaches for generating event logs and simulating process models. Section 3 introduces main notions used in the paper. In Section 4, we describe our approach to modeling multi-agent systems with the help of Petri nets. Implementation details are discussed in Section 5, and Section 6 concludes the paper.  the approach is implemented as a prototype software and evaluated. This paper is structured as follows. The next section gives an overview of existing approaches for generating event logs and simulating process models. Section 3 introduces main notions used in the paper. In Section 4, we describe our approach to modeling multi-agent systems with the help of Petri nets. 1. Introduction Process discovery has been actively developed over recent years [1]. Many algorithms for the automatic model synthesis from event logs have been proposed [2]–[7]. They produce process models in different notations. These can be Petri nets 285 1This work is supported by the Basic Research Program at the National Research University Higher School of Economics and Russian Foundation for Basic Research, project No. 16-01-00546. 285 Nesterov R.A., Mitsyuk A.A., Lomazova I.A. Simulating Behavior of Multi-Agent Systems with Acyclic Interactions of Agents. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 285-302 Nesterov R.A., Mitsyuk A.A., Lomazova I.A. Simulating Behavior of Multi-Agent Systems with Acyclic Interactions of Agents. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 285-302 [3], [6], [7], fuzzy models [2], heuristics nets [4] or BPMN models [5] and many others (see [8] for the comprehensive review of process discovery algorithms). [3], [6], [7], fuzzy models [2], heuristics nets [4] or BPMN models [5] and many others (see [8] for the comprehensive review of process discovery algorithms). Discovering process models from event logs helps to use information about users and system runtime behavior for proper specification, design, and maintenance of software systems [9], [10]. This topic is increasingly attracting the attention of researchers [11]–[14]. In particular, application of process mining techniques to distributed and multi-agent software systems [15], [16] is interesting and important. The main drawback of most algorithms is that they are not appropriate for modeling highly concurrent systems. In particular, these are multi-agent systems (MAS). Such a system consists of multiple agents executing their work independently and interacting via predefined interfaces. It makes sense to use compositional approaches to model MAS’s. Fortunately, such approaches have been proposed over recent years [17], [18]. The overwhelming majority of process discovery algorithms employ different heuristics. That is why testing is used to evaluate their efficiency and validity [8]. It is performed using real-life and artificially generated event logs with suitable characteristics. The latter are prepared using event log generators. The overwhelming majority of process discovery algorithms employ different heuristics. That is why testing is used to evaluate their efficiency and validity [8]. It is performed using real-life and artificially generated event logs with suitable characteristics. The latter are prepared using event log generators. In this paper, we describe a new event log generator that aims at preparing artificial event logs for MAS’s. 2. Related Work Process Logs Generator PLG2 [19] is one of the most popular tools for generating well-structured process models represented by dependency graphs. The tool constructs models using randomly generated context-free grammars. The user should specify desired characteristics of models: a size, a number of choices, hierarchy blocks etc. The obtained model can be used to generate an event log. Another tool that aims at randomized event log generation is PT and Log Generator [20]. It generates random process trees (well-structured models) containing desired number of specified workflow patterns. In particular, generated models can be constructed from sequences, AND/XOR/OR splits and joins, as well as structured loops. The algorithm can also randomly insert elements representing activities. The tool also generates the desired number of logs from automatically constructed models. The problem of the randomized process model generation has also been addressed by Yan, Dijkman, and Grefen in [21]. However, they have not considered event log generation within the context of their approach. The main goal of the tools discussed above is the randomized testing using sets of models and event logs. However, in some cases there is a need to generate event logs from specific process models that have been prepared on the basis of the real data or expert knowledge. If this is the case, one can use the tool GENA [22]. It aims at generating sets of event logs from a Petri net model. The approach allows users to use preferences to influence a control-flow and to artificially introduce a randomized noise into an event log. The improved version of GENA can generate event logs from BPMN 2.0 models [23]. Most basic BPMN constructs are supported: tasks, gateways, messages, pools, lanes, data objects. Colored Petri nets can be used to generate event logs [24]. Authors have developed the extension for CPN Tools that can generate randomized event logs based on a given colored Petri net. The main drawback of this approach is that it implies writing Standard ML scripts, which leads to possible problems during tool adaptation for a specific task. Moreover, this approach and GENA do not support multi-agent systems with independent asynchronous agents. Declarative process models might also be used to generate event logs [25]. This approach is based on construction of a finite automaton using a Declare process model. The tool can generate a specified number of strings accepted by this automaton. Нестеров Р.А., Мицюк А.А., Ломазова И.А. Симуляция поведения мультиагентных систем с ациклически взаимодействующими агентами. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 285-302 Нестеров Р.А., Мицюк А.А., Ломазова И.А. Симуляция поведения мультиагентных систем с ациклически взаимодействующими агентами. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 285-302 Нестеров Р.А., Мицюк А.А., Ломазова И.А. Симуляция поведения мультиагентных систем с ациклически взаимодействующими агентами. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 285-302 1. Introduction Implementation details are discussed in Section 5, and Section 6 concludes the paper. 286 2. Related Work Strings are generated using the automaton and its randomized execution. Afterwards, each string is transformed into a log trace with necessary attributes. This tool is useful, when the only information about the process is the set of constraints. This approach is also not appropriate for the MAS simulation as we suggest, because it does not support the imperative control-flow description of individual agents. In this paper, we propose an extension to the GENA tool that is supposed to be used for generating event logs by simulating MAS models, because the tools described above cannot fully support this feature. 287 Nesterov R.A., Mitsyuk A.A., Lomazova I.A. Simulating Behavior of Multi-Agent Systems with Acyclic Interactions of Agents. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 285-302 3. Preliminaries Let ℕ denote the set of all non-negative integers, A+ – the set of all finite non-empty sequences over a set A, and A* = A+∪{𝜀}, where 𝜀 is the empty sequence. For a subset B ⊂ A, the projection of 𝜎 ∈ A* on a set B, denoted 𝜎\|B, is the subsequence of 𝜎 including all elements belonging to B. 4. Modeling Multi-Agent Systems In this section, we present formalism for modeling multi-agent systems consisting of several asynchronously interacting agents. A model for a system of k agents will consist of k WF-nets N1, N2,…, Nk, representing behavior of individual agents (called agent nets), and constraints on their asynchronous interaction ℐ (called interface). We assume that transitions of agent nets have individual labels. In other words, different agents implement different activities. We also assume that agent interactions are acyclic, namely, activities in interaction constraints do not belong to cycles and therefore occur in each system run not more than once. We assume that transitions of agent nets have individual labels. In other words, different agents implement different activities. We also assume that agent interactions are acyclic, namely, activities in interaction constraints do not belong to cycles and therefore occur in each system run not more than once. Interfaces are defined as positive logical formulae over atomic constraints. Let us give the exact definitions. Interfaces are defined as positive logical formulae over atomic constraints. Let us give the exact definitions. Let N1, N2,…, Nk be agent nets with pairwise disjoint sets of activity names 𝜆1(T1), 𝜆2(T2),…, 𝜆k(Tk) respectively. We define two types of atomic constraints, namely A ◁ B and A ◁¯ B, where A and B are activity names from two different sets, i.e. A ∈ 𝜆i(Ti), B ∈ 𝜆j(Tj) and i≠j. ty of atomic constraints for a given trace 𝜎 over the set of activity names ∪𝜆(T )∪ ∪𝜆(T ) i d fi d f ll The validity of atomic constraints for a given trace 𝜎 over the set of activity names 𝒜 = 𝜆1(T1)∪𝜆2(T2)∪…∪𝜆k(Tk) is defined as follows: = 𝜆1(T1)∪𝜆2(T2)∪…∪𝜆k(Tk) is defined as follows:  𝜎 ⊨ A ◁ B ⇔ if B occurs in 𝜎, then A occurs before B;  𝜎 ⊨ A ◁ B ⇔ if B occurs in 𝜎, then A occurs before B;  𝜎 ⊨ A ◁¯ B ⇔ if A does not occur before B in 𝜎.  𝜎 ⊨ A ◁¯ B ⇔ if A does not occur before B in 𝜎. When 𝜎 ⊨ 𝜙, we say that 𝜙 is valid for 𝜎, and 𝜎 satisfies 𝜙. The validity of the atomic constraints has a natural interpretation. The constraint A ◁ B means that B should be always preceded by A, e.g. 3.1 Petri Nets Симуляция поведения мультиагентных систем с ациклически взаимодействующими агентами. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 285-302 Workflow nets (WF-nets) form a subclass of Petri nets used for business process modeling. A Petri net is a triple N = (P, T, F, m0) is a WF-net iff:  there is a single source place i and a single source place f, s.t. ˙i = f ˙ = ∅;  each node in P∪T lies on a path from i to f  there is a single source place i and a single source place f, s.t. ˙i = f ˙ = ∅;  there is a single source place i and a single source place f, s.t. ˙i = f ˙ = ∅;  each node in P∪T lies on a path from i to f.  each node in P∪T lies on a path from i to f.  each node in P∪T lies on a path from i to f. The initial marking m0 of a WF-net contains exactly one token in its source place i.  each node in P∪T lies on a path from i to f. The initial marking m0 of a WF-net contains exactly one token in its source place i. The initial marking m0 of a WF-net contains exactly one token in its source place i. 3.2 Event Logs A multiset over a set A is a map B: A → ℕ. The set of all multisets over A is denoted by ℬ(A). Let 𝒜 be a set of activity names. A trace 𝜎 over 𝒜 is defined as a finite non-empty sequence over 𝒜, i.e. 𝜎 ∈ 𝒜+. An event log L over 𝒜 is a finite multiset of traces, i.e. L ∈ ℬ(𝒜+). Let 𝒜 be a set of activity names. A trace 𝜎 over 𝒜 is defined as a finite non-empty sequence over 𝒜, i.e. 𝜎 ∈ 𝒜+. An event log L over 𝒜 is a finite multiset of traces, i.e. L ∈ ℬ(𝒜+). 3.1 Petri Nets A Petri net is a triple N = (P, T, F), where P and T are two disjoint sets of places and transitions, and F ⊆ (P×T)∪(T×P) is a flow relation. Pictorially, places are shown by circles, transitions – by boxes, whereas the flow relation is depicted using directed arcs (see Fig. 1 for an example). We suppose that transitions of a Petri net are labeled with activity names from 𝒜∪{𝜏}, where 𝒜 is a set of visible activity names, and 𝜏 is a label for an invisible action. Labels are assigned to transitions via a labeling function 𝜆: T → 𝒜∪{𝜏}. A marking (state) of a Petri net N is a function m: P → ℕ assigning numbers to places. A marking m is designated by putting m(p) black dots into each place p. By m0 we denote the initial marking. Let X = P ∪T. For x ∈ X, ˙x = {y ∈ X \| (y, x) ∈ F} is the set of input nodes of x in N, and x˙ = {y ∈ X \| (x, y) ∈ F} is the set of its output nodes. Fig. 1. A Petri net Fig. 1. A Petri net A marking m enables a transition t ∈ T iff there is at least one token in all places which are input for t. An enabled transition may fire yielding a new marking mʹ (denoted m[t⟩mʹ), consuming one token from each of its input places and producing a token into each of its output places (see Fig. 1b). A sequence w = t1t2...tn over T is a firing sequence iff m0[t1⟩m1[t2⟩…mn-1[tn⟩mn (denoted m0[w⟩mn). Let w = t1t2...tn be a firing sequence of the net N, 𝜆 – a labeling function over a set of activity names 𝒜. Define 𝜆(w) = 𝜆(t1)𝜆(t2)…𝜆(tn). Then 𝜆(w)\|𝒜 is called an (observable) run in N. A marking m is reachable iff ∃w ∈ T: m0[w⟩m. A reachable marking is called dead if it does not enable any transition. 288 Нестеров Р.А., Мицюк А.А., Ломазова И.А. Симуляция поведения мультиагентных систем с ациклически взаимодействующими агентами. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 285-302 Workflow nets (WF-nets) form a subclass of Petri nets used for business process modeling. A Petri net is a triple N = (P, T, F, m0) is a WF-net iff: Нестеров Р.А., Мицюк А.А., Ломазова И.А. 4. Modeling Multi-Agent Systems a message can be received only if it has already been sent. Thus, A ◁ B is valid for a trace 𝜎 = …A…B… and is not valid for a trace 𝜎 = …<except A>…B… The constraint A ◁¯ B means that B cannot occur if A has happened before, e.g. if a message has 𝜎 = …A…B… and is not valid for a trace 𝜎 = …<except A>…B… The constraint A ◁¯ B means that B cannot occur if A has happened before, e.g. if a message has 289 Nesterov R.A., Mitsyuk A.A., Lomazova I.A. Simulating Behavior of Multi-Agent Systems with Acyclic Interactions of Agents. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 285-302 been already sent by mail, we should not fax it again. A trace 𝜎 = …<except A>…B… satisfies this constraint, and a trace 𝜎 = …A…B… does not satisfy it. However, these atomic constraints are not negations of each other. Both A ◁ B and A ◁¯ B are valid for a trace that does not contain B. been already sent by mail, we should not fax it again. A trace 𝜎 = …<except A>…B… satisfies this constraint, and a trace 𝜎 = …A…B… does not satisfy it. However, these atomic constraints are not negations of each other. Both A ◁ B and A ◁¯ B are valid for a trace that does not contain B. nguage of interface constraints is defined by the following grammar rules: Atom ::= A ◁ B \| A ◁¯ B, Atom ::= A ◁ B \| A ◁ B, 𝜙 ::= Atom \| 𝜙 ∨ 𝜙 \| 𝜙 ∧ 𝜙, \| , 𝜙 ::= Atom \| 𝜙 ∨ 𝜙 \| 𝜙 ∧ 𝜙, where Atom is an atomic constraint, and 𝜙 is a constraint formula. , 𝜙 Validity of a constraint formula 𝜑 for a given trace 𝜎 is defined in a standard way: Validity of a constraint formula 𝜑 for a given trace 𝜎 is defined in a stand 𝜎 ⊨ 𝜙1 ∧ 𝜙2 ⇔ 𝜎 ⊨ 𝜙1 and 𝜎 ⊨ 𝜙2, 𝜎 ⊨ 𝜙1 ∧ 𝜙2 ⇔ 𝜎 ⊨ 𝜙1 and 𝜎 ⊨ 𝜙2, 𝜎 ⊨ 𝜙1 ∨ 𝜙2 ⇔ 𝜎 ⊨ 𝜙1 or 𝜎 ⊨ 𝜙2. 𝜎 ⊨ 𝜙1 ∨ 𝜙2 ⇔ 𝜎 ⊨ 𝜙1 or 𝜎 ⊨ 𝜙2. 4. Modeling Multi-Agent Systems Let L be an event log over a set 𝒜 of activity names, and 𝜙 be a constraint formula, then 𝜙 is valid for L iff 𝜙 is valid for each trace in L. Interface formulae allow us to express different useful interaction constraints, e.g. the formula 𝜙 = A ◁¯ B ∧ B ◁¯ A describes a conflict between A and B, i.e. A and B cannot occur in the same trace. Recall that a MAS model consists of k agent nets N1, N2,…, Nk, where Ni = (Pi, Ti, Fi, m0 i, 𝜆i), and a constraint formula ℐ (interface) with atomic constraints that defines the relations on activities of different agents. It is easy to see that the union of Petri nets (considering several disjoint graphs as one disconnected graph) is also a Petri net. Thus, we can consider k agent nets as a single Petri net N. Recall that a run for a Petri net N is a sequence of activity names, corresponding to a firing sequence of N, and a trace from the related event log. Then a run of a MAS model S = (N1, N2,…, Nk, ℐ) is defined as a run 𝜌 in N satisfying ℐ, i.e. 𝜌 ⊨ ℐ. y g 𝜌 he following proposition is the immediate consequence of the definitions. y g The following proposition is the immediate consequence of the definitions. Proposition 1: Let S = (N1, N2,…, Nk, ℐ) be a MAS model, and 𝜌 be a run in S. Then for all i the projection 𝜌\|𝜆i(Ti) on transition labels of an agent net Ni is a run in Ni. Fig. 2. A multi-agent system with two interacting systems Fig. 2. A multi-agent system with two interacting systems 290 Нестеров Р.А., Мицюк А.А., Ломазова И.А. Симуляция поведения мультиагентных систем с ациклически взаимодействующими агентами. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 285-302 Consider as an example the system shown in Fig. 2 with ℐ = A ◁¯ B ∧ B ◁¯ A meaning that A conflicts with B. Consider a run 𝜎 = x1By2x3 satisfying ℐ. Projecting 𝜎 on agent nets gives traces x1x3 and By2, which are runs of the corresponding agent nets. This property will be further used for designing the simulation algorithm presented in the next section. 5. Simulating MAS Process Models In this section, we describe an algorithm for simulating MAS models. It has been implemented as a ProM 6 plug-in extending GENA tool [22]. Step 2. Start with the initial marking m0 and 𝜀 for the current run 𝜎. 5.1 An Interface-Driven Firing Rule A constraint formula in a MAS model defines declarative restrictions on the model behavior. To simulate the model behavior, we need to define operational semantics for MAS models based on a special firing rule for selecting and executing the next step in the run of the model. We call this rule an interface-driven firing rule to distinguish it from the standard Petri net firing rule. Naturally, this rule should be consistent with the declarative definitions of MAS model behavior. Let S = (N, ℐ) be a MAS model, where a Petri net N = (P, T, F, m0, 𝜆) is a union of all agent nets. Firstly, we convert ℐ to a disjunctive normal form (DNF) using standard logical laws. Then, an interface ℐ = ∨Cj for j = 1, 2,…, n, where Cj = ∧ Sl, and Sl is an atomic constraint for l = 1, 2,…, m. By abuse of notation, we also denote by ℐ the set of its conjuncts, and by Cj – the set of atomic constraints in a conjunct Cj. Obviously, a trace 𝜎 satisfies ℐ iff ∃Cj ∈ ℐ: 𝜎 ⊨ Cj, i.e. it should satisfy at least one conjunct in ℐ. Thus, to generate a model run, we choose a conjunct Cj and fire transitions of N only if they do not violate Cj. Then we define Tℐ ⊆ T to be the set of transitions involved in agent interaction, i.e. t ∈ Tℐ iff 𝜆(t) occurs in ℐ. We call transitions from Tℐ interface transitions. Independent transitions from T\Tℐ fire according to the standard firing rule for Petri nets. The firing of interface transitions is restricted by the constraint formula. To check whether firing of a transition t violates Cj, we keep the current historical model run, i.e. a sequence of already fired activities. When a transition t ∈ Tℐ is enabled according to the standard Petri net firing rule at a current marking m, and an atomic constraint A ◁ 𝜆(t) occurs in Cj, then t is defined to be enabled only if A occurs in the current run. Similarly, if A ◁¯ 𝜆(t) occurs in Cj, then t is enabled only if A does not occur in the current run. Otherwise, a transition t is enabled in the model, when it is enabled in N. Step 1. Choose nondeterministically a conjunct C in ℐ. 5.1 An Interface-Driven Firing Rule Now the operational semantics of a MAS model S = (N, ℐ), where N = (P, T, F, m0, 𝜆) and ℐ = ∨Cj for j = 1, 2,…, n, is defined by the following procedure. Now the operational semantics of a MAS model S = (N, ℐ), where N = (P, T, F, m0, 𝜆) and ℐ = ∨Cj for j = 1, 2,…, n, is defined by the following procedure. Step 1. Choose nondeterministically a conjunct C in ℐ. Step 2. Start with the initial marking m0 and 𝜀 for the current run 𝜎. 291 Nesterov R.A., Mitsyuk A.A., Lomazova I.A. Simulating Behavior of Multi-Agent Systems with Acyclic Interactions of Agents. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 285-302 Nesterov R.A., Mitsyuk A.A., Lomazova I.A. Simulating Behavior of Multi-Agent Systems with Acyclic Interactions of Agents. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 285-302 Step 3. For a current marking m and a current run 𝜎 repeat while there are enabled transitions in N: Step 3. For a current marking m and a current run 𝜎 repeat while there are enabled transitions in N: Step 3. For a current marking m and a current run 𝜎 repeat while there are enabled transitions in N: 1) compute the set Tok of all transitions enabled at m and not violating constraints from C w.r.t. 𝜎; 1) compute the set Tok of all transitions enabled at m and not violating constraints from C w.r.t. 𝜎; 2) choose nondeterministically a transition t from Tok; 3) fire t by changing the current marking to mʹ, m[t⟩mʹ, and adding 𝜆(t) to 𝜎. 5.2 Event Log Generation This subsection presents an algorithm for generating an event log by simulating behavior of a MAS model. Let S = (N, ℐ) be a MAS model, where N = (P, T, F, m0, 𝜆) is a Petri net, and ℐ is in DNF. Firstly, for each conjunct C occurring in ℐ, we run (simulate) S to check if it is possible to obtain a trace 𝜎 satisfying C. If we cannot obtain such a trace, we exclude this conjunct. As a result, we come to a set of conjuncts ℐʹ ⊆ ℐ, which can be actually satisfied by traces of S or an empty set if ℐ cannot be satisfied by traces of S. If ℐʹ = ∅, then the simulation is terminated producing an empty event log L. That is why we can simulate S w.r.t. conjuncts occurring in ℐʹ only. Starting a new iteration of simulation, we randomly choose a conjunct from ℐʹ and fire transitions of N according to the interface-driven firing rule. The end user specifies the final marking mf, which is actually the set of sink places of agent nets. Apart from that, the log generation is regulated by the number of logs, the number of traces in a log, and by the maximum number of steps which can be executed while generating a single trace (denoted further by maxSteps). Algorithm 1 is used for generating a single trace that satisfies C from ℐʹ. 292 Algorithm 1. Single trace generation Input: N = (P, T, F, m0, 𝜆), ℐʹ, and mf Output: a trace 𝜎, s.t. 𝜎 ⊨ ℐʹ 𝜎 ⃪ 𝜀; m ⃪ m0; i ⃪ 1; C ⃪ pickRandomConjunct(ℐʹ) while (i ≤ maxSteps) ∧ (m ≠ mf) do Tok ⃪ findEnabledTransitions(N, m, C, 𝜎) if Tok ≠ ∅ then t ⃪ pickRandomTransition(Tok) m ⃪ fireTransition(N, m, t) if 𝜆(t) ≠ 𝜏 then 𝜎 ⃪ 𝜎 + 𝜆(t); i ⃪ i + 1 end else 𝜎 ⃪ 𝜀; break end end Algorithm 2 is used for finding enabled transitions, which do not violate constraints of C. Firstly, we find a set of transitions enabled at a reachable marking m according 292 Algorithm 1. Single trace generation Input: N = (P, T, F, m0, 𝜆), ℐʹ, and mf Output: a trace 𝜎, s.t. 2 ProM 6 Framework page: http://www.promtools.org 3 Fluxicon Disco page: https://fluxicon.com/disco/ 5.2 Event Log Generation C Tm ⃪ stEnabledTransitions(N, m) Tok ⃪ Tm\Tℐ foreach t ∈ Tm ∩ Tℐ do foreach S ∈ C do if S = X ◁ 𝜆(t) then if 𝜎 = uXv then Tok ⃪ Tok∪t else if S = X ◁¯ 𝜆(t) then if 𝜎≠uXv then Tok ⃪ Tok∪t end end end We do not show here how the transition firing is implemented. It is discussed in detail in [22] where the original GENA plug-in is described. Consider an example based on the system shown in Fig. 2. Assume ℐ = (A ◁ B) ∨ (y1 ◁ x1 ∧ x2 ◁¯ y1). C = y1 ◁ x1 ∧ x2 ◁¯ y1 is chosen. We are at the initial marking, i.e. 𝜎 = 𝜀. Enabled transitions are {A, x1, B, y1}. However, x1 cannot fire, since it should wait until y1 is executed. Then B fires nondetermenistically. Subsequently, the run is 𝜎 = B, and the enabled transitions are {A, x1, y2}, but x1 still cannot fire. We can choose A to fire. Then the run is 𝜎 = BA, and the enabled transitions are {x2, y2} firing of which is not influenced by C. As a result, we can obtain a trace 𝜎 = BAy2x2 satisfying C, and the projections of 𝜎 on agent transitions, Ax2 and By2, are the runs of corresponding agent nets. 5.2 Event Log Generation 𝜎 ⊨ ℐʹ 𝜎 ⃪ 𝜀; m ⃪ m0; i ⃪ 1; C ⃪ pickRandomConjunct(ℐʹ) while (i ≤ maxSteps) ∧ (m ≠ mf) do Tok ⃪ findEnabledTransitions(N, m, C, 𝜎) if Tok ≠ ∅ then t ⃪ pickRandomTransition(Tok) m ⃪ fireTransition(N, m, t) if 𝜆(t) ≠ 𝜏 then 𝜎 ⃪ 𝜎 + 𝜆(t); i ⃪ i + 1 end else 𝜎 ⃪ 𝜀; break end end Algorithm 2 is used for finding enabled transitions, which do not violate constraints of C. Firstly, we find a set of transitions enabled at a reachable marking m according 292 Algorithm 2 is used for finding enabled transitions, which do not violate constraints of C. Firstly, we find a set of transitions enabled at a reachable marking m according 292 Algorithm 2 is used for finding enabled transitions, which do not violate constraints of C. Firstly, we find a set of transitions enabled at a reachable marking m according 292 of C. Firstly, we find a set of transitions enabled at a reachable marking m according Нестеров Р.А., Мицюк А.А., Ломазова И.А. Симуляция поведения мультиагентных систем с ациклически взаимодействующими агентами. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 285-302 Нестеров Р.А., Мицюк А.А., Ломазова И.А. Симуляция поведения мультиагентных систем с ациклически взаимодействующими агентами. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 285-302 to the standard firing rule. Secondly, if m enables interface transitions, we check \| to the standard firing rule. Secondly, if m enables interface transitions, we check whether the current run 𝜎 = 𝜆(w)\|𝒜, s.t. m0[w⟩m, satisfies constraints of C using the interface-driven firing rule. A run 𝜎 is a trace to be recorded into an event log L. Algorithm 2. Function findEnabledTransitions Input: N = (P, T, F, m0, 𝜆), m ∈ [m0⟩, C ∈ ℐʹ, 𝜎 Output: a set Tok of transitions enabled w.r.t. C Tm ⃪ stEnabledTransitions(N, m) Tok ⃪ Tm\Tℐ foreach t ∈ Tm ∩ Tℐ do foreach S ∈ C do if S = X ◁ 𝜆(t) then if 𝜎 = uXv then Tok ⃪ Tok∪t else if S = X ◁¯ 𝜆(t) then if 𝜎≠uXv then Tok ⃪ Tok∪t end end end Algorithm 2. Function findEnabledTransitions Input: N = (P, T, F, m0, 𝜆), m ∈ [m0⟩, C ∈ ℐʹ, 𝜎 Output: a set Tok of transitions enabled w.r.t. 5.3 Experimental Simulation We have developed the extension to the ProM2 plug-in GENA implementing the proposed simulation algorithm and allowing users to obtain a set of event logs by simulating a given MAS model w.r.t. interaction constraints. We have prepared five use cases for evaluating the proposed simulation approach. In each case, we have generated event logs with 5000 traces. In addition, we provide a “filtered” version of a generated event log w.r.t. interacting actions, s.t. it is clear whether the corresponding interface is exactly observed. We have used Disco3 to visualize generated event logs. Insignificant parts of agent nets are shown by shaded ovals. 293 Nesterov R.A., Mitsyuk A.A., Lomazova I.A. Simulating Behavior of Multi-Agent Systems with Acyclic Interactions of Agents. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 285-302 A., Mitsyuk A.A., Lomazova I.A. Simulating Behavior of Multi-Agent Systems with Acyclic Interactions o dy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 285-302 a) Sequencing: Consider a system with three interacting agents (see Fig. 3). Each agent always executes one action. We have simulated it w.r.t. the interface ℐ = A ◁ B ∧ B ◁ C. Intuitively, in this case each interacting agent prepares resources needed for the other agent. a) Sequencing: Consider a system with three interacting agents (see Fig. 3). Each agent always executes one action. We have simulated it w.r.t. the interface ℐ = A ◁ B ∧ B ◁ C. Intuitively, in this case each interacting agent prepares resources needed for the other agent. Fig. 3. Sequential interaction Fig. 3. Sequential interaction Fig. 3. Sequential interaction Fig. 3. Sequential interaction b) Conditional sequencing: As opposed to sequencing, conditional sequencing allows for several execution options. In this case, a system consists of two agents, one of which has two alternative branches (see Fig. 4). The interface for the conditional sequencing is as follows: ℐ = A ◁ C ∨ C ◁ B. Fig. 4. Sequential interaction with options Fig. 4. Sequential interaction with options c) Alternative interaction: The alternative interaction implies that one of two interacting agents influences the choice done by the other agent. A system consists 294 Нестеров Р.А., Мицюк А.А., Ломазова И.А. Симуляция поведения мультиагентных систем с ациклически взаимодействующими агентами. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 285-302 of two interacting agents both having two alternative branches (see Fig. 5). Нестеров Р.А., Мицюк А.А., Ломазова И.А. Симуляция поведения мультиагентных систем с ациклически взаимодействующими агентами. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 285-302 Нестеров Р.А., Мицюк А.А., Ломазова И.А. Симуляция поведения мультиагентных систем с ациклически взаимодействующими агентами. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 285-302 6. Conclusion We have proposed the new approach to model and simulate multi-agent systems using Petri nets. Independent agents are modeled as a set of labeled workflow nets, and their interaction is described using a declarative interface. The interface is constructed as a logic formula over atomic constraints describing the order of internal agent actions. This study has considered only acyclic agent interactions described by two kinds of atomic constraints, s.t. interacting activities are implemented only once. If cyclic interactions are allowed, subtler relations on interacting activities are needed to express such constraints as “each B should be preceded by A” or “at least one B should be preceded by A”. This is a subject for further research. An algorithm for simulating process models of multi-agent systems with respect to the interface has been developed. We have implemented the algorithm within the existing ProM 6 plug-in GENA and have evaluated it using five different cases of agent interactions. The experiment results show how to obtain artificial event logs by simulating process models of multi-agent systems with a finite number of asynchronously interacting agents. 5.3 Experimental Simulation The interface formula for this case is as follows: ℐ = A ◁ C ∨ B ◁ D. Fig. 5. Alternative interaction Fig. 5. Alternative interaction d) Interaction using negative constraints: Assume we have a system of two interacting agents with two alternative branches as shown in Fig. 5a. The result of simulating this system w.r.t. the interface ℐ = A ◁¯ C is shown in Fig. 6. It is clear from the simulation result that C is never preceded by A. Intuitively, negative constraints allow for a more compact way of interface construction. d) Interaction using negative constraints: Assume we have a system of two interacting agents with two alternative branches as shown in Fig. 5a. The result of simulating this system w.r.t. the interface ℐ = A ◁¯ C is shown in Fig. 6. It is clear from the simulation result that C is never preceded by A. Intuitively, negative constraints allow for a more compact way of interface construction. Fig. 6. Interaction using negative constraints: an event log Fig. 6. Interaction using negative constraints: an event log 295 Nesterov R.A., Mitsyuk A.A., Lomazova I.A. Simulating Behavior of Multi-Agent Systems with Acyclic Interactions of Agents. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 285-302 Nesterov R.A., Mitsyuk A.A., Lomazova I.A. Simulating Behavior of Multi-Agent Systems with Acyclic Interactions of Agents. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 285-302 e) Complex interaction: In this case, we show several ways of interaction among three different agents (see Fig. 7a). For convenience, we have filtered the obtained log in two ways (see Fig. 8). We have used the following interface formula (given in a conjunctive normal form for the convenience of a reader): ℐ = B ◁ A ∧ H ◁ C ∧ (D ◁ F ∨ E ◁ G). a conjunctive normal form for the convenience of a reader): ℐ = B ◁ A ∧ H ◁ C ∧ (D ◁ F ∨ E ◁ G). Fig. 7. Complex interaction Fig. 8. Complex interaction: filtered event logs Fig. 7. Complex interaction Fig. 7. Complex interaction Fig. 7. Complex interaction mplex interaction Fig. 8. Complex interaction: filtered event logs Fig. 8. Complex interaction: filtered event logs 296 Нестеров Р.А., Мицюк А.А., Ломазова И.А. Симуляция поведения мультиагентных систем с ациклически взаимодействующими агентами. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 5.3 Experimental Simulation 285-302 References [1]. van der Aalst W.M.P. Process Mining – Data Science in Action. Springer, Heidelberg, 2016, 467 p. [2]. Günther C.W., van der Aalst W.M.P. Fuzzy mining: Adaptive process simplification based on multi-perspective metrics. BPM 2007. LNCS, vol. 4714, 2007, pp. 328-343. [3]. van der Werf J.M.E.M., van Dongen B.F., Hurkens C.A.J., Serebrenik A. Process Discovery using Integer Linear Programming. Fundamenta Informaticae, vol. 94, no. 3- 4, 2009, pp. 387-412. [4]. Weijters A.J.M.M., Ribeiro J.T.S. Flexible Heuristics Miner (FHM). In Proceedings of the IEEE Symposium on Computational Intelligence and Data Mining (CIDM), 2011, pp. 310-317. [5]. Kalenkova A.A., Lomazova I.A., van der Aalst W.M.P. Process Model Discovery: A Method Based on Transition System Decomposition. ICATPN 2014. LNCS, vol. 8489, 2014, pp. 71-90. [6]. Leemans S.J.J., Fahland D., van der Aalst W.M.P. Scalable Process Discovery with Guarantees. BPMDS 2015, EMMSAD 2015. LNBIP, vol 214, 2015, pp. 85-101. [7]. Begicheva A.K., Lomazova I.A. Discovering high-level process models from event logs. Modeling and Analysis of Information Systems, vol. 24, no. 2, 2017, pp. 125–140.. [8]. Augusto A., Conforti R., Dumas M., La Rosa M., Maria Maggi F., Marrella A., Mecella M., Soo A. Automated Discovery of Process Models from Event Logs: Review and Benchmark. CoRR, 2017, vol. abs/1705.02288. [9]. Rubin V.A., Mitsyuk A.A., Lomazova I.A., van der Aalst W.M.P. Process Mining can be applied to software too! In Proceedings of the 8th ACM/IEEE International Symposium on Empirical Software Engineering and Measurement (ESEM ’14), 2014, pp. 1-8. 297 Nesterov R.A., Mitsyuk A.A., Lomazova I.A. Simulating Behavior of Multi-Agent Systems with Acyclic Interactions of Agents. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 285-302 [10]. Leemans M., van der Aalst W.M.P. Process mining in software systems: Discovering real-life business transactions and process models from distributed systems. MODELS 2015, pp. 44-53. [11]. Leemans M., van der Aalst W.M.P., van den Brand M. Recursion Aware Modeling and Discovery for Hierarchical Software Event Log Analysis (Extended). CoRR, 2017, vol. abs/1710.09323. [12]. Liu C., van Dongen B.F., Assy N., van der Aalst W.M.P. Component behavior discovery from software execution data. In Proceedings of the IEEE Symposium Series on Computational Intelligence (SSCI), 2016, pp. 1-8. [13]. Davydova K.V., Shershakov S.A. Mining hybrid UML models from event logs of SOA systems. Trudy ISP RAN/Proc. ISP RAS, vol. 29, issue 4, 2017, pp. 155-174. DOI: 10.15514/ISPRAS-2017-29(4)-10. [14]. 3TU: Big software on the run. [Online]. Available: http://www.3tu-bsr.nl. Accessed: 09.06.2018. [15]. [25]. Di Ciccio C., Luca Bernardi M., Cimitile M., Maria Maggi F. Generating Event Logs Through the Simulation of Declare Models. EOMAS 2015. LNBIP, vol. 231, 2015, pp. 20-36. DOI: 10.15514/ISPRAS-2018-30(3)-20 Для цитирования: Нестеров Р.А., Мицюк А.А., Ломазова И.А. Симуляция поведения мультиагентных систем с ациклически взаимодействующими агентами. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 285-302 (на английском языке). DOI: 10.15514/ISPRAS-2018-30(3)-20 Симуляция поведения мультиагентных систем с ациклически взаимодействующим агентами ц y 1 И.А. Ломазова <ilomazova@hse.ru> 1 Национальный исследовательский университет «Высшая школа экономики», 101000, Россия, г. Москва, ул. Мясницкая, д. 20. 2 Департамент информатики, систем и коммуникаций, Миланский университет-Бикокка, 20126, Италия, г. Милан, Viale Sarca 336 – Edificio U14 ц y 1 И.А. Ломазова <ilomazova@hse.ru> 1 Национальный исследовательский университет «Высшая школа экономики», 101000, Россия, г. Москва, ул. Мясницкая, д. 20. 2 Департамент информатики, систем и коммуникаций, Миланский университет-Бикокка, 20126, Италия, г. Милан, Viale Sarca 336 – Edificio U14 1 И.А. Ломазова <ilomazova@hse.ru> 1 Национальный исследовательский университет «Высшая школа экономики», 101000, Россия, г. Москва, ул. Мясницкая, д. 20. 2 Департамент информатики, систем и коммуникаций, Миланский университет-Бикокка, 20126, Италия, г. Милан, Viale Sarca 336 – Edificio U14 Аннотация. В работе предложен подход для моделирования и симуляции поведения мультиагентных систем (МАС) с применением сетей Петри. МАС представляется как конечное множество сетей потоков работ. Асинхронные взаимодействия агентов описываются с помощью интерфейса, который определяется логической формулой над множеством атомарных ограничений. Эти ограничения задают порядок выполнения внутренних действий агентов. В статье рассматриваются только ациклические взаимодействия агентов. Также был разработан алгоритм симуляции поведения МАС с учетом ограничений взаимодействия агентов. Алгоритм реализован в виде подключаемого модуля для инструмента ProM 6. Предложенный подход может быть использован для оценки качества алгоритмов извлечения процессов (process discovery) с точки зрения характеристик получаемых моделей процессов. Ключевые слова: сети Петри; мультиагентные системы; взаимодействие; интерфейс; симуляция; журналы событий References Cabac L., Denz N. Net Components for the Integration of Process Mining into Agent- Oriented Software Engineering. Transactions on Petri Nets and Other Models of Concurrency I. LNCS, vol. 5100, 2008, pp. 86-103. [16]. Cabac L., Knaak N., Moldt D., Rölke H. Analysis of Multi-Agent Interactions with Process Mining Techniques. MATES 2006. LNCS, vol. 4196, 2006, pp. 12-23. [17]. Nesterov R.A., Lomazova I.A. Using Interface Patterns for Compositional Discovery of Distributed System Models. Trudy ISP RAN/Proc. ISP RAS, 2017, vol. 29, issue 4, pp. 21-38. DOI: 10.15514/ISPRAS-2017-29(4)-2. [18]. Nesterov R.A., Lomazova I.A. Compositional Process Model Synthesis Based on Interface Patterns. TMPA 2017. CCIS, vol. 779, 2018, pp. 151-162. [19]. Burattin A. PLG2: Multiperspective Process Randomization with Online and Offline Simulations. BPMD 2016. CEUR Workshop Proceedings, vol. 1789, 2016, pp. 1-6. [20]. Jouck T., Depaire B. PTandLogGenerator: A Generator for Artificial Event Data. BPMD 2016. CEUR Workshop Proceedings, vol. 1789, 2016, pp. 23-27. [21]. Yan Z., Dijkman R.M., Grefen P. Generating process model collections. Software and System Modeling, 2017, vol. 16, issue 4, pp. 979-995. [22]. Shugurov I.S., Mitsyuk A.A. Generation of a Set of Event Logs with Noise. In Proceedings of the 8th Spring/Summer Young Researchers Colloquium on Software Engineering (SYRCoSE 2014), 2014, pp. 88-95. DOI: 10.15514/SYRCOSE-2014-8-13. [23]. Mitsyuk A.A., Shugurov I.S., Kalenkova A.A., van der Aalst W.M.P. Generating event logs for high-level process models. Simulation Modelling Practice and Theory, vol. 74, 2017, pp. 1-16. [24]. de Medeiros A.K.A., Günther C.W. Process Mining: Using CPN Tools to Create Test Logs for Mining Algorithms. In Proceedings of CPN 2005. DAIMI, vol. 576, 2005, pp. 177-190. [25]. Di Ciccio C., Luca Bernardi M., Cimitile M., Maria Maggi F. Generating Event Logs Through the Simulation of Declare Models. EOMAS 2015. LNBIP, vol. 231, 2015, pp. 20-36. 298 Нестеров Р.А., Мицюк А.А., Ломазова И.А. Симуляция поведения мультиагентных систем с ациклически взаимодействующими агентами. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 285-302 Список литературы [1]. van der Aalst W.M.P. Process Mining – Data Science in Action. Springer, Heidelberg, 2016, 467 p. [2]. Günther C.W., van der Aalst W.M.P. Fuzzy mining: Adaptive process simplification based on multi-perspective metrics. BPM 2007. LNCS, vol. 4714, 2007, pp. 328-343. [3]. van der Werf J.M.E.M., van Dongen B.F., Hurkens C.A.J., Serebrenik A. Process Discovery using Integer Linear Programming. Fundamenta Informaticae, vol. 94, no. 3- 4, 2009, pp. 387-412. 299 Nesterov R.A., Mitsyuk A.A., Lomazova I.A. Simulating Behavior of Multi-Agent Systems with Acyclic Interactions of Agents. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 285-302 [4]. Weijters A.J.M.M., Ribeiro J.T.S. Flexible Heuristics Miner (FHM). In Proceedings of the IEEE Symposium on Computational Intelligence and Data Mining (CIDM), 2011, pp. 310-317. [5]. Kalenkova A.A., Lomazova I.A., van der Aalst W.M.P. Process Model Discovery: A Method Based on Transition System Decomposition. ICATPN 2014. LNCS, vol. 8489, 2014, pp. 71-90. pp [6]. Leemans S.J.J., Fahland D., van der Aalst W.M.P. Scalable Process Discovery with Guarantees. BPMDS 2015, EMMSAD 2015. LNBIP, vol 214, 2015, pp. 85-101. [7]. Begicheva A.K., Lomazova I.A. Discovering high-level process models from event logs. Modeling and Analysis of Information Systems, vol. 24, no. 2, 2017, pp. 125–140.. [8]. Augusto A., Conforti R., Dumas M., La Rosa M., Maria Maggi F., Marrella A., Mecella M., Soo A. Automated Discovery of Process Models from Event Logs: Review and Benchmark. CoRR, 2017, vol. abs/1705.02288. [9]. Rubin V.A., Mitsyuk A.A., Lomazova I.A., van der Aalst W.M.P. Process Mining can be applied to software too! In Proceedings of the 8th ACM/IEEE International Symposium on Empirical Software Engineering and Measurement (ESEM ’14), 2014, pp. 1-8. [10]. Leemans M., van der Aalst W.M.P. Process mining in software systems: Discovering real-life business transactions and process models from distributed systems. MODELS 2015, pp. 44-53. [11]. Leemans M., van der Aalst W.M.P., van den Brand M. Recursion Aware Modeling and Discovery for Hierarchical Software Event Log Analysis (Extended). CoRR, 2017, vol. abs/1710.09323. [12]. Liu C., van Dongen B.F., Assy N., van der Aalst W.M.P. Component behavior discovery from software execution data. In Proceedings of the IEEE Symposium Series on Computational Intelligence (SSCI), 2016, pp. 1-8. [13]. Davydova K.V., Shershakov S.A. Mining hybrid UML models from event logs of SOA systems. Trudy ISP RAN/Proc. ISP RAS, vol. 29, issue 4, 2017, pp. 155-174. DOI: 10.15514/ISPRAS-2017-29(4)-10. [14]. 3TU: Big software on the run. [Online]. Available: http://www.3tu-bsr.nl. Accessed: 09.06.2018. Nesterov R.A., Mitsyuk A.A., Lomazova I.A. Simulating Behavior of Multi-Agent Systems with Acyclic Interactions of Agents. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 285-302 Список литературы [15]. Cabac L., Denz N. Net Components for the Integration of Process Mining into Agent- Oriented Software Engineering. Transactions on Petri Nets and Other Models of Concurrency I. LNCS, vol. 5100, 2008, pp. 86-103. [16]. Cabac L., Knaak N., Moldt D., Rölke H. Analysis of Multi-Agent Interactions with Process Mining Techniques. MATES 2006. LNCS, vol. 4196, 2006, pp. 12-23. [17]. Nesterov R.A., Lomazova I.A. Using Interface Patterns for Compositional Discovery of Distributed System Models. Trudy ISP RAN/Proc. ISP RAS, 2017, vol. 29, issue 4, pp. 21-38. DOI: 10.15514/ISPRAS-2017-29(4)-2. [18]. Nesterov R.A., Lomazova I.A. Compositional Process Model Synthesis Based on Interface Patterns. TMPA 2017. CCIS, vol. 779, 2018, pp. 151-162. [19]. Burattin A. PLG2: Multiperspective Process Randomization with Online and Offline Simulations. BPMD 2016. CEUR Workshop Proceedings, vol. 1789, 2016, pp. 1-6. [20]. Jouck T., Depaire B. PTandLogGenerator: A Generator for Artificial Event Data. BPMD 2016. CEUR Workshop Proceedings, vol. 1789, 2016, pp. 23-27. [21]. Yan Z., Dijkman R.M., Grefen P. Generating process model collections. Software and System Modeling, 2017, vol. 16, issue 4, pp. 979-995. 300 Нестеров Р.А., Мицюк А.А., Ломазова И.А. Симуляция поведения мультиагентных систем с ациклически взаимодействующими агентами. Труды ИСП РАН, 2018, том 30, вып. 3, 2018 г., стр. 285-302 [22]. Shugurov I.S., Mitsyuk A.A. Generation of a Set of Event Logs with Noise. In Proceedings of the 8th Spring/Summer Young Researchers Colloquium on Software Engineering (SYRCoSE 2014), 2014, pp. 88-95. DOI: 10.15514/SYRCOSE-2014-8-13. [23]. Mitsyuk A.A., Shugurov I.S., Kalenkova A.A., van der Aalst W.M.P. Generating event logs for high-level process models. Simulation Modelling Practice and Theory, vol. 74, 2017, pp. 1-16. [24]. de Medeiros A.K.A., Günther C.W. Process Mining: Using CPN Tools to Create Test Logs for Mining Algorithms. In Proceedings of CPN 2005. DAIMI, vol. 576, 2005, pp. 177-190. [25]. Di Ciccio C., Luca Bernardi M., Cimitile M., Maria Maggi F. Generating Event Logs Through the Simulation of Declare Models. EOMAS 2015. LNBIP, vol. 231, 2015, pp. 20-36. 301 Nesterov R.A., Mitsyuk A.A., Lomazova I.A. Simulating Behavior of Multi-Agent Systems with Acyclic Interactions of Agents. Trudy ISP RAN /Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 285-302 302 DOI: 10.15514/ISPRAS-2018-30(3)-21 For citation: Gnatenko A.R., Zakharov V.A. On the model checking of finite state transducers over semigroups. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 303- 324. DOI: 10.15514/ISPRAS-2018-30(3)-21 For citation: Gnatenko A.R., Zakharov V.A. On the model checking of finite state transducers over semigroups. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 303- 324. DOI: 10.15514/ISPRAS-2018-30(3)-21 On the model checking of finite state transducers over semigroups 1 A.R. Gnatenko<gnatenko.cmc@gmail.com> 2 V.A. Zakharov<zakh@cs.msu.su> 1 Lomonosov Moscow State University, GSP-1, Leninskie Gory, Moscow, 119991, Russia 2 National Research University High School of Economics, 20, Myasnitskaya str., Moscow, 101000, Russia Abstract. Sequential reactive systems represent programs that interact with the environment by receiving signals or requests and react to these requests by performing operations with data. Such systems simulate various software like computer drivers, real-time systems, control procedures, online protocols etc. In this paper, we study the verification problem for the programs of this kind. We use finite state transducers over semigroups as formal models of reactive systems. We introduce a new specification language LP-CTL to describe the behavior of transducers. This language is based on the well-known temporal logic CTL* and has two distinguished features: 1) each temporal operator is parameterized with a regular expression over input alphabet of the transducer, and 2) each atomic proposition is specified by a regular expression over the output alphabet of the transducer. We develop a tabular algorithm for model checking of finite state transducers over semigroups against LP-CTL* formulae, prove its correctness, and estimate its complexity. We also consider a special fragment of LP-CTL* language, where temporal operators are parameterized with regular expressions over one-letter alphabet, and show that this fragment may be used to specify usual Kripke structures, while it is more expressive than usual CTL. Keywords: reactive program; transducer; verification; model checking; temporal logic; finite state automaton; regular language 1. Introduction 303 Finite state machines are widely used in the field of computer science and formal methods for various purposes. While finite automata represent regular sets, transducers stand for regular (or, rational) relations and, therefore, can serve as models of programs and algorithms that operate with input and output data. For Gnatenko A.R., Zakharov V.A. On the model checking of finite state transducers over semigroups. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 303-324 Gnatenko A.R., Zakharov V.A. On the model checking of finite state transducers over semigroups. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 303-324 example, transducers are used as formal models in software engineering to represent numerous algorithms, protocols and drivers that manipulate with strings, dataflows, etc. [1, 15, 25]. By extending the concept of ordinary transducers, we build a new formal model for sequential reactive systems. These systems are software programs or hardware devices that receive requests (control signals, commands) from the environment and perform in response some manipulations (actions, transformations) with data, interactions with the environment, mechanical movements, etc. While the flow of requests can be represented by finite or infinite words in some fixed alphabet, the sequence of actions of the system needs a more sophisticated interpretation. The key point here is that different sequences of actions may bring a computing system to the same result. To capture this effect the collection of actions performed by a reactive system can be viewed as a generating set of some algebraic structure (e.g. semigroup, group, ring, etc.) and particular algebraic properties of basic actions should be taken into account when choosing adequate formal models for this class of information processing systems. Let us illustrate this consideration by several examples.  A network switch with several input and output ports. A switch is a device, which receives data packets on its input port, modifies their heads and commutes them to one of the output ports. Once received a special control signal, this switch changes its packet-forwarding table and, thus, its behaviour. Since packets from different flows may be processed in any order, the switch can be modeled by a transducer, which operates over a free partially commutative semigroup, or a trace monoid. Trace monoids are commonly used as an algebraic foundation of concurrent computations and process calculi (see, e.g., [9]). 1. Introduction  A real-time device that control the operation of some industrial equipment (say, a boiling system). Such device receives data from temperature and pressure sensors and switches some processes on and off according to its instructions and the current state of the system. It seems reasonable that for some actions the order of their implementation is not important (routine actions), while others must follow in a strictly specified order (e.g. an execution of some complex operation). Moreover, there are also actions which bring system to certain predefined operation mode (set-up actions). These actions are implemented in the emergency situations. A partially commutative semigroup with right-zero elements 𝟎 which satisfy the equalities 𝑥𝟎= 𝟎 for every element 𝑥 provides an adequate interpretation for such operations.  A system supervisor that maintains a log-file. For each entry its date and time is recorded in the file and there is no way to delete entries from the log  only to append it. Thus, for any two basic actions (record operations to the log-file) it is crucial in which order they are performed and such a supervisor can be modeled by a transducer over a free semigroup. Verification techniques for such reactive  A system supervisor that maintains a log-file. For each entry its date and time is recorded in the file and there is no way to delete entries from the log  only to append it. Thus, for any two basic actions (record operations to the log-file) it is crucial in which order they are performed and such a supervisor can be modeled by a transducer over a free semigroup. Verification techniques for such reactive 304 Гнатенко А.Р., Захаров В.А. О верификации конечных автоматов-преобразователей над полугруппами. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 303-324 Гнатенко А.Р., Захаров В.А. О верификации конечных автоматов-преобразователей над полугруппами. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 303-324 systems are considered in [17]; this is the main topic of this paper as well.  A radio-controlled robot, that moves on a surface. It can make one step moves in any of direction. When it receives a control signal in a state 𝑞′ it must choose and carry out a sequence of steps and enter to the next state 𝑞′′. At some distinguished state 𝑞𝑓 the robot reports its current location. 1. Introduction Movements of the robot may be regarded as basic actions, and the simplest model of computation which is suitable for analyzing a behaviour of this robot is a nondeterministic finite state transducer operating on a free Abelian group of rank 2.  A radio-controlled robot, that moves on a surface. It can make one step moves in any of direction. When it receives a control signal in a state 𝑞′ it must choose and carry out a sequence of steps and enter to the next state 𝑞′′. At some distinguished state 𝑞𝑓 the robot reports its current location. Movements of the robot may be regarded as basic actions, and the simplest model of computation which is suitable for analyzing a behaviour of this robot is a nondeterministic finite state transducer operating on a free Abelian group of rank 2. To construct a reliable system or network it is crucial for its components to have a correct behaviour. For example, a network switch must process received data packets within a specified number of execution steps and the boiling system should never be overheated, that is, will never remain for a long time in a particular condition without appropriate responses from the control device. By using finite state transducers as formal models of reactive systems, one can develop verification algorithms for these models to solve such problems as equivalence checking, deductive verification or model checking. The study of the equivalence checking problem for classical transducers began in the early 60s. It was established that the equivalence checking problem for non- deterministic transducers is undecidable [13] even over 1-letter input alphabet [16]. However, the undecidability displays itself only in the case of unbounded transductions when an input word may have arbitrary many images. The equivalence checking problem was shown to be decidable for deterministic [4], functional (single-valued) [5, 19], and k-valued transducers [6, 26]. In a series of papers [20-22] techniques for checking bounded valuedness, k-valuedness and equivalence of finite state transducers over words were developed. Recently in [29] equivalence checking problem was shown to be decidable for finite state transducers that operate over finitely generated semigroups embeddable in decidable groups. There are also papers where equivalence checking problem for transducers is studiedin the framework of program verification. The authors of [23] proposed models of communication protocols as finite state transducers operating on bit strings. 1. Introduction From this viewpoint model checking problem for finite state transducers conforms well to standard model checking scheme for finite structures, and, hence, it is not worthy of any particular treatment. checking of parameterized distributed systems where configurations are represented as words over a finite alphabet. In such models, a transition relation on these configurations may be regarded as a rational relation and, thus, it may be specified by finite state transducers (see [7, 28]). In these papers, finite state transducers just play the role of verification instrument, but not an object of verification. However, as far as we know, a deeper investigation of the model checking problem for the reactive systems represented by transducers has not yet been carried out. We think that this is due the following main reason. A transducer is a model of computation which, given an input word, computes an output word. The letters of input and output alphabets can be regarded as valuations (tuples of truth values) of some set of basic predicates. Therefore, a transducer can be viewed as some special representation of a labeled transition system (Kripke structure) (see [2]). From this viewpoint model checking problem for finite state transducers conforms well to standard model checking scheme for finite structures, and, hence, it is not worthy of any particular treatment. However, our viewpoint is quite different. Transducer is a more complex model of computation than a finite state automaton (transition systems). Its behaviorism characterized by the correspondence between input and output words. A typical property of such behaviour to be checked is whether for every (some) input word from a given pattern a transducer outputs a word from another given pattern. Therefore, when formally expressing the requirements of this kind one needs not only temporal operators to specify an order in which events occur but also some means to refer to such patterns. Conventional Temporal Logics like 𝑳𝑻𝑳 or 𝑪𝑻𝑳 are not sufficient in this case; they should be modified in such a way as to acquire an ability to express correspondences between the sets (languages) of input words and the sets (languages) of output words. This could be achieved by supplying temporal operators with patterns as parameters. Every such pattern is a formal description of a language 𝐿 over an input alphabet𝒞; automata, formal grammars, regular expressions, language equations are suitable for this purpose. 1. Introduction They set up the verification problem as equivalence checking between the protocol transducer and the specification transducer. The authors of [25] extended finite state transducers with symbolic alphabets, which are represented as parametric theories. They showed that a number of classical problems for extended transducers, There are also papers where equivalence checking problem for transducers is studiedin the framework of program verification. The authors of [23] proposed models of communication protocols as finite state transducers operating on bit strings. They set up the verification problem as equivalence checking between the protocol transducer and the specification transducer. The authors of [25] extended finite state transducers with symbolic alphabets, which are represented as parametric theories. They showed that a number of classical problems for extended transducers, including equivalence checking problem, are decidable modulo underlying theories. In [1] a model of streaming transducers was proposed for programs that access and modify sequences of data items in a single pass. It was shown that a number of verification problems such as equivalence checking, assertion checking, and checking correctness with respect to pre/post conditions, are decidable for this program model. Meanwhile, very few papers on the model checking problem for transducers are known. Transducers can be conveniently used as auxiliary means in regular model 305 Gnatenko A.R., Zakharov V.A. On the model checking of finite state transducers over semigroups. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 303-324 checking of parameterized distributed systems where configurations are represented as words over a finite alphabet. In such models, a transition relation on these configurations may be regarded as a rational relation and, thus, it may be specified by finite state transducers (see [7, 28]). In these papers, finite state transducers just play the role of verification instrument, but not an object of verification. However, as far as we know, a deeper investigation of the model checking problem for the reactive systems represented by transducers has not yet been carried out. We think that this is due the following main reason. A transducer is a model of computation which, given an input word, computes an output word. The letters of input and output alphabets can be regarded as valuations (tuples of truth values) of some set of basic predicates. Therefore, a transducer can be viewed as some special representation of a labeled transition system (Kripke structure) (see [2]). 1. Introduction The basic properties of output words can be also represented by languages over an output alphabet. Then, for instance, an expression 𝑮𝐿𝑃 can be understood as the requirement that for every input word 𝑤 from the language 𝐿 the output word ℎ of a transducer belongs to the language 𝑃. The advantages of this approach are twofold. On the one hand, such extensions of Temporal Logics make it possible to express explicitly relationships between input and output words and specify thus desirable behaviours of transducers. On the other hand, it can be hoped that such extensions could rather easily assimilate some well- known model checking techniques (see [3, 8]) developed for conventional Temporal Logics. The first attempt to implement this approach was made in [17]. The authors of this paper introduced an 𝓛𝓟-𝑳𝑻𝑳 specification language based on 𝑳𝑻𝑳 temporal logic and developed a checking procedure for transducers over free monoids against specifications from 𝓛𝓟-𝑳𝑻𝑳. It was shown that this procedure has double exponential time complexity. 306 Гнатенко А.Р., Захаров В.А. О верификации конечных автоматов-преобразователей над полугруппами. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 303-324 Гнатенко А.Р., Захаров В.А. О верификации конечных автоматов-преобразователей над полугруппами. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 303-324 In this paper we continue this line of research and "raise'' the specification language introduced in [17] to the level of 𝓛𝓟-𝑪𝑻𝑳∗. We will focus only on one task related to the use of new logic for the verification of reactive systems, namely, the development of a general model checking algorithm for finite state transducers against specifications in 𝓛𝓟-𝑪𝑻𝑳∗. Such issues as expressive power of 𝓛𝓟-𝑪𝑻𝑳∗, complexity of model checking and satisfiability checking problems, the influence of types of languages used as parameters and basic predicates in 𝓛𝓟-𝑪𝑻𝑳∗ on decidability and complexity of model checking problem remain topic of our further research and will be covered in our subsequent works. We also leave beyond this work a number of applied questions, which are worthy of consideration in a separate paper. For example, it is important to understand to what extent the already developed model checking tools can be adapted to the new temporal logic. And, of course, in the future we will have a well-chosen series of examples that illustrate the new possibilities of using 𝓛𝓟-𝑪𝑻𝑳∗to describe the behavior of reactive systems. The paper is organized as follows. 1. Introduction In Section 2, we define the concept of finite state transducer over semigroup as a formal model of sequential reactive systems (see [29]) and in Section 3, we describe the syntax and the semantics of 𝓛𝓟-𝑪𝑻𝑳∗ as a formal language for specifying behaviour of sequential reactive systems. In Section 3 we also set up formally model checking problem for finite state transducers against 𝓛𝓟-𝑪𝑻𝑳∗ formulae. In Section 4, we present an 𝓛𝓟-𝑪𝑻𝑳∗ model checking algorithm for the case when parameters of temporal operators and basic predicates are regular languages represented by finite state automata. The model checking algorithm we designed has time complexity which is linear of the size of a transducer but exponential of the size of 𝓛𝓟-𝑪𝑻𝑳∗ formula. This complexity estimate is in contrast to the case of conventional 𝑪𝑻𝑳 model checking: its time complexity is linear of both the size of a model and the size of a 𝑪𝑻𝑳 formula. To explain this effect in Section 5 we show how 𝓛𝓟-𝑪𝑻𝑳∗ formulae can be also checked on the conventional Kripke structures. Finally, we compare 𝓛𝓟-𝑪𝑻𝑳∗ with some other known extensions Temporal Logics and discuss some topics for further research. 2. Finite state transducers as models of reactive systems In this section, we introduce a Finite State Transducer as a formal model of a reactive computing system which receives control signals from the environment and reacts to these signals by performing operations with data. Let 𝒞 be a finite set of signals. Finite words over 𝒞 are called signal flows; the set of all signal flows is denoted by𝒞∗. Given a pair of signal flows 𝑢 and 𝑣 we write 𝑢𝑣 for their concatenation, and denote by 𝜀 the empty flow. Let 𝒜= {𝑎1, … , 𝑎𝑛} be a finite set of elements called basic actions; these actions stand for the elementary operations performed by a reactive system. Finite words over 𝒜 are called compound actions; they denote sequential compositions of basic actions. Since different sequences of basic actions could produce the same result, 307 Gnatenko A.R., Zakharov V.A. On the model checking of finite state transducers over semigroups. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 303-324 one may interpret compound actions over a semigroup (𝑆, 𝑒, ∘) generated by a set of basic actions 𝒜. The elements of 𝑆 are called data states. Every compound action ℎ= 𝑎𝑖1𝑎𝑖2 … 𝑎𝑖𝑘 is evaluated by the data state [ℎ] = [𝑎𝑖1] ∘[𝑎𝑖2] ∘… ∘[𝑎𝑖𝑘]. For example, if a reactive system just keeps a track of input requests by adding certain records to a log-file then a free semigroup will be suitable for interpretation of these operations. But when a robot moves on a 2-dimensional surface then the actions (movements) performed by this robot may be regarded as generating elements of Abelian group 𝐺 of rank 2, and the positions on the surface occupied by this robot can be specified by the elements from 𝐺. In this paper we restrict ourselves to the consideration of free semigroups when [ℎ] = ℎ holds for every compound action ℎ, and ∘ is the word concatenation operation. Let 𝒞 be a set of signals and 𝒜 be a set of basic actions that are interpreted over a semigroup (𝑆, 𝑒,∘). Then a Finite State Transducer (in what follows, FST) is a quintuple Π = (𝑄, 𝒞, 𝒜, 𝑞𝑖𝑛𝑖𝑡, 𝑇), where  𝑄 is a finite set of control states;  𝑄 is a finite set of control states;  𝑞𝑖𝑛𝑖𝑡∈𝑄 is an initial control state;  𝑇⊆𝑄× 𝒞× 𝑄× 𝒜∗is a finite transition relation. 2. Finite state transducers as models of reactive systems  𝑇⊆𝑄× 𝒞× 𝑄× 𝒜∗is a finite transition relation. Each tuple (𝑞′, 𝑐, 𝑞′′, ℎ) in 𝑇 is called a transition: when a transducer is in a control state 𝑞′ and receives a signal 𝑐, it changes its state to 𝑞′′ and performs a compound action ℎ. We denote such transition by 𝑞′ 𝑐,ℎ → 𝑞′′. A run of a FST Π is any finite sequence of transitions  𝑇⊆𝑄× 𝒞× 𝑄× 𝒜∗is a finite transition relation. Each tuple (𝑞′, 𝑐, 𝑞′′, ℎ) in 𝑇 is called a transition: when a transducer is in a control state 𝑞′ and receives a signal 𝑐, it changes its state to 𝑞′′ and performs a compound action ℎ. We denote such transition by 𝑞′ 𝑐,ℎ → 𝑞′′. A run of a FST Π is any finite sequence of transitions 𝑞1 𝑐1,ℎ1 → 𝑞2 𝑐2,ℎ2 → 𝑞3 𝑐3,ℎ3 → ⋯ 𝑐𝑛,ℎ𝑛 → 𝑞𝑛+1; this run transduces a signal flow 𝑤= 𝑐1𝑐2 … 𝑐𝑛 into a data state [ℎ1ℎ2 … ℎ𝑛]. Th b h i f FST Π (𝑄𝒞𝒜𝑞 𝑇) i (d t ) this run transduces a signal flow 𝑤= 𝑐1𝑐2 … 𝑐𝑛 into a data state [ℎ1ℎ2 … ℎ𝑛]. The behaviour of a FST Π = (𝑄, 𝒞, 𝒜, 𝑞𝑖𝑛𝑖𝑡, 𝑇) over a semigroup (data space) (𝑆 ) i d f ll b 𝑆( 𝑆) ( 𝒞𝑑 ) run transduces a signal flow 𝑤= 𝑐1𝑐2 … 𝑐𝑛 into a data state [ℎ1ℎ2 … ℎ𝑛]. this run transduces a signal flow 𝑤= 𝑐1𝑐2 … 𝑐𝑛 into a data state [ℎ1ℎ2 … ℎ𝑛]. The behaviour of a FST Π = (𝑄, 𝒞, 𝒜, 𝑞𝑖𝑛𝑖𝑡, 𝑇) over a semigroup (data space) (𝑆, 𝑒,∘) is presented formally by a transition system 𝑇𝑆(Π, 𝑆) = (𝐷, 𝒞, 𝑑𝑖𝑛𝑖𝑡, 𝒯), where The behaviour of a FST Π = (𝑄, 𝒞, 𝒜, 𝑞𝑖𝑛𝑖𝑡, 𝑇) over a semigroup (data space) (𝑆, 𝑒,∘) is presented formally by a transition system 𝑇𝑆(Π, 𝑆) = (𝐷, 𝒞, 𝑑𝑖𝑛𝑖𝑡, 𝒯), where  𝐷 = 𝑄× 𝑆 is (in general case, infinite) set of states of computation,  𝑑𝑖𝑛𝑖𝑡 = (𝑞𝑖𝑛𝑖𝑡, 𝑒) is the initial state, and  𝒯⊆𝐷× 𝒞× 𝐷 is a transition relation such that for every states of computation 𝑑′ = (𝑞′, 𝑠′), 𝑑′′ = (𝑞′′, 𝑠′′) and every signal 𝑐∈𝒞 the relationship (𝑑′, 𝑐, 𝑑′′) ∈𝒯 ⇒ ∃ℎ∈𝒜∗(𝑞′, 𝑐, 𝑞′′, ℎ) ∈𝑇 and 𝑠′′ = 𝑠′ ∘[ℎ] holds. 3. 𝓛𝓟-𝑪𝑻𝑳∗ specification language When designing sequential reactive systems one should be provided with a suitable formalism to specify the requirements for their desirable behaviour. For example, one may expect that  a mobile robot, receiving an equal number of control signals "go_left'' and "go_right'', will always return to its original position,  a mobile robot, receiving an equal number of control signals "go_left'' and "go_right'', will always return to its original position,  a network switch will never commute data packets from different packet flows into the same output buffer,  it is not possible for the interrupt service routine to complete the processing of one interrupt before it receives a request to handle another.  it is not possible for the interrupt service routine to complete the processing of one interrupt before it receives a request to handle another. These and many other requirements which refer to the correspondences between control flows and compound actions in the course of FST runs can be specified by means of Temporal Logics. When choosing a suitable temporal logic as a formal specification language of FST behaviours one should take into account two principal features of our model of sequential reactive systems:  since a FST operates over a data space which is semigroup, the basic predicates must be interpreted over semigroups as well, and  since a FST operates over a data space which is semigroup, the basic predicates must be interpreted over semigroups as well, and  since a behaviour of a FST depends not on the time flow itself but on a signal flow which it receives as an input, temporal operators must be parameterized by certain descriptions of admissible signal flows. To adapt traditional temporal logic formalism to these specific features of FST behaviours the authors of [17] introduced a new variant of Linear Temporal Logic (LTL). We assume that in general case one may be interested in checking the correctness of FST's responses to arbitrary set of signal flows. Every set of control flows may be regarded as a language over the alphabet 𝒞 of signals. Therefore, it is reasonable to supply temporal operators ("globally'' 𝑮, "eventually'' 𝑭, etc.) with certain descriptions of such languages as parameters. In more specific cases we may confine ourselves with considering only a certain family of languages (finite, regular, context-free, etc.) ℒ used as parameters of temporal operators. 2. Finite state transducers as models of reactive systems  𝒯⊆𝐷× 𝒞× 𝐷 is a transition relation such that for every states of computation 𝑑′ = (𝑞′, 𝑠′), 𝑑′′ = (𝑞′′, 𝑠′′) and every signal 𝑐∈𝒞 the relationship As usual, a transition (𝑑′, 𝑐, 𝑑′′) ∈𝒯 is denoted by 𝑑′ 𝑐→𝑑′′. As usual, a transition (𝑑′, 𝑐, 𝑑′′) ∈𝒯 is denoted by 𝑑′ 𝑐→𝑑′′. A trajectory in a transition system 𝑇𝑆(Π, 𝑆) is a pair 𝑡𝑟= (𝑑0, 𝛼), where 𝑑0 ∈𝐷 and 𝛼= (𝑐1, 𝑑1), (𝑐2, 𝑑2), … , (𝑐𝑖, 𝑑𝑖), … is a sequence of pairs (𝑐𝑖, 𝑑𝑖) such that and 𝛼 (𝑐1, 𝑑1), (𝑐2, 𝑑2), … , (𝑐𝑖, 𝑑𝑖), … is a sequence of pairs (𝑐𝑖, 𝑑𝑖) such that 𝑑𝑖−1 𝑐𝑖→𝑑𝑖 holds for every 𝑖, 𝑖≥1. A trajectory represents a possible scenario of a behaviour of a sequential reactive system: when receiving a signal flow 𝑑𝑖−1 𝑐𝑖→𝑑𝑖 holds for every 𝑖, 𝑖≥1. A trajectory represents a possible scenario of a behaviour of a sequential reactive system: when receiving a signal flow 308 q y g g Гнатенко А.Р., Захаров В.А. О верификации конечных автоматов-преобразователей над полугруппами. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 303-324 Захаров В.А. О верификации конечных автоматов-преобразователей над полугруппами. Труды 30, вып. 3, 2018 г., стр. 303-324 𝑐1, 𝑐2, … , 𝑐𝑖, … the reactive system performs a sequence of basic actions ℎ and follows sequentially via the states of computation 𝑑1, 𝑑2, … , 𝑑𝑖, …. By 𝑡𝑟\|𝑖 we mean the trajectory(𝑑𝑖, 𝛼\|𝑖), where 𝛼\|𝑖 = (𝑐𝑖+1, 𝑑𝑖+1), (𝑐𝑖+2, 𝑑𝑖+2), … is a suffix of 𝛼, respectively. 3. 𝓛𝓟-𝑪𝑻𝑳∗ specification language These languages will be called environment behaviour patterns. A reactive system performs finite sequences of basic actions in response to control signals from the environment and thus follows in the course of its run via a sequence of data states, which are elements of a semigroup (𝑆, 𝑒, ∘), Therefore, basic predicates used in LTL formulae may be viewed as some sets of data states 𝑆′, 𝑆′ ⊆ 𝑆. These sets can be also specified in language-theoretic fashion. 309 Gnatenko A.R., Zakharov V.A. On the model checking of finite state transducers over semigroups. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 303-324 Gnatenko A.R., Zakharov V.A. On the model checking of finite state transducers over semigroups. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 303-324 Any language 𝑃 over the alphabet of basic actions 𝒜 corresponds to a predicate (set of data states) 𝑆𝑃= {[ℎ] \| ℎ∈𝑃}. As in the case of environment behaviour patterns we may distinguish a certain class 𝒫 of languages and use them as specifications of basic predicates. When these languages are used as parameters in temporal formulae then it will be assumed that they are defined constructively by means of automata, grammars, Turing machines, etc. Any language 𝑃 over the alphabet of basic actions 𝒜 corresponds to a predicate (set of data states) 𝑆𝑃= {[ℎ] \| ℎ∈𝑃}. As in the case of environment behaviour patterns we may distinguish a certain class 𝒫 of languages and use them as specifications of basic predicates. When these languages are used as parameters in temporal formulae then it will be assumed that they are defined constructively by means of automata, grammars, Turing machines, etc. Thus, we arrive at the concept of 𝓛𝓟-variants of Temporal Logics. In [17] the syntax and semantics of 𝓛𝓟-𝑳𝑻𝑳 was studied in some details in the case when both environment behaviour patterns and basic predicates are regular languages presented by finite automata. In this paper we make one step further and extend the concept of 𝓛𝓟-variants of Temporal Logics to 𝑪𝑻𝑳∗. Select an arbitrary family of environment behaviour patterns ℒ and a family of basic predicates 𝒫. 3. 𝓛𝓟-𝑪𝑻𝑳∗ specification language 303-324  𝑀, 𝑑⊨ 𝑨𝜑 ⟺ for any trajectory 𝑡𝑟′ = (𝑑, 𝛼′) in 𝑀 it is true that $𝑀, 𝑡𝑟′ ⊨𝜑;  𝑀, 𝑑⊨ 𝑨𝜑 ⟺ for any trajectory 𝑡𝑟′ = (𝑑, 𝛼′) in 𝑀 it is true that $𝑀, 𝑡𝑟′ ⊨𝜑;  if 𝜑 is a state formula then 𝑀, 𝑡𝑟⊨𝜑 ⟺ 𝑀, 𝑑0 ⊨𝜑;  𝑀, 𝑡𝑟⊨ ¬ 𝜓 ⟺ it is not true that 𝑀, 𝑡𝑟⊨ 𝜓;  𝑀, 𝑡𝑟⊨ 𝜓1 ∧ 𝜓2 ⟺ 𝑀, 𝑡𝑟⊨ 𝜓1 and 𝑀, 𝑡𝑟 \𝑚𝑜𝑑𝑒𝑙𝑠 𝜓2 ;  𝑀, 𝑡𝑟⊨ 𝑿𝑐𝜑 ⟺ 𝑐= 𝑐1 and 𝑀, 𝑑1 ⊨𝜑;  𝑀, 𝑡𝑟⊨ 𝒀𝑐𝜑 ⟺ either 𝑐≠ 𝑐1, or 𝑀, 𝑑1 ⊨𝜑;  𝑀, 𝑡𝑟⊨ 𝑭𝐿𝜑 ⟺∃ 𝑖 ≥ 0 ∶ 𝑐1𝑐2 … 𝑐𝑖∈ 𝐿 and 𝑀, 𝑡𝑟\|𝑖⊨𝜑; 𝑀 𝑮 𝑖 0 if 𝐿h 𝑀 \|𝑖  𝑀, 𝑡𝑟⊨ 𝑭𝐿𝜑 ⟺∃ 𝑖 ≥ 0 ∶ 𝑐1𝑐2 … 𝑐𝑖∈ 𝐿 and 𝑀, 𝑡𝑟\|𝑖⊨𝜑;  𝑀, 𝑡𝑟⊨ 𝑮𝐿𝜑 ⟺ ∀ 𝑖≥0 ∶ if 𝑐1𝑐2 … 𝑐𝑖∈ 𝐿 then 𝑀, 𝑡𝑟\|𝑖⊨𝜑;  𝑀, 𝑡𝑟⊨ 𝑮𝐿𝜑 ⟺ ∀ 𝑖≥0 ∶ if 𝑐1𝑐2 … 𝑐𝑖∈ 𝐿 then 𝑀, 𝑡𝑟\|𝑖⊨𝜑; 𝑀, 𝑡𝑟⊨𝜑 𝑼𝐿𝜓 ⟺∃ 𝑖≥ 0 ∶ 𝑐1𝑐2 … 𝑐𝑖∈ 𝐿, 𝑀, 𝑡𝑟\| ⊨ 𝜓 and ∀ 𝑗, 0 ≤ 𝑗 < 𝑖, if 𝑐1𝑐2 … 𝑐𝑖∈ 𝐿 then 𝑀, 𝑡𝑟\|𝑗⊨𝜑. and ∀ 𝑗, 0 ≤ 𝑗 < 𝑖, if 𝑐1𝑐2 … 𝑐𝑖∈ 𝐿 then 𝑀, 𝑡𝑟\|𝑗⊨𝜑. Observe, that operators 𝑿𝑐 and 𝒀𝑐, as well as 𝑭𝐿 and 𝑮𝐿, are dual to each other: Observe, that operators 𝑿𝑐 and 𝒀𝑐, as well as 𝑭𝐿 and 𝑮𝐿, are dual to each other: Proposition 1. For any 𝓛𝓟-𝑪𝑻𝑳∗ formula 𝜑, any 𝑐∈ 𝒞 and any 𝐿∈ ℒ, and for an arbitrary trajectory 𝑡𝑟 in 𝑀 Proposition 1. For any 𝓛𝓟-𝑪𝑻𝑳∗ formula 𝜑, any 𝑐∈ 𝒞 and any 𝐿∈ ℒ, and for an arbitrary trajectory 𝑡𝑟 in 𝑀  𝑡𝑟⊨ 𝑿𝑐𝜑 ⟺ 𝑡𝑟⊨ ¬ 𝒀𝑐 ¬ 𝜑,  𝑡𝑟⊨ 𝑭𝐿𝜑 ⟺ 𝑡𝑟⊨ ¬ 𝑮𝐿 ¬ 𝜑. As usual, other Boolean connectives like ∨, →, ≡ may be defined by means of ¬ and ∧. Some other 𝑪𝑻𝑳∗ operators like, for example, R (release) or 𝑾 (weak until) may be parametrized by environmental behaviour patterns in the same fashion. 3. 𝓛𝓟-𝑪𝑻𝑳∗ specification language The model checking problem we deal with is that of checking, given a finite state transducer Π operating over a semigroup (𝑆, 𝑒, ∘), and an 𝓛𝓟-𝑪𝑻𝑳∗ formula 𝜑, whether 𝑇𝑆(Π, 𝑆), 𝑑𝑖𝑛𝑖𝑡⊨ 𝜑 holds. When a semigroup is fixed then we use a brief notation Π ⊨𝜑. 3. 𝓛𝓟-𝑪𝑻𝑳∗ specification language The set of 𝓛𝓟-𝑪𝑻𝑳∗ formulae consists of state formulae and trajectory formulae, which are defined as follows:  each basic predicate 𝑃∈ 𝒫 is a state formula;  if 𝜑1, 𝜑2 are state formulae then ¬ 𝜑1, 𝜑1 ∧𝜑2 and 𝜑1 ∨𝜑2 are state formulae;  if 𝜓 is a trajectory formula then 𝑨𝜓 and 𝑬𝜓 are state formulae;  if 𝜑 is a state formula then 𝜑 is a trajectory formula;  if 𝜓1, 𝜓2 are trajectory formulae then ¬ 𝜓1, 𝜓1 ∧𝜓2 and 𝜓1 ∨𝜓2 are trajectory formulae;  if 𝜑, 𝜑1, 𝜑2 are state formulae, 𝑐∈ 𝒞, and 𝐿∈ ℒ then 𝑿𝑐𝜑, 𝒀𝑐𝜑, 𝑭𝐿𝜑, 𝑮𝐿𝜑 and 𝜑1𝑼𝜑2 are trajectory formulae. The specification language 𝓛𝓟-𝑪𝑻𝑳∗ is the set of all state formulae constructed as defined above. The specification language 𝓛𝓟-𝑪𝑻𝑳∗ is the set of all state formulae constructed as defined above. Now we introduce the semantics of 𝓛𝓟-𝑪𝑻𝑳∗ formulae. These formulae are interpreted over transition systems. Let 𝑀 = 𝑇𝑆(Π, 𝑆) be a transition system, 𝑑 be a state of computation in this system, and 𝑡𝑟 be a trajectory in 𝑀. Then for every state formula 𝜑 we write 𝑀, 𝑑⊨ 𝜑 to denote the fact that the assertion 𝜑 is true in the state 𝑑 of 𝑀, and for every trajectory formula 𝜓 we write 𝑀, 𝑡𝑟⊨𝜓 to denote the fact that the assertion 𝜓 holds for the trajectory 𝑡𝑟 in 𝑀. In the definition below it is assumed that 𝑀 is a transition system, 𝑑= (𝑞, 𝑠) is a state of computation in 𝑀, and 𝑡𝑟 = (𝑑0, 𝛼) is a trajectory in 𝑀 such that 𝛼= (𝑐1, 𝑑1), (𝑐2, 𝑑2), … , (𝑐𝑖, 𝑑𝑖), …. We define the satisfiability relation ⊨ by induction on the height of formulae:  𝑀, 𝑑⊨ ¬ 𝜑 ⟺ it is not true that 𝑀, 𝑑⊨𝜑;  𝑀, 𝑑⊨𝜑1 ∧𝜑2 ⟺ 𝑀, 𝑑⊨𝜑1 and 𝑀, 𝑑⊨𝜑2; 310  𝑀, 𝑑⊨ 𝑬𝜑 ⟺ there exists a trajectory 𝑡𝑟′ = (𝑑, 𝛼′) in 𝑀 such that 𝑀, 𝑡𝑟′ ⊨𝜑;  𝑀, 𝑑⊨ 𝑬𝜑 ⟺ there exists a trajectory 𝑡𝑟′ = (𝑑, 𝛼′) in 𝑀 such that 𝑀, 𝑡𝑟′ ⊨𝜑; 310 Гнатенко А.Р., Захаров В.А. О верификации конечных автоматов-преобразователей над полугруппами. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 303-324 Гнатенко А.Р., Захаров В.А. О верификации конечных автоматов-преобразователей над полугруппами. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 4. Model checking against 𝓛𝓟-𝑪𝑻𝑳∗ specifications 4. Model checking against 𝓛𝓟-𝑪𝑻𝑳∗ specifications In this paper, we discuss only the most simple case of model checking problem for finite state transducers against 𝓛𝓟-𝑪𝑻𝑳∗ formulae when g  the semigroup (𝑆, ∘, 𝑒) the transducers operate over is a free monoid, which means that 𝑆 is the set of all finite words in the alphabet 𝒜, the binary operation ∘ is concatenation of words, and the neutral element 𝑒 is the empty word 𝜀;  the semigroup (𝑆, ∘, 𝑒) the transducers operate over is a free monoid, which means that 𝑆 is the set of all finite words in the alphabet 𝒜, the binary operation ∘ is concatenation of words, and the neutral element 𝑒 is the empty word 𝜀;  the family of environment behaviour patterns ℒ is the family of regular languages in the alphabet 𝒞;  the family of environment behaviour patterns ℒ is the family of regular languages in the alphabet 𝒞;  all basic predicates in 𝒫 are specified by regular languages in the alphabet 𝒜. All regular languages used as environment behaviour patterns and basic predicate specifications are defined by means of deterministic finite state automata (DFAs). Therefore, the size of a 𝓛𝓟-𝑪𝑻𝑳∗ formula is the number of Boolean connectives and temporal operators occurred in 𝜑 plus the total size of automata used in 𝜑 to specify environment behaviour patterns and basic predicates. 311 Gnatenko A.R., Zakharov V.A. On the model checking of finite state transducers over semigroups. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 303-324 Gnatenko A.R., Zakharov V.A. On the model checking of finite state transducers over semigroups. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 303-324 Let us first describe a model checking algorithm for 𝓛𝓟-𝑪𝑻𝑳 fragment of 𝓛𝓟- 𝑪𝑻𝑳∗, which consists of all 𝓛𝓟-𝑪𝑻𝑳∗ formulae such that every temporal operator 𝐗𝒄, 𝐘𝒄, 𝐅𝑳, 𝐆𝑳 𝐔𝑳 is immediately preceded by a trajectory quantifier 𝐄 or 𝐀. In our algorithm, we involve an explicit iterative model checking techniques for the ordinary 𝑪𝑻𝑳 (see [8, 10]). Following this approach satisfiability checking of a formula 𝝋 in a state 𝒅 of a model 𝑴 is reduced to satisfiability checking of the largest subformulae of 𝝋 in the state 𝒅 and in the neighboring states of 𝑴. 4. Model checking against 𝓛𝓟-𝑪𝑻𝑳∗ specifications Therefore, to obtain an effective model checking procedure we need a construction that will model the behaviour of 𝑴 w.r.t. a target formula 𝝋. For every basic predicate 𝑷∈ 𝓟 let 𝑨𝑷 = (𝑸𝑷, 𝓐, 𝒊𝒏𝒊𝒕𝑷, 𝜹𝑷 , 𝑭𝑷) be a minimal DFA recognizing this language. Here 𝑸𝑷 is a finite set of states, 𝒊𝒏𝒊𝒕𝑷 is an initial state, 𝑭𝑷 is a set of accepting states and 𝜹𝑷 ∶ 𝑸𝑷× 𝓐→ 𝑸𝑷 is a transition function. The latter can be extended to the set 𝓐∗ in the usual fashion: 4. Model checking against 𝓛𝓟-𝑪𝑻𝑳∗ specifications In other words, a model checking procedure incrementally labels all states of a model by those subformulae of 𝝋 which are satisfied in these states. Let 𝚷 = (𝑸, 𝓒, 𝓐, 𝒒𝒊𝒏𝒊𝒕, 𝑻) be a finite state transducer over the free semigroup (𝓐∗, ⋅ , 𝜺) and let 𝝋 be an 𝓛𝓟-𝑪𝑻𝑳 formula. There are five pairs of coupled 𝓛𝓟- 𝑪𝑻𝑳 temporal operators: 𝐀𝐗𝒄 and 𝐄𝐗𝒄, 𝐀𝐘𝒄 and 𝐄𝐘𝒄, 𝐀𝐅𝑳 and 𝐄𝐅𝑳, 𝐀𝐆𝑳 and 𝐄𝐆𝑳, 𝐀𝐔𝑳 and 𝐄𝐔𝑳. As in the case of “ordinary” 𝑪𝑻𝑳 (see ), each of these couple can be expressed in terms of four main coupled operators 𝐄𝐗𝒄, 𝐄𝐘𝒄, 𝐄𝐆𝑳 and 𝐄𝐔𝑳: Proposition 2. For every formula 𝝋 the following equalities hold Proposition 2. For every formula 𝝋 the following equalities hold 1. ⊨𝐀𝐗𝑐𝜑 ≡ ¬𝐄𝐘𝑐 ¬𝜑, 2. ⊨𝐀𝐘𝑐𝜑 ≡ ¬𝐄𝐗𝑐 ¬𝜑, 3. ⊨𝐀𝐅𝐿𝜑 ≡ ¬𝐄𝐆𝐿 ¬𝜑, 4. ⊨𝐄𝐅𝐿𝜑 ≡ 𝐄[𝑡𝑟𝑢𝑒 𝐔𝐿𝜑], 5. ⊨𝐀𝐆𝐿𝜑 ≡ ¬𝐄𝐅𝐿 ¬𝜑, 6. ⊨𝐀[𝜑 𝐔𝐿𝜓] ≡ ¬𝐄[¬𝜓 𝐔𝐿 (¬𝜑∧¬𝜓)] ∧ ¬𝐄𝐆𝐿 ¬𝜓. Certainly, some other relationships like fixed-point identities are also valid in 𝓛𝓟- 𝑪𝑻𝑳∗ (see [17]) but they will not be involved in this paper. Certainly, some other relationships like fixed-point identities are also valid in 𝓛𝓟- 𝑪𝑻𝑳∗ (see [17]) but they will not be involved in this paper. We can now bound our consideration with those 𝓛𝓟-𝑪𝑻𝑳 formulae which are constructed using only ¬, ∧, 𝐄𝐗𝒄, 𝐄𝐘𝒄, 𝐄𝐆𝑳 and 𝐄𝐔𝑳. Let 𝑴 be a transition system 𝑻𝑺(𝚷, 𝓐∗) = (𝑫, 𝓒, 𝒅𝒊𝒏𝒊𝒕, 𝓣) of 𝜫 over 𝓐∗. It should be noticed that 𝑴 is, in general, infinite. Therefore, to obtain an effective model checking procedure we need a construction that will model the behaviour of 𝑴 w.r.t. a target formula 𝝋. For every basic predicate 𝑷∈ 𝓟 let 𝑨𝑷 = (𝑸𝑷, 𝓐, 𝒊𝒏𝒊𝒕𝑷, 𝜹𝑷 , 𝑭𝑷) be a minimal DFA recognizing this language. Here 𝑸𝑷 is a finite set of states, 𝒊𝒏𝒊𝒕𝑷 is an initial state, 𝑭𝑷 is a set of accepting states and 𝜹𝑷 ∶ 𝑸𝑷× 𝓐→ 𝑸𝑷 is a transition We can now bound our consideration with those 𝓛𝓟-𝑪𝑻𝑳 formulae which are constructed using only ¬, ∧, 𝐄𝐗𝒄, 𝐄𝐘𝒄, 𝐄𝐆𝑳 and 𝐄𝐔𝑳. Let 𝑴 be a transition system 𝑻𝑺(𝚷, 𝓐∗) = (𝑫, 𝓒, 𝒅𝒊𝒏𝒊𝒕, 𝓣) of 𝜫 over 𝓐∗. It should be noticed that 𝑴 is, in general, infinite. 𝜹𝑷(𝒒𝑷, 𝜺) = 𝒒𝑷 𝐚𝐧𝐝 𝜹𝑷(𝒒𝑷, 𝜸𝒂) = 𝜹𝑷(𝜹𝑷(𝒒𝑷, 𝜸), 𝒂). As it can be seen from the definition of ∼, if two states of computation 𝒅′ and 𝒅″ are equivalent and there is a trajectory 𝒕𝒓′ = (𝒅′, 𝜶′) in 𝑴, where 𝜶′ = (𝒄𝟏, 𝒅′𝟏), (𝒄𝟐, 𝒅′𝟐), …, from one of these states, then there is also a corresponding trajectory 𝒕𝒓″ = (𝒅″, 𝜶″), where 𝜶″ = (𝒄𝟏, 𝒅″𝟏), (𝒄𝟐, 𝒅″𝟐), … from the other state, such that 𝒅′𝒊∼𝒅″𝒊 holds for every 𝒊, 𝒊≥𝟏. Actually, this means that ∼ is a bisimulation relation on the state space of the transition system 𝑴. It is well known (see [3, 8]) that bisimulation preserves the satisfiability of 𝑪𝑻𝑳 formulae. The Proposition below shows that the same is true for 𝓛𝓟-𝑪𝑻𝑳. This means that the checking machine provides a finite contraction of the infinite transition system 𝑴= 𝑻𝑺(𝚷, 𝓐∗) w.r.t. satisfiability of 𝓛𝓟-𝑪𝑻𝑳 formulae. The relation ∼ is clearly an equivalence relation of finite index, and every equivalence class of states of computation in 𝑴 corresponds to a metastate of the checking machine 𝓜. As it can be seen from the definition of ∼, if two states of computation 𝒅′ and 𝒅″ are equivalent and there is a trajectory 𝒕𝒓′ = (𝒅′, 𝜶′) in 𝑴, where 𝜶′ = (𝒄𝟏, 𝒅′𝟏), (𝒄𝟐, 𝒅′𝟐), …, from one of these states, then there is ( 𝟏 𝟏) ( 𝟐 𝟐) also a corresponding trajectory 𝒕𝒓″ = (𝒅″, 𝜶″), where 𝜶″ = (𝒄𝟏, 𝒅″𝟏), (𝒄𝟐, 𝒅″𝟐), … from the other state, such that 𝒅′𝒊∼𝒅″𝒊 holds for every 𝒊, 𝒊≥𝟏. Actually, this means that ∼ is a bisimulation relation on the state space of the transition system 𝑴. It is well known (see [3, 8]) that bisimulation preserves the satisfiability of 𝑪𝑻𝑳 formulae. The Proposition below shows that the same is true for 𝓛𝓟-𝑪𝑻𝑳. This means that the checking machine provides a finite contraction of the infinite transition system 𝑴= 𝑻𝑺(𝚷, 𝓐∗) w.r.t. satisfiability of 𝓛𝓟-𝑪𝑻𝑳 formulae. Proposition 3. Suppose that 𝒅′ and 𝒅″ are two states of computation in 𝑴 such that 𝒅′ ∼𝒅″. Then 𝑴, 𝒅′ ⊨𝝋 ⇔ 𝑴, 𝒅″ ⊨𝝋. Proposition 3. Suppose that 𝒅′ and 𝒅″ are two states of computation in 𝑴 such that 𝒅′ ∼𝒅″. Then 𝑴, 𝒅′ ⊨𝝋 ⇔ 𝑴, 𝒅″ ⊨𝝋. Proof. It is carried out by induction on the nesting depth of 𝝋. When 𝝋 is a basic predicate the assertion follows from the definition of ∼. 𝜹𝑷(𝒒𝑷, 𝜺) = 𝒒𝑷 𝐚𝐧𝐝 𝜹𝑷(𝒒𝑷, 𝜸𝒂) = 𝜹𝑷(𝜹𝑷(𝒒𝑷, 𝜸), 𝒂). Let 𝑷𝟏, 𝑷𝟐, … , 𝑷𝒌 be all basic predicates occurred in the formula 𝝋. Given a transducer 𝚷 = (𝑸, 𝓒, 𝓐, 𝒒𝒊𝒏𝒊𝒕, 𝑻) and a formula 𝝋, we build a checking machine ̂ ̂ — a transducer 𝓜= (𝑸̂ , 𝓒, 𝓐, 𝒒̂𝒊𝒏𝒊𝒕, 𝑻̂), where ̂ — a transducer 𝓜= (𝑸̂ , 𝓒, 𝓐, 𝒒̂𝒊𝒏𝒊𝒕, 𝑻̂), where — a transducer 𝓜= (𝑸 , 𝓒, 𝓐, 𝒒̂𝒊𝒏𝒊𝒕, 𝑻), where  𝑄̂ = 𝑄× 𝑄𝑃1 × … × 𝑄𝑃𝑘 is a set of states (to avoid misunderstanding we will call them metastates);  𝑄̂ = 𝑄× 𝑄𝑃1 × … × 𝑄𝑃𝑘 is a set of states (to avoid misunderstanding we will call them metastates);  𝑞̂𝑖𝑛𝑖𝑡= (𝑞𝑖𝑛𝑖𝑡, 𝑖𝑛𝑖𝑡𝑃1, … , 𝑖𝑛𝑖𝑡𝑃𝑘) is an initial metastate;  𝑇̂ ⊆ 𝑄̂ × 𝒞× 𝑄̂ × 𝒜∗ is a transition relation, such that: 312 Гнатенко А.Р., Захаров В.А. О верификации конечных автоматов-преобразователей над полугруппами. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 303-324 Гнатенко А.Р., Захаров В.А. О верификации конечных автоматов-преобразователей над полугруппами. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 303-324 (𝐪̂′, 𝒄, 𝒒̂″, 𝒉) ∈𝒒̂ ⇔ { (𝒒′𝝅, 𝒄, 𝒒″𝝅, 𝒉) ∈𝑻 𝐚𝐧𝐝 𝜹𝑷𝒋(𝒒′𝑷𝒋, 𝒉) = 𝒒″𝑷𝒋 𝐟𝐨𝐫 𝐚𝐥𝐥 𝒋, 𝟏≤𝒋≤𝒌. Thus, every metastate is a tuple 𝑞̂ = (𝑞0, 𝑞1, … , 𝑞𝑘) such that 𝑞0 ∈𝑄 and 𝑞𝑗∈𝑄𝑃𝑗 for every 𝑗, 1 ≤𝑗≤𝑘, and the transition relation 𝑇̂ synchronizes transitions of 𝛱 and the automata 𝐴𝑃1, … , 𝐴𝑃𝑘 in response to every signal 𝑐. Recall that the elements of the free monoid are words 𝑠 from 𝒜∗. The checking machine ℳ induces a binary relation ∼ on the set 𝐷: for an arbitrary pair 𝑑′ = (𝑞′, 𝑠′) and 𝑑″ = (𝑞″, 𝑠″) of states of computation of 𝛱 over 𝒜∗ 𝒅′ ∼𝒅″ ⇔ { 𝒒′ = 𝒒″ 𝐚𝐧𝐝 𝜹𝑷𝒋(𝒊𝒏𝒊𝒕𝑷𝒋, 𝒔′) = 𝜹𝑷𝒋(𝒊𝒏𝒊𝒕𝑷𝒋, 𝒔″) 𝐟𝐨𝐫 𝐚𝐥𝐥 𝒋. The relation ∼ is clearly an equivalence relation of finite index, and every equivalence class of states of computation in 𝑴 corresponds to a metastate of the checking machine 𝓜. 1. 𝑐1𝑐2 … 𝑐𝑖∈𝐿 and 𝑀, 𝑡𝑟′\|𝑖⊨𝜒; 2. for all 𝑗< 𝑖 if 𝑐1𝑐2 … 𝑐𝑗∈𝐿 then 𝑀, 𝑡𝑟′\|𝑗⊨𝜓. E[𝜓 𝐔𝐿𝜒]. ∎ Each metastate 𝒒̂ = (𝒒𝟎, 𝒒𝟏, … , 𝒒𝒌) of the checking machine 𝓐 represents an equivalence class 𝑫𝒒̂ which includes all states 𝒅= (𝒒, 𝒉) ∈𝑫 such that 𝒒= 𝒒𝟎 and 𝜹𝑷𝒋(𝒊𝒏𝒊𝒕𝑷𝒋, 𝒉) = 𝒒𝒋 for all 𝒋, 𝟏≤𝒋≤𝒌. By using Proposition 3, we can correctly introduce a new satisfiability relation ⊨𝟎 on the metastates of the checking machine: 𝒒̂ ⊨ 𝟎𝝋 ⇔ 𝐟𝐨𝐫 𝐬𝐨𝐦𝐞 𝒅∈𝑫𝒒̂: 𝑴, 𝒅⊨𝝋 . 𝟎 Not only the states of the transition system 𝑴= 𝑻𝑺(𝜫, 𝑺) correspond to the metastates of the checking machine 𝓜, but also there is a relationship between the trajectories in 𝑴 and the traces in 𝓜 (they can be quite naturally called metatrajectories). More formally, every trajectory 𝒕𝒓= (𝒅𝟎, 𝜶) in 𝑴 with 𝜶= (𝒄𝟏, 𝒅𝟏)(𝒄𝟐, 𝒅𝟐) …, corresponds to a metatrajectory 𝒕𝒓 ̂ = (𝒒̂𝟎, 𝜶̂), where 𝜶̂ = (𝒄𝟏, 𝒒̂𝟏)(𝒄𝟐, 𝒒̂𝟐) … is such that for all 𝒊≥𝟎: 𝒅𝒊∈𝑫𝒒^𝒊. It is easy to see that every metatrajectory 𝒕𝒓 ̂ = (𝒒̂𝟎, 𝜶̂) corresponds to the only trajectory 𝒕𝒓= (𝒅𝟎, 𝜶), which originates in a given state 𝒅𝟎 from 𝑫𝒒̂𝟎. The well-known labeling algorithm for conventional 𝑪𝑻𝑳 and ordinary Kripke structures can be now adapted in such a way as to cope with model checking problem for 𝓛𝓟-𝑪𝑻𝑳. The algorithm operates as follows. For every metastate 𝒒̂ ∈𝑸̂ of the checking machine 𝓜 it computes a set 𝒍𝒂𝒃𝒆𝒍(𝒒̂) of all subformulae of 𝝋 satisfied in 𝒒̂. More formally, let 𝑺𝒖𝒃(𝝋) be the minimal set of 𝓛𝓟-𝑪𝑻𝑳 formulae such that: 1. 𝜑∈𝑆𝑢𝑏(𝜑); 2. if ¬𝜓∈𝑆𝑢𝑏(𝜑) then 𝜓∈𝑆𝑢𝑏(𝜑); 3. if 𝜓∧𝜒∈𝑆𝑢𝑏(𝜑) then 𝜓, 𝜒∈𝑆𝑢𝑏(𝜑); 4. if 𝐄𝐗𝑐𝜓∈𝑆𝑢𝑏(𝜑), 𝐄𝐘𝑐𝜓∈𝑆𝑢𝑏(𝜑) or 𝐄𝐆𝐿𝜓 ∈ 𝑆𝑢𝑏(𝜑) then 𝜓∈𝑆𝑢𝑏(𝜑); 5. if 𝐄[𝜓 𝐔𝐿𝜒] ∈𝑆𝑢𝑏(𝜑) then 𝜓, 𝜒∈𝑆𝑢𝑏(𝜑). The algorithm builds incrementally the sets 𝑙𝑎𝑏𝑒𝑙(𝒒̂) of all those 𝜓∈𝑆𝑢𝑏(𝜑) for which 𝒒̂ ⊨0 𝜓 holds. At the first step every 𝑙𝑎𝑏𝑒𝑙(𝒒̂) contains only basic predicates, i. e. 𝑙𝑎𝑏𝑒𝑙(𝒒̂) ⊆𝑆𝑢𝑏(𝜑) ∩ 𝒫. Then, at 𝑠𝑡𝑒𝑝 𝑖 the algorithm processes those subformulae 𝜓 whose nesting depth is 𝑖−1. Every time when the algorithm adds a subformula 𝜓 to 𝑙𝑎𝑏𝑒𝑙(𝑞̂) it thus detects that 𝑞̂ ⊨0 𝜓. 𝜓 (𝑞) 𝑞 0 𝜓 All we need now is to describe how the algorithm should process formulae of 7 types: basic predicate 𝑷, ¬𝝍, 𝝍𝟏∧𝝍𝟐, 𝑬𝑿𝒄𝝍, 𝐄𝐘𝒄𝝍, 𝑬𝑮𝑳𝝍 and 𝐄[𝝍 𝐔𝑳𝝌]. 𝜹𝑷(𝒒𝑷, 𝜺) = 𝒒𝑷 𝐚𝐧𝐝 𝜹𝑷(𝒒𝑷, 𝜸𝒂) = 𝜹𝑷(𝜹𝑷(𝒒𝑷, 𝜸), 𝒂). The cases when 𝝋= ¬𝝍 and 𝝋= 𝝍𝟏∧𝝍𝟐 are obvious. We focus only on the case of 𝝋= 𝐄[𝝍 𝐔𝑳𝝌]; the other cases when 𝝋 is of the form 𝐄𝐗𝒄𝝍, 𝐄𝐘𝒄𝝍, or 𝐄𝐆𝑳𝝍 can be treated similarly. Suppose that 𝑴, 𝒅′ ⊨𝐄[𝝍 𝐔𝑳𝝌]. Then, by the definition of 𝓛𝓟-𝑪𝑻𝑳 semantics, there exists a trajectory 𝒕𝒓′ = (𝒅′, 𝜶′), such that 𝑴, 𝒕𝒓′ ⊨𝝍 𝐔𝑳𝝌 and 𝜶′ = (𝒄𝟏, 𝒅′𝟏), (𝒄𝟐, 𝒅′𝟐), …. As it was noticed above, there is also a corresponding trajectory 𝒕𝒓″ = (𝒅″, 𝜶″) in 𝑴, where 𝜶″ = (𝒄𝟏, 𝒅″𝟏), (𝒄𝟐, 𝒅″𝟐), …, such that 𝒅′𝒊∼𝒅″𝒊 holds for every 𝒊, 𝒊≥𝟏. Then, by induction hypotheses, 𝑴, 𝒅′𝒊⊨𝝍⇔ 𝑴, 𝒅″𝒊⊨𝝍 and 𝑴, 𝒅′𝒊⊨𝝌⇔𝑴, 𝒅″𝒊⊨𝝌 hold for every 𝒊, 𝒊≥𝟏. Since 𝑴, 𝒕𝒓′ ⊨𝝍 𝐔𝑳𝝌, there exists 𝒊 such that Proof. It is carried out by induction on the nesting depth of 𝝋. When 𝝋 is a basic predicate the assertion follows from the definition of ∼. The cases when 𝝋= ¬𝝍 and 𝝋= 𝝍𝟏∧𝝍𝟐 are obvious. We focus only on the case of 𝝋= 𝐄[𝝍 𝐔𝑳𝝌]; the other cases when 𝝋 is of the form 𝐄𝐗𝒄𝝍, 𝐄𝐘𝒄𝝍, or 𝐄𝐆𝑳𝝍 can be treated similarly. 2. for all 𝑗< 𝑖 if 𝑐1𝑐2 … 𝑐𝑗∈𝐿 then 𝑀, 𝑡𝑟′\|𝑗⊨𝜓. 313 Gnatenko A.R., Zakharov V.A. On the model checking of finite state transducers over semigroups. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 303-324 Gnatenko A.R., Zakharov V.A. On the model checking of finite state transducers over semigroups. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 303-324 However, taking into account the fact that 𝜓 and 𝜒 are state formulas, we must recognize that 𝑀, 𝑡𝑟″\|𝑖⊨𝜒 and that 𝑀, 𝑡𝑟″\|𝑗⊨𝜓 every time when 𝑀, 𝑡𝑟′\|𝑗⊨𝜓. Thus, we arrive at the conclusion that 𝑀, 𝑡𝑟″ ⊨𝜓 𝐔𝐿𝜒 and, hence, 𝑀, 𝑑″ ⊨ E[𝜓 𝐔𝐿𝜒]. ∎ A subformula 𝜓1 ∧𝜓2 is added to 𝑙𝑎𝑏𝑒𝑙(𝑞̂) iff both 𝜓1, 𝜓2 ∈𝑙𝑎𝑏𝑒𝑙(𝑞̂); E[𝜓 𝐔𝐿𝜒]. ∎ As the result we obtain the reduced DLG Γ′_𝐺 (ℳ, 𝐿). For a subformula 𝐸𝐺𝐿𝜓 we construct a DLG Γ𝐺(ℳ, 𝐿) in the same fashion and delete all the nodes (𝑞̂, 𝑞𝐿) for which the relations 𝜓∉𝑙𝑎𝑏𝑒𝑙(𝑞̂) and 𝑞𝐿∈𝐹𝐿 hold simultaneously. As the result we obtain the reduced DLG Γ′_𝐺 (ℳ, 𝐿). The subformula 𝐄𝐆𝐿𝜓 is added to the set 𝑙𝑎𝑏𝑒𝑙(𝒒̂) iff Γ𝐺 ′(ℳ, 𝐿) includes the node (𝒒̂, 𝑖𝑛𝑖𝑡𝐿) and there exists a directed path in this graph from this node to some nontrivial strongly connected component (SCC), that is, to a subgraph, every node of which is reachable from any other node by some non-empty path. The subformula 𝐄𝐆𝐿𝜓 is added to the set 𝑙𝑎𝑏𝑒𝑙(𝒒̂) iff Γ𝐺 ′(ℳ, 𝐿) includes the node (𝒒̂, 𝑖𝑛𝑖𝑡𝐿) and there exists a directed path in this graph from this node to some nontrivial strongly connected component (SCC), that is, to a subgraph, every node of which is reachable from any other node by some non-empty path. As soon as all the subformulae from 𝑆𝑢𝑏(𝜑) (including the formula 𝜑) are processed we obtain the result of the model checking as Proposition 4. Let 𝒒̂𝟎∈𝑸̂ be an arbitrary metastate in 𝓜. Then 𝒒̂𝟎⊨𝟎𝑬[𝝍 𝑼𝑳𝝌] iff some node (𝒒̂′, 𝒒𝑳′) in DLG 𝜞𝑼 ′ (𝓜, 𝑳), such that 𝒒̂′ ⊨𝟎𝝌 and 𝒒𝑳′ ∈𝑭𝑳, is reachable from the node (𝒒̂𝟎, 𝒊𝒏𝒊𝒕𝑳) by a directed path. E[𝜓 𝐔𝐿𝜒]. ∎ A subformula 𝐸[𝜓 𝑈𝐿𝜒] is added to the set 𝑙𝑎𝑏𝑒𝑙(𝑞̂) iff Γ𝑈 ′ (ℳ, 𝐿) includes the node (𝑞̂ 𝑖𝑛𝑖𝑡) and there exists a directed path in this graph from this node to  To handle a subformula 𝑬[𝜓 𝑼𝐿𝜒] we construct a directed labeled graph (DLG) Γ𝑈(ℳ, 𝐿) as follows. Let 𝐴𝐿= (𝑄𝐿, 𝒞, 𝑖𝑛𝑖𝑡𝐿, 𝛿𝐿, 𝐹𝐿) be a minimal DFA that recognizes the language 𝐿. Then the nodes of Γ𝑈(ℳ, 𝐿) are all pairs (𝑞̂, 𝑞𝐿) ∈𝑄̂ × 𝑄𝐿. This DLG has an arc of the form (𝑞̂′, 𝑞𝐿′) 𝑐,ℎ → (𝑞̂′′, 𝑞𝐿′′) iff 𝑞̂′ 𝑐,ℎ → 𝑞̂'' is a transition of ℳ and 𝛿𝐿(𝑞𝐿′, 𝑐) = 𝑞𝐿″. We then delete all those nodes (𝑞^ , 𝑞𝐿) of Γ𝑈(ℳ, 𝐿) for which the relations 𝜓∉𝑙𝑎𝑏𝑒𝑙(𝑞̂), 𝜒∉𝑙𝑎𝑏𝑒𝑙(𝑞̂) and 𝑞𝐿∈𝐹𝐿 hold simultaneously and discard all arcs incoming to or outcoming from such nodes. A DLG thus reduced is denoted by Γ𝑈 ′ (ℳ, 𝐿). A subformula 𝐸[𝜓 𝑈𝐿𝜒] is added to the set 𝑙𝑎𝑏𝑒𝑙(𝑞̂) iff Γ𝑈 ′ (ℳ, 𝐿) includes the node (𝑞̂, 𝑖𝑛𝑖𝑡𝐿) and there exists a directed path in this graph from this node to some node (𝑞̂′, 𝑞𝐿′) such that 𝜒∈𝑙𝑎𝑏𝑒𝑙(𝑞̂′) and 𝑞𝐿′ ∈𝐹𝐿.  For a subformula 𝐸𝐺𝐿𝜓 we construct a DLG Γ𝐺(ℳ, 𝐿) in the same fashion and delete all the nodes (𝑞̂, 𝑞𝐿) for which the relations 𝜓∉𝑙𝑎𝑏𝑒𝑙(𝑞̂) and 𝑞𝐿∈𝐹𝐿 hold simultaneously. As the result we obtain the reduced DLG Γ′ 𝐺(ℳ𝐿) We then delete all those nodes (𝑞^ , 𝑞𝐿) of Γ𝑈(ℳ, 𝐿) for which the relations 𝜓∉𝑙𝑎𝑏𝑒𝑙(𝑞̂), 𝜒∉𝑙𝑎𝑏𝑒𝑙(𝑞̂) and 𝑞𝐿∈𝐹𝐿 hold simultaneously and discard all arcs incoming to or outcoming from such nodes. A DLG thus reduced is denoted by Γ𝑈 ′ (ℳ, 𝐿). A subformula 𝐸[𝜓 𝑈𝐿𝜒] is added to the set 𝑙𝑎𝑏𝑒𝑙(𝑞̂) iff Γ𝑈 ′ (ℳ, 𝐿) includes the node (𝑞̂, 𝑖𝑛𝑖𝑡𝐿) and there exists a directed path in this graph from this node to some node (𝑞̂′, 𝑞𝐿′) such that 𝜒∈𝑙𝑎𝑏𝑒𝑙(𝑞̂′) and 𝑞𝐿′ ∈𝐹𝐿. A subformula 𝐸[𝜓 𝑈𝐿𝜒] is added to the set 𝑙𝑎𝑏𝑒𝑙(𝑞̂) iff Γ𝑈 ′ (ℳ, 𝐿) includes the node (𝑞̂, 𝑖𝑛𝑖𝑡𝐿) and there exists a directed path in this graph from this node to some node (𝑞̂′, 𝑞𝐿′) such that 𝜒∈𝑙𝑎𝑏𝑒𝑙(𝑞̂′) and 𝑞𝐿′ ∈𝐹𝐿. For a subformula 𝐸𝐺𝐿𝜓 we construct a DLG Γ𝐺(ℳ, 𝐿) in the same fashion and delete all the nodes (𝑞̂, 𝑞𝐿) for which the relations 𝜓∉𝑙𝑎𝑏𝑒𝑙(𝑞̂) and 𝑞𝐿∈𝐹𝐿 hold simultaneously. E[𝜓 𝐔𝐿𝜒]. ∎  A basic predicate 𝑃𝑖 is added to 𝑙𝑎𝑏𝑒𝑙(𝑞̂) iff 𝑞̂ = (𝑞0, 𝑞1, … , 𝑞𝑖, … , 𝑞𝑘) and 𝑞̂𝑖∈𝐹𝑃𝑖, 𝑖≥1;  A subformula ¬𝜓 is added to 𝑙𝑎𝑏𝑒𝑙(𝑞̂) iff 𝜓∉𝑙𝑎𝑏𝑒𝑙(𝑞̂);  A subformula 𝜓1 ∧𝜓2 is added to 𝑙𝑎𝑏𝑒𝑙(𝑞̂) iff both 𝜓1, 𝜓2 ∈𝑙𝑎𝑏𝑒𝑙(𝑞̂); 314 Гнатенко А.Р., Захаров В.А. О верификации конечных автоматов-преобразователей над полугруппами. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 303-324  A subformula 𝑬𝑿𝑐𝜓 is added to 𝑙𝑎𝑏𝑒𝑙(𝑞̂) iff there exists a transition 𝑞̂ 𝑐,ℎ → 𝑞̂′ such that 𝜓∈𝑙𝑎𝑏𝑒𝑙(𝑞̂′);  A subformula 𝑬𝑿𝑐𝜓 is added to 𝑙𝑎𝑏𝑒𝑙(𝑞̂) iff there exists a transition 𝑞̂ 𝑐,ℎ → 𝑞̂′ such that 𝜓∈𝑙𝑎𝑏𝑒𝑙(𝑞̂′);  A subformula 𝑬𝑿𝑐𝜓 is added to 𝑙𝑎𝑏𝑒𝑙(𝑞̂) iff there exists a transition 𝑞̂ 𝑐,ℎ → 𝑞̂′ such that 𝜓∈𝑙𝑎𝑏𝑒𝑙(𝑞̂′);  A subformula 𝑬𝒀𝑐𝜓 is added to 𝑙𝑎𝑏𝑒𝑙(𝑞^) iff there exists a transition 𝑞̂ 𝑐,ℎ → 𝑞̂′ such that 𝜓∈𝑙𝑎𝑏𝑒𝑙(𝑞^ ′) or a transition 𝑞̂ 𝑐,ℎ → 𝑞̂′ such that 𝑐′ ≠𝑐;  To handle a subformula 𝑬[𝜓 𝑼𝐿𝜒] we construct a directed labeled graph (DLG) Γ𝑈(ℳ, 𝐿) as follows. Let 𝐴𝐿= (𝑄𝐿, 𝒞, 𝑖𝑛𝑖𝑡𝐿, 𝛿𝐿, 𝐹𝐿) be a minimal DFA that recognizes the language 𝐿. Then the nodes of Γ𝑈(ℳ, 𝐿) are all pairs (𝑞̂, 𝑞𝐿) ∈𝑄̂ × 𝑄𝐿. This DLG has an arc of the form (𝑞̂′, 𝑞𝐿′) 𝑐,ℎ → (𝑞̂′′, 𝑞𝐿′′) iff 𝑞̂′ 𝑐,ℎ → 𝑞̂'' is a transition of ℳ and 𝛿𝐿(𝑞𝐿′, 𝑐) = 𝑞𝐿″.  To handle a subformula 𝑬[𝜓 𝑼𝐿𝜒] we construct a directed labeled graph (DLG) Γ𝑈(ℳ, 𝐿) as follows. Let 𝐴𝐿= (𝑄𝐿, 𝒞, 𝑖𝑛𝑖𝑡𝐿, 𝛿𝐿, 𝐹𝐿) be a minimal DFA that recognizes the language 𝐿. Then the nodes of Γ𝑈(ℳ, 𝐿) are all pairs (𝑞̂, 𝑞𝐿) ∈𝑄̂ × 𝑄𝐿. This DLG has an arc of the form (𝑞̂′, 𝑞𝐿′) 𝑐,ℎ → (𝑞̂′′, 𝑞𝐿′′) iff 𝑞̂′ 𝑐,ℎ → 𝑞̂'' is a transition of ℳ and 𝛿𝐿(𝑞𝐿′, 𝑐) = 𝑞𝐿″. We then delete all those nodes (𝑞^ , 𝑞𝐿) of Γ𝑈(ℳ, 𝐿) for which the relations 𝜓∉𝑙𝑎𝑏𝑒𝑙(𝑞̂), 𝜒∉𝑙𝑎𝑏𝑒𝑙(𝑞̂) and 𝑞𝐿∈𝐹𝐿 hold simultaneously and discard all arcs incoming to or outcoming from such nodes. A DLG thus reduced is denoted by Γ𝑈 ′ (ℳ, 𝐿). Proof of Proposition 5 (Sketch). (⇒) Suppose, that 𝒒̂𝟎⊨𝟎𝐄𝐆𝑳𝝍. Consider an arbitrary state 𝒅𝟎∈𝑫𝒒̂𝟎. Then, by definition of ⊨𝟎 and by Proposition 3, it is true that 𝑴, 𝒅𝟎⊨𝐄𝐆𝑳𝝍. This means that there is a trajectory 𝒕𝒓= (𝒅𝟎, 𝜶), where 𝜶= (𝒄𝟏, 𝒅𝟏), (𝒄𝟐, 𝒅𝟐), …, such that 𝑴, 𝒕𝒓⊨𝐆𝑳𝝍. By the semantics of 𝓛𝓟-𝑪𝑻𝑳∗, 𝑴, 𝒅𝒊⊨𝝍 holds for every 𝒊 such that 𝒄𝟏𝒄𝟐… 𝒄𝒊∈𝑳. Consider now the corresponding metatrajectory 𝒕𝒓 ̂ = (𝒒̂𝟎, 𝜶̂) in the checking machine, where 𝜶̂ = (𝒄𝟏, 𝒒̂𝟏), (𝒄𝟐, 𝒒̂𝟐), …, and let 𝝅 = (𝒒̂𝟎, 𝒊𝒏𝒊𝒕𝑳) 𝒄𝟏,𝒉𝟏 → (𝒒̂𝟏, 𝒒𝟏𝑳) 𝒄𝟐,𝒉𝟐 → (𝒒̂𝟐, 𝒒𝟐𝑳) 𝒄𝟑,𝒉𝟑 → … , be the respective path in the DLG 𝚪𝑮(𝓜, 𝑳) which originates in the node (𝒒̂𝟎, 𝒊𝒏𝒊𝒕𝑳). Relying on Proposition 3 and taking into account the fact that 𝒒𝒊𝑳= 𝜹𝑳(𝒊𝒏𝒊𝒕𝑳, 𝒄𝟏𝒄𝟐… 𝒄𝒊) for every 𝒊, 𝒊≥𝟎, we may conclude that 𝒒̂𝒊⊨𝟎𝝍 holds for every 𝒊 such that 𝒒𝒊𝑳∈𝑭. By induction hypothesis, 𝒒̂𝒊⊨𝟎𝝍 is equivalent to 𝝍∈ 𝒍𝒂𝒃𝒆𝒍(𝒒̂𝒊). Therefore, by definition of DLG 𝚪𝑮(𝓜, 𝑳) the path 𝝅 is the infinite path which is entirely contained in the 𝚪𝑮 ′(𝓜, 𝑳). Due to the finiteness of 𝚪𝑮 ′(𝓜, 𝑳), this path may be represented as a concatenation 𝝅= 𝝅𝟏𝝅𝟐, where 𝝅𝟏 is a finite path, and 𝝅𝟐 is an infinite path passing through each of its nodes infinitely often. It is clear that the set 𝑽(𝝅𝟐) of all nodes of 𝝅𝟐 is included in some strongly connected component. Thus, a nontrivial strongly connected component is reachable from the node (𝒒̂𝟎, 𝒊𝒏𝒊𝒕𝑳) in DLG 𝚪𝑮 ′(𝓜, 𝑳). (⇐) Suppose, that a nontrivial strongly connected component is reachable from the node (𝒒̂𝟎, 𝒊𝒏𝒊𝒕𝑳) in DLG 𝚪𝑮 ′(𝓜, 𝑳). Then there exists an infinite path 𝝅 = (𝒒̂𝟎, 𝒊𝒏𝒊𝒕𝑳) 𝒄𝟏,𝒉𝟏 → (𝒒̂𝟏, 𝒒𝟏𝑳) 𝒄𝟐,𝒉𝟐 → (𝒒̂𝟐, 𝒒𝟐𝑳) 𝒄𝟑,𝒉𝟑 → … , in 𝚪𝑮 ′(𝓜, 𝑳) from the node (𝒒̂𝟎, 𝒊𝒏𝒊𝒕𝑳) Consider now the sequence of the first components 𝒒̂𝒊 of all nodes (𝒒̂𝒊, 𝒒𝒊𝑳), 𝒊≥𝟎, occurred in this path. By the definition of the DLG 𝚪𝑮 ′(𝓜, 𝑳), 𝑮( ) 1. this sequence is a metatrajectory 𝑡𝑟̂ in the checking machine ℳ, 𝑮 1. this sequence is a metatrajectory 𝑡𝑟̂ in the checking machine ℳ, 2. 𝜓∈𝑙𝑎𝑏𝑒𝑙(𝑞̂𝑖) holds for every node (𝑞̂𝑖, 𝑞𝑖𝐿) such that 𝑞𝑖𝐿∈𝐹𝐿. 2. 𝜓∈𝑙𝑎𝑏𝑒𝑙(𝑞̂𝑖) holds for every node (𝑞̂𝑖, 𝑞𝑖𝐿) such that 𝑞𝑖𝐿∈𝐹𝐿. 𝜫⊨𝝋 ⇔ 𝝋∈𝒍𝒂𝒃𝒆𝒍(𝒒̂𝒊𝒏𝒊𝒕). The correctness of this assertion is based on the following relationship: 𝒒̂ ⊨𝟎𝝋 ⇔ 𝝋∈𝒍𝒂𝒃𝒆𝒍(𝒒̂). It can be proved by applying induction on the nesting depth of formulae with the help of Proposition 3. We also need Propositions 4 and 5 to justify the induction step for formulae of the form 𝐄[𝝍 𝐔𝑳𝝌] and 𝐄𝐆𝑳𝝍. Suppose, that for every metastate 𝒒̂ ∈𝑸̂ it is true that 𝒒̂ ⊨𝟎 𝝍⇔𝝍∈𝒍𝒂𝒃𝒆𝒍(𝒒̂) and 𝒒̂ ⊨𝟎 𝝌⇔𝝌∈𝒍𝒂𝒃𝒆𝒍(𝒒̂). This statement is used as an inductive hypothesis. The correctness of this assertion is based on the following relationship: 𝒒̂ ⊨𝟎𝝋 ⇔ 𝝋∈𝒍𝒂𝒃𝒆𝒍(𝒒̂). It can be proved by applying induction on the nesting depth of formulae with the help of Proposition 3. We also need Propositions 4 and 5 to justify the induction step for formulae of the form 𝐄[𝝍 𝐔𝑳𝝌] and 𝐄𝐆𝑳𝝍. ̂ Suppose, that for every metastate 𝒒̂ ∈𝑸̂ it is true that 𝒒̂ ⊨𝟎 𝝍⇔𝝍∈𝒍𝒂𝒃𝒆𝒍(𝒒̂) and 𝒒̂ ⊨𝟎 𝝌⇔𝝌∈𝒍𝒂𝒃𝒆𝒍(𝒒̂). This statement is used as an inductive hypothesis. Proposition 4. Let 𝒒̂𝟎∈𝑸̂ be an arbitrary metastate in 𝓜. Then 𝒒̂𝟎⊨𝟎𝑬[𝝍 𝑼𝑳𝝌] iff some node (𝒒̂′, 𝒒𝑳′) in DLG 𝜞𝑼 ′ (𝓜, 𝑳), such that 𝒒̂′ ⊨𝟎𝝌 and 𝒒𝑳′ ∈𝑭𝑳, is reachable from the node (𝒒̂𝟎, 𝒊𝒏𝒊𝒕𝑳) by a directed path. 315 Gnatenko A.R., Zakharov V.A. On the model checking of finite state transducers over semigroups. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 303-324 Gnatenko A.R., Zakharov V.A. On the model checking of finite state transducers over semigroups. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 303-324 Proposition 5. Let 𝒒̂𝟎∈𝑸̂ be an arbitrary metastate in 𝓜 . Then 𝒒̂ ⊨𝟎𝑬𝑮𝑳𝝍 iff some nontrivial strongly connected component is reachable from the node (𝒒̂, 𝒊𝒏𝒊𝒕𝑳) in DLG 𝜞𝑮 ′ (𝓜, 𝑳) by a directed path. The proofs of these Propositions are straightforward adaptations of the correctness proof of the tabular model checking algorithm for 𝑪𝑻𝑳 which is discussed in much details in [8]. However, for completeness of the exposition we give here a proof of Proposition 5. The proof of Proposition 4 follows the similar line of reasoning. 𝒊≤𝒔) which is 𝑶(∥𝜫∥⋅𝟐∥𝝋∥). 𝒊≤𝒔) which is 𝑶(∥𝜫∥⋅𝟐∥𝝋∥). Because of these considerations, we get the following Theorem 1. Model checking of a finite state transducer 𝜫 operating over a free monoid against a formula 𝝋∈𝓛𝓟-𝑪𝑻𝑳 can be performed in time 𝑶(∥𝜫∥⋅𝟐∥𝝋∥). When a more general case of model checking problem of FSTs against 𝓛𝓟-𝑪𝑻𝑳∗ formulae is concerned we can rely on the well-known combining approach which is based on the interleaving application of model checking algorithms for 𝑪𝑻𝑳 and 𝑳𝑻𝑳. The details can be found in [8]. The similar procedure for 𝓛𝓟-𝑪𝑻𝑳∗ can be obtained in the same fashion by means of 𝓛𝓟-𝑪𝑻𝑳 model checking algorithm described above and 𝓛𝓟-𝑳𝑻𝑳 model checking algorithm developed in . Since this approach does not take into account any specific features of 𝓛𝓟-𝑪𝑻𝑳∗ formulae, we will not give a complete description of it. Proof of Proposition 5 (Sketch). By the induction hypothesis, the latter implies 𝑞̂𝑖⊨0 𝜓 for every metastate 𝒒̂𝑖 in this trajectory such that 𝑐1𝑐2 … 𝑐𝑖∈𝐿. Consider an arbitrary state 𝑑0 ∈𝐷𝒒̂0 and a 316 Захаров В.А. О верификации конечных автоматов-преобразователей над полугруппами. Труды 30, вып. 3, 2018 г., стр. 303-324 Гнатенко А.Р., Захаров В.А. О верификации конечных автоматов-преобразователей над полугруппами. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 303-324 trajectory 𝑡𝑟= (𝑑0, 𝛼) in 𝑀, where 𝛼= (𝑐1, 𝑑1), (𝑐2, 𝑑2), …, which corresponds to 𝑡𝑟̂ . By definition of ⊨0 and Proposition 3, 𝑀, 𝑑𝑖⊨𝜓 holds for every 𝑖 such that 𝑐1𝑐2 … 𝑐𝑖∈𝐿. Then, according to the semantics of 𝓛𝓟-𝑪𝑻𝑳∗, 𝑀, 𝑡𝑟⊨𝐆𝐿𝜓, and, h 𝑀𝑑⊨𝐄𝐆𝜓Th b f i i t d fi iti f ⊨ i Гнатенко А.Р., Захаров В.А. О верификации конечных автоматов-преобразователей над полугруппами. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 303-324 trajectory 𝑡𝑟= (𝑑0, 𝛼) in 𝑀, where 𝛼= (𝑐1, 𝑑1), (𝑐2, 𝑑2), …, which corresponds to 𝑡𝑟̂ . By definition of ⊨0 and Proposition 3, 𝑀, 𝑑𝑖⊨𝜓 holds for every 𝑖 such that 𝑐1𝑐2 … 𝑐𝑖∈𝐿. Then, according to the semantics of 𝓛𝓟-𝑪𝑻𝑳∗, 𝑀, 𝑡𝑟⊨𝐆𝐿𝜓, and, hence, 𝑀, 𝑑0 ⊨𝐄𝐆𝐿𝜓. Thus, by referring once again to definition of ⊨0, we arrive at the conclusion that 𝒒̂0 ⊨0 𝐄𝐆𝐿𝜓. ∎ Now we estimate the complexity of the model checking algorithm for 𝓛𝓟-𝑪𝑻𝑳 described above. By the size of a transducer 𝚷= (𝑸, 𝓒, 𝓐, 𝒒𝒊𝒏𝒊𝒕, 𝑻) we will mean the sum ∥𝜫∥= \|𝑸\| + \|𝑻\|. The size of a formula 𝝋 is defined as follows. Suppose that basic predicates {𝑷𝒊}𝒊=𝟏 𝒌 occurred in 𝝋 are recognized by minimal DFAs {𝑨𝑷𝒊 = 𝑸𝑷𝒊, 𝓐, 𝒊𝒏𝒊𝒕𝑷𝒊, 𝜹𝑷𝒊, 𝑭𝑷𝒊)}𝒊=𝟏 𝒌 . Suppose also that environment patterns {𝑳𝒊}𝒊=𝟏 𝒔 used in 𝝋 are recognized by minimal DFAs {𝐀𝐋𝐢 = (𝐐𝐋𝐢𝓐, 𝐢𝐧𝐢𝐭𝐋𝐢, 𝛅𝐋𝐢, 𝐅𝐋𝐢)}𝒊=𝟏 𝒔 . Then the size of 𝝋 is the sum ∥𝝋∥ = \|𝑺𝒖𝒃(𝝋)\| + ∑ \| 𝒌 𝒊=𝟏𝑸𝑷𝒊\| + ∑ \| 𝒔 𝒊=𝟏𝑸𝑳𝒊\|. 𝒊 𝒊 As it can be seen from the description of our model checking algorithm, the size of auxiliary graphs 𝚪𝑼 ′ (𝓜, 𝑳) and 𝚪𝑮 ′(𝓜, 𝑳) used in this algorithm does not exceed the value ∥𝜫∥⋅( ∏ 𝒊=𝟎 𝒌\|𝑸𝑷𝒊\|) ⋅𝐦𝐚𝐱(\|𝑸𝑳𝒊\|: 𝟏≤𝒊≤𝒔). These graphs are processed in no value ∥𝜫∥⋅( ∏ 𝒊=𝟎 𝒌\|𝑸𝑷𝒊\|) ⋅𝐦𝐚𝐱(\|𝑸𝑳𝒊\|: 𝟏≤𝒊≤𝒔). These graphs are processed in no more than \|𝑺𝒖𝒃(𝝋)\| steps. Proof of Proposition 5 (Sketch). So, the total time complexity of our model checking algorithm does not exceed the value ∥𝜫∥⋅\|𝑺𝒖𝒃(𝝋)\| ( ∏ 𝒊=𝟎 𝒌\|𝑸𝑷𝒊\|) ⋅𝐦𝐚𝐱(\|𝑸𝑳𝒊\|: 𝟏≤ algorithm does not exceed the value ∥𝜫∥⋅\|𝑺𝒖𝒃(𝝋)\| ( ∏ 𝒊=𝟎 𝒌\|𝑸𝑷𝒊\|) ⋅𝐦𝐚𝐱(\|𝑸𝑳𝒊\|: 𝟏≤ 𝒊≤𝒔) which is 𝑶(∥𝜫∥⋅𝟐∥𝝋∥). 5. 𝓛𝓟-𝑳𝑻𝑳∗ and ordinary Kripke structures In this section, we consider the model checking problem for two subfamilies of 𝓛𝓟-𝑪𝑻𝑳∗ whose semantics can be defined on ordinary Kripke structures. 317 Gnatenko A.R., Zakharov V.A. On the model checking of finite state transducers over semigroups. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 303-324 Gnatenko A.R., Zakharov V.A. On the model checking of finite state transducers over semigroups. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 303-324 Recall, that a Kripke structure over a finite set 𝑨𝑷 of atomic propositions is a quadruple 𝑴= (𝑸, 𝒒𝒊𝒏𝒊𝒕, 𝑹, 𝝆), where 𝑸 is a finite set of states which includes an initial state 𝒒𝒊𝒏𝒊𝒕, 𝑹⊆𝑸× 𝑸 is a transition relation and 𝝆: 𝑸→𝟐𝑨𝑷 is a labeling function which for each state 𝒒 gives a matching set 𝝆(𝒒) ⊆𝑨𝑷 of all atomic propositions that are evaluated to 𝒕𝒓𝒖𝒆 in this state. As usual, the size of 𝑴 is the sum ∥𝑴∥= \|𝑸\| + \|𝑹\|. Below we present two modifications of 𝓛𝓟-𝑪𝑻𝑳∗ that are well suited for model checking of Kripke structures. Recall, that a Kripke structure over a finite set 𝑨𝑷 of atomic propositions is a quadruple 𝑴= (𝑸, 𝒒𝒊𝒏𝒊𝒕, 𝑹, 𝝆), where 𝑸 is a finite set of states which includes an initial state 𝒒𝒊𝒏𝒊𝒕, 𝑹⊆𝑸× 𝑸 is a transition relation and 𝝆: 𝑸→𝟐𝑨𝑷 is a labeling function which for each state 𝒒 gives a matching set 𝝆(𝒒) ⊆𝑨𝑷 of all atomic propositions that are evaluated to 𝒕𝒓𝒖𝒆 in this state. As usual, the size of 𝑴 is the sum ∥𝑴∥= \|𝑸\| + \|𝑹\|. Below we present two modifications of 𝓛𝓟-𝑪𝑻𝑳∗ that are well suited for model checking of Kripke structures. Given a Kripke structure 𝑴= (𝑸, 𝒒𝒊𝒏𝒊𝒕, 𝑹, 𝑳), consider a set of 𝓛𝓟-𝑪𝑻𝑳∗ formulae where 𝓛is a family of regular languages over one-letter alphabet {𝒄} and 𝓟= 𝑨𝑷 (we denote this formulae by 𝓛𝓟-1-𝑪𝑻𝑳∗) and a transition system 𝑴𝒄= (𝑸, {𝒄}, 𝒒𝒊𝒏𝒊𝒕, 𝑹𝒄, 𝑳) where (𝒒′, 𝒄, 𝒒″) ∈𝑹𝒄 iff (𝒒′, 𝒒″) ∈𝑹. Then for 𝒒∈𝑸 the relation 𝒒⊨𝑷 holds iff 𝑷∈𝝆(𝒒). The semantics of more complex formulae is defined exactly as in Section 3. Some 𝓛𝓟-1-𝑪𝑻𝑳∗ formulae have an ability to keep track of the number of steps of the run. 5. 𝓛𝓟-𝑳𝑻𝑳∗ and ordinary Kripke structures For example, an 𝓛𝓟-1-𝑳𝑻𝑳 formula 𝐀𝐆𝑳 𝝋, where 𝑳= {𝒄𝟐𝒏} is a regular language which contains all 1-letter words of even length, expresses the assertion that 𝝋 holds at every even step of a run. By using the techniques of Ehrenfeucht- Fraisse games for Temporal Logics developed and studied in [11] one can prove that this property cannot be specified by means of usual 𝑳𝑻𝑳. This certifies that 𝓛𝓟- 1-𝑪𝑻𝑳∗ is more expressive than 𝑪𝑻𝑳∗ and justifies its use as a new specification language for finite state transducers and Kripke structures. Observe, that given a set 𝑨𝑷 of all atomic propositions used in formulae we can use the 𝑴𝒄 directly as a checking machine 𝓜 for the algorithm described in Section 4. Suppose that formula 𝝋 refers to 1-letter regular languages 𝑳𝟏, 𝑳𝟐, … , 𝑳𝒔 as the parameters of temporal operators, and every language 𝑳𝒊, 𝟏≤𝒊≤𝒔, is recognized by a DFA with a set of states 𝑸𝑳𝒊. Then the size of the graphs used in this algorithm does not exceed the value ∥𝑴∥⋅𝐦𝐚𝐱(\|𝑸𝑳𝒊\|: 𝟏≤𝒊≤𝒔) which is 𝑶(∥𝑴∥⋅∥𝝋∥), where ∥𝝋∥= \|𝑺𝒖𝒃(𝝋)\| + ∑ \| 𝒔 𝒊=𝟎𝑸𝑳𝒊\|. Another modification of the Kripke structure 𝑴 allows one to encode more detailed information of the computation flow. Let 𝜮= 𝟐𝑨𝑷. For each state 𝒒 in 𝑴 there exists a letter 𝝈𝝆(𝒒) ∈𝜮 corresponding to the label 𝝆(𝒒) assigned to this state. Let 𝑴𝑨𝑷= (𝑸∪{𝒆𝒓𝒓}, 𝒒𝒊𝒏𝒊𝒕, 𝑹𝑨𝑷, 𝝆𝑨𝑷) be a transition system for 𝑴, where for every 𝒒∈𝑸the following equalities hold: 𝝆𝑨𝑷(𝒒) = 𝝆(𝒒) 𝝆𝑨𝑷(𝒆𝒓𝒓) = {𝒆𝒓𝒓} Another modification of the Kripke structure 𝑴 allows one to encode more detailed information of the computation flow. Let 𝜮= 𝟐𝑨𝑷. For each state 𝒒 in 𝑴 there exists a letter 𝝈𝝆(𝒒) ∈𝜮 corresponding to the label 𝝆(𝒒) assigned to this state. information of the computation flow. Let 𝜮= 𝟐 . For each state 𝒒 in 𝑴 there exists a letter 𝝈𝝆(𝒒) ∈𝜮 corresponding to the label 𝝆(𝒒) assigned to this state. Then consider a specification language 𝓛𝓟-𝒏-𝑪𝑻𝑳∗ which is a set of all such formulae where ℒ is a family of regular languages over 𝛴 and 𝒫 = AP. To model Theorem 2. 1. There exists an algorithm for model checking of a Kripke structure 𝑀 against a formula 𝜑∈𝓛𝓟-1-𝑪𝑻𝑳 with time complexity 𝑂(∥𝑀∥⋅∥𝜑∥2). 2. There exists an algorithm for model checking of a Kripke structure 𝑀 against a formula 𝜑∈𝓛𝓟-n-𝑪𝑻𝑳with time complexity 𝑂(∥𝑀∥⋅∥𝜑∥2⋅2\|𝐴𝑃\|) 1. There exists an algorithm for model checking of a Kripke structure 𝑀 against a formula 𝜑∈𝓛𝓟-1-𝑪𝑻𝑳 with time complexity 𝑂(∥𝑀∥⋅∥𝜑∥2). 1. There exists an algorithm for model checking of a Kripke structure 𝑀 against a formula 𝜑∈𝓛𝓟-1-𝑪𝑻𝑳 with time complexity 𝑂(∥𝑀∥⋅∥𝜑∥2). 2. There exists an algorithm for model checking of a Kripke structure 𝑀 against a formula 𝜑∈𝓛𝓟-n-𝑪𝑻𝑳 with time complexity 𝑂(∥𝑀∥⋅∥𝜑∥2⋅2\|𝐴𝑃\|). 2. There exists an algorithm for model checking of a Kripke structure 𝑀 against a formula 𝜑∈𝓛𝓟-n-𝑪𝑻𝑳 with time complexity 𝑂(∥𝑀∥⋅∥𝜑∥2⋅2\|𝐴𝑃\|). As it can be seen from this theorem, the exponential complexity of model checking procedure described in Section 4 is due to the language-theoretic nature of basic predicates used in 𝓛𝓟-𝑪𝑻𝑳∗. 5. 𝓛𝓟-𝑳𝑻𝑳∗ and ordinary Kripke structures Let 𝑴𝑨𝑷= (𝑸∪{𝒆𝒓𝒓}, 𝒒𝒊𝒏𝒊𝒕, 𝑹𝑨𝑷, 𝝆𝑨𝑷) be a transition system for 𝑴, where for every 𝒒∈𝑸 the following equalities hold: 𝝆𝑨𝑷(𝒒) = 𝝆(𝒒), 𝝆𝑨𝑷(𝒆𝒓𝒓) = {𝒆𝒓𝒓} and 𝑹𝑨𝑷 ⊆ 𝑸 × 𝟐𝑨𝑷 × 𝑸 is a minimal transition relation such that: Let 𝑴𝑨𝑷= (𝑸∪{𝒆𝒓𝒓}, 𝒒𝒊𝒏𝒊𝒕, 𝑹𝑨𝑷, 𝝆𝑨𝑷) be a transition system for 𝑴, where for every 𝒒∈𝑸 the following equalities hold: 𝝆𝑨𝑷(𝒒) = 𝝆(𝒒), 𝝆𝑨𝑷(𝒆𝒓𝒓) = {𝒆𝒓𝒓} and 𝑹𝑨𝑷 ⊆ 𝑸 × 𝟐𝑨𝑷 × 𝑸 is a minimal transition relation such that:  for each transition (𝑞′, 𝑞″) of the Kripke structure 𝑀 there exists a fair transition (𝑞′, 𝜎𝜌(𝑞″), 𝑞″) and erroneous transitions (𝑞′, 𝜎, 𝑒𝑟𝑟) for each 𝜎≠𝜎𝜌(𝑞″);  (𝑒𝑟𝑟, 𝜎, 𝑒𝑟𝑟) ∈𝑅𝐴𝑃 holds for each 𝜎∈𝛴 and (𝑒𝑟𝑟, 𝜎, 𝑞) ∉𝑅𝐴𝑃 holds for each 𝑞 ≠ 𝑒𝑟𝑟. Then consider a specification language 𝓛𝓟-𝒏-𝑪𝑻𝑳∗ which is a set of all such formulae where ℒ is a family of regular languages over 𝛴 and 𝒫 = AP. To model Then consider a specification language 𝓛𝓟-𝒏-𝑪𝑻𝑳∗ which is a set of all such formulae where ℒ is a family of regular languages over 𝛴 and 𝒫 = AP. To model 318 Гнатенко А.Р., Захаров В.А. О верификации конечных автоматов-преобразователей над полугруппами. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 303-324 check a transition system 𝑀𝐴𝑃 against these formulae one needs to process only the states in 𝑄 and only the fair transitions. To do so, we replace all state formulae of type 𝐀𝜑 with 𝑨(𝑮 ¬ 𝑒𝑟𝑟 ⟶𝜑) and all state formulae of type 𝑬𝜑 with 𝑬(𝑮 ¬ 𝑒𝑟𝑟∧𝜑). The transition system 𝑀𝐴𝑃 thus obtained may as well be used as a checking machine for the model checking algorithm described in Section 4. Th b th f ll i th h ld Thereby, the following theorem holds. 6. Related papers and conclusion Actually, the idea of providing parameterization of temporal operators is not new. In [27] right-linear grammar patterns were offered to define new temporal operators. The same kind of temporal patterns but specified by means of finite state automata were introduced in [18, 24]. For these extensions it was proved that they have the same expressiveness as S𝟏S and that satisfiability checking problem in these logics is PSPACE-complete. We did not pursue a goal of merely expanding the expressive possibilities of 𝑪𝑻𝑳∗; our aim was to make 𝑪𝑻𝑳∗ more adequate for describing the behaviour of reactive systems. Almost the same kind of parametrization is used in Dynamic LTL. However, our extension of 𝑪𝑻𝑳∗ differs from that which was developed in [14], since in our logic basic predicates are also parameterized. The 𝓛𝓟-𝑪𝑻𝑳∗ formulae allows one to specify and verify the behaviour of finite state transducers that operate over semigroups as well as classical Kripke structures. Moreover, when Kripke structures are concerned 𝓛𝓟-𝑪𝑻𝑳∗ has more expressive power than conventional temporal logics. But the place of 𝓛𝓟-𝑪𝑻𝑳∗ in the expressive hierarchy of specification languages, such as S1S, PDL or 𝝁-calculus, has not yet been established and remains a matter for our further research. The results of this paper combined with the results of [17] provide positive solution to model checking for transducers over free semigroups. Free semigroups is the most simple algebraic structure which can be used for interpretation of basic actions performed by transducers when they are regarded as formal models of sequential reactive systems. Next, we are going to find out whether model checking algorithms could be built for transducers operating over more specific semigroups. Some preliminary results showed that this is not an easy problem. In [12] we proved that it is undecidable for the case of Abelian groups and free commutative semigroups. Actually, the idea of providing parameterization of temporal operators is not new. In [27] right-linear grammar patterns were offered to define new temporal operators. Actually, the idea of providing parameterization of temporal operators is not new. In [27] right-linear grammar patterns were offered to define new temporal operators. The same kind of temporal patterns but specified by means of finite state automata were introduced in [18, 24]. For these extensions it was proved that they have the same expressiveness as S𝟏S and that satisfiability checking problem in these logics is PSPACE-complete. 6. Related papers and conclusion We did not pursue a goal of merely expanding the expressive possibilities of 𝑪𝑻𝑳∗; our aim was to make 𝑪𝑻𝑳∗ more adequate for describing the behaviour of reactive systems. Almost the same kind of parametrization is used in Dynamic LTL. However, our extension of 𝑪𝑻𝑳∗ differs from that which was developed in [14], since in our logic basic predicates are also parameterized. The 𝓛𝓟-𝑪𝑻𝑳∗ formulae allows one to specify and verify the behaviour of finite state transducers that operate over semigroups as well as classical Kripke structures. Moreover, when Kripke structures are concerned 𝓛𝓟-𝑪𝑻𝑳∗ has more expressive power than conventional temporal logics. But the place of 𝓛𝓟-𝑪𝑻𝑳∗ in the expressive hierarchy of specification languages, such as S1S, PDL or 𝝁-calculus, has not yet been established and remains a matter for our further research. The results of this paper combined with the results of [17] provide positive solution to model checking for transducers over free semigroups. Free semigroups is the most simple algebraic structure which can be used for interpretation of basic actions performed by transducers when they are regarded as formal models of sequential reactive systems. Next, we are going to find out whether model checking algorithms could be built for transducers operating over more specific semigroups. Some preliminary results showed that this is not an easy problem. In [12] we proved that it is undecidable for the case of Abelian groups and free commutative semigroups. 319 Gnatenko A.R., Zakharov V.A. On the model checking of finite state transducers over semigroups. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 303-324 Gnatenko A.R., Zakharov V.A. On the model checking of finite state transducers over semigroups. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 303-324 It is also interesting how much the complexity of model checking algorithms for 𝓛𝓟-𝑪𝑻𝑳∗ depends on languages that are used as parameters of temporal operators. We assume that model checking problem becomes undecidable when context-free languages are allowed for this purpose. The complexity issues of model checking for regular variant of 𝓛𝓟-𝑪𝑻𝑳∗ also need further research. We assume that even for regular 𝓛𝓟-𝑪𝑻𝑳 this problem is PSPACE-complete. 6. Related papers and conclusion As for practical application of the results obtained, the most important issue is that of adapting the existing means of working with finite automata to widely known model checking tools (like SPIN, 𝝂-SMV, etc.) in order to be able to effectively implement the proposed model checking algorithms for 𝓛𝓟-𝑪𝑻𝑳∗. Acknowledgments The authors of the article thank the anonymous reviewers for their valuable comments and advice on improving the article. This work was supported by the Russian Foundation for Basic Research, Grant N 18-01-00854. References 303-324 [13]. Griffiths T. The unsolvability of the equivalence problem for free nondeterministic generalized machines. Journal of the ACM, vol. 15, 1968, pp. 409-413. [14]. HenriksenJ. G., Thiagarajan P.S. Dynamic linear time temporal logic. Annals of Pure and Applied Logic, vol. 96, 1999, pp.187-207. [15]. Hu Q., D'Antoni L. Automatic Program Inversion using Symbolic Transducers. In Proceedings of the 38-th ACM SIGPLAN Conference on Programming Language Design and Implementation, 2017, pp. 376-389. [16]. Ibarra O. The unsolvability of the equivalence problem for Efree NGSM’s with unary input (output) alphabet and applications. SIAM Journal on Computing, vol. 4, 1978, pp. 524-532. [17]. Kozlova D. G., Zakharov V. A. On the model checking of sequential reactive systems. Proceedings of the25-th International Workshop on Concurrency, Specification and Programming (CS&P 2016), CEUR Workshop Proceedings, vol. 1698, 2016, pp. 233- 244. [18]. Kupferman O., Piterman N., Vardi M.Y. Extended Temporal Logic Revisited. In Proceedings of 12-th International Conference on Concurrency Theory, 2001, pp. 519- 535. [19]. Schutzenberger M. P. Sur les relations rationnelles. In Proceedings of Conference on Automata Theory and Formal Languages, 1975, pp. 209-213. [20]. Sakarovitch J., de Souza R. On the decomposition of k-valued rational relations. In Proceedings of 25-th International Symposium on Theoretical Aspects of Computer Science, 2008, pp. 621-632. [21]. Sakarovitch J., de Souza R. On the decidability of bounded valuedness for transducers. In Proceedings of the 33-rd International Symposium on Mathematical Foundations of Computer Science, 2008, pp. 588-600. [22]. De Souza R. On the decidability of the equivalence for k-valued transducers. In Proceedings of 12-th International Conference on Developments in Language Theory, 2008, pp. 252-263. [23]. Thakkar J., Kanade A., Alur R. A transducer-based algorithmic verification of retransmission protocols over noisy channels. In Proceedings of IFIP Joint International Conference on Formal Techniques for Distributed Systems, 2013, pp. 209-224. q y pp [24]. Vardi M.Y., Wolper P. Yet Another Process Logic. Logic of Programs, 1983, pp. 501- 512. [25]. Veanes M., Hooimeijer P., Livshits B. et al. Symbolic finite state transducers: algorithms and applications. In Proceedings of the 39-th ACM SIGACT-SIGPLAN Symposium on Principles of Programming Languages, ACM SIGPLAN Notices, vol. 147, 2012, pp. 137-150. [26]. Weber A. Decomposing finite-valued transducers and deciding their equivalence. SIAM Journal on Computing. vol. 22, 1993, pp. 175-202. [27]. Wolper P. Temporal Logic Can Be More Expressive. Information and Control, vol. 56, N 1/2, 1983, pp. 72-99. [28]. Wolper P., Boigelot B. References [1]. Alur R., Cerny P. Streaming transducers for algorithmic verification of single-pass list- processing programs. In Proceedings of 38-th ACM SIGACT-SIGPLAN Symposium on Principles of Programming Languages, 2011, pp. 599-610 [2]. Alur R., Moarref S., and Topcu U. Pattern-based refinement of assume-guarantee specifications in reactive synthesis. In Proceedings of 21-st International Conference on Tools and Algorithms for the Construction and Analysis of Systems, 2015, pp. 501-516. [3] Baier C Katoen J Principles of Model Checking 2008 MIT Press [3]. Baier C., Katoen J. Principles of Model Checking, 2008, MIT Press. [4]. Blattner M., Head T. The decidability of equivalence for deterministic finite transducers. Journal of Computer and System Sciences, 1979, vol. 1, pp. 45-49. p y pp [5]. Blattner M., T. Head T. Single-valued a-transducers. Journal of Computer and System Sciences, vol. 15, 1977, pp. 310-327. [6]. Culik K, Karhumaki J. The equivalence of finite-valued transducers (on HDTOL languages) is decidable. Theoretical Computer Science, 1986, vol. 47, pp. 71-84. [7]. Bouajjani A., Jonsson B., Nilsson M., Touili T. Regular Model Checking. Proceedings of 12-th International Conference on Computer Aided Verification, 2000, p. 403-418. 8]. Clarke (Jr.) E. M., Grumberg O., Peled D. A. Model Checking. MIT Press, 1999 ( ) g g [9]. Diekert V., Rozenberg G. eds. The Book of Traces, 1995, World Scientific, Singapore. [10]. Emerson E.A., Halpern J.Y. Decision procedures and expressiveness in the temporal l i f b hi i J l f C d S S i l 30 N 1 1985 g g [9]. Diekert V., Rozenberg G. eds. The Book of Traces, 1995, World Scientific, Singapore. 10 A l i i d d i i h l Diekert V., Rozenberg G. eds. The Book of Traces, 1995, World Scientific, Singap g g p [10]. Emerson E.A., Halpern J.Y. Decision procedures and expressiveness in the temporal logic of branching time. Journal of Computer and System Sciences, vol. 30, N 1, 1985, pp. 1–24. [11]. Etessami K., WilkeT. An Until Hierarchy and Other Applications of and Ehrenfeucht- Fraisse Game for Temporal Logic. Information and Computation, vol. 160, 2000, pp. 88-108. [12]. Gnatenko A.R., Zakharov V. A. On the complexity of verification of finite state machines over commutative semigroups. In Proceedings of the 18-th International Conference "Problems of Theoretical Cybernetics'', 2017, pp. 68-70 (in Russian). 320 Гнатенко А.Р., Захаров В.А. О верификации конечных автоматов-преобразователей над полугруппами. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. DOI: 10.15514/ISPRAS-2018-30(3)-21 DOI: 10.15514/ISPRAS-2018-30(3)-21 Для цитирования: Гнатенко А.Р., Захаров В.А. О верификации конечных автоматов- преобразователей над полугруппами. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 303-324 (на английском языке). DOI: 10.15514/ISPRAS-2018-30(3)-21 Для цитирования: Гнатенко А.Р., Захаров В.А. О верификации конечных автоматов- преобразователей над полугруппами. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 303-324 (на английском языке). DOI: 10.15514/ISPRAS-2018-30(3)-21 Gnatenko A.R., Zakharov V.A. On the model checking of finite state transducers over semigroups. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 303-324 natenko A.R., Zakharov V.A. On the model checking of finite state transducers over semigroups. Trudy ISP AN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 303-324 References Verifying systems with infinite but regular state spaces. In Proceedings of the 10-th Int. Conf. on Computer Aided Verification (CAV-1998), 1998, pp. 88-97. [29]. Zakharov V.A. Equivalence checking problem for finite state transducers over semigroups. In Proceedings of the~6-th International Conference on Algebraic Informatics (CAI-2015), 2015, pp. 208-221. 321 1 А.Р. Гнатенко<gnatenko.cmc@gmail.com> 2 В.А. Захаров<zakh@cs.msu.su> 1 Московский государственный университет им. М. В. Ломоносова, Московский государственный университет им. М. В. Ломоносова, 119991, Российская Федерация, Москва, Ленинские горы, д. 1 119991, Российская Федерация, Москва, Ленинские горы, д. 1 2 Национальный исследовательский университет Высшая школа экономики, 101000, Российская Федерация, Москва, ул. Мясницкая, д. 20 2 Национальный исследовательский университет Высшая школа экономики, 101000, Российская Федерация, Москва, ул. Мясницкая, д. 20 Аннотация. Последовательные реагирующие системы – это программы, которые взаимодействуют с окружением, получая от него сигналы или запросы, и реагируют на эти запросы, проводя операции с данными. Подобные системы могут служить моделью для многих программ: драйверов, систем реального времени, сетевых протоколов и др. В статье исследуются задача верификации программ такого вида. В качестве формальных моделей для реагирующих систем мы используем конечные автоматы-преобразователи, работающие над полугруппами. Для описания поведения автоматов-преобразователей введён новый язык спецификаций LP-CTL. В его основу положена темпоральная логика CTL. Этот язык спецификаций имеет две характерные особенности: 1) каждый темпоральный оператор снабжён регулярным выражением над входным алфавитом автомата, и 2) каждое атомарное высказывание задается регулярным выражением над выходным алфавитом автомата-преобразователя. В данной работе представлен табличный алгоритм проверки выполнимости формул LP-CTL на моделях конечных автоматов-преобразователей, работающих над свободными полугруппами. Доказана корректность предложенного алгоритма и получена оценка его сложности. Кроме того, рассмотрен специальный фрагмент языка LP-CTL, содержащий в качестве параметров темпоральных операторов только регулярные выражения над однобуквенным алфавитом. Показано, что этот фрагмента применим для спецификаций обычных моделей Крипке, и при этом его выразительные возможности превосходят обычную логику CTL. Ключевые слова: реагирующая система, автомат-преобразователь, верификация, проверка на модели, темпоральная логика, конечный автомат, регулярный язык. Список литературы [1]. Alur R., Cerny P. Streaming transducers for algorithmic verification of single-pass list- processing programs. In Proceedings of 38-th ACM SIGACT-SIGPLAN Symposium on Principles of Programming Languages, 2011, pp. 599-610 [1]. Alur R., Cerny P. Streaming transducers for algorithmic verification of single-pass list- processing programs. In Proceedings of 38-th ACM SIGACT-SIGPLAN Symposium on Principles of Programming Languages, 2011, pp. 599-610 322 Гнатенко А.Р., Захаров В.А. О верификации конечных автоматов-преобразователей над полугруппами. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 303-324 Гнатенко А.Р., Захаров В.А. О верификации конечных автоматов-преобразователей над полугруппами. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 303-324 [2]. Alur R., Moarref S., and Topcu U. Pattern-based refinement of assume-guarantee specifications in reactive synthesis. In Proceedings of 21-st International Conference on Tools and Algorithms for the Construction and Analysis of Systems, 2015, pp. 501-516. g y y [3]. Baier C., Katoen J. Principles of Model Checking, 2008, MIT Press. [4]. Blattner M., Head T. The decidability of equivalence for deterministic finite transducers. Journal of Computer and System Sciences, 1979, vol. 1, pp. 45-49. [5]. Blattner M., T. Head T. Single-valued a-transducers. Journal of Computer and System Sciences, vol. 15, 1977, pp. 310-327. [6]. Culik K, Karhumaki J. The equivalence of finite-valued transducers (on HDTOL languages) is decidable. Theoretical Computer Science, 1986, vol. 47, pp. 71-84. [7]. Bouajjani A., Jonsson B., Nilsson M., Touili T. Regular Model Checking. Proceedings of 12-th International Conference on Computer Aided Verification, 2000, p. 403-418. Clarke (Jr.) E. M., Grumberg O., Peled D. A. Model Checking. MIT Press, 1999. [9]. Diekert V., Rozenberg G. eds. The Book of Traces, 1995, World Scientific, Sin [10]. Emerson E.A., Halpern J.Y. Decision procedures and expressiveness in the temporal logic of branching time. Journal of Computer and System Sciences, vol. 30, N 1, 1985, pp. 1–24. [11]. Etessami K., WilkeT. An Until Hierarchy and Other Applications of and Ehrenfeucht- Fraisse Game for Temporal Logic. Information and Computation, vol. 160, 2000, pp. 88-108. [12]. Гнатенко А.Р., Захаров В.А. О сложности верификации конечных автоматов- преобразователей над коммутативными полугруппами. Материалы XVIII международной конференции «Проблемы теоретической кибернетики»' (Пенза, 20-24 июня, 2017), стр. 68-70. [13]. Griffiths T. The unsolvability of the equivalence problem for free nondeterministic generalized machines. Journal of the ACM, vol. 15, 1968, pp. 409-413. g pp [14]. HenriksenJ. G., Thiagarajan P.S. Dynamic linear time temporal logic. Annals of Pure and Applied Logic, vol. 96, 1999, pp.187-207. [15]. Список литературы Hu Q., D'Antoni L. Automatic Program Inversion using Symbolic Transducers. In Proceedings of the 38-th ACM SIGPLAN Conference on Programming Language Design and Implementation, 2017, pp. 376-389. [16]. Ibarra O. The unsolvability of the equivalence problem for Efree NGSM’s with unary input (output) alphabet and applications. SIAM Journal on Computing, vol. 4, 1978, pp. 524-532. [17]. Kozlova D. G., Zakharov V. A. On the model checking of sequential reactive systems. Proceedings of the25-th International Workshop on Concurrency, Specification and Programming (CS&P 2016), CEUR Workshop Proceedings, vol. 1698, 2016, pp. 233- 244. [18]. Kupferman O., Piterman N., Vardi M.Y. Extended Temporal Logic Revisited. In Proceedings of 12-th International Conference on Concurrency Theory, 2001, pp. 519- 535. [19]. Schutzenberger M. P. Sur les relations rationnelles. In Proceedings of Conference on Automata Theory and Formal Languages, 1975, pp. 209-213. [20]. Sakarovitch J., de Souza R. On the decomposition of k-valued rational relations. In Proceedings of 25-th International Symposium on Theoretical Aspects of Computer Science, 2008, pp. 621-632. 323 Gnatenko A.R., Zakharov V.A. On the model checking of finite state transducers over semigroups. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 303-324 Gnatenko A.R., Zakharov V.A. On the model checking of finite state transducers over semigroups. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 303-324 Gnatenko A.R., Zakharov V.A. On the model checking of finite state transducers over semigroups. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 303-324 [21]. Sakarovitch J., de Souza R. On the decidability of bounded valuedness for transducers. In Proceedings of the 33-rd International Symposium on Mathematical Foundations of Computer Science, 2008, pp. 588-600. In Proceedings of the 33-rd International Symposium on Mathematical Foundations of Computer Science, 2008, pp. 588-600. [22]. De Souza R. On the decidability of the equivalence for k-valued transducers. In Proceedings of 12-th International Conference on Developments in Language Theory, 2008, pp. 252-263. pp [23]. Thakkar J., Kanade A., Alur R. A transducer-based algorithmic verification of retransmission protocols over noisy channels. In Proceedings of IFIP Joint International Conference on Formal Techniques for Distributed Systems, 2013, pp. 209-224. [24]. Vardi M.Y., Wolper P. Yet Another Process Logic. Logic of Programs, 1983, pp. 501- 512. [25]. Veanes M., Hooimeijer P., Livshits B. et al. Symbolic finite state transducers: algorithms and applications. In Proceedings of the 39-th ACM SIGACT-SIGPLAN Symposium on Principles of Programming Languages, ACM SIGPLAN Notices, vol. pp [29]. Zakharov V.A. Equivalence checking problem for finite state transducers over semigroups. In Proceedings of the~6-th International Conference on Algebraic Informatics (CAI-2015), 2015, pp. 208-221. Список литературы 147, 2012, pp. 137-150. [26]. Weber A. Decomposing finite-valued transducers and deciding their equivalence. SIAM Journal on Computing. vol. 22, 1993, pp. 175-202. [27]. Wolper P. Temporal Logic Can Be More Expressive. Information and Control, vol. 56, N 1/2, 1983, pp. 72-99. [28]. Wolper P., Boigelot B. Verifying systems with infinite but regular state spaces. In Proceedings of the 10-th Int. Conf. on Computer Aided Verification (CAV-1998), 1998, pp. 88-97. [29]. Zakharov V.A. Equivalence checking problem for finite state transducers over semigroups. In Proceedings of the~6-th International Conference on Algebraic Informatics (CAI-2015), 2015, pp. 208-221. 324 324 On the verification of strictly deterministic behavior of Timed Finite State Machines E.M. Vinarskii <vinevg2015@gmail.com> V.A. Zakharov <zakh@cs.msu.su> Lomonosov Moscow State University, GSP-1, Leninskie Gory, Moscow, 119991, Russia Abstract. Finite State Machines (FSMs) are widely used as formal models for solving numerous tasks in software engineering, VLSI design, development of telecommunication systems, etc. To describe the behavior of a real-time system one could supply FSM model with clocks — a continuous time parameters with real values. In a Timed FSM (TFSM) inputs and outputs have timestamps, and each transition is equipped with a timed guard and an output delay to indicate time interval when the transition is active and how much time does it take to produce an output. A variety of algorithms for equivalence checking, minimization and test generation were developed for TFSMs in many papers. A distinguishing feature of TFSMs studied in these papers is that the order in which output letters occur in an output timed word does not depend on their timestamps. We think that such behavior of a TFSM is not realistic from the point of view of an outside observer. In this paper we consider a more advanced and adequate TFSM functioning; in our model the order in which outputs become visible to an outsider is determined not only by the order of inputs, but also by de lays required for their processing. When the same sequence of transitions is performed by a TFSM modified in a such way, the same outputs may follow in different order depending on the time when corresponding inputs become available to the machine. A TFSM is called strictly deterministic if every input timed word activates no more than one sequence of transitions (trace) and for any input timed word which activates this trace the letters in the output words always follows in the same order (but, maybe, with different timestamps). We studied the problem of checking whether a behavior of an improved model of TFSM is strictly deterministic. To this end we showed how to verify whether an arbitrary given trace in a TFSM is steady, i.e. preserves the same order of output letters for every input timed word which activates this trace. Further, having the criterion of trace steadiness, we developed an exhaustive algorithm for checking the property of strict determinacy of TFSMs. Exhaustive search in this case can hardly be avoided: we proved that determinacy checking problem for our model of TFSM is co-NP-hard. Keywords: Timed Finite State Machines; strictly deterministic behavior DOI: 10.15514/ISPRAS-2018-30(3)-22 For citation: Vinarskii E.M., Zakharov V.A. On the verification of strictly deterministic behaviour of Timed Finite State Machines. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 325-340. DOI: 10.15514/ISPRAS-2018-30(3)-22 1. Introduction Finite State Machines (FSMs) are widely used as formal models for analysis and synthesis of information processing systems in software engineering, VLSI design, telecommunication, etc. The most attractive feature of this model of computation is its simplicity — many important synthesis and analysis problems (equivalence checking, minimization, test derivation, etc.) for classical FSMs can be solved in time which is almost linear or quadratic of the size of an FSM under consideration. The concept of FSM is rather flexible. Since in many applications time aspects such as durations, delays, timeouts are very important, FSMs can be augmented with some additional features to describe the dependence of the behavior of a system on events occurring in real time. One of the most advanced timed extension of FSMs is the concept of Timed Automata which was developed and studied in [1]. Timed Automata are supplied with clocks (timers) for indicating real time moments, measuring durations of events, providing timeout effects. Transitions in such automata depends not only on the incoming of the outside messages and signals but also on the values of clocks. Further research showed that this model of computation is very expressive and captures many important features of real-time systems behavior. On the other side, Timed Automata in the full scope of their computing power are very hard for analysis and transformations. The reachability problem for Timed Automata is decidable [2], and, therefore, this model of computation is suitable for formal verification of real-time computer systems. But many other problems such as universality, inclusion, determinability, etc. are undecidable (see [2], [8]), and this hampers considerably formal analysis of Timed Automata. Finite State Machines (FSMs) are widely used as formal models for analysis and synthesis of information processing systems in software engineering, VLSI design, telecommunication, etc. The most attractive feature of this model of computation is its simplicity — many important synthesis and analysis problems (equivalence checking, minimization, test derivation, etc.) for classical FSMs can be solved in time which is almost linear or quadratic of the size of an FSM under consideration. When a Timed Automaton is capable to selectively reset timers, it can display rather sophisticated behavior which is very difficult for understanding and analysis. In some cases, such ability is very important; see, e.g. [9]. On the verification of strictly deterministic behavior of Timed Finite State Machines Keywords: Timed Finite State Machines; strictly deterministic behavior DOI: 10.15514/ISPRAS-2018-30(3)-22 325 Vinarskii E.M., Zakharov V.A. On the verification of strictly deterministic behaviour of Timed Finite State Machines. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 325-340 Vinarskii E.M., Zakharov V.A. On the verification of strictly deterministic behaviour of Timed Finite State Machines. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 325-340 1. Introduction Therefore, it can happen that a user will detect first a notification «File Management System is closed» and then, some time later, he/she will be surprised to find an announcement «File is saved». Of course, the user may regard such behavior of the system enigmatic. But much worse if the order in which these notifications appear may vary in different sessions of the system. If a File Management System interacts with other service programs such an interaction will almost certainly lead to errors. However, if a behavior of TFSMs is defined as in the papers referred above then such a model can not adequately capture behavioral defects of real-time systems, similar to the one that was considered in the example. To avoid this shortcoming of conventional TFSMs and to make their behavior more “realistic” from the point of view of an outside observer we offer some technical change to this model. We will assume that an output timed word consists of timestamped letters, and these letters always follow in ascending order of their timestamps regardless of an order in which the corresponding input letters entered a TFSM. In this model it may happen so that an input 𝒃 follows an input 𝒂 but a response to 𝒃 appears before a response to 𝒂 is computed. Clearly, the defect with File Management System discussed above becomes visible to an outside observer “through” the model of TFSMs thus modified. At first sight, it may seem that this change only slightly complicates the analysis of the behavior of such models. But this is a false impression. In the initial model of TFSM the formation of an output timed word is carried out by local means for each state of the system. In our model this is a global task since to find the proper position of a timestamped output letter one should consider the run of TFSM as a whole. Therefore, even the problem of checking whether a behavior of an improved model of TFSM is deterministic can not be solved as easy and straightforwardly as in the case of the initial model of TFSM. It should be noticed that the property of deterministic behavior is very important in theory real-time machines. 1. Introduction Therefore, it can happen that a user will detect first a notification «File Management System is closed» and then, some time later, he/she will be surprised to find an announcement «File is saved». Of course, the user may regard such behavior of the system enigmatic. But much worse if the order in which these notifications appear may vary in different sessions of the system. If a File Management System interacts with other service programs such an interaction will almost certainly lead to errors. However, if a behavior of TFSMs is defined as in the papers referred above then such a model can not adequately capture behavioral defects of real-time systems, similar to the one that was considered in the example. To avoid this shortcoming of conventional TFSMs and to make their behavior more “realistic” from the point of view of an outside observer we offer some technical change to this model. We will assume that an output timed word consists of timestamped letters, and these letters always follow in ascending order of their timestamps regardless of an order in which the corresponding input letters entered a TFSM. In this model it may happen so that an input 𝒃 follows an input 𝒂 but a response to 𝒃 appears before a response to 𝒂 is computed. Clearly, the defect with File Management System discussed above becomes visible to an outside observer “through” the model of TFSMs thus modified. As it was noticed above a behavior of a TFSM is characterized by a pair sequences: an input timed word and a corresponding output timed word. A distinguishing feature of TFSMs studied in [5], [10], [13], [14], [15] is that an output timed word is formed of timestamped output letters that follows in the same order as the corresponding input letters regardless of their timestamps. Meanwhile, suppose that a user of some file management system gives a command «Save» and immediately after that a command «Exit». Then if a file to be saved is small then the user will observe first a response «File is saved» and then a notification «File Management System is closed». But if a file has a considerable size then it takes a lot of time to close it. 1. Introduction But a great deal of real-time programs and devices operate with timers much more simply: as soon as such a device switches to a new mode of operation (new state), it resets all timers. Timed Finite State Machines (TFSM) of this kind were studied in [5], [10], [13], [14]. TFSM has the only timer which it resets "automatically” as soon as it moves from one state to another. On the other hand, TFSMs, in contrast to Timed Automata introduced in [1], operate like transducers: they receive a sequence of input signals augmented with their timestamps (input timed word) and output a sequence of responses also labeled by timestamps (output timed word). The timestamps are real numbers which indicate the time when an input signal becomes available to a TFSM or an output response is generated. Transitions of a TFSM are equipped with time guards to indicate time intervals when transitions are active. Therefore, a reaction of a TFSM to an input signal depends not only on the signal but also on its timestamp. Some algorithms for equivalence checking, minimization and test generation were developed for TFSMs in [6], [5], [13], [14], [15]. It can be recognized that this model of TFSM combines a sufficient expressive power for modeling a wide class of real-time information processing systems and a developed algorithmic support. 326 Винарский Е.М., Захаров В.А. К проверке строго детерминированного поведения временных конечных автоматов. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 325-340 автоматов. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 325 340 As it was noticed above a behavior of a TFSM is characterized by a pair sequences: an input timed word and a corresponding output timed word. A distinguishing feature of TFSMs studied in [5], [10], [13], [14], [15] is that an output timed word is formed of timestamped output letters that follows in the same order as the corresponding input letters regardless of their timestamps. Meanwhile, suppose that a user of some file management system gives a command «Save» and immediately after that a command «Exit». Then if a file to be saved is small then the user will observe first a response «File is saved» and then a notification «File Management System is closed». But if a file has a considerable size then it takes a lot of time to close it. 1. Introduction As it was said above, universality, inclusion and equivalence checking problems are undecidable for Timed Automata in general case [2] but all these problems have been shown to be decidable for deterministic Timed Automata [3], [11]. However, testing whether a Timed Automaton is determinable has been proved undecidable [8]. Understanding and coping with these weaknesses have attracted lots of research, and classes of timed automata have been exhibited, that can be effectively determinized [3], [12]. A generic construction that is 327 Vinarskii E.M., Zakharov V.A. On the verification of strictly deterministic behaviour of Timed Finite State Machines. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 325-340 applicable to every Timed Automaton, and which, under certain conditions, yields a deterministic Timed Automaton, which is language-equivalent to the original timed automaton, has been developed in [4]. We studied the determinacy checking problem for improved TFSMs and present the results of our research in this paper. First, we offer a criterion to determine whether a given sequence of transition (trace) in a TFSM is steady, i.e. for any input timed word which activates this trace the letters of output words always follow in the same order (but, maybe, with different timestamps). Then, using this criterion we developed an exhaustive algorithm for checking the property of strict determinacy of TFSMs. This property means that every input timed word activates no more than one trace and all traces in a TFSM are steady. Exhaustive search, although been time consuming, can hardly be avoided in this case: we proved that determinacy checking problem for improved version of TFSMs is co-NP-hard by polynomially reducing to its complement the subset-sum problem [7] which is known to be NP- complete. The structure of the paper is as follows. In Section II we define the basic notions and introduce an improved concept of TFSM (or, it would be better said, a concept of TFSM with an improved behavior). In Section III we present necessary and sufficient conditions for steadiness of traces in a TFSM and show how to use this criterion to check whether a given TFSM is strictly deterministic. Section IV contains the results on the complexity of checking the properties of strictly deterministic behavior of TFSM. In the Conclusion we briefly outline the consequences of our results and topics for further research. 2. Formatting overview Consider two non-empty finite alphabets 𝐼 and 𝑂; the alphabet 𝐼 is an input alphabet and the alphabet 𝑂 is an output alphabet. The letters from 𝐼 can be regarded as control signals received by some real-time computing system, whereas the letters from 𝑂 may be viewed as responses (actions) generated by the system. A finite sequence 𝑤= 𝑖1, 𝑖2, … , 𝑖𝑛 of input letters is called an input word, whereas a sequence 𝑧= 𝑜1, 𝑜2, … , 𝑜𝑛 of output letters is called an output word. As usual, the time domain is represented by the set of non-negative reals ℝ0 +. The set of all positive real numbers will be denoted by ℝ+. When such a system receives a control signal (a letter 𝑖) its output depends not only on the input signal 𝑖 but also on  a current internal state of the system,  a current internal state of the system,  a time instance when 𝒊 becomes available to a system, and  time required to process the input (output delay). These aspects of real-time behavior can be formalized with the help of timestamps, time guards and delays. A timestamp as well as a delay is a real number from ℝ+. A timestamp indicates a time instance when the system receives an input signal or generates a response to it. A delay is time the system needs to generate an output response after receiving an input signal. A time guard is an interval 𝑔= ⟨𝑢, 𝑣⟩, 328 Винарский Е.М., Захаров В.А. К проверке строго детерминированного поведения временных конечных автоматов. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 325-340 where ⟨∈{(, [}, ⟩∈{), ]}, and 𝑢, 𝑣 are timestamps such that 0 < 𝑢< 𝑣. Time intervals indicate the periods of time when transitions of a system are active for processing input signals. As usual, the term time sequences is reserved for an increasing sequence of timestamps. For the sake of simplicity we will deal only with time guards of the form (𝑢, 𝑣]: all the results obtained in this paper can be adapted with minor changes to arbitrary time guards. Let 𝒘= 𝒙𝟏, 𝒙𝟐, … 𝒙𝒏 and 𝝉= 𝒕𝟏, 𝒕𝟐, … , 𝒕𝒏 be an input (output) word and a time sequence, respectively, of the same length. Then a pair (𝒘, 𝝉) is called a timed word. 2. Formatting overview Every pair of corresponding elements 𝒙𝒋 and 𝒕𝒋, 𝟏≤𝒋≤𝒏, indicates that an input signal (or an output response) 𝒙𝒋 appears at time instance 𝒕𝒋. In order to make this correspondence clearer we will often write timed words as sequences of pairs (𝒘, 𝝉) = (𝒊𝟏, 𝒕𝟏), (𝒊𝟐, 𝒕𝟐), … , (𝒊𝒏, 𝒕𝒏) whose components are input signals (or output responses) and their timestamps. A Finite State Machine (FSM) over the alphabets 𝑰 and 𝑶 is a triple 𝑴= ⟨𝑺, 𝒔𝒊𝒏, 𝝆⟩ where 𝑺 is a finite non-empty set of states, 𝒔𝒊𝒏 is an initial state, 𝝆⊆(𝑺× 𝑰× 𝑶× 𝑺) is a transition relation. A transition (𝒔, 𝒊, 𝒐, 𝒔′) means that FSM 𝑴 when being at the state 𝒔 and receiving an input signal 𝒊 moves to the state 𝒔′ and generates the output response 𝒐. A Finite State Machine (FSM) over the alphabets 𝑰 and 𝑶 is a triple 𝑴= ⟨𝑺, 𝒔𝒊𝒏, 𝝆⟩ where 𝑺 is a finite non-empty set of states, 𝒔𝒊𝒏 is an initial state, 𝝆⊆(𝑺× 𝑰× 𝑶× 𝑺) is a transition relation. A transition (𝒔, 𝒊, 𝒐, 𝒔′) means that FSM 𝑴 when being at the state 𝒔 and receiving an input signal 𝒊 moves to the state 𝒔′ and generates the output response 𝒐. FSMs can not measure time and, therefore, they are unsuitable for modeling the behavior of real-time systems. The authors of [1] proposed to equip FSMs with clocks — variables which take non-negative real values. To manipulate with clocks machines use reset instructions, timed guards and output delays. Time guards indicate time intervals when transitions are active for processing input signals. An output delay indicates how much time does it take to process an input. Thus, every transition in such a machine is a quadruple ⟨𝒊𝒏𝒑𝒖𝒕, 𝒕𝒊𝒎𝒆𝒅 𝒈𝒖𝒂𝒓𝒅, 𝒐𝒖𝒕𝒑𝒖𝒕, 𝒅𝒆𝒍𝒂𝒚⟩. Input signals and output responses are accompanied by timestamps. If an input is marked by a timestamp which satisfies the time guard then the transition fires, the machine moves to the next state and generates the output. This output is marked by a timestamp which is equal to the timestamp of the input plus the delay. For real-time machines of this kind usual problems from automata theory (equivalence and containment checking, minimization, etc.) may be set up and solved.  𝝆⊆(𝑺× 𝑰× 𝑶× 𝑺× 𝑮× ℝ+) is a transition relation.  𝝆⊆(𝑺× 𝑰× 𝑶× 𝑺× 𝑮× ℝ+) is a transition relation. 𝝆⊆(𝑺× 𝑰× 𝑶× 𝑺× 𝑮× ℝ+) is a transition relation. A transition (𝑠, 𝑖, 𝑜, 𝑠′, 𝑔, 𝑑) should be understood as follows. Suppose that TFSM receives the input letter 𝑖 marked by a timestamp 𝑡 when being at the state 𝑠. If the previous letter has been delivered to the TFSM at time 𝑡^ such that 𝛥𝑡= 𝑡−𝑡^ ∈𝑔 then the TFSM moves to the state 𝑠′ and outputs the letter 𝑜 marked with the timestamp 𝜏= 𝑡+ 𝑑. When algorithmic and complexity issues of TFSM’s analysis and synthesis are concerned then we assume that time guards and delays are rational numbers, and the size of a TFSM is the length of a binary string which encodes all transitions in the TFSM. A transition (𝑠, 𝑖, 𝑜, 𝑠′, 𝑔, 𝑑) should be understood as follows. Suppose that TFSM receives the input letter 𝑖 marked by a timestamp 𝑡 when being at the state 𝑠. If the previous letter has been delivered to the TFSM at time 𝑡^ such that 𝛥𝑡= 𝑡−𝑡^ ∈𝑔 then the TFSM moves to the state 𝑠′ and outputs the letter 𝑜 marked with the A trace 𝒕𝒓 in TFSM 𝑴 is a sequence of transitions (𝒔𝟎, 𝒂𝟏, 𝒃𝟏, 𝒔𝟏, (𝒖𝟏, 𝒗𝟏], 𝒅𝟏), … , (𝒔𝒏−𝟏, 𝒂𝒏, 𝒃𝒏, 𝒔𝒏, (𝒖𝒏, 𝒗𝒏], 𝒅𝒏), where every state 𝒔𝒋, 𝟎< 𝒋< 𝒏, is an arrival state of one transition and a departure state of the next transition. We say that the trace 𝒕𝒓 converts an input timed word 𝜶= (𝒂𝟏, 𝒕𝟏), (𝒂𝟐, 𝒕𝟐), … , (𝒂𝒏, 𝒕𝒏) to the timed output word 𝜷= (𝒃𝒋𝟏, 𝝉𝟏), (𝒃𝒋𝟐, 𝝉𝟐), … , (𝒃𝒋𝒏, 𝝉𝒏), iff • 𝑡𝑗−𝑡𝑗−1 ∈(𝑢𝑗, 𝑣𝑗] holds for all 𝑗, 1 ≤𝑖≤𝑛 (it is assumed that 𝑡0 = 0); • 𝑡𝑗−𝑡𝑗−1 ∈(𝑢𝑗, 𝑣𝑗] holds for all 𝑗, 1 ≤𝑖≤𝑛 (it is assumed that 𝑡0 = 0); • 𝑡𝑗−𝑡𝑗−1 ∈(𝑢𝑗, 𝑣𝑗] holds for all 𝑗, 1 ≤𝑖≤𝑛 (it is assumed that 𝑡0 = 0); • 𝛽 is such a permutation of the sequence 𝛾= (𝑏1, 𝑡1 + 𝑑1), (𝑏2, 𝑡2 + 𝑑2), … , (𝑏𝑛, 𝑡𝑛+ 𝑑𝑛) that the second components of the pairs 𝜏1, 𝜏2, … , 𝜏𝑛 constitute a time sequence. 2. Formatting overview The minimization problem for real- time machines is very important, since the complexity of many analysis and synthesis algorithms depend on the size of machines. In [14] this problem was studied under the so called "slow environment assumption”: next input becomes available only after an output response to the previous one is generated. In this paper, we consider a more advanced real-time machine; in this model the order in which outputs become visible to an outside observer is determined not only by the order in which inputs follow, but also by the delay required for their processing. When the same sequence of transitions is performed by such a machine the same outputs may follow in different order depending on the arriving time of the corresponding inputs. Our main goal is to develop equivalence checking and 329 Vinarskii E.M., Zakharov V.A. On the verification of strictly deterministic behaviour of Timed Finite State Machines. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 325-340 Vinarskii E.M., Zakharov V.A. On the verification of strictly deterministic behaviour of Timed Finite State Machines. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 325-340 minimization algorithms for real-time machines of this kind. But, as the results of Automata Theory show, these problems may have efficient solution only for deterministic machines. Thus, our first step toward the solution of these problems is to find a way to check if the behavior of a machine is deterministic. But there is also another reason to study the problem of checking the determinism of the behavior of real-time machines. Unlike traditional discrete models of computation, the behavior of real-time machines depends not only on the control signals as such, but also on the time of their arrival. However, the latter factor has a greater degree of uncertainty. In most cases, in practice, it is desirable to reduce the effect of this uncertainty to a minimum. Therefore, the determinacy checking problem for real-time machines can be considered as a special version of the verification problem — checking that the time factor does not have an unforeseen influence on the behavior of the system. Formally, by Timed FSM (TFSM) over the alphabets 𝑰 and 𝑶 we mean a quadruple 𝑴= (𝑺, 𝒔𝒊𝒏, 𝑮, 𝝆) where:  𝑺 is a finite non-empty set of states,  𝒔𝒊𝒏 is an initial state.  𝒔𝒊𝒏 is an initial state. 2. Formatting overview  𝑮 is a set of timed guards,  𝑮 is a set of timed guards, in this TFSM. Then this trace 1. accepts an input timed word 𝛼1 = (𝑖, 1), (𝑖, 2.7), (𝑖, 4.1) and converts it to the output timed word 𝛽1 = (𝑜1, 5), (𝑜3, 5.1), (𝑜2, 5.7); thus, the plain response of 𝑀 to 𝛼1 is 𝑤1 = 𝑜1, 𝑜3, 𝑜2; 1. accepts an input timed word 𝛼1 = (𝑖, 1), (𝑖, 2.7), (𝑖, 4.1) and converts it to the output timed word 𝛽1 = (𝑜1, 5), (𝑜3, 5.1), (𝑜2, 5.7); thus, the plain response of 𝑀 to 𝛼1 is 𝑤1 = 𝑜1, 𝑜3, 𝑜2; 2. accepts an input timed word 𝛼2 = (𝑖, 1.5), (𝑖, 3.2), (𝑖, 4.3) and converts it to the output timed word 𝛽2 = (𝑜3, 5.3), (𝑜1, 5.5), (𝑜2, 6.2), and the plain response of 𝑀 to 𝛼2 is 𝑤2 = 𝑜3, 𝑜1, 𝑜2 which is different from 𝑤1; 2. accepts an input timed word 𝛼2 = (𝑖, 1.5), (𝑖, 3.2), (𝑖, 4.3) and converts it to the output timed word 𝛽2 = (𝑜3, 5.3), (𝑜1, 5.5), (𝑜2, 6.2), and the plain response of 𝑀 to 𝛼2 is 𝑤2 = 𝑜3, 𝑜1, 𝑜2 which is different from 𝑤1; does not accept an input timed word 𝛼3 = (𝑖, 2.3), (𝑖, 4), (𝑖, 6). does not accept an input timed word 𝛼3 = (𝑖, 2.3), (𝑖, 4), (𝑖, 6).  𝝆⊆(𝑺× 𝑰× 𝑶× 𝑺× 𝑮× ℝ+) is a transition relation. • 𝛽 is such a permutation of the sequence 𝛾= (𝑏1, 𝑡1 + 𝑑1), (𝑏2, 𝑡2 + 𝑑2), … , (𝑏𝑛, 𝑡𝑛+ 𝑑𝑛) that the second components of the pairs 𝜏1, 𝜏2, … , 𝜏𝑛 constitute a time sequence. 𝛽 p q 𝛾 ( 1, 1 1), ( 2, 2 𝑑2), … , (𝑏𝑛, 𝑡𝑛+ 𝑑𝑛) that the second components of the pairs 𝜏1, 𝜏2, … , 𝜏𝑛 constitute a time sequence. Clearly, for every trace 𝑡𝑟 and an input timed word 𝛼 its conversion 𝛽 (if any) is determined uniquely; such a conversion will be denoted as 𝑐𝑜𝑛𝑣(𝑡𝑟, 𝛼). If Clearly, for every trace 𝑡𝑟 and an input timed word 𝛼 its conversion 𝛽 (if any) is determined uniquely; such a conversion will be denoted as 𝑐𝑜𝑛𝑣(𝑡𝑟, 𝛼). If Clearly, for every trace 𝑡𝑟 and an input timed word 𝛼 its conversion 𝛽 (if any) is determined uniquely; such a conversion will be denoted as 𝑐𝑜𝑛𝑣(𝑡𝑟, 𝛼). If 330 Винарский Е.М., Захаров В.А. К проверке строго детерминированного поведения временных конечных автоматов. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 325-340 𝑐𝑜𝑛𝑣(𝑡𝑟, 𝛼) is defined then we say that the input timed word 𝛼 activates the trace 𝑡𝑟. We will say that the output word 𝑏𝑗1, 𝑏𝑗2, … , 𝑏𝑗𝑛 is a plain response to the input timed word 𝛼 on the trace 𝑡𝑟; it will be denoted as 𝑟𝑒𝑠𝑝(𝑡𝑟, 𝛼). 3. Steady traces and strictly deterministic TFSMs As can be seen from the above example, a pair of input timed words that differ only in timestamps of input signals may activate the same trace in a TFSM, although plain responses of TFSM to these words are different. Generally speaking, there is nothing unusual in this: in real-time models not only the input signals, but also the values of timers influence a run of a model. Nevertheless, in many applications it is critically important to be sure that the behavior of a real-time system is predictable: once a system choose a mode of computation (i.e. a trace in TFSM) it will behave in a similar way (i.e. give the same plain response) in all computations of this mode. Traditionally, computer systems in which for any input data the processing mode is uniquely determined by the system are called deterministic. But for our model of real-time systems this requirement should be clarified and strengthened. For this purpose, we introduce the notion of steady traces and the property of strict determinacy of a real-time system. A trace 𝒕𝒓 in TFSM 𝑴 is called steady if 𝒓𝒆𝒔𝒑(𝒕𝒓, 𝜶𝟏) = 𝒓𝒆𝒔𝒑(𝒕𝒓, 𝜶𝟐) holds for every pair of input timed words 𝜶𝟏 and 𝜶𝟐 that activate 𝒕𝒓. Thus, the order of the output letters generated by a steady trace does not depend on the small deviations of the timestamps of the input signals. A TFSM 𝑴= (𝑺, 𝒔𝒊𝒏, 𝑮, 𝝆) is called deterministic iff for every pair of transitions (𝒔, 𝒊𝟏, 𝒐𝟏, 𝒔′, (𝒖𝟏, 𝒗𝟏], 𝒅𝟏) and (𝒔, 𝒊𝟐, 𝒐𝟐, 𝒔″, (𝒖𝟐, 𝒗𝟐], 𝒅𝟐) in 𝝆 either 𝒊𝟏≠𝒊𝟐, or (𝒖𝟏, 𝒗𝟏] ∩(𝒖𝟐, 𝒗𝟐] = ∅. This requirement means that every timestamped input letter can activate no more than 331 Vinarskii E.M., Zakharov V.A. On the verification of strictly deterministic behaviour of Timed Finite State Machines. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 325-340 one transition from an arbitrary given state 𝒔. It also implies that every input timed word can activate no more than one trace in 𝑴. A deterministic TFSM is called strictly deterministic iff every initial trace in 𝑴 which starts from the initial state 𝒔𝒊𝒏 is steady. It is easy to see that TFSM, depicted in Fig. 1, is not strictly deterministic. The Strict Determinacy Checking Problem (in what follows, SDCP) is that of checking, given a TFSM, if it is strictly deterministic. 3. Steady traces and strictly deterministic TFSMs 3, 2018 г., стр. 325-340 Винарский Е.М., Захаров В.А. К проверке строго детерминированного поведения временных конечных автоматов. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 325-340 Hence, 𝒓𝒆𝒔𝒑(𝒕𝒓, 𝜶′) = ⋯, 𝒐𝒎, … , 𝒐𝒌, …. Hence, 𝒓𝒆𝒔𝒑(𝒕𝒓, 𝜶′) = ⋯, 𝒐𝒎, … , 𝒐𝒌, …. On the other hand, the trace 𝒕𝒓 converts the timed input word 𝜶″ to the timed output word 𝒄𝒐𝒏𝒗(𝒕𝒓, 𝜶″) = ⋯, (𝒐𝒌, 𝑻″𝒌), … , (𝒐𝒎, 𝑻″𝒎), … such that 𝑻″𝒌= ∑ 𝒗𝒋 𝒌 𝒋=𝟏 + 𝒅𝒌 and 𝑻″𝒎= ∑ 𝒗𝒋 𝒎 𝒋=𝟏 + 𝒅𝒎. In this timed output word the output letter 𝒐𝒎 follows the output letter 𝒐𝒌 since 𝑻″𝒎−𝑻″𝒌= 𝒅𝒎−𝒅𝒌= ∑𝒗𝒋 𝒎 𝒋=𝒌+𝟏 ≥𝟎 Therefore, 𝒓𝒆𝒔𝒑(𝒕𝒓, 𝜶″) = ⋯, 𝒐𝒌, … , 𝒐𝒎, …. 3. Steady traces and strictly deterministic TFSMs It is easy to check whether a TFSM is deterministic by considering one by one all pairs of transitions that emerge from the same state. But local means alone are not enough to check whether a given trace in a TFSM is steady. A simple criterion for steadiness of traces is presented as a Theorem below. Let a sequence of transitions (𝒔𝟎, 𝒊𝟏, 𝒔𝟏, 𝒐𝟏, ⟨𝒖𝟏, 𝒗𝟏⟩, 𝒅𝟏), … , (𝒔𝒏−𝟏, 𝒊𝒏, 𝒔𝒏, 𝒐𝒏, ⟨𝒖𝒏, 𝒗𝒏⟩, 𝒅𝒏) (𝒔𝟎, 𝒊𝟏, 𝒔𝟏, 𝒐𝟏, ⟨𝒖𝟏, 𝒗𝟏⟩, 𝒅𝟏), … , (𝒔𝒏−𝟏, 𝒊𝒏, 𝒔𝒏, 𝒐𝒏, ⟨𝒖𝒏, 𝒗𝒏⟩, 𝒅𝒏) be a trace 𝒕𝒓 in a TFSM 𝑴. Then the following theorem holds. Theorem 1. A trace 𝒕𝒓 is steady iff for all pairs of integers 𝒌, 𝒎 such that 𝟏≤𝒌< 𝒎≤𝒏 at least one of the two inequalities 𝒅𝒌−𝒅𝒎≤∑ 𝒖𝒋 𝒎 𝒋=𝒌+𝟏 or 𝒅𝒌−𝒅𝒎> ∑ 𝒗𝒋 𝒎 𝒋=𝒌+𝟏 holds. Proof. (⇒) Suppose that there exists a pair 𝒌, 𝒎 such that 𝟏≤𝒌< 𝒎≤𝒏, and a double inequality holds: ∑𝒖𝒋 𝒎 𝒋=𝒌+𝟏 < 𝒅𝒌−𝒅𝒎≤∑𝒗𝒋 𝒎 𝒋=𝒌+𝟏 . Then we use two positive numbers 𝒓= 𝒅𝒌−𝒅𝒎−∑ 𝒖𝒋 𝒎 𝒋=𝒌+𝟏 and 𝜺= 𝒓 𝒏 and consider a behaviour of a TFSM 𝑴 in the input timed words 𝜶′ = (𝒊𝟏, 𝒗𝟏), … , (𝒊𝒌, ∑𝒗𝒋 𝒌 𝒋=𝟏 ), (𝒊𝒌+𝟏, ∑𝒗𝒋 𝒌 𝒋=𝟏 + 𝒖𝒌+𝟏+ 𝜺), … , (𝒊𝒎, ∑𝒗𝒋 𝒌 𝒋=𝟏 + ∑𝒖𝒋 𝒎 𝒋=𝒌+𝟏 + 𝜺), 𝜶″ = (𝒊𝟏, 𝒗𝟏), … , (𝒊𝒌, ∑𝒗𝒋 𝒌 𝒋=𝟏 ), (𝒊𝒌+𝟏, ∑𝒗𝒋 𝒌+𝟏 𝒋=𝟏 ), … , (𝒊𝒎, ∑𝒗𝒋 𝒎 𝒋=𝟏 ). It is easy to see that both words activate 𝒕𝒓. It is easy to see that both words activate 𝒕𝒓. It is easy to see that both words activate 𝒕𝒓. The trace 𝒕𝒓 converts the timed input word 𝜶𝟏 to the timed output word trace 𝒕𝒓 converts the timed input word 𝜶𝟏 to the timed output word 𝒄𝒐𝒏𝒗(𝒕𝒓, 𝜶′) = ⋯, (𝒐𝒎, 𝑻′𝒎), … , (𝒐𝒌, 𝑻′𝒌), … 𝒄𝒐𝒏𝒗(𝒕𝒓, 𝜶′) = ⋯, (𝒐𝒎, 𝑻′𝒎), … , (𝒐𝒌, 𝑻′𝒌), … such that 𝑻′𝒎= ∑ 𝒗𝒋 𝒌 𝒋=𝟏 + ∑ ( 𝒎 𝒋=𝒌+𝟏𝒖𝒋+ 𝜺) + 𝒅𝒎, and 𝑻′𝒌= ∑ 𝒗𝒋 𝒌 𝒋=𝟏 + 𝒅𝒌. In this timed output word, the output letter 𝒐𝒌 follows the output letter 𝒐𝒎 since 𝑻′𝒌−𝑻′𝒎= 𝒅𝒌−𝒅𝒎−∑𝒖𝒋 𝒎 𝒋=𝒌+𝟏 + (𝒎−𝒌)𝜺= 𝒓−𝒓(𝒎−𝒌) 𝒏 > 𝟎. 332 Винарский Е.М., Захаров В.А. К проверке строго детерминированного поведения временных конечных автоматов. Труды ИСП РАН, том 30, вып. Therefore, 𝒓𝒆𝒔𝒑(𝒕𝒓, 𝜶″) = ⋯, 𝒐𝒌, … , 𝒐𝒎, …. Thus, we got evidence that the trace 𝒕𝒓 is not steady. (⇐) Suppose that the trace 𝒕𝒓 is not steady. Then there exists a pair of timed input words 𝜶′ = (𝒊𝟏, 𝒕′𝟏), … , (𝒊𝒏, 𝒕′𝒏) and 𝜶″ = (𝒊𝟏, 𝒕″𝟏), … , (𝒊𝒏, 𝒕″𝒏) such that both words activate the trace 𝒕𝒓 and 𝒓𝒆𝒔𝒑(𝒕𝒓, 𝜶′) ≠𝒓𝒆𝒔𝒑(𝒕𝒓, 𝜶″). Consequently, there exists a pair of output letters 𝒐𝒎 and 𝒐𝒌 such that 𝒄𝒐𝒏𝒗(𝒕𝒓, 𝜶′) = ⋯, (𝒐𝒌, 𝑻′𝒌), … , (𝒐𝒎, 𝑻′𝒎), … 𝒄𝒐𝒏𝒗(𝒕𝒓, 𝜶″) = ⋯, (𝒐𝒎, 𝑻″𝒎), … , (𝒐𝒌, 𝑻″𝒌), … . Such permutation of output letters is possible iff the following inequalities hold 𝒕′𝒌+ 𝒅𝒌= 𝑻′𝒌< 𝑻′𝒎= 𝒕′𝒎+ 𝒅𝒎, 𝒕″𝒌+ 𝒅𝒌= 𝑻″𝒌> 𝑻″𝒎= 𝒕″𝒎+ 𝒅𝒎 . But since both input timed words 𝜶′ and 𝜶″ activate 𝒕𝒓, we have the following chain of inequalities: ∑𝒖𝒋 𝒎 𝒋=𝒌+𝟏 < 𝑻″𝒎−𝑻″𝒌< 𝒅𝒌−𝒅𝒎< 𝑻′𝒎−𝑻′𝒌≤∑𝒗𝒋 𝒎 𝒋=𝒌+𝟏 . Thus, if 𝒕𝒓 is not steady then there exists a pair of integers such that 𝟏≤𝒌< 𝒎≤ 𝒏 and Thus, if 𝒕𝒓 is not steady then there exists a pair of integers such that 𝟏≤𝒌< 𝒎≤ 𝒏 and ∑𝒖𝒋 𝒎 𝒋=𝒌+𝟏 < 𝒅𝒌−𝒅𝒎≤∑𝒗𝒋 𝒎 𝒋=𝒌+𝟏 holds. holds. End proof. Now, having the criterion for steadiness of traces, we can give a solution to SDCP for TFSMs. Let TFSM 𝑴= (𝑺, 𝒔𝒊𝒏, 𝑮, 𝝆) be a deterministic TFSM. Denote by 𝒖𝒎𝒊𝒏 the greatest lower bound of all left boundaries used in the time guards of 𝑴. In our model of TFSM 𝒖𝒎𝒊𝒏> 𝟎. Let 𝒅𝒎𝒊𝒏 and 𝒅𝒎𝒂𝒙 be the minimum and the maximum 333 Vinarskii E.M., Zakharov V.A. On the verification of strictly deterministic behaviour of Timed Finite State Machines. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 325-340 Vinarskii E.M., Zakharov V.A. On the verification of strictly deterministic behaviour of Timed Finite State Machines. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 325-340 output delays occurred in the transitions of 𝑴. A theorem below gives necessary and sufficient conditions for the behaviour of 𝑴 to be strictly deterministic. Theorem 2. A deterministic TFSM 𝑴 is strictly deterministic iff all its traces of length 𝒑, where 𝒑= ⌈ 𝒅𝒎𝒂𝒙−𝒅𝒎𝒊𝒏 𝒖𝒎𝒊𝒏 ⌉, are steady. Proof. The necessity of conditions is obvious. We prove the sufficiency of conditions by contradiction. Suppose that all traces of length less or equal 𝒑 are steady but TFSM 𝑴 is not. Then there exists such a trace 𝒕𝒓 in 𝑴 which is not steady. Then, by Theorem 1, this trace is a sequence of transitions (𝒔𝒋−𝟏, 𝒊𝒋, 𝒔𝒋, 𝒃𝒋, (𝒖𝒋, 𝒗𝒋], 𝒅𝒋), 𝟏≤𝒋≤𝒏, such that for some pair of integers 𝒎 and 𝒌, where 𝟏≤𝒌< 𝒎≤𝒏, two inequalities ∑𝒖𝒋 𝒎 𝒋=𝒌+𝟏 ≤𝒅𝒌−𝒅𝒎≤∑𝒗𝒋 𝒎 𝒋=𝒌+𝟏 hold. It should be noticed, that, by the same Theorem 1, the trace 𝒕𝒓′ which includes only the transitions (𝒔𝒋−𝟏, 𝒊𝒋, 𝒔𝒋, 𝒃𝒋, (𝒖𝒋, 𝒗𝒋], 𝒅𝒋), 𝒎≤𝒋≤𝒌, is not steady as well. Hence, 𝒎−𝒌> 𝒑, and we have the following sequence of inequalities hold. It should be noticed, that, by the same Theorem 1, the trace 𝒕𝒓′ which includes only the transitions (𝒔𝒋−𝟏, 𝒊𝒋, 𝒔𝒋, 𝒃𝒋, (𝒖𝒋, 𝒗𝒋], 𝒅𝒋), 𝒎≤𝒋≤𝒌, is not steady as well. Hence, 𝒎−𝒌> 𝒑, and we have the following sequence of inequalities 𝒅𝒎𝒂𝒙−𝒅𝒎𝒊𝒏≥𝒅𝒎−𝒅𝒌≥∑𝒖𝒋 𝒎 𝒋=𝒌+𝟏 > 𝒑∗𝒖𝒎𝒊𝒏 ts our choice of 𝒑= ⌈ 𝒅𝒎𝒂𝒙−𝒅𝒎𝒊𝒏 𝒖𝒎𝒊𝒏 ⌉. 𝒖𝒎𝒊𝒏 End of proof. As it follows from Theorems 1 and 2, to guarantee that a given TFSM 𝑴= (𝑺, 𝒔𝒊𝒏, 𝑮, 𝝆) is strictly deterministic it is sufficient to consider all traces (𝒔𝟎, 𝒂𝟏, 𝒃𝟏, 𝒔𝟏, (𝒖𝟏, 𝒗𝟏], 𝒅𝟏), … , (𝒔𝒏−𝟏, 𝒂𝒏, 𝒃𝒏, 𝒔𝒏, (𝒖𝒏, 𝒗𝒏], 𝒅𝒏) in 𝑴, whose length 𝒏 does not exceed the value 𝒑= ⌈ 𝒅𝒎𝒂𝒙−𝒅𝒎𝒊𝒏 𝒖𝒎𝒊𝒏 ⌉ defined in Theorem 2, and for every such trace check that one of the inequalities 𝒅𝟏−𝒅𝒏< ∑ 𝒖𝒋 𝒏 𝒋=𝟐 or 𝒅𝟏−𝒅𝒏> ∑ 𝒗𝒋 𝒏 𝒋=𝟐 holds. Thus, we arrive at orollary 1. Strict Determinacy Checking Problem for TFSMs is decidable. 4. Strict Determinacy Checking Problem for TFSMs is co-NP- hard Clearly, the decision procedure, based on Theorem 2, is time consuming since 𝑝 may be exponential of the size of 𝑀 and the number of traces of length 𝑝 in TFSM 𝑀 is exponential of 𝑝. In this section we show that such an exhaustive search can hardly be avoided because SDCP for improved version of TFSMs is co-NP-hard. We are aimed to show that the complement of SDCP is NP-hard. To this end we consider the Subset-Sum Problem (see [7]) which is known to be NP-complete and Clearly, the decision procedure, based on Theorem 2, is time consuming since 𝑝 may be exponential of the size of 𝑀 and the number of traces of length 𝑝 in TFSM 𝑀 is exponential of 𝑝. In this section we show that such an exhaustive search can hardly be avoided because SDCP for improved version of TFSMs is co-NP-hard. We are aimed to show that the complement of SDCP is NP-hard. To this end we consider the Subset-Sum Problem (see [7]) which is known to be NP-complete and 334 арский Е.М., Захаров В.А. К проверке строго детерминированного поведения временных конечных матов. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 325-340 demonstrate that this problem can be reduced in polynomial time to the complement of SDCP for TFSMs. The Subset-Sum Problem (SSP) is that of checking, given a set of integers 𝑸 and an integer 𝑳, whether there is any subset 𝑸′, 𝑸′ ⊆𝑸, such that the sum of all its elements is equal to 𝑳. More formally, the variant of the SSP we are interested in is defined as follows. Let 𝑸= 𝒎𝟏, 𝒎𝟐, … , 𝒎𝑵 be a sequence of positive integers, and 𝑳 be also a positive integer. A solution to (𝑸, 𝑳)-instance of SSP is a binary tuple 𝒛= ⟨𝝈𝟏, 𝝈𝟐, … , 𝝈𝑵⟩ such that ∑ 𝝈𝒋 𝑵 𝒋=𝟏 𝒎𝒋= 𝑳. In [7] it was proved that the problem of checking the existence of a solution to a given (𝑸, 𝑳)-instance of SSP is NP- complete. p Fig.2 TFSM 𝑀 Fig.2 TFSM 𝑀 Fig.2 TFSM 𝑀 Now, given a (𝑸, 𝑳)-instance of SSP, we show how to build a deterministic TFSM 𝑴𝑸,𝑳 such that it has an initial trace which is not strictly determined iff this instance of SSP has a solution. 4. Strict Determinacy Checking Problem for TFSMs is co-NP- hard On the verification of strictly deterministic behaviour of Timed Finite State Machines. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 325-340 2) Since 𝜹= 𝒐(𝟏/𝑵𝟒) and 𝟎< 𝜺= 𝒐(𝟏/𝑵𝟐), for every 𝒌, 𝟏< 𝒌≤𝑵, and a binary tuple 𝒛= ⟨𝝈𝒌, 𝝈𝒌+𝟏, … , 𝝈𝑵⟩ the inequalities 2) Since 𝜹= 𝒐(𝟏/𝑵𝟒) and 𝟎< 𝜺= 𝒐(𝟏/𝑵𝟐), for every 𝒌, 𝟏< 𝒌≤𝑵, and a binary tuple 𝒛= ⟨𝝈𝒌, 𝝈𝒌+𝟏, … , 𝝈𝑵⟩ the inequalities 𝜹−𝑫< 𝟎< 𝑵𝜹≤∑( 𝑵 𝒋=𝒌+𝟏 𝝈𝒋(𝒎𝒋−𝜺) + (𝟏−𝝈𝒋)𝜹) hold. By Theorem 1, this implies that every trace from a state 𝒔𝒌, 𝟏≤𝒌≤𝑵, to the state 𝒔𝑵+𝟏 is strictly deterministic. hold. By Theorem 1, this implies that every trace from a state 𝒔𝒌, 𝟏≤𝒌≤𝑵, to the state 𝒔𝑵+𝟏 is strictly deterministic. 3) For the same reason the inequalities 3) For the same reason the inequalities 𝑫+ 𝑳−𝜹> ∑𝒎𝒋 𝒌 𝒋=𝟏 + 𝒌𝜺= ∑( 𝒌 𝒋=𝟏 𝝈𝒋(𝒎𝒋+ 𝜺) + (𝟏−𝝈𝒋)𝜺) hold for every 𝒌, 𝟏≤𝒌< 𝑵, and a binary tuple 𝒛= ⟨𝝈𝟏, 𝝈𝟐, … , 𝝈𝒌⟩. By Theorem 1, this guarantees that every initial trace leading to a state 𝒔𝒌, 𝟏≤𝒌≤𝑵 is strictly deterministic. hold for every 𝒌, 𝟏≤𝒌< 𝑵, and a binary tuple 𝒛= ⟨𝝈𝟏, 𝝈𝟐, … , 𝝈𝒌⟩. By Theorem 1, this guarantees that every initial trace leading to a state 𝒔𝒌, 𝟏≤𝒌≤𝑵 is strictly deterministic. As for the initial traces that lead to the state 𝒔𝑵+𝟏, due to our choice of 𝜺 and 𝜹, we can trust the following chain of reasoning. By definition, a (𝑸, 𝑳)-instance of SSP has a solution 𝒛= ⟨𝝈𝟏, 𝝈𝟐, … , 𝝈𝑵⟩ iff ∑ 𝝈𝒋 𝑵 𝒋=𝟏 𝒎𝒋= 𝑳. 4. Strict Determinacy Checking Problem for TFSMs is co-NP- hard Let 𝑫= ∑ 𝒎𝒋 𝑵 𝒋=𝟏 , and 𝜺 and 𝜹 be positive rational numbers such that 𝜺= 𝒐(𝟏/𝑵𝟐) and 𝜹= 𝒐(𝜺/𝑵𝟐). Consider a TFSM depicted in Fig. 2. This machine operates over alphabets 𝑰= 𝑶= {𝟎, 𝟏}. It has 𝑵+ 𝟐 states 𝒔𝟎, 𝒔𝟏, … , 𝒔𝒏, 𝒔𝑵+𝟏. The only transition (𝒔𝟎, 𝟎, 𝟎, 𝒔𝟏, (𝟏, 𝟐], 𝑳+ 𝑫) leads from the initial state 𝒔𝟎 to 𝒔𝟏. From each state 𝒔𝒋, 𝟏≤𝒋< 𝑵, two transitions (𝒔𝒋, 𝟏, 𝟏, 𝒔𝒋+𝟏, (𝒎𝒋−𝜺, 𝒎𝒋+ 𝜺], 𝜹) and (𝒔𝒋, 𝟎, 𝟎, 𝒔𝒋+𝟏, (𝜹, 𝜺], 𝜹) lead to the state 𝒔𝒋+𝟏. The state 𝒔𝑵 is different: two transitions (𝒔𝑵, 𝟏, 𝟏, 𝒔𝑵+𝟏, (𝒎𝑵−𝜺, 𝒎𝑵+ 𝜺], 𝑫) and (𝒔𝑵, 𝟎, 𝟎, 𝒔𝑵+𝟏, (𝜹, 𝜺], 𝑫) lead this state to 𝒔𝑵+𝟏. First, we make some observations. Now, given a (𝑸, 𝑳)-instance of SSP, we show how to build a deterministic TFSM 𝑴𝑸,𝑳 such that it has an initial trace which is not strictly determined iff this instance of SSP has a solution. Let 𝑫= ∑ 𝒎𝒋 𝑵 𝒋=𝟏 , and 𝜺 and 𝜹 be positive rational numbers such that 𝜺= 𝒐(𝟏/𝑵𝟐) and 𝜹= 𝒐(𝜺/𝑵𝟐). Consider a TFSM depicted in Fig. 2. This machine operates over alphabets 𝑰= 𝑶= {𝟎, 𝟏}. It has 𝑵+ 𝟐 states 𝒔𝟎, 𝒔𝟏, … , 𝒔𝒏, 𝒔𝑵+𝟏. The only transition (𝒔𝟎, 𝟎, 𝟎, 𝒔𝟏, (𝟏, 𝟐], 𝑳+ 𝑫) leads from the initial state 𝒔𝟎 to 𝒔𝟏. From each state 𝒔𝒋, 𝟏≤𝒋< 𝑵, two transitions (𝒔𝒋, 𝟏, 𝟏, 𝒔𝒋+𝟏, (𝒎𝒋−𝜺, 𝒎𝒋+ 𝜺], 𝜹) and (𝒔𝒋, 𝟎, 𝟎, 𝒔𝒋+𝟏, (𝜹, 𝜺], 𝜹) lead to the state 𝒔𝒋+𝟏. The state 𝒔𝑵 is different: two transitions (𝒔𝑵, 𝟏, 𝟏, 𝒔𝑵+𝟏, (𝒎𝑵−𝜺, 𝒎𝑵+ 𝜺], 𝑫) and (𝒔𝑵, 𝟎, 𝟎, 𝒔𝑵+𝟏, (𝜹, 𝜺], 𝑫) lead this state to 𝒔𝑵+𝟏. First we make some observations 1) Since all transitions outgoing from the states 𝒔𝒋, 𝟏≤𝒋< 𝑵, have the same delay 𝜹, every trace from a state 𝒔𝒌 to a state 𝒔𝓵, where 𝟎< 𝒌< 𝓵≤𝑵, is strictly deterministic. 1) Since all transitions outgoing from the states 𝒔𝒋, 𝟏≤𝒋< 𝑵, have the same delay 𝜹, every trace from a state 𝒔𝒌 to a state 𝒔𝓵, where 𝟎< 𝒌< 𝓵≤𝑵, is strictly deterministic. 335 Vinarskii E.M., Zakharov V.A. On the verification of strictly deterministic behaviour of Timed Finite State Machines. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 325-340 Vinarskii E.M., Zakharov V.A. 5. Conclusion Therefore, it is reasonable to confine ourselves to considering only such TFSMs, in which the time guards and the delays are chosen from some fixed finite set. As it follows from Theorem 2, for this class of TFSMs SDCP is decidable in polynomial time. One may wonder what is the degree of such a polynomial, or, in other words, how efficiently the strict determinacy property can be checked for TFSMs corresponded to real systems. 3. 3. In the model of TFSM besides the usual transitions there are also possible timeout transitions. A timeout transition fires when a timestamped input letter (𝒊, 𝒕) can not activate any usual transition from a current state. In it was shown that in some cases such timeout transitions can not be replaced by any combination of ordinary transitions. In the future we are going to study how SDCP can be solved for TFSMs with timeouts. 4. Strict Determinacy Checking Problem for TFSMs is co-NP- hard The latter is possible iff two following inequalities hold: q ∑𝝈𝒋 𝑵 𝒋=𝟏 𝒎𝒋 − 𝜺+ 𝑵𝜹< 𝑳< ∑𝝈𝒋 𝑵 𝒋=𝟏 (𝒎𝒋) + 𝑵𝜺 ∑𝝈𝒋 𝑵 𝒋=𝟏 𝒎𝒋 − 𝜺+ 𝑵𝜹< 𝑳< ∑𝝈𝒋 𝑵 𝒋=𝟏 (𝒎𝒋) + 𝑵𝜺 (1) (1) By taking into account the relationships below ∑( 𝑵 𝒋=𝟏 𝝈𝒋(𝒎𝒋−𝜺) + (𝟏−𝝈𝒋)𝜹) < ∑𝝈𝒋 𝑵 𝒋=𝟏 𝒎𝒋 − 𝜺+ 𝑵𝜹 ∑𝝈𝒋 𝑵 𝒋=𝟏 (𝒎𝒋) + 𝑵𝜺= ∑( 𝑵 𝒋=𝟏 𝝈𝒋(𝒎𝒋+ 𝜺) + (𝟏−𝝈𝒋)𝜺), we can conclude that (1) holds iff another pair of inequalities hold: we can conclude that (1) holds iff another pair of inequalities hold: ∑( 𝑵 𝒋=𝟏 𝝈𝒋(𝒎𝒋−𝜺) + (𝟏−𝝈𝒋)𝜹) < 𝑳< ∑( 𝑵 𝒋=𝟏 𝝈𝒋(𝒎𝒋+ 𝜺) + (𝟏−𝝈𝒋)𝜺) But in the context of observations 1) – 3) above, the latter inequalities, as it follows from Theorem 1, provide the necessary and sufficient conditions that the initial trace in TFSM 𝑴𝑸,𝑳 activated by the input word 𝒛= ⟨𝝈𝟏, 𝝈𝟐, … , 𝝈𝑵⟩ is not strictly deterministic. Thus, a (𝑸, 𝑳)-instance of SSP has a solution iff TFSM 𝑴𝑸,𝑳 is not strictly deterministic. The considerations above bring us to Theorem 3. SDCP for TFSMs is co-NP-hard. Theorem 3. SDCP for TFSMs is co-NP-hard. 336 Винарский Е.М., Захаров В.А. К проверке строго детерминированного поведения временных конечных автоматов. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 325-340 Винарский Е.М., Захаров В.А. К проверке строго детерминированного поведения временных конечных автоматов. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 325-340 Винарский Е.М., Захаров В.А. К проверке строго детерминированного поведения временных конечных автоматов. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 325-340 5. Conclusion The main contributions of this paper are 1. the development of a modified version of TFSM which, in our opinion, provides a more adequate model of real-time computing systems; 2. the introduction of the notion of strict deterministic behaviour of TFSM and setting up the Strict Determinacy Checking Problem (SDCP) for a modified version of TFSMs; 3. the establishing of an effectively verifiable criterion for the strict determinacy property of TFSMs; 4. the proving that SDCP for TFSMs is co-NP-hard. However, some problems concerning strict deterministic behavior of TFSMs still remain open. They will be topics for our further research. 1. In Sections [Sect3] and [Sect4] it was shown that SDCP for TFSMs is co-NP- hard and in the worst case it can be solved in double exponential time by means of a naive exhaustive searching algorithm based on Theorems 1 and 2. We think that this complexity upper bound estimate is too much high. The question arises, for what complexity class C SDCP for TFSMs is a C- complete problem. By some indications we assume that SDCP for TFSMs is PSPACE-complete problem. 2. 2. As it can be seen from the proof of Theorem 3, SDCP for TFSMs is intractable only if timed parameters of transitions (time guards and delays) depend on the number of states in TFSM. But this is not a typical phenomenon in real-time systems since in practice the performance of individual components of a system does not depend on the size of the system. Therefore, it is reasonable to confine ourselves to considering only such TFSMs, in which the time guards and the delays are chosen from some fixed finite set. As it follows from Theorem 2, for this class of TFSMs SDCP is decidable in polynomial time. One may wonder what is the degree of such a polynomial, or, in other words, how efficiently the strict determinacy property can be checked for TFSMs corresponded to real systems. 2. As it can be seen from the proof of Theorem 3, SDCP for TFSMs is intractable only if timed parameters of transitions (time guards and delays) depend on the number of states in TFSM. But this is not a typical phenomenon in real-time systems since in practice the performance of individual components of a system does not depend on the size of the system. References [1]. Alur R., Dill D. A Theory of Timed Automata. Theoretical Computer Science, vol. 126, 1994, pp. 183-235. [2]. Alur R., Madhusudan P. Decision Problems for Timed Automata: A Survey. In Proceedings of the 4-th International School on Formal Methods for the Design of Computer, Communication, and Software Systems (SFM’04), 2004, pp. 1-24. [3]. Alur R., Fix L., Henzinger Th. A. A Determinizable Class of Timed Automata. In Proceedings of the 6-th International Conference on Computer Aided Veriﬁcation (CAV’94), 1994, p 1-13. [4]. Baier C., Bertrand N., Bouyer P., Brihaye T. When are Timed Automata Determinizable? In Proceedings of the 36-th International Colloquium on Automata, Languages, and Programming (ICALP 2009), 2009, p. 43-54. [5]. Bresolin D., El-Fakih K., Villa T., Yevtushenko N. Deterministic Timed Finite State Machines: Equivalence Checking and Expressive Power. In Proceedings of the International Conference GANDALF, 2014, p. 203-216. [6]. Cardell-Oliver R. Conformance Tests for Real-Time Systems with Timed Automata Specifications. Formal Aspects of Computing, vol. 12, no. 5, 2000, p. 350–371. [7]. Cormen T. H., Leiserson C. E., Rivest R. L., Stein C. 35.5: The subset-sum problem. Introduction to Algorithms (2-nd ed.), 2001. [8]. Finkel O. Undecidable Problems about Timed Automata. In Proceedings of 4th International Conference on Formal Modeling and Analysis of Timed Systems (FORMATS’06), 2006, p. 187-199. ( ) p [9]. Fletcher J. G., Watson R. W. Mechanism for Reliable Timer-Based Protocol. Computer Networks, vol. 2, 1978, pp. 271-290. [10]. Merayo M.G., Nuunez M., Rodriguez I. Formal Testing from Timed Finite State Machines. Computer Networks, vol. 52, no 2, 2008, pp. 432-460. [11]. Ouaknine J., Worrell J. On the Language Inclusion Problem for Timed Automata: Closing a Decidability Gap. In Proceedings of the 19-th Annual Symposium on Logic in Computer Science (LICS’04), 2004, pp. 54-63. [12]. Suman P.V., Pandya P.K., Krishna S.N., Manasa L. Timed Automata with Integer Resets: Language Inclusion and Expressiveness. In Proceedings of the 6-th International Conference on Formal Modeling and Analysis of Timed Systems (FORMATS’08), 2008, pp. 78–92. [13]. Tvardovskii A., Yevtushenko N. Minimizing Timed Finite State Machines. Tomsk State University Journal of Control and Computer Science, No 4 (29), 2014, pp. 77-83 (in Russian). [14]. Tvardovskii A., Yevtushenko N., M. Gromov. Minimizing Finite State Machines with Time Guards and Timeouts. Trudy ISP RAN/Proc. ISP RAS, vol. 29, issue 4, 2017, pp. 139-154 (in Russian). [15]. Zhigulin M., Yevtushenko N., Maag S., Cavalli A. Acknowledgments The authors of the article express their deep gratitude to V.V. Podymov and the anonymous reviewers for their valuable comments and advice on improving the article. This work was supported by the Russian Foundation for Basic Research, Grant N 18-01-00854. 337 Vinarskii E.M., Zakharov V.A. On the verification of strictly deterministic behaviour of Timed Finite State Machines. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 325-340 Винарский Е.М., Захаров В.А. К проверке строго детерминированного поведения временных конечных автоматов. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 325-340 арский Е.М., Захаров В.А. К проверке строго детерминированного поведения временных конечных матов. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 325-340 К проверке строго детерминированного поведения временных конечных автоматов Е.М.Винарский <vinevg2015@gmail.com> В.А. Захаров < zakh@cs.msu.su >. Московский государственный университет имени М.В. Ломоносова, 119991, Россия, Москва, Ленинские горы, д. 1 Аннотация. Конечные автоматы широко применяются в качестве математических моделей при решении многочисленных задач в области программирования, проектирования микроэлектронных схем и телекоммуникационных систем. Для описания поведения систем реального времени модель конечного автомата может быть расширена добавлением в неё часов - параметра непрерывного времени, моделируемого вещественной переменной. В автоматах реального времени для входных и выходных сигналов указывается время их поступления и выдачи, а переходы автомата снабжены описанием задержек, связанных с ожиданием входных сигналов и формированием выходных сигналов. Так же, как и для классических автоматов дискретного времени, задача минимизации конечных автоматов реального времени возникает во многих приложениях этой модели вычислений. Для классической модели автоматов реального времени эта задача уже подробно рассмотрена. В нашей работе мы предлагаем более сложную модель: в ней порядок следования выходных сигналов определяется не только порядком поступления входных сигналов, но также и задержкой, связанной с их обработкой. В этой модели при выполнении одной и той же последовательности переходов выходные сигналы могут выдаваться в разном порядке в зависимости от времени поступления входных сигналов. В новой модели автоматов реального времени решению задачи минимизации должно предшествовать изучение вопроса строгой детерминированности - однозначности поведения автомата на одних и тех же последовательностях переходов. В представленной статье приведены и обоснованы необходимые и достаточные условия строгой детерминированности автоматов реального времени, а также исследованы вопросы, связанные с решением задачи минимизации этой разновидности автоматов. Ключевые слова: конечные временные автоматы; строго детерминированное поведение DOI: 10.15514/ISPRAS-2018-30(3)-22 Ключевые слова: конечные временные автоматы; строго детерминированное поведение Для цитирования: Винарский Е.М., Захаров В.А. К проверке строго детерминированного поведения временных конечных автоматов. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 325-340 (на английском языке). DOI: 10.15514/ISPRAS-2018- 30(3)-22 References FSM-Based Test Derivation Strategies for Systems with Timeouts. In Proceedings of the 11-th International Conference on Quality Software, 2011, p. 141-149. 338 DOI: 10.15514/ISPRAS-2018-30(3)-22 DOI: 10.15514/ISPRAS-2018-30(3)-22 Для цитирования: Винарский Е.М., Захаров В.А. К проверке строго детерминированного поведения временных конечных автоматов. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 325-340 (на английском языке). DOI: 10.15514/ISPRAS-2018- 30(3)-22 339 Vinarskii E.M., Zakharov V.A. On the verification of strictly deterministic behaviour of Timed Finite State Machines. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 325-340 Vinarskii E.M., Zakharov V.A. On the verification of strictly deterministic behaviour of Timed Finite State Machines. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 325-340 Список литературы [1]. Alur R., Dill D. A Theory of Timed Automata. Theoretical Computer Science, vol. 126, 1994, pp. 183-235. [2]. Alur R., Madhusudan P. Decision Problems for Timed Automata: A Survey. In Proceedings of the 4-th International School on Formal Methods for the Design of Computer, Communication, and Software Systems (SFM’04), 2004, pp. 1-24. [3]. Alur R., Fix L., Henzinger Th. A. A Determinizable Class of Timed Automata. In Proceedings of the 6-th International Conference on Computer Aided Veriﬁcation (CAV’94), 1994, p 1-13. ( ) p [4]. Baier C., Bertrand N., Bouyer P., Brihaye T. When are Timed Automata Determinizable? In Proceedings of the 36-th International Colloquium on Automata, Languages, and Programming (ICALP 2009), 2009, p. 43-54. [5]. Bresolin D., El-Fakih K., Villa T., Yevtushenko N. Deterministic Timed Finite State Machines: Equivalence Checking and Expressive Power. In Proceedings of the International Conference GANDALF, 2014, p. 203-216. [6]. Cardell-Oliver R. Conformance Tests for Real-Time Systems with Timed Automata Specifications. Formal Aspects of Computing, vol. 12, no. 5, 2000, p. 350–371. [7]. Cormen T. H., Leiserson C. E., Rivest R. L., Stein C. 35.5: The subset-sum problem. Introduction to Algorithms (2-nd ed.), 2001. [8]. Finkel O. Undecidable Problems about Timed Automata. In Proceedings of 4th International Conference on Formal Modeling and Analysis of Timed Systems (FORMATS’06), 2006, p. 187-199. [9]. Fletcher J. G., Watson R. W. Mechanism for Reliable Timer-Based Protocol. Computer Networks, vol. 2, 1978, pp. 271-290. [10]. Merayo M.G., Nuunez M., Rodriguez I. Formal Testing from Timed Finite State Machines. Computer Networks, vol. 52, no 2, 2008, pp. 432-460. [11]. Ouaknine J., Worrell J. On the Language Inclusion Problem for Timed Automata: Closing a Decidability Gap. In Proceedings of the 19-th Annual Symposium on Logic in Computer Science (LICS’04), 2004, pp. 54-63. [12]. Suman P.V., Pandya P.K., Krishna S.N., Manasa L. Timed Automata with Integer Resets: Language Inclusion and Expressiveness. In Proceedings of the 6-th International Conference on Formal Modeling and Analysis of Timed Systems (FORMATS’08), 2008, pp. 78–92. [13]. А.С. Твардовский, Н.В. Евтушенко. К минимизации автоматов с временными ограничениями. Вестник Томского государственного университета. Управление, вычислительная техника и информатика, vol. 29, no 4, 2014, pp. 77-83. [14]. Твардовский А.С., Евтушенко Н.В., Громов М.Л. Минимизация автоматов с таймаутами и временными ограничениями. Труды ИСП РАН, том 29, вып. 4, 2017 г., стр. 139-154. DOI: 10.15514/ISPRAS-2017-29(4)-8.. [15]. Zhigulin M., Yevtushenko N., Maag S., Cavalli A. FSM-Based Test Derivation Strategies for Systems with Timeouts. Список литературы In Proceedings of the 11-th International Conference on Quality Software, 2011, p. 141-149. 340 DOI: 10.15514/ISPRAS-2018-30(3)-23 DOI: 10.15514/ISPRAS-2018-30(3)-23 Для цитирования: Словохотов Ю.Л., Неретин И.С. К построению модульной модели распределенного интеллекта. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 341-362. DOI: 10.15514/ISPRAS-2018-30(3)-23 К построению модульной модели распределенного интеллекта Аннотация. Описание и моделирование динамики мультиагентных социальных систем методами, заимствованными из статистической физики «неживых» многочастичных систем, не отражает принципиальную особенность совокупности взаимодействующих автономных агентов: способность воспринимать, обрабатывать и использовать внешнюю информацию. Распределенный интеллект социальных систем следует непосредственно учитывать в их экспериментальных и теоретических исследованиях. В работе предложена «модульная» модель интеллектуальной деятельности, включающая производство новой информации и пригодная для описания как индивидуального, так и распределенного интеллекта, перечислены возможные области ее использования. «Количественную оценку» эффективности распределенного интеллекта иллюстрирует компьютерная модель искусственной социальной системы; обсуждаются полученные результаты. Ключевые слова: мультиагентные социальные системы; распределенный интеллект; моделирование интеллектуальной деятельности 1 В рамках этой работы понятия «индивид», «индивидуум» и «агент» применительно к единичному «актору» социальной системы используются как синонимы. 1. Введение Изучению и прогнозированию динамики реально существующих мультиагентных социальных систем – таких как пешеходные потоки [1], объекты управления [2], участники политического противоборства [3], избирательные кампании [4] и многих других – посвящена обширная литература. Основным формальным инструментом анализа служит теория 341 Slovokhotov Yu.L., Neretin I.S. Toward construction of a modular model of distributed intelligence. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 341-362 Slovokhotov Yu.L., Neretin I.S. Toward construction of a modular model of distributed intelligence. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 341-362 игр, призванная типологизировать поведение агентов1 и установить возможные равновесия в системе. Структуру совокупности взаимосвязанных агентов исследуют методами теории графов, представляя достаточно большие системы в виде сложных сетей [5]. Это позволяет выделить характерные структурные элементы (концентраторы, клики, сообщества, деревья), определить количественные характеристики сети (диаметр, распределение порядков вершин, степень кластеризации и др.), а также моделировать динамику различных процессов на сетевых субстратах [6–8]. При большом количестве положительных результатов и новых эмпирических наблюдений, в теоретико-игровых и теоретико-графовых исследованиях мультиагентных систем до сих пор не выработано их общепринятой модели. Моделирование мультиагентных процессов методами теории игр сталкивается с необходимостью учитывать как многоуровневую рефлексию агентов, обладающих интеллектом [9], так и «ограниченно рациональное» поведение людей в реальных условиях [10]. Это усложняет модели, делает их стохастическими и зависящими от эмпирических параметров, снижая предсказательную способность. Даже для систем с простыми стратегиями агентов, которые удается выразить квази-динамическими уравнениями, расчетный краткосрочный прогноз требует больших вычислительных мощностей (транспортный поток на шоссе [11]) либо оказывается невозможным за пределами нескольких идеализированных режимов (биржа [12]). С другой стороны, результаты моделирования социальных процессов на сетях во многих случаях показывают отсутствие сильного влияния структуры субстрата (которое неявно постулируется в большинстве «сетевых» работ). Так, основные качественные результаты математической социологии, полученные во 2-й половине ХХ века на квадратных решетках, изображавших систему социальных связей [13, 14], в целом воспроизводятся и на сложные сетях; различия в основном состоят в замедлении процесса и неоднородных «промежуточных» распределениях состояний агентов [14, 15]. С формальной точки зрения, большинство современных мультиагентных моделей построены по аналогии с моделями многочастичных физических систем [14]. Подобно взаимодействиям в ансамбле «неживых» частиц, взаимодействия агентов основаны на балансе количественно или качественно оцениваемых факторов (выигрышей, намерений, интересов и т.д.), а динамика каждого агента направляется стремлением к максимуму энергоподобной функции полезности и «размывается» стохастическим шумом, нередко называемым температурой. 1. Введение Усложнения агентных моделей, по сравнению с их физическими прототипами, сводятся к использованию нечетко определенных характеристик (начиная с «полезности»), неоднозначных зависимостей 1 В рамках этой работы понятия «индивид», «индивидуум» и «агент» применительно к единичному «актору» социальной системы используются как синонимы. 342 Словохотов Ю.Л., Неретин И.С. К построению модульной модели распределенного интеллекта. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 341-362 Словохотов Ю.Л., Неретин И.С. К построению модульной модели распределенного интеллекта. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 341-362 Словохотов Ю.Л., Неретин И.С. К построению модульной модели распределенного интеллекта. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 341-362 (рефлексии) и сложной дискретной конфигурации взаимодействий (сети). Предсказуемый результат состоит в сильном увеличении объема расчетов и неопределенности прогноза в таких системах, где динамика процесса не задается самой сетевой структурой (как в эпидемиологии, энергетических и транспортных сетях, некоторых других областях). Имеющиеся результаты мультиагентного моделирования позволяют предположить, что многочастичные модели, заимствованные из статистической термодинамики и физики стохастических процессов, не вполне адекватны фундаментальным особенностям социальных систем и процессов в таких системах. (Утверждение о несводимости социума к физике является общим местом в работах гуманитарного направления). Интригующим дополнительным обстоятельством служат эмпирические аналогии коллективного поведения людей и динамики, наблюдаемой для систем индивидов со значительно меньшими интеллектуальными возможностями (муравейник, термитник [16]), а также для формаций программируемых автоматов [17]. Таким образом, наличие индивидуального сознания у людей не является принципиальным отличием социальной мультиагентной системы от физической системы «неживых» частиц (молекул и атомов при сколь угодно детальном описании их состояний и взаимодействий): сходную динамику могут проявлять социальные системы, сильно различающиеся по уровню интеллекта у их агентов. Мы предполагаем, что главной особенностью, отличающей систему взаимодействующих агентов от ансамбля частиц физической среды, является распределенный интеллект: способность мультиагентной системы воспринимать внешнюю информацию, обрабатывать и использовать ее в своей коллективной динамике. Наличие распределенного интеллекта следует непосредственно учитывать как в эмпирическом описании социальных мультиагентных систем, так и в их моделировании. Далее будут рассмотрены некоторые возможные подходы к исследованиям в этом направлении. 2. Определения Мультиагентная социальная система (МСС) – это динамическая совокупность автономных агентов, которые воспринимают информацию и взаимодействуют с внешней средой и другими агентами в ходе собственной деятельности (динамики). Взаимодействующие биологические существа одного вида составляют социальную систему в узком смысле слова. «Неживые» программируемые агенты (роботы, беспилотные аппараты и т.д.) при наличии взаимодействия между ними образуют искусственную МСС. Индивидуальные агенты составляют МСС 1-го уровня (сообщество живых организмов, автомобили на шоссе и т.д.). Взаимодействующие мультиагентные системы (экономические субъекты, научные школы, политические «акторы») сами могут выступать в роли агентов в социальной системе более высокого уровня. 343 Slovokhotov Yu.L., Neretin I.S. Toward construction of a modular model of distributed intelligence. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 341-362 Каждая МСС (1) обладает структурой, (2) осуществляет некоторую суммарную («системную») деятельность, (3) в ее ходе воспринимает, обрабатывает и использует информацию. Структуру МСС определяют воздействия окружающей среды на агентов («внешнее поле») и взаимодействия между агентами (внутренняя структура, в общем случае – взвешенный ориентированный граф, зависящий от внешнего поля и от времени). Динамика МСС определяется изменениями внешней среды, собственной динамикой агентов и эволюцией внутренней структуры во времени. Системная динамика МСС складывается из индивидуальных динамик агентов, однако ее результаты не сводятся к сумме индивидуальных результатов агентов (перемещение стада животных из засушливой местности в благоприятную, биржевой крах, автоматизация производства и т.д.). Поскольку агенты-индивиды в МСС 1-го уровня воспринимают информацию и преследуют определенный набор целей, социальным системам как «акторам» можно приписать системное целеполагание. Объективной целью МСС на заданном интервале времени является оптимальный для системы результат ее динамики. Разнообразные МСС с вариативной динамикой агентов, не предполагающей сложной интеллектуальной деятельности, воспроизводимо и гибко преследуют системные цели (пчелиный рой, муравейник, «невидимая рука рынка»). Одной из объективных целей социальной системы в большинстве случаев является сохранение либо увеличение количества агентов и поддержка их функционирования: самосохранение системы. Восприятие, обработка и использование информации социальной системой являются результатом восприятия информации ее агентами, взаимодействий между ними и целеполагания агентов. (Так, формация автоматов, движущихся в заданном направлении по алгоритму «следовать за соседом» и «избегать столкновений», огибает препятствия). Таким образом, социальные системы любого уровня не только преследуют определенные цели, но также, в процессе их достижения, анализируют информацию – как внешнюю (сила стимула или угрозы извне), так и внутреннюю (состояния агентов и их соседей в зоне восприятия). Иными словами, динамика МСС существенно включает распределенный интеллект: восприятие, анализ и использование информации социальной системой как единым «актором» и, возможно, как агентом в МСС более высокого уровня. 3. Моделирование интеллекта: состояние проблемы Toward construction of a modular model of distributed intelligence. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 341-362 Slovokhotov Yu.L., Neretin I.S. Toward construction of a modular model of distributed intelligence. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 341-362 2. спонтанное возникновение правильного решения («озарение», или инсайт) у людей и животных, поставленных в «проблемную ситуацию»; 2. спонтанное возникновение правильного решения («озарение», или инсайт) у людей и животных, поставленных в «проблемную ситуацию»; 3. повышение вероятности инсайта при активизации случайного перебора вариантов ответа и ослаблении ограничений на варьирование компонентов «проблемной ситуации». 3. повышение вероятности инсайта при активизации случайного перебора вариантов ответа и ослаблении ограничений на варьирование компонентов «проблемной ситуации». Перечисленные положения использует, в частности, эвристическая схема стимулирования изобретательской деятельности (ТРИЗ). В ее рамках проблемная ситуация названа «изобретательской ситуацией», а изобретение (инсайт) состоит во «взгляде с новой стороны» на компоненты проблемной ситуации и в изменении логической связи между ними [23]. 3. Моделирование интеллекта: состояние проблемы Мы полагаем, что описание, типология и моделирование социальных систем не могут быть корректными без анализа распределенного интеллекта, которым обладают различные типы МСС. В этом случае особую важность приобретает формальный анализ индивидуального интеллекта – прежде всего 344 Словохотов Ю.Л., Неретин И.С. К построению модульной модели распределенного интеллекта. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 341-362 человеческого мышления – и степень переносимости его результатов на «коллективный разум». человеческого мышления – и степень переносимости его результатов на «коллективный разум». Анализу умственной деятельности людей и элементов мышления животных посвящена большая область когнитивных наук – «междисциплинарных исследований познания, понимаемого как совокупность процессов приобретения, хранения, преобразования и использования знаний живыми и искусственными системами» [18, 19]. В этой области получен огромный фактический материал по психологическим механизмам мышления, структуре сознания и нейрофизиологическим процессам («коррелятам»), сопровождающим работу мозга у человека и животных. К когнитивным наукам примыкает область конструирования и исследования искусственного интеллекта [17, 20]. Одним из наиболее известных достижением здесь являются искусственные нейронные сети (ИНС), успешно используемые в аппроксимации функций, распознавании образов, финансовом прогнозировании и поиске многопараметровых корреляций. Важное место в исследованиях ИНС, до настоящего времени остающихся перспективным, но не вполне объясненным техническим средством, занимают изучение механизма их функционирования и осуществляемой ими переработки информации [21]. Несмотря на многочисленные положительные результаты когнитивных исследований и компьютерного моделирования познавательных процессов, общепринятой формальной модели интеллектуальной деятельности до настоящего времени также не существует. В частности, детально разработанный алгоритм «обучения» нейронных сетей позволяет ИНС распознавать лишь такие виды внешнего воздействия, которые были заданы в ходе обучения. За пределами теории остается главное содержание мышления: производство новой, ранее не известной информации. В то же время в психологии мышления имеется общепринятая описательная схема умственной деятельности, приводящей к открытию новых знаний [22], и выработаны эмпирические рекомендации («эвристики»), повышающие вероятность ее успеха – такие, как «мозговой штурм» [23]. Рядом авторов – в частности, в работах Д.С.Чернавского [24] – был предложен «генетический» механизм отбора полезной информации, который закрепляет новое знание, генерируемое в случайном процессе. В исследованиях динамики головного мозга методами электроэнцефалографии (ЭЭГ) и функциональной магнитно- резонансной томографии (фМРТ) экспериментально установлены локализация информации в определенных участках коры [25], «хранение» человеческих знаний и эмоций в различных областях мозга [26]. Характерными особенностями творческой умственной деятельности, установленными в психологии мышления [22, 27], являются: Характерными особенностями творческой умственной деятельности, установленными в психологии мышления [22, 27], являются: 1. обучение как необходимое условие мышления; желательность возможно более широкой «базы знаний» для плодотворного мышления; 345 Slovokhotov Yu.L., Neretin I.S. 4. Модульная модель интеллекта Термин «модульность» (modularity) весьма распространен в описании структуры интеллекта и его функций (познания), однако в когнитивных науках он используется скорее на вербальном уровне [28]. Мы предлагаем «модульную» модель индивидуального интеллекта, которая воспроизводит выработку новой информации в процессе мышления и может быть распространена на описание распределенного интеллекта МСС. Наша модель не использует математических соотношений теории информации, ограничиваясь «наивным» представлением об информации как имеющемся (или возникающем) человеческом знании. Ее основой служат перечисленные выше характеристики интеллектуальной творческой деятельности (п.п. 1 – 3) и «угадывание» (а не строгий логический вывод) правильного решения в проблемной ситуации. Суть модели заключается в блочном (модульном) характере схематичного «опосредованного отражения», или внутренней репрезентации [29], вырабатываемой сознанием в ответ на внешнее воздействие, и в генетическом алгоритме подбора модулей, составляющих репрезентацию. Мы предполагаем, что внешнее воздействие на сознание вызывает в нем «отпечаток», интерпретируемый (т.е. воспринимаемый) путем сравнения с некоторой схемой, ранее выработанной для сходных «отпечатков» и вызываемой из памяти. Предполагается, что эта схема (далее – «образ-схема») состоит из небольшого числа знаковых модулей-«иероглифов», библиотека которых составляется в процессе обучения и хранится в памяти (рис. 1). Разные «отпечатки» интерпретируются разными комбинациями модулей. «Единицами хранения» в памяти являются не образы и не аппроксимирующие их схемы: в памяти хранится библиотека модулей и содержатся «ключи» (наборы идентификаторов модулей), позволяющие вызвать ранее построенные из них схемы (рис. 2). 346 Словохотов Ю.Л., Неретин И.С. К построению модульной модели распределенного интеллекта. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 341-362 ов Ю.Л., Неретин И.С. К построению модульной модели распределенного интеллекта. П РАН, том 30, вып. 3, 2018 г., стр. 341-362 Рис. 1. Компоненты восприятия сознанием внешних стимулов: (1) «отпечаток» внешнего воздействия, (2) его интерпретация («образ- схема»), (3) сопоставление «отпечатка» и «образа- схемы» Рис. 1. Компоненты восприятия сознанием внешних стимулов: (1) «отпечаток» внешнего воздействия, (2) его интерпретация («образ- схема»), (3) сопоставление «отпечатка» и «образа- схемы» р Fig. 1. Components of perception of external stimulus: (1) «imprint» of the external action, (2) its interpretation («image-scheme»), (3) comparison of the «image-scheme» with the «imprint» Fig. 1. Components of perception of external stimulus: (1) «imprint» of the external action, (2) its interpretation («image-scheme»), (3) comparison of the «image-scheme» with the «imprint» 347 g p f p p f ( ) p f , (2) its interpretation («image-scheme»), (3) comparison of the «image-scheme» with the «imprint» Рис. 2. Алгоритм восприятия: построение «образа-схемы» из библиотеки модулей Fig. 2. Building of «image-scheme» from the library of modules Рис. 2. 4. Модульная модель интеллекта Алгоритм восприятия: построение «образа-схемы» из библиотеки модулей Fig. 2. Building of «image-scheme» from the library of modules 347 Slovokhotov Yu.L., Neretin I.S. Toward construction of a modular model of distributed intelligence. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 341-362 Slovokhotov Yu.L., Neretin I.S. Toward construction of a modular model of distributed intelligence. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 341-362 По аналогии со знаками иероглифической письменности, модули, составляющие схему, в процессе их приобретения (обучения), вероятно, характеризуют определенные признаки внешних воздействий, но используются далее как универсальные компоненты «образов-схем»: комбинаций модулей, индивидуальных для каждого «отпечатка». Соответствие вызванной модульной схемы «отпечатку» внешнего воздействия проверяется в мозге по некоторому быстрому «автоматическому», не вполне установленному механизму [26]. По аналогии со знаками иероглифической письменности, модули, составляющие схему, в процессе их приобретения (обучения), вероятно, характеризуют определенные признаки внешних воздействий, но используются далее как универсальные компоненты «образов-схем»: комбинаций модулей, индивидуальных для каждого «отпечатка». Соответствие вызванной модульной схемы «отпечатку» внешнего воздействия проверяется в мозге по некоторому быстрому «автоматическому», не вполне установленному механизму [26]. При хорошем соответствии происходит узнавание. Плохое соответствие вызывает неудовлетворенность (в терминах ТРИЗ – «обостренное противоречие»). Стремление преодолеть неудовлетворенность стимулирует построение новой схемы, лучше соответствующей «отпечатку», из блоков, имеющихся в библиотеке. Создание новой модульной схемы и запоминание ее ключа соответствует производству новой информации в ходе умственной деятельности. Число модулей, составляющих «образ-схему» (m), можно оценить сверху числом параметров или факторов, которые человек в состоянии воспринимать параллельно («72»). Разумной оценкой общего количества N модулей в библиотеке (безусловно, определяемого продолжительностью и глубиной обучения) может служить количество иероглифов, которыми оперирует грамотный носитель японского или китайского языка (до нескольких тысяч). Оценки m~10 и N~1000 дают практически бесконечное число возможных сочетаний модулей в схеме (𝑚 𝑁)~1023. Вместе с тем идентификатор схемы, позволяющий вызвать ее модули из памяти в заданном порядке, может весьма экономно состоять из «номеров» или иных признаков модулей в библиотеке – для чего в предельном случае достаточно одного нейрона, локализованного в коре головного мозга, на один модуль. Предлагаемая модель на качественном уровне отражает главное содержание интеллектуальной деятельности: «придумывание нового» в форме подбора новой комбинации модулей, которая аппроксимирует воспринимаемый образ лучше всех комбинаций, хранящихся в памяти. 4. Модульная модель интеллекта Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 341-362 Slovokhotov Yu.L., Neretin I.S. Toward construction of a modular model of distributed intelligence. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 341-362 обсуждении сетевых структур мозга, физическим носителем модуля является ког: «распределенная группа нейронов, сцепленная единым когнитивным опытом», т.е. кодирующая определенный аспект восприятия внешних воздействий [25]. Следует заметить, что нейроны, входящие в состав «паттерна активации» [31], могут быть локализованы в коре головного мозга без определенного алгоритма, ситуативно – подобно записи информации на магнитный диск. В этом случае тесная связь мышления с конкретными сетевыми структурами мозга (коннектомом) и сознания (когнитомом), постулируемая в данной области [25], подобно предполагаемой связи сетевой структуры с динамикой социальных систем, может не соответствовать действительности (см. разд. 1). обсуждении сетевых структур мозга, физическим носителем модуля является ког: «распределенная группа нейронов, сцепленная единым когнитивным опытом», т.е. кодирующая определенный аспект восприятия внешних воздействий [25]. Следует заметить, что нейроны, входящие в состав «паттерна активации» [31], могут быть локализованы в коре головного мозга без определенного алгоритма, ситуативно – подобно записи информации на магнитный диск. В этом случае тесная связь мышления с конкретными сетевыми структурами мозга (коннектомом) и сознания (когнитомом), постулируемая в данной области [25], подобно предполагаемой связи сетевой структуры с динамикой социальных систем, может не соответствовать действительности (см. разд. 1). 4. Модульная модель интеллекта Из невозможности, при принятых оценках чисел m и N, построить образ-схему прямым перебором модулей следует достаточно правдоподобный алгоритм творческой интеллектуальной деятельности: 1) поиск наилучшего образа-схемы для «отпечатка» внешнего воздействия по имеющимся в памяти ключам; 2) выявление в найденной схеме модуля (или модулей) с наихудшим соответствием «отпечатку»; 3) замена «плохого» модуля на другие модули, хранящиеся в памяти, ограниченным случайным перебором; 4) нахождение модуля, улучшающего соответствие схемы «отпечатку»; 4) нахождение модуля, улучшающего соответствие схемы «отпечатку»; 5) запоминание новой комбинации модулей: введение в память ее ключа (рис. 3). 5) запоминание новой комбинации модулей: введение в память ее ключа (рис. 3). 348 348 Словохотов Ю.Л., Неретин И.С. К построению модульной модели распределенного интеллекта. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 341-362 Рис. 3. Замена модуля в образе-схеме, улучшающая соответствие «отпечатку» Fig. 3. Replacement of a module in the «image-scheme» improving its correspondence to “imprint” Рис. 3. Замена модуля в образе-схеме, улучшающая соответствие «отпечатку» Fig. 3. Replacement of a module in the «image-scheme» improving its correspondence to “imprint” Наша модель согласуется с такими известными условиями успешного решения творческих задач, как предварительное обучение, эрудиция (создание и расширение библиотеки), опыт распознавания проблемных ситуаций (алгоритм нахождения наилучшего образа в памяти), свободный ассоциативный поиск («ключи») и умственные способности, т.е. способность без ошибок обрабатывать большие объемы информации. В эмпирических описаниях творческой деятельности пункту (2) нашего алгоритма соответствует «проблемная», или «изобретательская» ситуация, п. (3) – состояние «нерешенной проблемы», которому сопутствуют процессы «расшатывания» образа (и снятие ограничений в поиске как одно из условий успеха), п. (4) – «озарение» (инсайт), а п. (5) – рождение новой информации. В бесструктурной схеме Д.С.Чернавского [24] аналогом перебора модулей является хаотический «перемешивающий слой», или странный аттрактор, в фазовом пространстве континуальной математической модели, а ценность новой информации, полученной случайным путем, определяется ее пригодностью для решения текущих задач. В нашей модели генетический подбор модулей контролируется сопоставлением получаемых схем с «отпечатком» воздействия; хорошее согласие предположительно вызывает биохимически контролируемое удовлетворение [26, 30]. Обсуждаемая модель согласуется с наличием локализованных нейронов, воспроизводимо активируемых в коре головного мозга при предъявлении ( й й й Обсуждаемая модель согласуется с наличием локализованных нейронов, воспроизводимо активируемых в коре головного мозга при предъявлении определенного стимула («когнитивной специализацией нейронов в мозге человека» [25]). Локализованные нейроны в этом случае являются частью физического «субстрата» модуля, или воспроизводимо активируемого ансамбля нейронов [31], в составе «образа-схемы», возникающей при распознавании внешнего воздействия. В терминах, используемых при 349 Slovokhotov Yu.L., Neretin I.S. Toward construction of a modular model of distributed intelligence. 5. К моделированию распределенного интеллекта A comparison of activity of individual human’s intellect with a distributed intelligence of an enterprise индивидуальный интеллект человека распределенный интеллект предприятия 1-й этап восприятия: «отпечаток» внешнего воздействия сумма системных и индивидуальных реакций работников на внешнее воздействие библиотека модулей и «хранилище ключей» рабочая информация, должностные инструкции 2-й этап восприятия: выбор модульной «схемы-образа» оценки параметров воздействия и его предварительная характеристика; действия по инструкциям поиск лучшей имеющейся приближенной схемы, оценка ее расхождений с образом оценка успешности стандартных действий («создание комиссии») случайная замена модулей корректировка действий; испытание новых предложений («работа комиссии») «генетический контроль» оценка успешности скорректированных действий («выводы комиссии») запоминание новой схемы дополнение и изменение инструкций Представленный модельный пример иллюстрирует как сложность и неоднородность структурированной социальной системы, так и наличие нескольких уровней ее распределенного интеллекта (так, при потере управления персонал предприятия будет действовать группами, преследующими собственные цели). Существенно, что рациональность системной динамики увеличивается при ограничении набора возможных действий агентов и при наличии «библиотеки» стандартных реакций руководства предприятия и персонала. Тесную связь «интеллектуальности» организационной системы с Табл. 1. Сравнение компонентов индивидуальной интеллектуальной деятельности и распределенного интеллекта организационной системы Table 1. A comparison of activity of individual human’s intellect with a distributed intelligence of an enterprise индивидуальный интеллект человека распределенный интеллект предприятия 1-й этап восприятия: «отпечаток» внешнего воздействия сумма системных и индивидуальных реакций работников на внешнее воздействие библиотека модулей и «хранилище ключей» рабочая информация, должностные инструкции 2-й этап восприятия: выбор модульной «схемы-образа» оценки параметров воздействия и его предварительная характеристика; действия по инструкциям поиск лучшей имеющейся приближенной схемы, оценка ее расхождений с образом оценка успешности стандартных действий («создание комиссии») случайная замена модулей корректировка действий; испытание новых предложений («работа комиссии») «генетический контроль» оценка успешности скорректированных действий («выводы комиссии») запоминание новой схемы дополнение и изменение инструкций Представленный модельный пример иллюстрирует как сложность и неоднородность структурированной социальной системы, так и наличие нескольких уровней ее распределенного интеллекта (так, при потере управления персонал предприятия будет действовать группами, преследующими собственные цели). Существенно, что рациональность системной динамики увеличивается при ограничении набора возможных действий агентов и при наличии «библиотеки» стандартных реакций руководства предприятия и персонала. Тесную связь «интеллектуальности» организационной системы с формализацией ее структуры и действий иллюстрирует табл. 2. Рациональность индивидуального поведения людей также возрастает с сокращением возможностей выбора (переход улицы по светофору, покупка или продажа акций на бирже по фиксированной цене) и становится «ограниченной» при выборе из многих возможностей (переход улицы без светофора, произвольная цена акций на бирже). Табл. 2. Формальные условия функционирования разных организационных систем Table 2. 5. К моделированию распределенного интеллекта Предлагаемая «модульная» модель индивидуального интеллекта естественно распространяется на описание распределенного интеллекта (РИ) мультиагентных социальных систем. Наличие интеллекта у МСС определяется способностью агентов воспринимать информацию, наличием целей у агентов и их взаимодействием (см. Введение). Уровень РИ соответственно зависит от когнитивных возможностей агентов, от преследуемых ими целей, от характера и структуры межагентных взаимодействий, а также от внешних условий. При распаде структуры МСС (отсутствие ресурсов, изобилие ресурсов либо «война всех против всех») коллективная обработка информации не реализуется. Распределенный интеллект системы в общем случае возрастает с увеличением числа агентов и с усложнением структуры их взаимодействий. Анализ и использование информации организационными системами достаточно высокого уровня, функционирующими в человеческом обществе, имеет очевидные аналогии с описанием индивидуального интеллекта по «модульной» схеме (табл. 1). В качестве «отпечатка» внешнего воздействия на систему (например, аварии на производстве) здесь выступает гетерогенная совокупность реакций системы как целого и индивидуальных реакций агентов. (Так, по должностным инструкциям, работники предприятия при аварии обязаны сообщить о ней руководству, однако серьезная авария может вызвать панику и потерю управления). Составными единицами при интерпретации воздействия системой служат инструкции и должностные обязанности. Сильные внешние воздействия изменяют структуру системы (при аварии – эвакуация персонала, создание аварийного штаба и комиссии). «Образом-схемой» в данном случае является стандартный ответ системы (действие по инструкциям), корректировкой схемы – оценка успешности действий и испытание новых предложений (работа комиссии), заменой модуля (модулей) в «образе-схеме» – выбор оптимальных действий (выводы комиссии), запоминанием новой информации – дополнение и изменение инструкций. 350 Словохотов Ю.Л., Неретин И.С. К построению модульной модели распределенного интеллекта. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 341-362 ов Ю.Л., Неретин И.С. К построению модульной модели распределенного интеллекта. П РАН, том 30, вып. 3, 2018 г., стр. 341-362 Табл. 1. Сравнение компонентов индивидуальной интеллектуальной деятельности и распределенного интеллекта организационной системы Table 1. 5. К моделированию распределенного интеллекта Formal conditions of activity for different organizing systems «идеальная комиссия» митинг «коллективный идиот» общие знания по специальности общие намерения ничего общего целевой отбор участников по критерию квалификации случайный отбор участников по близости свободный вход Тесную связь «интеллектуальности» организационной системы с формализацией ее структуры и действий иллюстрирует табл. 2. Рациональность индивидуального поведения людей также возрастает с сокращением возможностей выбора (переход улицы по светофору, покупка или продажа акций на бирже по фиксированной цене) и становится «ограниченной» при выборе из многих возможностей (переход улицы без светофора, произвольная цена акций на бирже). 351 Табл. 2. Формальные условия функционирования разных организационных систем Table 2. Formal conditions of activity for different organizing systems «идеальная комиссия» митинг «коллективный идиот» общие знания по специальности общие намерения ничего общего целевой отбор участников по критерию квалификации случайный отбор участников по близости свободный вход 351 Slovokhotov Yu.L., Neretin I.S. Toward construction of a modular model of distributed intelligence. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 341-362 настроений сильное управление (председатель с решающим голосом) слабое управление нет управления формализованный обмен информацией и мнениями, исключение эмоций неформальный обмен мнениями и эмоциями случайный обмен эмоциями количественное сравнение значимости мнений (голосование) декларации мнений (призывы) нет формулируемых мнений подчинение меньшинства большинству неподчинение меньшинства большинству нет большинства обязательность исполнения решений необязательность исполнения решений нет решений Общеизвестные схемы коллективного поведения «живых» социальных систем с более примитивными агентами (улей, муравейник, гнездо ос) также неплохо согласуются с основными этапами «модульного» алгоритма интеллектуальной деятельности [16]. Как и в организационной системе, состоящей из людей, в поведении насекомых трудно разделить формирование «отпечатка» (фиксацию системой внешнего воздействия) и стандартную реакцию на него («образ-схему»); примером может служить атака муравьев или ос на чужеродный объект вблизи гнезда. (Разделение воздействия, его отражения в сознании и репрезентаций образа в психологии также является непростой задачей [19]). «Танцы» пчел в улье и «тропинки» муравьев, которые можно рассматривать как модули, направляющие движения особей, изменяются при изменении внешних условий. Перемещение муравьиного гнезда из неблагоприятной зоны в благоприятную, восстановление поврежденного муравейника и многие другие действия коллективных насекомых разбиваются на несколько стандартных воспроизводимых режимов (возбуждение, несогласованные действия, согласованные действия, «успокоение» при достигнутой цели) – этот перечень легко продолжить. Slovokhotov Yu.L., Neretin I.S. Toward construction of a modular model of distributed intelligence. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 5. К моделированию распределенного интеллекта 341-362 настроений сильное управление (председатель с решающим голосом) слабое управление нет управления формализованный обмен информацией и мнениями, исключение эмоций неформальный обмен мнениями и эмоциями случайный обмен эмоциями количественное сравнение значимости мнений (голосование) декларации мнений (призывы) нет формулируемых мнений подчинение меньшинства большинству неподчинение меньшинства большинству нет большинства обязательность исполнения решений необязательность исполнения решений нет решений L., Neretin I.S. Toward construction of a modular model of distributed intelligence. Trudy ISP RAS, vol. 30, issue 3, 2018, pp. 341-362 Общеизвестные схемы коллективного поведения «живых» социальных систем с более примитивными агентами (улей, муравейник, гнездо ос) также неплохо согласуются с основными этапами «модульного» алгоритма интеллектуальной деятельности [16]. Как и в организационной системе, состоящей из людей, в поведении насекомых трудно разделить формирование «отпечатка» (фиксацию системой внешнего воздействия) и стандартную реакцию на него («образ-схему»); примером может служить атака муравьев или ос на чужеродный объект вблизи гнезда. (Разделение воздействия, его отражения в сознании и репрезентаций образа в психологии также является непростой задачей [19]). «Танцы» пчел в улье и «тропинки» муравьев, которые можно рассматривать как модули, направляющие движения особей, изменяются при изменении внешних условий. Перемещение муравьиного гнезда из неблагоприятной зоны в благоприятную, восстановление поврежденного муравейника и многие другие действия коллективных насекомых разбиваются на несколько стандартных воспроизводимых режимов (возбуждение, несогласованные действия, согласованные действия, «успокоение» при достигнутой цели) – этот перечень легко продолжить. 6. «Измерение» распределенного интеллекта модельной системы Toward construction of a modular model of distributed intelligence. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 341-362 Slovokhotov Yu.L., Neretin I.S. Toward construction of a modular model of distributed intelligence. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 341-362 Slovokhotov Yu.L., Neretin I.S. Toward construction of a modular model of distributed intelligence. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 341-362 Рис.5. (а) Изменение среднего времени прохода t при обучении» системы, (б) изменения коэффициентов движения Fig. 5. (a) Evolution of average passing time t during ‘learning’ (b) evolution of movement parameters Fig. 5. (a) Evolution of average passing time t during ‘learning’ (b) evolution of movement parameters В ходе эволюции искусственной МСС среднее время прохода t за первые 150- 200 циклов уменьшалось от 1040–1050 до 880–870 шагов и далее оставалось приблизительно постоянным (рис. 5 а). Параметры движения агентов при этом продолжали изменяться (рис. 5 б). Характер движения в результате эволюции изменялся от хаотического дрейфа при многочисленных столкновениях с препятствиями и между собой в «необученной» системе к согласованному движению группами, огибающими препятствия, в «обученной» системе (рис. 4 б). Время прохождения коридора агентами с фиксированными параметрами движения (k1, 0.01, 0.01, 0.01) при k1≤0.05 уменьшалось с увеличением радиуса восприятия, проходя через минимум (рис. 6). (Дальнейшее увеличение k1 приводило к достижению всеми агентами максимальной скорости v0 за первые 10-15 шагов дискретного времени и сопровождалось увеличением t с его небольшим линейным ростом при увеличении R). Таким образом, расчеты наглядно продемонстрировали способность модельной МСС из искусственных агентов воспринимать информацию и оптимизировать динамику в ходе «обучения» (т.е. генетического отбора «быстрых» агентов). 6. «Измерение» распределенного интеллекта модельной системы Компьютерное моделирование динамики МСС неявно включает оценку уровня распределенного интеллекта модельной системы (например, транспортного потока [11] – по скорости движения, пропускной способности дороги, предсказываемому числу аварий и т. д.) и максимизацию этого уровня. Мерой РИ в таких расчетах является близость достигаемого результата к объективной цели системы – в случае транспортного потока к максимально быстрому безаварийному движению. Для компьютерной иллюстрации этих положений мы использовали анимационную модель, в которой агенты с одинаковым радиусом r0 перемещались по коридору с препятствиями, блокирующими сквозное движение (рис. 4). Состояние агента 352 Словохотов Ю.Л., Неретин И.С. К построению модульной модели распределенного интеллекта. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 341-362 задавали пять числовых параметров (R(n), k1, k2, k3, k4), где R = nr0 – радиус восприятия агентом его окружения, а остальные 0≤ki≤0.1 – параметры движения агента: k1 – ускорение вправо: приращение скорости на каждом шаге дискретного времени как доля максимальной скорости v0 (целеполагание агента), k2 – торможение перед препятствием в радиусе восприятия, k2 – торможение перед препятствием в радиусе восприятия, k2 – торможение перед препятствием в радиусе восприятия, k4 и k3 – соответственно следование за другими агентами в радиусе восприятия и коррекция собственной скорости по скорости их движения. Столкновение агента с препятствиями и другими агентами происходили как абсолютно упругие с соответствующими изменениями направления движения и скоростей. Более детальное описание схемы расчета будет представлено в отдельной публикации. Рис.4. (а) Схема «коридора» с препятствиями и параметры агента, (б) движение в «обученной» системе Fig. 4. (a) Pathway with obstacles and agent’s parameters; (b) movement of agents in a «trained» system Рис.4. (а) Схема «коридора» с препятствиями и параметры агента, (б) движение в «обученной» системе Fig. 4. (a) Pathway with obstacles and agent’s parameters; (b) movement of agents in a «trained» system В первом цикле расчета 24 агента одинакового радиуса r0 с фиксированным радиусом восприятия R=3r0 и параметрами движения (0.01, 0, 0, 0), случайным образом размещенные в полосе шириной 5r0 перед «стартовой» линией, перемещались по коридору слева направо до «финиша», т.е. прохождения полной длины коридора. Направления движения и скорости агентов корректировались на каждом шаге дискретного времени в соответствии с параметрами {ki} по объектам (препятствиям и другим агентам) в радиусе восприятия. В последующих циклах параметры движения 12 «медленных» агентов, преодолевших коридор на предыдущем цикле позже 12 «быстрых», заменяли параметрами 12 «быстрых» агентов со случайными вариациями {ki =0.001} (генетический отбор). Количественной оценкой РИ системы служило среднее время t прохождения коридора агентом в каждом цикле. 353 Slovokhotov Yu.L., Neretin I.S. 7. Обсуждение 30, issue 3, 2018, pp. 341-362 составляющих ее людей, вряд ли способна принять какое-либо общее решение. составляющих ее людей, вряд ли способна принять какое-либо общее решение. «Модульный» подход представляется продуктивным также для конструирования искусственного интеллекта (ИИ) в вычислительной компьютерной среде. Известно, что при «обучении» многослойных нейронных сетей векторные реплики распознаваемых объектов в промежуточных слоях ИНС преобразуются к модульной структуре [21]. В общем случае распознаваемому объекту может сопоставляться набор модулей, фиксирующих определенные качества объекта – что открывает путь к созданию самообучающихся программ. В отличие от индивидуального человеческого интеллекта, число модулей в «образе-схеме», при хороших вычислительных мощностях, может быть достаточно большим, а перебор «плохих» модулей – быстрым и эффективным. Параметры «модульного» ИИ, таким образом, могут существенно превысить любые человеческие возможности, что указывает на вероятные риски при его реализации. «Модульный» подход представляется продуктивным также для конструирования искусственного интеллекта (ИИ) в вычислительной компьютерной среде. Известно, что при «обучении» многослойных нейронных сетей векторные реплики распознаваемых объектов в промежуточных слоях ИНС преобразуются к модульной структуре [21]. В общем случае распознаваемому объекту может сопоставляться набор модулей, фиксирующих определенные качества объекта – что открывает путь к созданию самообучающихся программ. В отличие от индивидуального человеческого интеллекта, число модулей в «образе-схеме», при хороших вычислительных мощностях, может быть достаточно большим, а перебор «плохих» модулей – быстрым и эффективным. Параметры «модульного» ИИ, таким образом, могут существенно превысить любые человеческие возможности, что указывает на вероятные риски при его реализации. 7. Обсуждение Проявлениям РИ в социальных системах, включая различные аспекты коллективной человеческой деятельности, посвящена обширная литература (см. [1, 10, 30, 32, 33]). Однако в большинстве современных публикаций интеллектуальная деятельность МСС трактуется весьма ограниченно: как обмен информацией между агентами [1, 33] или как выработка общего мнения 354 Словохотов Ю.Л., Неретин И.С. К построению модульной модели распределенного интеллекта. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 341-362 Словохотов Ю.Л., Неретин И.С. К построению модульной модели распределенного интеллекта. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 341-362 Словохотов Ю.Л., Неретин И.С. К построению модульной модели распределенного интеллекта. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 341-362 в совокупности людей («crowd wisdom» [34]). За пределами обсуждения, аналогично теориям индивидуального интеллекта, оказывается способность МСС не только воспринимать и обрабатывать информацию, но и генерировать новую информацию – например, в форме муравьиных «тропинок», ведущих к новому, не использованному ранее источнику пищи. При этом «обучаемая» система может состоять из агентов с нулевыми когнитивными возможностями – как в построенной нами модели движения по коридору, где динамику каждого агента задавали пять числовых параметров. Рис.6. Зависимость времени прохода t от радиуса восприятия R при разных коэффициентах k1 Fig 6 Passing time t vs perception radius R at different values of movement parameter k Рис.6. Зависимость времени прохода t от радиуса восприятия R при разных коэффициентах k1 6 P i i i di R diff l f g. 6. Passing time t vs. perception radius R at different values of movement parame g g p p ff f p 1 Организационные системы, состоящие из людей, также способны не только к обмену индивидуальными мнениями, приводящему к «консенсусу» (что воспроизводится большинством социологических моделей, см. [14, 15]), но и к производству новой информации «коллективным разумом». РИ таких систем отличается от человеческого сознания как принципиально меньшим числом «узлов» (т.е. людей) в сетевых структурах, так и способностью этих узлов к глубокой переработке информации. Текущая информация о состоянии, генерируемая в системе, направляет ее коллективную динамику в переменных внешних условиях и фиксируется в виде административных «образов-схем» – т.е. распоряжений, правил и инструкций. При этом «глубина» РИ явно возрастает с повышением формализации связей в структуре МСС. Так, в модельных примерах таблицы 2, строгая регламентация «идеальной комиссии» (левая колонка) является эвристикой, стимулирующей ее содержательную работу, тогда как полностью деформализованная МСС (правая колонка таблицы), независимо от интеллектуального уровня 355 Slovokhotov Yu.L., Neretin I.S. Toward construction of a modular model of distributed intelligence. Trudy ISP RAN/Proc. ISP RAS, vol. 8. Перспективные направления исследований Исследования распределенного интеллекта социальных систем будут проводиться в нескольких основных направлениях. Исследования распределенного интеллекта социальных систем будут проводиться в нескольких основных направлениях. 9. Заключение Мы полагаем, что предложенный нами «модульный» алгоритм восприятия, обработки, использования и хранения информации применим к описанию индивидуального интеллекта человека и (в скорректированном виде) к моделированию распределенного интеллекта социальных систем. Оба вида информационной динамики реализуются в определенных сетевых структурах (см. [1, 8, 14, 17, 25]), взаимосвязь которых с уровнем интеллекта следует дополнительно изучать. Очевидными направлениями дальнейших исследований в данной области являются детализация и формализация модульной модели интеллекта, эмпирические оценки «глубины» распределенного интеллекта МСС и его непосредственный учет в моделировании мультиагентных систем. Ю.Л.С. признателен доктору психологических наук М.В. Фаликман (НИУ ВШЭ) за предоставленную литературу по когнитивным исследованиям и плодотворное обсуждение «модульной» модели. Исследования распределенного интеллекта социальных систем будут проводиться в нескольких основных направлениях. (1) Разработка модульной модели индивидуального интеллекта, пока сформулированной на качественном уровне, наполнение ее количественным содержанием (структура модулей, ключей и библиотек, модульный состав «образов-схем», механизмы сравнения «отпечатка» с модульной схемой, замены модулей в схеме, обновления модулей в библиотеке и т.д.). (2) Эмпирические оценки уровня распределенного интеллекта для социальных систем разной природы (организаций, партий, политических течений и т.д.). Поиск корреляций «уровня интеллекта» со структурой социальной системы. (3) Формальное описание процессов восприятия и переработки информации индивидуальным сознанием и «коллективным разумом». Исследование математических свойств информационного пространства, объектами которого служат модули восприятия и «образы-схемы». (4) Модификация существующих моделей коллективного поведения (динамика пешеходных и автомобильных потоков, распространение мнений в онлайновых социальных сетях, политические кампании и др.) с непосредственным учетом распределенного интеллекта у системы взаимодействующих агентов. Сопоставление результатов моделирования с эмпирическими данными. 356 Словохотов Ю.Л., Неретин И.С. К построению модульной модели распределенного интеллекта. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 341-362 Словохотов Ю.Л., Неретин И.С. К построению модульной модели распределенного интеллекта. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 341-362 Словохотов Ю.Л., Неретин И.С. К построению модульной модели распределенного интеллекта. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 341-362 Список литературы [1] Moussaïd M, Helbing D, Theraulaz G, How simple rules determine pedestrian behavior and crowd disasters. PNAS, 108 (17), 2011, pp. 6884-6888 [2] Губко М.В., Новиков Д.А. Теория игр в управлении организационными системами. 2-е издание. М.: Синтег, 2005, 138 с. [3] Galam S. Sociophysics: a physicist's modeling of phycho-polytical phenomena. Springer, 2012, 439 p. [4] Захаров А.В. Модели политической конкуренции: обзор литературы. Экономика и математические методы, том 45, вып. 1, 2009 г., стр. 110-128 [5] Dorogovtsev S.N. Lectures on Complex Networks. Clarendon: Oxford, 2010, 134 p. [6] Newman M.E.J. The structure and functions of complex networks, SIAM Review, 45(2), 2003, pp. 167-225. [7] Берновский М.М., Кузюрин Н.Н. Случайные графы, модели и генераторы безмасштабных графов. Труды ИСП РАН, 2012, том 22, стр. 419-432. DOI: 10.15514/ISPRAS-2012-22-22. [8] Евин И.А. Введение в теорию сложных сетей. Компьютерные исследования и моделирование, том 2, вып. 2, 2010 г., стр. 121-141. [9] Новиков Д.А. Модели стратегической рефлексии. Автоматика и телемеханика, том 73, вып.1, 2012 г., стр. 1 -19 [10] Kahneman D. Maps of Bounded Rationality: Psychology for Behavioral Economics Amer. Econ. Rev., 2003, 93(5), pp. 1449-1475. [11] Гасников А.В. (ред.), Введение в математическое моделирование транспортных потоков. М.: МЦНМО, 2013, 428 с. [12] Адамчук А.Н., Есипов С.Е. Коллективно флуктуирующие активы при иналичии арбитражных возможностей и оценка платежных обязательств. Усп. физ. наук, том 167, вып. 12, 1997 г., стр. 1295-1306. р [13] Schelling T. Dynamic models of segregation. J. Math. Sociol., 1 (2), 1971, pp. 143-186 [14] Castellano C., Fortunato S., Loreto V. Statistical physics of social dynamics. Rev. Mod. Phys. 81 (2), 2009, pp. 591-646. 357 Slovokhotov Yu.L., Neretin I.S. Toward construction of a modular model of distributed intelligence. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 341-362 [15] Словохотов Ю.Л. Физика и социофизика. Проблемы управления, 2012, вып. 3, стр. 2–34. тков В.Е. Мир общественных насекомых, 3-е изд., М.: Либроком, 2009, 408 с. [16] Кипятков В.Е. Мир общественных насекомых, 3-е изд., М.: Либроком, 2 [16] Кипятков В.Е. Мир общественных насекомых, 3-е изд., М.: Либроком, 2009, 408 с. [17] Engelbtecht A.P. Fundamentals of computational swarm intelligence. N.-Y.: Wiley, 2005, 672 p. р р [17] Engelbtecht A.P. Fundamentals of computational swarm intelligence. N.-Y.: Wiley, 2005, 672 p. [18] Фаликман М.В. Основные подходы в когнитивной науке. http://www.soc- phys.chem.msu.ru/rus/prev/zas-2017-02-09/presentation.pdf. Дата обращения: 02.04.2018. [19] Величковский Б.М. Когнитивная наука: основы психологии познания. В 2-х т.. М.: Смысл; Академия, 2006. [20] Shaib-Draa B, Moulin B., Mandiau P., Millot P. Словохотов Ю.Л., Неретин И.С. К построению модульной модели распределенного интеллекта. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 341-362 ов Ю.Л., Неретин И.С. К построению модульной модели распределенного интеллекта. П РАН, том 30, вып. 3, 2018 г., стр. 341-362 DOI: 10.15514/ISPRAS-2018-30(3)-23 For citation: Slovokhotov Yu.L., Neretin I.S. Toward construction of a modular model of distributed intelligence. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 341-362 (in Russian). DOI: 10.15514/ISPRAS-2018-30(3)-23 For citation: Slovokhotov Yu.L., Neretin I.S. Toward construction of a modular model of distributed intelligence. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 341-362 (in Russian). DOI: 10.15514/ISPRAS-2018-30(3)-23 Toward construction of a modular model of distributed intelligence 1,2Yu.L. Slovokhotov <slov@phys.chem.msu.ru> 3I.S. Neretin <ivan@neretin.ru> 1Lomonosov Moscow State University, GSP-1, 1 Leninskie Gory, Moscow, 119991, Russia 2Institute of Organoelement Compounds, Russian Academy of Sciences, 28 Vavilov St., Moscow, 119991, Russia 3Rock Flow Dynamics, 25A Profsoyuznaya St., Moscow, 117418, Russia Abstract. Multi-agent social systems (MASS) in general are systems of autonomous interdependent agents each pursuing its own goals interacting with other agents and environment. Dynamics of MASS cannot be adequately modeled by the methods borrowed from statistical physics since these methods do not reflect the main feature of social systems, viz. their ability to percept, process and use the external information. This important quality of distributed (“swarm”) intelligence has to be directly taken into account in a correct theoretical description of social systems. However, discussion of distributed intelligence (DI) in the literature is mostly restricted to distributed tasks, information exchange and aggregated judgment – i.e. to ‘sum’ or ‘average’ of independent intellectual activities. This approach ignores empirically well-known option of a ‘collective insight’ in a group as a special demonstration of MASS’s DI . It this paper, a state of art in modeling social systems and studies of intelligence per se are briefly characterized, and a new modular model of intelligence is suggested. The model allows to reproduce the most important result of intellectual activity, i.e. creation of new information, which is not reflected in the contemporary theoretical schemes (e.g. neural networks). Using the “modular” approach, a correspondence between individual intelligence and DI of MASS is discussed, and prospective directions for future studies are suggested. Efficiency of DI was estimated numerically by computer simulations of a simple system of agents with variable kinematic parameters {ki}, moving through a pathway with obstacles. Selection of fast agents with ‘positive mutation’ of parameters gives ca. 20% reduction of average passing time after 200- 300 cycles and creates a swarm movement where agents follow a leader and cooperatively avoid obstacles. Keywords: multi-agent social systems; swarm intelligence; models of intelligence. Список литературы Trends in distributed artificial intelligence. Artific. Intelligence Rev., 6 (1), 1992, pp. 35-66. g g pp [21] Хайкин С. Нейронные сети. Полный курс. 2-е издание. Вильямс, 2016, 1104 с. [22] Петухов В.В. Психология мышления. Учебно-методическое пособие. М.: МГУ, 1987, 99 с. [23] Альтшуллер Г.С. Найти идею. Введение в ТРИЗ – теорию решения изобретательских задач. 4-е изд. М.: Альпина Паблишерз, 2011, 400 с. [24] Чернавский Д.С. Синергетика и информация: динамическая теория информации. 3-е изд. М.: Либроком, 2009, 304 с. [25] Анохин К.В. Когнитом: разум как физическая и математическая структура. http://www.soc-phys.chem.msu.ru/rus/prev/zas-2016-09-27/presentation.pdf. Дата обращения: 02.04.2018. [26] Шумский С.А. Моделирование работы мозга: состояние и перспективы. http://www.soc-phys.chem.msu.ru/rus/prev/zas-2015-03-31/presentation.pdf. Дата обращения: 02.04.2018. [27] Ohlsson S. Information-processing explanations of insight and related phenomena. In M. T. Keane & K. J. Gilhooly (Eds.), Advances in the psychology of thinking. New York, NY: Harvester Wheatsheaf, 1992, pp. 1– 44. [28] Fodor J.A. The Modularity of Mind. MIT Press 1983,142 p. [28] Fodor J.A. The Modularity of Mind. MIT Press 1983,142 p. [29] Курганский А.В. Понятие внутренней репрезентации в когнитивной нейронауке. http://www.soc-phys.chem.msu.ru/rus/prev/zas-2017-02-28/presentation.pdf. Дата обращения: 02.04.2018. р [30] Dandurand F., Shultz T.R., Rivest F. Complex problem solving with reinforcement learning. Proc. 6th IEEE Internat. Conf. on Development and Learning, 2007, pp. 157- 162. [31] Hebb D.O. The Organization of Behavior: a Neuropsychological Theory, Wiley, 1949, 335 p. [32] Mossaid M., Garnieer S., Theraulaz G., Helbing D. Collective information processing and pattern formation in swarms, flocks and crowds. Topics Cogn. Sci., 1, 2009, pp 469- 497. [33] Becker J., Brackbill D., Centola D. PNAS, 114, 2017, pp. E5070-E5076. 358 [1] Moussaïd M, Helbing D, Theraulaz G, How simple rules determine pedestrian behavior and crowd disasters PNAS, 108 (17), 2011, pp. 6884-6888 [2] Goubko M V Novikov D A Game theory in control of organizing systemsТ 2 nd ed [1] Moussaïd M, Helbing D, Theraulaz G, How simple rules determine pedestrian behavior and crowd disasters PNAS, 108 (17), 2011, pp. 6884-6888 [2] Goubko M.V.,Novikov D.A. Game theory in control of organizing systemsТ. 2-nd ed.. Moscow, Sinteg, 2005, 138 p. (in Russian) , ( ), , pp [2] Goubko M.V.,Novikov D.A. Game theory in control of organizing systemsТ. 2-nd ed.. Moscow, Sinteg, 2005, 138 p. (in Russian) References [1] Moussaïd M, Helbing D, Theraulaz G, How simple rules determine pedestrian behavior and crowd disasters PNAS, 108 (17), 2011, pp. 6884-6888 [1] Moussaïd M, Helbing D, Theraulaz G, How simple rules determine pedestrian behavior and crowd disasters PNAS, 108 (17), 2011, pp. 6884-6888 [1] Moussaïd M, Helbing D, Theraulaz G, How simple rules determine pedestrian behavior and crowd disasters PNAS, 108 (17), 2011, pp. 6884-6888 pp [2] Goubko M.V.,Novikov D.A. Game theory in control of organizing systemsТ. 2-nd ed.. Moscow, Sinteg, 2005, 138 p. (in Russian) [2] Goubko M.V.,Novikov D.A. Game theory in control of organizing systemsТ. 2-nd ed.. Moscow, Sinteg, 2005, 138 p. (in Russian) 359 Slovokhotov Yu.L., Neretin I.S. Toward construction of a modular model of distributed intelligence. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 341-362 [3] Galam S. Sociophysics: a physicist's modeling of phycho-polytical phenomena. Springer, 2012, 439 p. [4] Zakharov A.V. Models of political competition^ a review. Economics and Mathematical Methods [Ekonomika I matematicheskiye metody], 45 (1), , 2009 pp. 110-128 (in Russian) [5] Dorogovtsev S.N. Lectures on Complex Networks. Clarendon: Oxford, 2010, 134 p. [6] Newman M.E.J., The structure and functions of complex networks, SIAM Review, 45(2), 2003, pp. 167-225. [7] Bernovskiy M.M., Kuzyurin N.N. Random graphs, models and generators of scale-free graphs. Trudy ISP RAN/Proc. ISP RAS, 2012, v. 22, pp. 419-432 (in Russian). DOI: 10.15514/ISPRAS-2012-22-22. [8] Yevin I.A. Introduction to a theory of complex networks. Computer Research and Modeling [Kompyuternye issledovaniya I modelirovanie], 2 (2), 2010, pp. 121-141 (in Russian). [9] Novikov D.A. Models of strategic behavior. Automation and Remote Control, 73 (1), 2012, pp. 1 -19 [10] Kahneman D. Maps of Bounded Rationality: Psychology for Behavioral Economics Amer. Econ. Rev., 2003, 93(5), pp. 1449-1475. [11] Gasnikov A.V. (Ed.). Introduction into mathematic modeling of traffic flows. Moscow, MTsIMO, 2013, 428 p. (in Russian) [12] Adamchuk A.N., Esipov C.E. Collectively fluctuating assets in the presence of arbitrage opportunities, and option pricing. Phys. Usp., vol. 40, 1997, pp. 1239–1248 [13] Schelling T. Dynamic models of segregation. J. Math. Sociol., 1 (2), 1971, pp. 143-186 [14] Castellano C., Fortunato S., Loreto V. Statistical physics of social dynamics. Rev. Mod. Phys. 81 (2), 2009, pp. 591-646. [15] Slovokhotov Yu.L. Physics vs. Sociophysics. Control Science [Problemy upravleniya], 2012 (3), pp. 2-34 (in Russian). Kipyatkov V.E. World of social insects, 3rd Ed., Moscow, Librokom, 2009, 308 p. [17] Engelbtecht A.P. References Fundamentals of computational swarm intelligence. N.-Y.: Wiley, 2005, 672 p. [18] Falikman M.V. The principal approaches in cognitive science http://www.soc- phys.chem.msu.ru/rus/prev/zas-2017-02-09/presentation.pdf (in Russian). Accessed 02.04.2018. [19] Velichkovsky B.M. Cognitive science: foundation of cognition psychology. In 2 volumes.Moscow, Smysl, Akademiya, 2006 (in Russian). [20] Shaib-Draa B, Moulin B., Mandiau P., Millot P. Trends in distributed artificial intelligence. Artific. Intelligence Rev., 6 (1), 1992, pp. 35-66. [21] Khaikin S. Neural networks. A comprehensive course. 2nd edition. Viljams, 2016, 1104 p. (in Russian) [22] Petukhov V.V. Psykhology of thinking A textbook. Moscow, MGU,. 1987, 99 p. (in Russian). [23] Altshuler G.S. To find the idea. Introduction into TRIZ: a theory of solving inventional problems. 4ЕР ed. Moscow, Alpina Publishers, 2011, 400 p. (in Russian) [24] Chernavsky D.S. Synergetics and information: Moscow, Librokom, 2009, 304 p. (in Russian) [24] Chernavsky D.S. Synergetics and information: dynamic theory of information. 3rd ed. Moscow, Librokom, 2009, 304 p. (in Russian) [25] Anokhin K,V. Cognitom: mind as a physical and mathematical structure. http://www.soc-phys.chem.msu.ru/rus/prev/zas-2016-09-27/presentation.pdf (in Russian). Accessed 02.04.2018. 360 Словохотов Ю.Л., Неретин И.С. К построению модульной модели распределенного интеллекта. Труды ИСП РАН, том 30, вып. 3, 2018 г., стр. 341-362 [26] Shumsky S.A. Modeling of brain activity: state of art and prospects. http://www.soc- phys.chem.msu.ru/rus/prev/zas-2015-03-31/presentation.pdf (in Russian). Accessed 02.04.2018. [27] Ohlsson S. Information-processing explanations of insight and related phenomena. In M. T. Keane & K. J. Gilhooly (Eds.), Advances in the psychology of thinking, New York, NY: Harvester Wheatsheaf, 1992, pp. 1– 44. pp [28] Fodor J.A. The Modularity of Mind. MIT Press 1983,142 p. [29] Kurgansky A.V. Internal representation in cognitive neuroscience. http://www.soc- phys.chem.msu.ru/rus/prev/zas-2017-02-28/presentation.pdf (in Russian). Accessed 02.04.2018. [30] Dandurand F., Shultz T.R., Rivest F. Complex problem solving with reinforcement learning Proc. 6th IEEE Internat. Conf. on Development and Learning, 2007, pp 157- 162. [31] Hebb D.O. The Organization of Behavior: a Neuropsychological Theory, Wiley, 1949, 335 p. [32] Mossaid M., Garnieer S., Theraulaz G., Helbing D. Collective information processing and pattern formation in swarms, flocks and crowds. Topics Cogn. Sci., 1, 2009, pp. 469-497. [33] Becker J., Brackbill D., Centola D. PNAS, 114, 2017, pp. E5070-E5076. 361 Slovokhotov Yu.L., Neretin I.S. Toward construction of a modular model of distributed intelligence. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 341-362 Slovokhotov Yu.L., Neretin I.S. Toward construction of a modular model of distributed intelligence. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. Slovokhotov Yu.L., Neretin I.S. Toward construction of a modular model of distributed intelligence. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 341-362 References 341-362 Slovokhotov Yu.L., Neretin I.S. Toward construction of a modular model of distributed intelligence. Trudy ISP RAN/Proc. ISP RAS, vol. 30, issue 3, 2018, pp. 341-362 362
https://openalex.org/W2596523885	https://europepmc.org/articles/pmc5428303?pdf=render	English	null	Palmitoylethanolamide induces microglia changes associated with increased migration and phagocytic activity: involvement of the CB2 receptor	Scientific reports	2,017	cc-by	11,365	Palmitoylethanolamide induces microglia changes associated with increased migration and phagocytic activity: involvement of the CB2 receptor F. Guida1,5, L. Luongo1,5, S. Boccella1, M. E. Giordano1, R. Romano1, G. Bellini2, I. Manzo2, A. Furiano1, A. Rizzo3, R. Imperatore4,6, F. A. Iannotti4,5, E. D’Aniello4,5, F. Piscitelli4,5, F. sca Rossi2, L. Cristino4,5, V. Di Marzo4,5, V. de Novellis1,5 & S. Maione1,5 Received: 15 September 2016 Accepted: 22 February 2017 Published: xx xx xxxx The endogenous fatty acid amide palmitoylethanolamide (PEA) has been shown to exert anti- inflammatory actions mainly through inhibition of the release of pro-inflammatory molecules from mast cells, monocytes and macrophages. Indirect activation of the endocannabinoid (eCB) system is among the several mechanisms of action that have been proposed to underlie the different effects of PEA in vivo. In this study, we used cultured rat microglia and human macrophages to evaluate whether PEA affects eCB signaling. PEA was found to increase CB2 mRNA and protein expression through peroxisome proliferator-activated receptor-α (PPAR-α) activation. This novel gene regulation mechanism was demonstrated through: (i) pharmacological PPAR-α manipulation, (ii) PPAR-α mRNA silencing, (iii) chromatin immunoprecipitation. Moreover, exposure to PEA induced morphological changes associated with a reactive microglial phenotype, including increased phagocytosis and migratory activity. Our findings suggest indirect regulation of microglial CB2R expression as a new possible mechanism underlying the effects of PEA. PEA can be explored as a useful tool for preventing/ treating the symptoms associated with neuroinflammation in CNS disorders. Microglial cells are the resident immune cells responsible for maintaining homeostasis in the Central Nervous System (CNS). The activation of microglia is part of an early defense mechanism following injury or disease1. In pathological conditions, these cells assume an activated state characterized by morphological rearrangement, proliferation, chemotaxis towards the site of damage, and release of mediators2. Increasing evidence indicates that microglial activation is heterogeneous. In fact, depending on the phenotype assumed, microglia can elicit either cytotoxic or neuroprotective effects3. In addition to the inflammatory component, microglia are also considered to be the professional phagocytes in the CNS, responsible for removing dying or apoptotic cells, myelin debris, and bacteria4. The majority of pathogens that reach the brain are physiologically eliminated or sequestered in a latent form by microglia. However, highly virulent strains can produce a progressive inflammatory response associated with increased expression of cytokines and dysregulation of microglia cells, which is an hallmark of early stage neurodegeneration5. y g g It is well known that microglia express components of the endocannabinoid (eCB) system, including recep- tors, ligands, and metabolic enzymes. In particular, the cannabinoid (CB) type 2 receptor (CB2R) is tightly 1Department of Experimental Medicine, Section of Pharmacology L. Donatelli, Università degli Studi della Campania “Luigi Vanvitelli” (Ex SUN), 80138, Naples, Italy. www.nature.com/scientificreports www.nature.com/scientificreports www.nature.com/scientificreports Received: 15 September 2016 Accepted: 22 February 2017 Published: xx xx xxxx Results h PEA enhances CB2R expression via PPAR-α activation. Control experiments verified that PEA treat- ment (1–100 nM for 24 hours) did not induce significant disruption of cellular integrity, as measured by release of the cytoplasmic enzyme lactate dehydrogenase (LDH) (not shown).f y p y y g We therefore examined the effect of a 24 hour incubation of rat microglia with 1, 10, and 100 nM PEA on CB2R expression. IFNγ (300 U/mL for 24 hours) was used as a positive control23. No difference between untreated or vehicle-treated cells (0.01% DMSO) was observed (not shown). We found that the highest concentration of PEA (100 nM) significantly increased CB2 mRNA and protein levels in microglia compared to the levels in vehicle-treated cells (Fig. 1A and B). The involvement of PPAR-α in the PEA-mediated effect was evaluated by assessing the ability of the synthetic PPAR-α selective agonist GW7647 to modulate CB2R expression. CB2R expression was significantly higher in microglia treated with GW7647 (1 µM for 24 hours) than in vehicle-treated cells (Fig. S1A). Pre-treatment (15 min before) with the PPAR-α antagonist GW6471 (10 µM for 24 hours) com- pletely blocked the enhancement of CB2R protein expression induced by both PEA and GW7647 (Figs 1B and S1A). The treatment with GW6471 alone did not significantly change CB2R expression, as compared with the vehicle (Fig. 1B). By immunocytochemical analysis, we also showed CB2R expression in cultured microglia. In particular, we observed the expression of CB2R in the cytosol of microglia incubated with PEA, GW7647 or IFNγ, whereas little staining was detected in vehicle-treated cells (Fig. 1C). Likewise, we found the presence of PPAR-α in the cytosol and, as punctate staining, also in the perinuclear area of resting or treated microglia (Fig. 1D). y p g p g g g Finally, we observed that all treatments induced morphological changes that are typically associated with an activated microglial state, such as increased soma diameters and shorter process lengths24. In contrast, vehicle-treated cells were in a ‘resting’ condition (Fig. 1F).f g ( g ) To investigate the involvement of PPAR-α in PEA effect on CB2R expression, we silenced PPAR-α mRNA expression in microglia and measured CB2R expression in the presence of PEA. Silencing PPAR-α gene expres- sion with a specific siRNA led to a significant reduction in PPAR-α expression (81% reduction), whereas a non-targeting siRNA did not affect PPAR-α gene expression (Fig. 1E’). Palmitoylethanolamide induces microglia changes associated with increased migration and phagocytic activity: involvement of the CB2 receptor F. Guida1,5, L. Luongo1,5, S. Boccella1, M. E. Giordano1, R. Romano1, G. Bellini2, I. Manzo2, A. Furiano1, A. Rizzo3, R. Imperatore4,6, F. A. Iannotti4,5, E. D’Aniello4,5, F. Piscitelli4,5, F. sca Rossi2, L. Cristino4,5, V. Di Marzo4,5, V. de Novellis1,5 & S. Maione1,5 2Department of Women, Child and General and Specialistic Surgery, Università degli Studi della Campania “Luigi Vanvitelli” (Ex SUN), 80138, Naples, Italy. 3Department of Experimental Medicine, Section of Microbiology and Clinical Microbiology, Università degli Studi della Campania “Luigi Vanvitelli” (Ex SUN), 80138, Naples, Italy. 4Institute of Biomolecular Chemistry, Consiglio Nazionale delle Ricerche, Pozzuoli, Italy. 5Endocannabinoid Research Group, Institute of Biomolecular Chemistry, C.N.R., Pozzuoli, Italy. 6Department of Science and Technology, University of Sannio, Benevento, Italy. F. Guida and L. Luongo contributed equally to this work. Correspondence and requests for materials should be addressed to S.M. (email: sabatino.maione@unina2.it) Scientific Reports \| 7: 375 \| DOI:10.1038/s41598-017-00342-1 1 www.nature.com/scientificreports/ regulated on microglial surfaces during several pathological states6, 7. Indeed, selective CB2R stimulation inhibits microglial reactivity and promotes a neuroprotective phenotype8, 9. g y p p p yp Palmitoylethanolamide (PEA) is an eCB-like compound, which belongs to the class of long chain fatty acid ethanolamides that have been shown to have cytoprotective and anti-inflammatory activity. This is achieved by inhibition of the release of pro-inflammatory molecules from mast cells, monocytes, and macrophages10, 11. PEA is produced by neurons and glial cells in the CNS and is involved in the endogenous neuroprotective mechanisms that are activated following tissue damage or inflammation12–14. PEA exerts anti-inflammatory and analgesic effects through several molecular and cellular mechanisms15–17. Although it does not bind to CB receptors with high affinity18, selective blockers of CB1 or CB2 receptors have been shown to inhibit a multitude of its effects in vivo19, 20. In fact, PEA has been shown to act on several targets, including GPR55, transient receptor potential vannilloid type-1 (TRPV1) channels, fatty acid amide hydrolase (FAAH), and, last but not least, peroxisome proliferator-activated receptor-α (PPAR-α)21, 22. However, the mechanism by which PEA acts on these receptors is highly debated. The apparent multitarget action of PEA inspired us to investigate the possible mechanisms underlying the role of CBRs activation in response to PEA-mediated effects.f y g pf In the present study, we aimed to determine if PEA affects eCB signaling, by focusing on the expression of CB2R in microglia cells, because of the well established role of this receptor in immune cells6. Using pharmaco- logical, biochemical, and computational approaches, we investigated the involvement of PPAR-α in CB2R reg- ulation in response to PEA. Palmitoylethanolamide induces microglia changes associated with increased migration and phagocytic activity: involvement of the CB2 receptor F. Guida1,5, L. Luongo1,5, S. Boccella1, M. E. Giordano1, R. Romano1, G. Bellini2, I. Manzo2, A. Furiano1, A. Rizzo3, R. Imperatore4,6, F. A. Iannotti4,5, E. D’Aniello4,5, F. Piscitelli4,5, F. sca Rossi2, L. Cristino4,5, V. Di Marzo4,5, V. de Novellis1,5 & S. Maione1,5 Additionally, the capability of PEA to regulate the ability of microglia to migrate and phagocytose as a possible consequence of CB2R activation was also evaluated. The subsequent discovery of a further mechanism of action for PEA paves the way to investigate its possible implication in several chronic neuroinflammatory diseases. Results h (C) CB2R expression (red) in Iba-1 labeled microglia (green) following incubation with vehicle, PEA (100 nM), GW7647 (1 µM), IFNγ (300 U/ml). Scale bar = 25 μm. (D) PPAR-α expression (green) in microglia cells (red) following incubation with vehicle, PEA (100 nM), GW7647 (1 µM), IFNγ (300 U/ml). Scale bar = 25 μm. (E) Representative western blot image and related quantification showing the expression of CB2R in microglial cell lysates incubated for 24 hours with vehicle, PEA (100 nM), GW7647 (1 µM), IFNγ (300 U/ml), in presence Figure 1. PEA induces CB2R expression via PPAR-α activation. (A) Real-time PCR analyses of CB2m-RNA in cultured microglia following incubation with PEA (1, 10 and 100 nM for 24 hours), using 18S as a loading control. Data are shown as mean ± SD (n = 6–8). P < 0.05 compared to vehicle-treated cells. One way ANOVA, post-hoc Tukey’s. (B) Representative western blot image and related quantification showing the expression of CB2R in microglial cell lysates following incubation with vehicle, PEA (100 nM), GW6471 (10 µM) + PEA, or GW6471 alone, using β-actin as a loading control. Data are shown as mean ± SD (n = 6–8). P < 0.05 compared to vehicle-treated cells, °P < 0.05 compared to PEA-treated cells. One way ANOVA, post-hoc Tukey’s. (C) CB2R expression (red) in Iba-1 labeled microglia (green) following incubation with vehicle, PEA (100 nM), GW7647 (1 µM), IFNγ (300 U/ml). Scale bar = 25 μm. (D) PPAR-α expression (green) in microglia cells (red) following incubation with vehicle, PEA (100 nM), GW7647 (1 µM), IFNγ (300 U/ml). Scale bar = 25 μm. (E) Representative western blot image and related quantification showing the expression of CB2R in microglial cell lysates incubated for 24 hours with vehicle, PEA (100 nM), GW7647 (1 µM), IFNγ (300 U/ml), in presence of siRNA or non-targeting siRNA. Data are shown as mean ± SD (n = 6). P < 0.05 compared to vehicle, °P < 0.05 compared to relative control within non-targeting siRNA group. One way ANOVA, post-hoc Tukey’s. (E’) Real-time PCR analyses of PPAR-α/HPRT ratio in PPAR-α silencing (siRNA) or non-targeting siRNA conditions, compared with vehicle. Data are shown as means ± SD (n = 6). P < 0.05, One way ANOVA, post- hoc Tukey’s. Results h The receptor modulation observed in cells treated with the non-targeting siRNA was comparable to that observed in the absence of any siRNA treatment. In PPAR-α-silenced cells, incubation with PEA or GW7647 failed to increase CB2R expression, as compared to the levels of CB2R in cells treated with a non-targeting siRNA (Fig. 1E). CB2R antibody specificity was assessed in presence of the relative blocking peptide by western blotting and immunocytochemistry (Fig. S1B and C). PEA increases CB2R expression in a human macrophage cell culture. Unable to investigate micro- glia in humans, we explored the possibility of PEA-induced CB2R modulation in human macrophages, based on the evidence that resident microglia and blood-derived-macrophages are derived from the same cell line- age. Similar to the results from our microglia studies, we observed that under basal conditions (untreated or vehicle-treated cells), macrophages expressed low CB2R levels and treatment with PEA (100 nM) significantly increased both CB2 mRNA and protein expression (Fig. S2A and B). In particular, we observed a cyclic expres- sion of mRNA transcripts in 2 and 12 hours-treated cells which was correlated with increased protein levels detected at 4 hours and possibly in the time interval between the 12 and 24 hours.if p y Treatment with GW7647 (1 µM for 4 hours) significantly enhanced CB2R expression and this effect was inhib- ited by GW6471 pre-treatment (10 µM for 4 hours) (Fig. S2C). Interestingly, a decrease in IFNγ-induced CB2R expression was observed in the presence of GW6471, suggesting possible cross-talk between PPAR-α and the IFNγ pathway in CB2R modulation (Fig. S2C). The quantification of CB2R protein expression was complemented by a double immunofluorescence experiment through which we demonstrated a merged fluorescence signal of CB2R and vimentin-labeled macrophages (Fig. S2D). Scientific Reports \| 7: 375 \| DOI:10.1038/s41598-017-00342-1 2 www.nature.com/scientificreports/ Figure 1. PEA induces CB2R expression via PPAR-α activation. (A) Real-time PCR analyses of CB2m-RNA in cultured microglia following incubation with PEA (1, 10 and 100 nM for 24 hours), using 18S as a loading control. Data are shown as mean ± SD (n = 6–8). P < 0.05 compared to vehicle-treated cells. One way ANOVA, post-hoc Tukey’s. (B) Representative western blot image and related quantification showing the expression of CB2R in microglial cell lysates following incubation with vehicle, PEA (100 nM), GW6471 (10 µM) + PEA, or GW6471 alone, using β-actin as a loading control. Data are shown as mean ± SD (n = 6–8). Results h P < 0.05 compared to vehicle-treated cells, °P < 0.05 compared to PEA-treated cells. One way ANOVA, post-hoc Tukey’s. (C) CB2R expression (red) in Iba-1 labeled microglia (green) following incubation with vehicle, PEA (100 nM), GW7647 (1 µM), IFNγ (300 U/ml). Scale bar = 25 μm. (D) PPAR-α expression (green) in microglia cells (red) following incubation with vehicle, PEA (100 nM), GW7647 (1 µM), IFNγ (300 U/ml). Scale bar = 25 μm. (E) Representative western blot image and related quantification showing the expression of CB2R in microglial cell lysates incubated for 24 hours with vehicle, PEA (100 nM), GW7647 (1 µM), IFNγ (300 U/ml), in presence of siRNA or non-targeting siRNA. Data are shown as mean ± SD (n = 6). P < 0.05 compared to vehicle, °P < 0.05 compared to relative control within non-targeting siRNA group. One way ANOVA, post-hoc Tukey’s. (E’) Real-time PCR analyses of PPAR-α/HPRT ratio in PPAR-α silencing (siRNA) or non-targeting siRNA conditions, compared with vehicle. Data are shown as means ± SD (n = 6). P < 0.05, One way ANOVA, post- hoc Tukey’s. (F) Representative staining and related quantification of microglia cells in Iba-1 labeled microglia incubated for 24 hours with PEA (100 nM), GW7647 (1 µM) or IFNγ (300 U/ml), compared with vehicle-treated cells. Morphological evaluations measured as a percentage of activated cells over the total cell number. Data are shown as mean ± SEM (n = 6). **Indicates statistically significant values (P < 0.0001) vs. vehicle-treated cells. One way ANOVA, post-hoc Tukey’s. Full-length gels are showed in Supplementary Figs S6 and S7. Figure 1. PEA induces CB2R expression via PPAR-α activation. (A) Real-time PCR analyses of CB2m-RNA in cultured microglia following incubation with PEA (1, 10 and 100 nM for 24 hours), using 18S as a loading control. Data are shown as mean ± SD (n = 6–8). P < 0.05 compared to vehicle-treated cells. One way ANOVA, post-hoc Tukey’s. (B) Representative western blot image and related quantification showing the expression of CB2R in microglial cell lysates following incubation with vehicle, PEA (100 nM), GW6471 (10 µM) + PEA, or GW6471 alone, using β-actin as a loading control. Data are shown as mean ± SD (n = 6–8). P < 0.05 compared to vehicle-treated cells, °P < 0.05 compared to PEA-treated cells. One way ANOVA, post-hoc Tukey’s. Results h Data are shown as mean ± SEM (n = 3) of three independent experiments conducted in triplicate. p value ≤ 0.05 vscontrol condition (pcDNA-transfected cells), obtained using the unpaired T-TEST for statistical analyses. Figure 2. Cnr2 is directly regulated by PEA-mediated PPAR-α activation: bioinformatic analysis and chromatin immunoprecipitation assay. (A) A schematic alignment of the human and rat Cnr2. The black boxes correspond to regions of high sequence homology between the two species contaning the putative PPARα/ RXR sites indicated as small black rectangles in the lower part below the schematic. The PPARα/RXR matrix used for the analysis is also shown in the upper part. (B) Average data of the relative amount of the PPARα -immunoprecipitated DNA in HEK293 cells transfected with either control (pCDNA3) or murine PPARα and RXR encoding plasmids. Data are from three to six separate experiments and normalized relative to the input DNA, using the 2−ΔΔCt formula. The inset shows a representative agarose gel electrophoresis of the qPCR products obtained from PAX7-immunoprecipitated DNA for each experimental condition. Data are shown as mean ± SEM (n = 3) of three independent experiments conducted in triplicate. p value ≤ 0.05 vscontrol condition (pcDNA-transfected cells), obtained using the unpaired T-TEST for statistical analyses. Human embryonic kidney (HEK293) cells were transiently transfected with control (pcDNA3) or PPARα/RXR encoding plasmids. After 24 hours, transfected HEK293 cells were treated with PEA (0.1 or 0.3 µM for 24 hours). Afterwards, the genomic DNA was isolated from control and PPARα/RXR transfected cells. Then, the DNA was sonicated to obtain fragments between 400–500 bp and used for the immune-affinity reaction using a specific anti-PPAR-α antibody (see experimental procedure section). Subsequently, PPAR-α-immunoprecipitated DNA fragments were used for qPCR analysis using a specific pair of primers able to amplify the genomic portions of interest within the Cnr2 sequence containing the putative PPAR-α binding sites (black arrows, Fig. 2A). We found Human embryonic kidney (HEK293) cells were transiently transfected with control (pcDNA3) or PPARα/RXR encoding plasmids. After 24 hours, transfected HEK293 cells were treated with PEA (0.1 or 0.3 µM for 24 hours). Afterwards, the genomic DNA was isolated from control and PPARα/RXR transfected cells. Then, the DNA was sonicated to obtain fragments between 400–500 bp and used for the immune-affinity reaction using a specific anti-PPAR-α antibody (see experimental procedure section). Results h Data are from three to six separate experiments and normalized relative to the input DNA, using the 2−ΔΔCt formula. The inset shows a representative agarose gel electrophoresis of the qPCR products obtained from PAX7-immunoprecipitated DNA for each experimental condition. Data are shown ±SEM ( 3) f th i d d t i t d t d i t i li t l ≤0 05 t l Figure 2. Cnr2 is directly regulated by PEA-mediated PPAR-α activation: bioinformatic analysis and chromatin immunoprecipitation assay. (A) A schematic alignment of the human and rat Cnr2. The black boxes correspond to regions of high sequence homology between the two species contaning the putative PPARα/ RXR sites indicated as small black rectangles in the lower part below the schematic. The PPARα/RXR matrix used for the analysis is also shown in the upper part. (B) Average data of the relative amount of the PPARα -immunoprecipitated DNA in HEK293 cells transfected with either control (pCDNA3) or murine PPARα and RXR encoding plasmids. Data are from three to six separate experiments and normalized relative to the input DNA, using the 2−ΔΔCt formula. The inset shows a representative agarose gel electrophoresis of the qPCR products obtained from PAX7-immunoprecipitated DNA for each experimental condition. Data are shown as mean ± SEM (n = 3) of three independent experiments conducted in triplicate. p value ≤ 0.05 vscontrol condition (pcDNA-transfected cells), obtained using the unpaired T-TEST for statistical analyses. Figure 2. Cnr2 is directly regulated by PEA-mediated PPAR-α activation: bioinformatic analysis and Figure 2. Cnr2 is directly regulated by PEA-mediated PPAR-α activation: bioinformatic analysis and chromatin immunoprecipitation assay. (A) A schematic alignment of the human and rat Cnr2. The black boxes correspond to regions of high sequence homology between the two species contaning the putative PPARα/ RXR sites indicated as small black rectangles in the lower part below the schematic. The PPARα/RXR matrix used for the analysis is also shown in the upper part. (B) Average data of the relative amount of the PPARα -immunoprecipitated DNA in HEK293 cells transfected with either control (pCDNA3) or murine PPARα and RXR encoding plasmids. Data are from three to six separate experiments and normalized relative to the input DNA, using the 2−ΔΔCt formula. The inset shows a representative agarose gel electrophoresis of the qPCR products obtained from PAX7-immunoprecipitated DNA for each experimental condition. Results h (F) Representative staining and related quantification of microglia cells in Iba-1 labeled microglia incubated for 24 hours with PEA (100 nM), GW7647 (1 µM) or IFNγ (300 U/ml), compared with vehicle-treated cells. Morphological evaluations measured as a percentage of activated cells over the total cell number. Data are shown as mean ± SEM (n = 6). *Indicates statistically significant values (P < 0.0001) vs. vehicle-treated cells. One way ANOVA, post-hoc Tukey’s. Full-length gels are showed in Supplementary Figs S6 and S7. Cnr2 is directly regulated by PEA-mediated PPAR-α activation: bioinformatic analysis and chromatin immunoprecipitation assay. In view of the PEA-induced increased levels of CB2R mRNA and protein observed in microglia, we explored the possibility of a direct regulation of the transcription of Cnr2 (the gene encoding for CB2R) by PPAR-α. As nuclear receptors, PPARs forms heterodimers with the retinoid X receptor (RXR)25. Therefore, we used bioinformatic tools to search for putative consensus sequences for PPARα/ RXR within the Cnr2 gene, as well in its upstream untranslated 3′ and 5′-UTR region.hf g p g The Cnr2 sequence shows many differences between the two species26, and for this reason we searched for putative PPARα/RXR sequence-rich sites located within conserved regions, and identified putative binding sequences within two conserved regions located at the 5′UTR region of Cnr2 (named sequence 1 and 2, Fig. 2A). Besides the two conserved regions identified at 5′UTR portion of the Cnr2, we found other putative PPARα/RXR sequences in non-conserved regions of both human and rat CB2. q g Chromatin immunoprecipitation (Chip) analysis was performed to prove that, once stimulated with PEA, the nuclear receptor PPARα/RXR effectively binds the putative responsive sequences identified within Cnr2. Scientific Reports \| 7: 375 \| DOI:10.1038/s41598-017-00342-1 3 www.nature.com/scientificreports/ Figure 2. Cnr2 is directly regulated by PEA-mediated PPAR-α activation: bioinformatic analysis and chromatin immunoprecipitation assay. (A) A schematic alignment of the human and rat Cnr2. The black boxes correspond to regions of high sequence homology between the two species contaning the putative PPARα/ RXR sites indicated as small black rectangles in the lower part below the schematic. The PPARα/RXR matrix used for the analysis is also shown in the upper part. (B) Average data of the relative amount of the PPARα -immunoprecipitated DNA in HEK293 cells transfected with either control (pCDNA3) or murine PPARα and RXR encoding plasmids. Results h Subsequently, PPAR-α-immunoprecipitated DNA fragments were used for qPCR analysis using a specific pair of primers able to amplify the genomic portions of interest within the Cnr2 sequence containing the putative PPAR-α binding sites (black arrows, Fig. 2A). We found Scientific Reports \| 7: 375 \| DOI:10.1038/s41598-017-00342-1 4 www.nature.com/scientificreports/ Figure 3. PEA enhances phagocytosis and intracellular killing of P. gingivalis by microglial cells. Phagocytosis assay (A) and intracellular survival evaluation (B) were performed in the presence of PEA (100 nM), or GW6471 (10 µM) + PEA, or AM630 (100 nM) + PEA, or GW6471 alone. The number of bacteria ingested (at 90 min) or number of bacteria recovered (at 270 min) by the control group (vehicle) were considered to be 100%. Data are shown as means ± SEM (n = 6). P < 0.05 compared to vehicle, °P < 0.05 compared to PEA. One way ANOVA, post-hoc Tukey’s. Figure 3. PEA enhances phagocytosis and intracellular killing of P. gingivalis by microglial cells. Phagocytosis assay (A) and intracellular survival evaluation (B) were performed in the presence of PEA (100 nM), or GW6471 (10 µM) + PEA, or AM630 (100 nM) + PEA, or GW6471 alone. The number of bacteria ingested (at 90 min) or number of bacteria recovered (at 270 min) by the control group (vehicle) were considered to be 100%. Data are shown as means ± SEM (n = 6). P < 0.05 compared to vehicle, °P < 0.05 compared to PEA. One way ANOVA, post-hoc Tukey’s. that upon stimulation with PEA 0.3 µM, PPAR-α binds with high affinity to only one of the two conserved regions identified in Cnr2 (Fig. 2B). Furthermore, we found a similar result upon exposure of PPARα/RXR transfected HEK293 cells to PEA 0.1 µM, although to a lesser extent, thus suggesting a plausible dose dependent effect exter- ted on PPARa/RXR heterodimerization/activation (data not shown). PEA enhances phagocytosis and intracellular killing of P. gingivalis by microglial cells. To assess the effect of PEA on the phagocytosis of P. gingivalis by microglia, cells were cultured with P. gingivalis and incubated with PEA or vehicle. No significant difference was observed between untreated and vehicle-treated cells (not shown). The numbers of ingested P. gingivalis after 24 hours of PEA treatment are shown in Fig. 3. We observed that PEA (100 nM) induced a significant enhancement of ingestion of bacteria as compared to vehicle-treated cells (Fig. 3A). Results h In particular, in cells pretreated with PEA, the amount of phagocytosed P. gingiv- alis was 60% higher than that in control. When blocking PPAR-α with GW6471 (10 µM), the phagocytic activity was normalized; similar results were obtained by blocking CB2R (AM630, 100 nM) (Fig. 3A). Additionally, we observed that PEA treatment resulted in a decreased number of surviving P. gingivalis after 4 hours of incubation compared to the control (~46%). The pretreatment with GW6471 or AM630 significantly reduced PEA-mediated effects compared to untreated cells (Fig. 3B). GW6471 alone (10 µM) did not change the phagocytosis or intracel- lular killing of P. gingivalis as compared with vehicle (Fig. 3A and B). PEA induces microglia migration through CB2R. To assess the possible capability of PEA to induce and/or potentate microglia motility/migration we used a time-lapse experiment in which microglia migrate towards a source of 2-arachidonoylglycerol (2-AG) [100 µM, one of the two major eCBs]. Minimum microglial migration was observed under physiological conditions. PEA incubation (100 nM for 24 hours) significantly increased the migratory activity as compared to vehicle-treated cells. Interestingly, the pre-treatment with AM630 (100 nM) or GW6471 (10 µM) blocked 2-AG-induced chemotaxis, confirming the involvement of CB2R and PPAR-α in this process. In a similar fashion to PEA, the synthetic PPAR-α agonist also improved microglial migration, suggesting that the cannabinoid-induced microglial chemotaxis through CB2R could be selectively mediated by PPAR-α (Fig. 4). Conversely, we observed that pre-treatment with the CB1R antagonist AM251 (500 nM for 24 hours) did not modify PEA-mediated migration (Fig. S4). Microglial cells produce sufficient amounts of 2-AG to tonically activate the PEA-induced upregulated CB2R in the presence of P. gingivalis. Next, we measured by isotope-dilution LC-MS the amounts of 2-AG produced by microglial cells stimulated with vehicle, P. gingivalis and P. gingivalis plus PEA Scientific Reports \| 7: 375 \| DOI:10.1038/s41598-017-00342-1 5 www.nature.com/scientificreports/ Figure 4. PEA induces microglia migration towards a 2-AG source. (A) Image-based detection of 2-AG (100 µM) chemoattractive effect on cultured microglial cells after different treatments. Panel shows the migratory effect induced by PEA (100 nM) or GW7647 (1 µM), alone or in the presence of GW6471 (10 µM) or AM630 (100 nM), as compared to the control group (vehicle). Representative data of microglial cell time-lapse migration recorded at starting point (0 min), 30 and 90 min from the 2-AG exposure. Results h (B) The quantification indicates the percentage of microglia cells affected by 2-AG chemoattractive movements vs total cell number at 90 min. Data are shown as means ± SEM (n = 6–8). P < 0.05 compared to vehicle and °P < 0.05 compared to PEA or GW7647. One way ANOVA, post-hoc Tukey’s. Scale bar = 50 µm. Figure 4. PEA induces microglia migration towards a 2-AG source. (A) Image-based detection of 2-AG (100 µM) chemoattractive effect on cultured microglial cells after different treatments. Panel shows the migratory effect induced by PEA (100 nM) or GW7647 (1 µM), alone or in the presence of GW6471 (10 µM) or AM630 (100 nM), as compared to the control group (vehicle). Representative data of microglial cell time-lapse migration recorded at starting point (0 min), 30 and 90 min from the 2-AG exposure. (B) The quantification indicates the percentage of microglia cells affected by 2-AG chemoattractive movements vs total cell number at 90 min. Data are shown as means ± SEM (n = 6–8). P < 0.05 compared to vehicle and °P < 0.05 compared to PEA or GW7647. One way ANOVA, post-hoc Tukey’s. Scale bar = 50 µm. Figure 4. PEA induces microglia migration towards a 2-AG source. (A) Image-based detection of 2-AG (100 µM) chemoattractive effect on cultured microglial cells after different treatments. Panel shows the migratory effect induced by PEA (100 nM) or GW7647 (1 µM), alone or in the presence of GW6471 (10 µM) or AM630 (100 nM), as compared to the control group (vehicle). Representative data of microglial cell time-lapse migration recorded at starting point (0 min), 30 and 90 min from the 2-AG exposure. (B) The quantification indicates the percentage of microglia cells affected by 2-AG chemoattractive movements vs total cell number at 90 min. Data are shown as means ± SEM (n = 6–8). P < 0.05 compared to vehicle and °P < 0.05 compared to PEA or GW7647. One way ANOVA, post-hoc Tukey’s. Scale bar = 50 µm. (100 nM). We found that levels of 2-AG were much higher than those of the other endocannabinoid, anandamide, and potentially sufficient to tonically activate CB2R (39.6 ± 8.0 vs. 1.2 ± 0.2 pmol/mg lipid extract, equivalent to 17.9 ± 3.4 vs. Results h 0.5 ± 0.06 nM in 1 ml of culture, means ± SD, n = 3; each ml of culture contained 5 × 105 cells), par- ticularly if these receptors are upregulated (Fig. S5). Following P. gingivalis or P. gingivalis plus PEA incubation, we found that 2-AG levels were not significantly increased (up to 53.6 ± 12.7 and 74.04 ± 12.7 pmol/mg of extracted lipids, respectively, means ± SD, n = 3), whereas anandamide levels were significantly increased only following co-incubation with P. gingivalis plus PEA (up to 1.0 ± 0.16 and 2.7 ± 0.2 pmol/mg of extracted lipids, respectively, means ± SD, n = 3). Importantly, PEA was also detected and abundant in microglia under our experimental con- ditions (20.1 ± 2.3 and 17.6 ± 5.6 pmol/mg of extracted lipids, respectively in vehicle and P. gingivalis-stimulated cells, means ± SD, n = 3, equivalent to 9.2 ± 1.3 and 6.1 ± 1.8 nM, respectively, per 1 ml of cultured cells) (Fig. S5). i i (100 nM). We found that levels of 2-AG were much higher than those of the other endocannabinoid, anandamide, and potentially sufficient to tonically activate CB2R (39.6 ± 8.0 vs. 1.2 ± 0.2 pmol/mg lipid extract, equivalent to 17.9 ± 3.4 vs. 0.5 ± 0.06 nM in 1 ml of culture, means ± SD, n = 3; each ml of culture contained 5 × 105 cells), par- ticularly if these receptors are upregulated (Fig. S5). Following P. gingivalis or P. gingivalis plus PEA incubation, we found that 2-AG levels were not significantly increased (up to 53.6 ± 12.7 and 74.04 ± 12.7 pmol/mg of extracted lipids, respectively, means ± SD, n = 3), whereas anandamide levels were significantly increased only following co-incubation with P. gingivalis plus PEA (up to 1.0 ± 0.16 and 2.7 ± 0.2 pmol/mg of extracted lipids, respectively, means ± SD, n = 3). Importantly, PEA was also detected and abundant in microglia under our experimental con- ditions (20.1 ± 2.3 and 17.6 ± 5.6 pmol/mg of extracted lipids, respectively in vehicle and P. gingivalis-stimulated cells, means ± SD, n = 3, equivalent to 9.2 ± 1.3 and 6.1 ± 1.8 nM, respectively, per 1 ml of cultured cells) (Fig. S5). i i www.nature.com/scientificreports/ To obtain conclusive evidence for this hypothesis, we performed ChiP assay by transfecting PPAR-α in HEK293 cells, a common cell model used to perform heterologous expression studies40, and evaluated the pos- sible induction of a physical interaction between PPAR-α and Cnr2 by PEA. We found that upon stimulation with PEA, PPAR-α binds with high affinity a specific region identified in the Cnr2 gene. Thus, our results clearly show a straight gene regulation operated by PEA, although this drug could modulate several other pharmacological targets. Interestingly, compared with microglia, a higher dose of PEA (300 nM) was necessary to induce appreci- able changes in HEK293 cells. This could be explained by assuming that primary microglia contain constitutively a more efficacious PPAR-α signaling system. fi g g y Unable to investigate resident microglia cells in humans, we decided to evaluated possible the PEA-induced CB2R modulation in blood-derived human macrophages, because of the similarity of the derived-cell lineage between microglia and macrophages. Indeed, similar to microglia, in human macrophage cells cultures we also observed that PEA and GW7647 increased CB2R expression. However, in these cells, the enhancement of protein expression was observed after a shorter incubation period (4 hours) than that observed in microglia (24 hours). This different time course of CB2R expression might be consequence of the different cell type used (human mac- rophages vs rat microglia) and, hence, of the possible different PPAR-α expression levels, as well as, localization/ expression of the CB2R protein. Interestingly, decreased IFNγ-induced CB2R expression was observed in the presence of the PPAR-α antagonist, suggesting the existence of a cross-talk between the PPAR-α and IFNγ path- ways in CB2R modulation in human macrophages.l y p g In agreement with its anti-inflammatory/neuroprotective activities, previous studies proved the capability of PEA to increase resistance against systemic infections in vivo41–44. Indeed, short-term PEA exposure (30 min) stimulates E. coli uptake in macrophages and microglia45–47. Our findings confirm the ability of PEA to increase microglia phagocytosis of the periodontal P. gingivalis pathogens. We have choosen P. gingivalis because these bacteria or their virulent factors are not only a source of chronic infections and inflammation, but can also reach the brain systemically over time and, by priming the resident immune cells (such as microglia), trigger neuroin- flammatory processes associated with early stages of dementia47. In this study, we showed that treatment with PEA increases phagocytosis and enhances intracellular bacteria killing by microglia. www.nature.com/scientificreports/ mechanism by which PEA interacts with CB2R is unknown. Here we show, for the first time, that PEA increases CB2R expression through a genomic mechanism involving the activation of the best established direct target for this lipid compound, i.e. PPAR-α. Because selective stimulation of CB2R is associated with analgesia and the release of anti-inflammatory cytokines30–32, our findings may explain the involvement of CB2R in the therapeutic effects of PEA, which are often observed in pathologies involving specific immune responses. In fact, the presence of CB2R in immune cells, including microglia, has been conclusively established6. Microglia closely resemble tissue macrophages and are key sensors of CNS injury caused by different diseases33–37. They are known to rapidly proliferate under activation and exert both pro- and anti-inflammatory effects38 in the inflammatory response. Milligan and Watkins39 suggested that the neuroprotective microglia (M2 phenotype) overexpress both CB1R and CB2R, which concurrently control the release of pro- and anti-inflammatory cytokines. However, the role of CBRs in microglia physiology has not been fully clarified. Here, we show that microglia increase CB2R expres- sion following PEA treatment under physiological conditions, thereby assuming a hypertrophic morphology. In fact, using IFNγ as a positive control23, we found that treatment with PEA significantly increased CB2R levels in microglia, in association with cytoskeletal rearrangements that are typical of morphological cell activation. Indeed, as shown by immunofluorescence, PEA-treated cells exhibited larger somata (diameter) and shorter pro- cesses than control cells. A stringent analysis of surface antigens may identify possible changes in the M1/M2 cell profile induced by PEA. pi y The involvement of PPAR-α in the effects of PEA was evaluated by assessing the ability of the synthetic PPAR-α-selective agonist GW7647 to up-regulate CB2R. We found that GW7647 significantly increases CB2R expression levels compared to those in non-treated cells, and pre-treatment with the PPAR-α antagonist GW6471 blocks this effect. The requirement for PPAR-α in PEA-induced CB2R up-regulation was shown by silencing PPAR-α mRNA. Indeed, in PPAR-α-silenced microglia (which exhibit a ~80% reduction in PPAR-α mRNA levels), incubation with PEA or GW7647 failed to increase CB2R expression. p Our results suggest that the mechanism underlying PEA-induced CB2R up-regulation involves the activation of PPAR-α. Discussion Th The present study adds another piece to the intricate mosaic of the proposed mechanisms for the pleiotropic effects of PEA. In particular, we show that this endogenous fatty acid ethanolamide affects cannabinoid signaling by upregulating CB2R expression in mononuclear phagocytic cells. A significant increase in CB2R expression was observed in cultured rat microglia and human macrophages following incubation with PEA. Pharmacological, biochemical and computational analyses revealed that CB2R upregulation occurs through a PPAR-α-mediated genomic mechanism. Furthermore, we provide evidence for the involvement of CB2R in PEA-induced microglial phagocytosis and migration in vitro.ff p g y g Previous evidence suggests a role for CB2R in PEA-mediated effects under different pathological conditions with particular regard to chronic pain states19, 20, 27. PEA, which does not directly activate CB1R or CB2R28, shows anti-inflammatory properties that can be blocked by the selective CB2 antagonist SR14452822. These findings have suggested that these PEA-mediated effects may result from direct stimulation of yet uncharacterized CB2-like receptors. Moreover, an ‘entourage hypothesis’, which states that the effects of PEA are due to enhancement of eCBs, and thus are a consequence of CBR or TRPV1 stimulation, has been also proposed29. However, the Scientific Reports \| 7: 375 \| DOI:10.1038/s41598-017-00342-1 6 www.nature.com/scientificreports/ p Cell cultures and transfection Cell cultures and transfection. -Microglial Cells. Primary microglia were prepared according to our pre- vious studies61, 62 (Supplemental Experimental Procedures). Briefly, cortices from SD postnatal rats were dis- sociated and suspended in Dulbecco’s modified Eagle medium (DMEM). The cell suspension was filtered and plated in tissue culture flasks precoated with poly-D-lysine. Cells were harvested as floating cell suspensions following shaking after 10–12 days. Cell viability was determined by measuring the lactate dehydrogenase (LDH assay kit, Sigma, UK) according to the manufacturers’ instructions. In all experiments no significant difference was observed between untreated and vehicle-treated cells (DMSO 0.01%). The experimental procedures were conducted in conformity with protocols approved by the Animal Ethics Committee of the Second University of Naples. Animal care was in compliance with the IASP and European Community (E.C. L358/1 18/12/86) guide- lines on the use and protection of animals in experimental research. - Human Embryonic Kidney-293 (HEK293) cells were grown in 60-mm plastic Petri dishes in Eagle’s Minimum Essential Medium (EMEM) containing 10% FBS, nonessential amino acids (0.1 mM), penicillin (50 U/ ml), and streptomycin (100 µg/ml) in a humidified atmosphere at 95% O2, 5% CO2 at 37 °C. After plating, the cells were transfected on the next day with plasmids encoding murine PPAR-α (Addgene, Plasmid #22751) and reti- noid X receptor (RXR Addgene, Plasmid #8882) cDNAs by use of Lipofectamine LTX (Life Tecnology) following manufacturer’s instructions. One day after transfection, PEA was added to the culture medium for 24 hours. t - Macrophages. Eight healthy volunteers were enrolled as whole blood donors for macrophages cultures after written informed consent with the approval of the Ethic Committee of the Second University of Naples (Supplemental Experimental Procedures). All procedures were carried out in accordance with the national legislation and the Declaration of Helsin Immunocytochemistry. Cells were fixed in 4% paraformaldehyde for 20 min, followed by ice-cold 100% methanol for 3 min. Fixed cells were incubated for 2 h with primary antibodies against Iba-1 (rabbit anti-ionized calcium binding adapter molecule 1; 1:1000 dilution; Wako Chemicals, Germany), or Vimentin (mouse anti-Vimentin; 1:500 dilution; Santa Cruz Biotechnology), CB2R (goat anti-CB2 1:200 dilution; Santa Cruz Biotechnology) or PPAR-α (mouse anti PPAR-α 1:400 diluition, Abcam). Following this incubation, cells were washed and incubated for 45 min with the secondary antibody solution (donkey anti-rabbit or donkey anti-goat or donkey anti-mouse IgG-conjugated Alexa Fluor TM 488 and 568; 1:1000 dilution; Molecular Probes). www.nature.com/scientificreports/ Moreover, we demonstrated that the effects of PEA were mediated by both PPAR-α and CB2R activation. Our findings suggest that PEA or CB2R agonists may prevent some forms of brain infection, and, consequently, inflammatory neurodegeneration, by promoting microglial phagocytosis. Conversely, a recent study by Redlich and colleagues (2012)46 showed that the ability of PEA to induce phagocytosis does not depend on PPAR-α stimulation. This discrepancy may be due to the different bacterial strains (Streptococcus pneumonia or E. coli vs P. gingivalis) and cell lines (mouse vs rat) used in the two studies. Additionally, prolonged PEA treatment might promote CB2R activation, as shown here and elsewhere, also by merely enhancing eCB levels48, and thus induce phagocytosis through a PPAR-α-independent mechanism. At any rate, the present findings provide for the first time a molecular mechanism for PEA capability to increase resistance against systemic infections, an effect that has been known for decades and has been already exploited in the clinic in the past when PEA was marketed as “impulsin”49.hl p p p The recruitment of phagocytic cells to sites of damage is a crucial step in inflammation and antimicrobial immune responses. Microglia cell migration can be triggered by a pool of chemoattractant factors acting at G-coupled receptors, which include chemokines, nucleotides, and bioactive lipids (i.e. eCBs)50–53. In this context, CB receptor modulators (synthetic compounds or phytocannabinoids) have been shown to regulate macrophage/ microglial cell activity and migration in several conditions54–57. Our present data reveal that the increased CB2R expression mediated by PEA manifests also an enhanced capability of microglia to migrate towards a high (μM) local concentration of extracellular 2-AG source. On the other hand, we found that, following stimulation with PEA and P. gingivalis, lower (nM) concentrations of 2-AG are present in microglia, which, alone or together with the other eCB, anandamide, might activate the PEA-upregulated CB2R and trigger phagocytosis, but would not Scientific Reports \| 7: 375 \| DOI:10.1038/s41598-017-00342-1 7 www.nature.com/scientificreports/ be sufficiently high to prevent microglia migration towards an external source of eCBs. www.nature.com/scientificreports/ Indeed, although it is always difficult to predict to what extent the global concentrations of very lipophilic compounds (such as the eCBs and PEA) that are detected in an experimental set-up reflect the local concentration at their targets (CBRs and PPAR-α, respectively) in a biological setting, we can surmise, from our present findings and previous data on its functional activity at CB2R58, 59, that only microglial 2-AG might tonically stimulate the CB2R upregulated by PEA, and that also PEA coming from neighboring cells is necessary to exert the upregulation of CB2R. In conclusion, the present study provides evidence for a new mechanism of action for PEA. Our data indicate that PEA may act indirectly on CB2R, not only by increasing endogenous ligand content29, 60, but also by reg- ulating receptor expression. Moreover, we demonstrated that PEA induces a microglial morphological change associated with enhanced phagocytosis and migratory activity in which CB2R plays a role. These findings suggest that PEA, by targeting CB2R, may be a useful tool for preventing the consequences of neuroinflammation in CNS disorders, and explain the clinical use in the late 1970’s of this compound against infections. p Cell cultures and transfection The genomic DNA was sonicated, immunoprecipitated with an anti-PPARα antibody certified for ChiP analysis (Abcam; ab2779) and subsequently released from histones by following the previously described procedure63. Immunoprecipitated DNA was then analyzed by qPCR (40–45 cycles) using specific primers able to amplify each of PPAR/RXR-rich regions identified (Forward: GTC AGG TAC CTC TCA GCT CC; Reverse CTG TAA GGG GAT AAT GCA CC the sequence 1; Forward: GGT GCA TTA TCC CCA TTT TAC AG; Reverse: GTG AGT GTG AGG AGG TCT GG for the sequence 2). In addition, for each primer pair, non-immunoprecipitated DNA was used as input for normalization. All ChIP assays were performed in triplicate for at least three different biological preparations (Fig. S3). Phagocytosis assay and intracellular survival assay. Pretreated microglia with PEA (100 nM), and/ or GW6471 (10 µM), and/or AM630 (100 nM) for 24 hours were infected with P. gingivalis cells. Ingested bacteria were serially diluted and spread on plates for viable counts. To monitor survival, microglial cells were incu- bated with P. gingivalis for 90 min. Thereafter, cells were washed and the extracellular bacteria destroyed. The amounts of intracellular bacteria were determined as CFU by quantitative plating of serial dilutions on TS agar (Supplemental Experimental Procedures). Migration assay. The migration assay was performed in PEA or PPAR-α agonist-treated cells, in presence of the PPAR-α selective antagonist GW6471 (10 µM) or the CB2R selective antagonist AM630 100 nM), or the CB1 selective antagonist AM251 (500 nM). The movement of microglia was recorded each 5 min for 90 min starting from the application of 2-AG as the chemo-attractant ligand (Supplemental Experimental Procedures). Quantification of eCBs and related mediators in microglial cells. Cell samples were dounce-homogenized and extracted in 5 vol of chloroform/methanol/Tris–HCl 50 mM (2:1:1) containing 5 pmol of [2H]8 anandamide and 50 pmol of [2H]4 palmitoylethanolamide (PEA) and [2H]5 2-AG (Cayman Chemicals, MI, USA). The lipid-containing organic phase was dried down in a rotating evaporator, weighed, and pre-purified by open-bed chromatography on silica gel. Lyophilized extracts were resuspended in chloroform/methanol 99:1 by vol. The solutions were then purified by open bed chromatography on silica as described64. Fractions eluted with chloroform/methanol 9:1 by vol. p Cell cultures and transfection Scientific Reports \| 7: 375 \| DOI:10.1038/s41598-017-00342-1 8 www.nature.com/scientificreports/ A semi-quantitative analysis of mRNA levels was carried out by the Gel Doc EZ UV System (Bio-Rad, Hercules, CA). Protein Extraction and Western Blotting Analysis. For protein extraction, cells or tissues were col- lected in RIPA buffer (Thermo Fisher Scientific Inc., Rockford, IL, USA), and proteins were extracted according to the manufacturer’s protocol. Protein concentrations were determined using the Bradford reagent (Sigma-Aldrich, St Louis, MO, USA). Samples were run on a 12% or 15% sodium dodecyl sulfate–polyacrylamide gel and trans- ferred to nitrocellulose membranes. Membranes strips were blocked in 5% milk, 13 Tris-buffered saline, and 0.05% Tween-20 and incubated, firstly overnight with primary antibodies to detect CB2R (1:200; Abcam, Cambridge, UK) or PPARα (PPAR-α 1:400 diluition, Abcam) and then with the relative secondary antibody. A blocking peptide for the anti-CB2R antibody (1:100 diluition, Abcam) in control microglia cells or spleen was used to validate the specificity of the antibody. An anti-β-tubulin monoclonal antibody (1:5000 dilution; Santa Cruz Biotechnology Santa Cruz, CA, USA) was used to check for identical protein loading. Reactive bands were detected by chemiluminescence (SuperSignal, West Femto, Pierce, USA) and visualized on a X-ray film (Fuji Corporation, Tokyo, Japan) or on a C-DiGit® Blot Scanner (LI-COR Biosciences, Nebraska, USA) or on a ChemiDoc station (Bio-Rad, Hercules, CA, USA). Images were captured, stored, and analyzed using “Image studio Digits ver. 5.0” software. Gene silencing. PPAR-α mRNA was silenced by using a human PPAR-α siRNA reagent (PPAR-a silencer select Pre-designed siRNA, Ambion, USA). Cell transfection was performed according to manufacturer’s proto- col. Briefly, microglial cells were obtained as described above and plated on 24-multiwell plates with D-MEM and 10% FBS at density 500.000 cells/well. Cells were then transfected by using 1 μM non-targeting or experimental siRNA combined in 1%-supplemented FBS medium. Cells were incubated at 37 °C in a CO2 incubator for 3 days, after which they were collected for bio-molecular examination or treated with PEA (100 nM), GW7647 (1 μM), and IFNγ (300 U/mL) for 24 hours in order to evaluate CB2R protein expression. Transfection minus controls were carried out using non-targeting siRNA in order to check the efficiency of transfection. Bioinformatic analysis and chromatin immunoprecipitation assays. Putative PPAR/RXR con- sensus sequence identification was performed by using Genomatix software. The ChIP assay was performed by the use of ChiP Kit (Sigma-Aldrich MI, IT; cat. n. CHP1) following the manufacturer’s instructions. p Cell cultures and transfection Coverslips were mounted using Vectashield mounting media (Vector Laboratories, Burlingame, CA). Activated cells were detected by measuring the cell soma and processes length. Resting and activated microglia were clas- sified based on the following criteria: resting microglia displayed small somata bearing long, thin, ramified pro- cesses, whereas activated microglia exhibited marked cellular hypertrophy and retraction of processes such that the process length was lower than the diameter of the soma compartment. Data were expressed as percentage of activated cells. All experiments were performed in triplicate. Real Time Polymerase Chain Reaction. To quantify the expression levels of CB2R mRNA, three serial complementary DNA (cDNA) dilutions (1:5) obtained from reverse transcription of 100 ng total mRNA (High Capacity cDNA Reverse Transcription kit; Applied Biosystems, Foster City, CA, USA) were amplified by Real Time polymerase chain reaction (Real Time PCR or qPCR) with SYBR green as fluorophore. The housekeeping gene β-actin was used as endogenous control. Real Time-PCR products were analyzed using the SDS software (Applied Biosystems, Foster City, CA, USA) (Supplemental Experimental Procedures). RNA extraction and RT-PCR. Total RNA was extracted from homogenized cells using a RNA Tri-Reagent (Molecular Research Center Inc., Cincinnati, OH) according to the manufacturer’s protocol. The extracted RNA was subjected to DNase I treatment at 37 °C for 30 min. The total RNA concentration was determined by UV spectrophotometry. The mRNA levels of the genes under analysis were measured by RT-PCR amplification. RT minus controls were carried out to check potential genomic DNA contamination. These RT minus controls were performed without using the reverse transcriptase enzyme in the reaction mix. Sequences for the mRNAs from GeneBank (DNASTAR INC., Madison, WI) were used to design primer pairs for RT-PCRs (OLIGO 4.05 software, National Biosciences Inc., Plymouth, MN). Each RT-PCR was repeated at least three times to achieve the best possible reproducibility data. The measured mRNA levels were normalized with respect to hypoxanthine-guanine phosphoribosyltransferase (HPRT), chosen as housekeeping gene. The HPRT gene expression did not change in several experimental conditions. To our knowledge there is no molecular evidence for variation in HPRT mRNA-levels in this model. The gene expression values were expressed as arbitrary units ± SE. Amplification of genes of interest and HPRT were performed simultaneously. PCR products were resolved into 2.0% agarose gel. p Cell cultures and transfection (containing 2-AG, anandamide and PEA) were collected and the excess solvent evaporated with a rotating evaporator, and aliquots analyzed by isotope dilution-liquid chromatogra- phy/atmospheric pressure chemical ionisation/mass spectrometry (LC-APCI–MS) carried out under conditions described previously65 and allowing the separations of 2-AG, anandamide and PEA. MS detection was carried out in the selected ion monitoring mode using m/z values of 356 and 348 (molecular ion +1 for deuterated and undeuterated anandamide), 384.35 and 379.35 (molecular ion +1 for deuterated and undeuterated 2-AG), 304 and 300 (molecular ion +1 for deuterated and undeuterated PEA). The amounts of all mediators were expressed as pmol/mg of cells. 1. Kettenmann, H., Hanisch, U. K., Noda, M. & Verkhratsky, A. Physiology of microglia. Physiol. Rev. 91, 461 553 (2011). 2. McMahon, S. B. & Malcangio, M. Current challenges in glia-pain biology. Neuron 64(1), 46–54 (2009). 3. Franco, R. & Fernàndez-Suàrezs, D. Alternatively activated microglia and macrophages in the central nervous system. Prog. Neurobiol. 131, 65–86 (2015). g g g p gy ( ) ( ) 3. Franco, R. & Fernàndez-Suàrezs, D. Alternatively activated microglia and macrophages in the central nervous system. Prog Neurobiol. 131, 65–86 (2015). 1. Kettenmann, H., Hanisch, U.-K., Noda, M. & Verkhratsky, A. Physiology of microglia. Physiol. Rev. 91, 461–553 (2011). www.nature.com/scientificreports/ www.nature.com/scientificreports/ 4. Sierra, A., Tremblay, M. E. & Wake, H. Never-resting microglia: physiological roles in the healthy brain and pathologica implications. Front. Cell. Neurosci 8, 240 (2014). 5. von Bernhardi, R., von Bernhardi Eugenín, L. & Eugenín. Microglial cell dysregulation in brain aging and neurodegeneration. Fron Aging Neurosci 7, 124 (2015). g g ( ) 6. Stella, N. Cannabinoid and cannabinoid-like receptors in microglia, astrocytes, and astrocytomas. Glia 58(9), 1017–1030 (2010). g ( ) lla, N. Cannabinoid and cannabinoid-like receptors in microglia, a 7. Luongo, L. et al. 1-(2′,4′-dichlorophenyl)−6-methyl-N-cyclohexylamine-1,4-dihydroindeno[1,2-c]pyrazole-3-carboxamide, a nove CB2 agonist, alleviates neuropathic pain through functional microglial changes in mice. Neurobiol. Dis. 37, 177–185 (2010). g p p g g g 8. Romero-Sandoval, E. A., Horvath, R., Landry, R. P. & DeLeo, J. A. Cannabinoid receptor type 2 activation induces a microglial 8. Romero-Sandoval, E. A., Horvath, R., Landry, R. P. & DeLeo, J. A. Cannabinoid receptor type 2 activation induces a microglial anti- inflammatory phenotype and reduces migration via MKP induction and ERK dephosphorylation. Mol Pain. 28, 5–25 (2009).l R , , , R , y, R , J p yp g inflammatory phenotype and reduces migration via MKP induction and ERK dephosphorylation. Mol Pain. 28, 5–25 (2009).l t al. Stimulation of cannabinoid receptor 2 (CB2) suppresses micro l Ehrhart, J. et al. Stimulation of cannabinoid receptor 2 (CB2) suppre l 9. Ehrhart, J. et al. Stimulation of cannabinoid receptor 2 (CB2) suppresses microglial activation. Neuroinflammation 12(2), 29 (2005). 10. Facci, L. et al. Mast cells express a peripheral cannabinoid receptor with differential sensitivity to anandamide and palmitoylethanolamide. Proc. Natl. Acad. Sci. USA 92, 3376–3380 (1995). , J p ( ) pp g fl ( ), ( ) 0. Facci, L. et al. Mast cells express a peripheral cannabinoid receptor with differential sensitivity to anandamide and palmitoylethanolamide. Proc. Natl. Acad. Sci. USA 92, 3376–3380 (1995). p y 11. Scarampella, F., Abramo, F. & Noli, C. Clinical and histological evaluation of an analogue of palmitoylethanolamide, PLR 120 (comicronized Palmidrol INN) in cats with eosinophilic granuloma and eosinophilic plaque: a pilot study. Vet. Dermatol. 12(1), 29–39 (2001). 12. Skaper, S. D., Facci, L. & Giusti, P. Glia and mast cells as targets for palmitoylethanolamide, an anti-inflammatory neuroprotective lipid mediator. Mol. Neurobiol. 48(2), 340–452 (2013). 13. Mattace Raso, G., Russo, R., Calignano, A. & Meli, R. Palmitoylethanolamide in CNS health and disease. Pharmacol. Res. 86, 32–41 (2014). 4. Koch, M. www.nature.com/scientificreports/ et al. Palmitoylethanolamide protects dentate gyrus granule cells via peroxisome proliferator-activated receptor-α Neurotox Res. 19(2), 330–40 (2011).h 15. Costa, B., Comelli, F., Bettoni, I., Colleoni, M. & Giagnoni, G. The endogenous fatty acid amide, palmitoylethanolamide, has anti- allodynic and anti-hyperalgesic effects in a murine model of neuropathic pain: involvement of CB(1), TRPV1 and PPARgamma receptors and neurotrophic factors. Pain 139, 541–550 (2008). p p ( ) 16. Di Cesare Mannelli, L. et al. Palmitoylethanolamide is a disease-modifying agent in peripheral neuropathy: pain relief neuroprotection share a PPAR-alpha-mediated mechanism. Mediators Inflamm. 2013, 328797 (2013). 17 L L t l P l it l th l id d f li i d d thi lik b h i th h i l li l/ i fl 17. Luongo, L. et al. Palmitoylethanolamide reduces formalin-induced neuropathic-like behaviour through spinal glial/microglial phenotypical changes in mice. CNS Neurol. Disord. Drug Targets 12, 45–54 (2013). l 17. Luongo, L. et al. Palmitoylethanolamide reduces formalin-induced neuropathic-like behavi phenotypical changes in mice. CNS Neurol. Disord. Drug Targets 12, 45–54 (2013). 17. Luongo, L. et al. Palmitoylethanolamide reduces formalin-induced neuropathic-like behaviour through spinal glial/m phenotypical changes in mice. CNS Neurol. Disord. Drug Targets 12, 45–54 (2013). 18. Mackie, K. & Stella, N. Cannabinoid receptors and endocannabinoids: evidence for new players. AAPS J. 8, 298–306 (2006). l ff d ll l f b d b d 19. Calignano, A., La Rana, G., Giuffrida, A. & Piomelli, D. Control of pain initiation by endogenous cannabinoids. Nat 394, 277 (1998). 20. Calignano, A., La Rana, G., Loubet-Lescoulié, P. & Piomelli, D. A role for the endogenous cannabinoid system in the peripheral control of pain initiation. Prog. Brain. Res. 129, 471–482 (2000).h p g 1. LoVerme, J., La Rana, G., Russo, R., Calignano, A. & Piomelli, D. The search for the palmitoylethanolamide receptor. Life. Sci. 77 1685–1698 (2005). 2. Iannotti, F. A., Di Marzo, V. & Petrosino, S. Endocannabinoids and endocannabinoid-related mediators: Targets, metabolism and role in neurological disorders. Prog. Lipid. Res. 62, 107–28 (2016). g g p 3. Cabral, G. A. & Marciano-Cabral, F. Cannabinoid receptors in microglia of the central nervous system: immune functiona relevance. J. Leukoc. Biol. 78(6), 1192–1197 (2005).t 4. Hains, B. & Waxman, S. Activated microglia contribute to the maintenance of chronic pain after spinal cord injury. J. Neurol. Sci. 26 4308–4317 (2006).f 25. Chandra, V. et al. References References 1. Kettenmann, H., Hanisch, U.-K., Noda, M. & Verkhratsky, A. Physiology of microglia. Physiol. Rev. 91, 461–553 (2011). 2. McMahon, S. B. & Malcangio, M. Current challenges in glia-pain biology. Neuron 64(1), 46–54 (2009). 3. Franco, R. & Fernàndez-Suàrezs, D. Alternatively activated microglia and macrophages in the central nervous system. Prog. Neurobiol. 131, 65–86 (2015). Scientific Reports \| 7: 375 \| DOI:10.1038/s41598-017-00342-1 9 www.nature.com/scientificreports/ Structure of the intact PPAR-gamma-RXR- nuclear receptor complex on DNA Species differences in cannabinoid receptor 2 (CNR2 gene): identification of novel human and rodent. Nature 456(7220), 350–6 (2008). handra, V. et al. Structure of the intact PPAR-gamma-RXR- nuclear receptor complex on DNA Species differences in cannabinoid 2 (CNR2 ) id ifi i f l h d d N 6( 220) 3 0 6 (2008) handra, V. et al. Structure of the intact PPAR-gamma-RXR- nuclear receptor complex on DNA Species differences in cannabinoid ceptor 2 (CNR2 gene): identification of novel human and rodent. Nature 456(7220), 350–6 (2008). i 26. Liu, Q. R. et al. Species differences in cannabinoid receptor 2 (CNR2 gene): identification of novel human and rodent isoforms,differential tissue expression and regulation by cannabinoid receptor ligands. Genes Brain Behav 8(5), 519–30 (2009) f 27. Farquhar-Smith, W.-P. & Rice, A. S. A novel neuroimmune mechanism in cannabinoid-mediated attenuation of nerve gr factor-induced hyperalgesia. Anesthesiology 99, 1391–1401 (2003). yp g gy ( ) 28. Showalter, V., Compton, D. R., Martin, B. R. & Abood, M. E. Evaluation of binding in a transfected cell line expressing a peripheral cannabinoid receptor (CB2): identification of cannabinoid receptor subtype selective ligands. J. Pharmacol. Exp. Ther. 278(3), 989–999 (1996). 29. Re, G., Barbero, R., Miolo, A. & Di Marzo, V. Palmitoylethanolamide, endocannabinoids and related cannabimimetic compo in protection against tissue inflammation and pain: potential use in companion animals. Vet. J. 173(1), 21–30 (2007). 30 R I t l C i l l f CB(2) bi id t i th l ti f t l i d i thi 0. Racz, I. et al. Crucial role of CB(2) cannabinoid receptor in the regulation of central immune responses during neuropathic pain. J Neurosci. 28(46), 12125–12135-a (2008). 31. Pasquini, S. et al. Design, synthesis, and pharmacological characterization of indol-3-ylacetamides, indol-3-yloxoacetamides, and indol-3-ylcarboxamides: potent and selective CB2 cannabinoid receptor inverse agonists. J. Med. Chem. 55(11), 5391–402 (2012).i 31. Pasquini, S. et al. Design, synthesis, and pharmacological characterization of indol-3-ylacetamides, indol-3-yloxoacetamides, and i d l 3 l b id t t d l ti CB2 bi id t i i t J M d Ch 55(11) 5391 402 (2012) dol-3-ylcarboxamides: potent and selective CB2 cannabinoid receptor inverse agonists. J. Med. Chem. 55(11), 5391–402 (2012). Mugnaini, C. et al. Investigations on the 4-quinolone-3-carboxylic acid motif part 5: modulation of the physicochemical profile of a y p p g J ( ), ( 32. www.nature.com/scientificreports/ Mugnaini, C. et al. Investigations on the 4-quinolone-3-carboxylic acid motif part 5: modulation of the physicochemical profil y p p g ( ) ( ) 32. Mugnaini, C. et al. Investigations on the 4-quinolone-3-carboxylic acid motif part 5: modulation of the physicochemical profile of a set of potent and selective cannabinoid-2 receptor ligands through a bioisosteric approach. Chem. Med. Chem. 7(5), 920–934 (2012). 32. Mugnaini, C. et al. Investigations on the 4 quinolone 3 carboxylic acid motif part 5: modulation of the physicochemical profile of a set of potent and selective cannabinoid-2 receptor ligands through a bioisosteric approach. Chem. Med. Chem. 7(5), 920–934 (2012). i set of potent and selective cannabinoid-2 receptor ligands through a bioisosteric approach. Chem. Med. Chem. 7(5), 920–934 (2012). 33 Luongo L et al 5′ Chloro 5′ deoxy (±) ENBA a potent and selective adenosine A(1) receptor agonist alleviates neuropathic pain i set of potent and selective cannabinoid-2 receptor ligands through a bioisosteric approach. Chem. Med. Chem. 7(5), 920–934 (2012). 33. Luongo, L. et al. 5′-Chloro-5′-deoxy-(±)-ENBA, a potent and selective adenosine A(1) receptor agonist, alleviates neuropathic pain in mice through functional glial and microglial changes without affecting motor or cardiovascular functions. Molecules 17, 13712–26 (2012). 4. Guida, F. et al. Salvinorin A reduces mechanical allodynia and spinal neuronal hyperexcitability induced by peripheral formalin injection. Mol. Pain. 8, 60 (2012). j 35. Guida, F. et al. Palmitoylethanolamide reduces pain-related behaviors and restores glutamatergic synapses homeostasis in the medial prefrontal cortex of neuropathic mice. Mol. Brain 8, 47-a (2015).f 6. Rinaldi, B. et al. Effect of Prolonged Moderate Exercise on the Changes of Nonneuronal Cells in Early Myocardial Infarction. Neura Plast. 2015, 265967 (2015).t 7. Spaziano, G. et al. Exposure to Allergen Causes Changes in NTS Neural Activities after Intratracheal Capsaicin Application, in Endocannabinoid Levels and in the Glia Morphology of NTS. Biomed. Res. 980983 (2015). 38. Romero-Sandovalm, E. A., Horvathm, R. J. & De Leom, J. A. Neuroimmune interactions and pain: focus on glial-modulating targets. Curr. Opin. Investig. Drugs. 9(7), 726–734 (2008). 39 Milli E & W ki L P h l i l d i l f li i h i i N R N i 10(1) 23 36 (2009) g p g g 39. Milliganm, E. & Watkins, L. Pathological and protective roles of glia in chronic pain. Nat. Rev. Neurosci. 10(1), 23–36 (2009). www.nature.com/scientificreports/ 4. Redlich, S., Ribes, S., Schutze, S. & Nau, R. Palmitoylethanolamide stimulates phagocytosis of Escherichia coli K1 by macrophage and increases the resistance of mice against infections. J. Neuroinflammation 11, 108 (2014).fi l 45. Masek, K., Perlík, F., Klíma, J. & Kahlich, R. Prophylactic efficacy of N-2-hydroxyethyl palmitamide (impulsin) in acute respir tract infections. Eur. J. Clin. Pharmacol. 7, 415–9 (1974). 46. Redlich, S., Ribes, S., Scutze, S., Czesnik, D. & Nau, R. Palmitoylethanolamide stimulates phagocytosis of Escherichia coli K1 and Streptococcus pneumoniae R6 by microglial cells. J. Neuroimmunol. 244(1–2), 32–4 (2012). p p y g 7. Singhrao, S. K., Harding, A., Poole, S., Kesavalu, L. & Crean, S. Porphyromonas gingivalis Periodontal Infection and Its Putative Links with Alzheimer’s Disease. Mediators Infiamm. 137357 (2015).hifi fi 48. Petrosino, S. & Di Marzo, V. The pharmacology of palmitoylethanolamide and first data on the therapeutic efficacy of some of its new formulations. Br J Pharmacol. 13580 (2016).fi 9. Kahlich, R. et al. Studies on prophylactic efficacy of N-2-hydroxyethyl palmitamide (Impulsin) in acute respiratory infections Serologically controlled field trials. J. Hyg. Epidemiol. Microbiol. Immunol. 23, 11–24 (1979). Serologically controlled field trials. J. Hyg. Epidemiol. Microbiol. Immunol. 23, 11–24 (1979) g yi J yg p , ( 50. Inoue, K. Microglial activation by purines and pyrimidines. Glia 40, 156–163 (2002). , K g y p py , ( ) 51. Miller, A. M. & Stella, N. CB2 receptor-mediated migration of immune cells: it can go either way. BR. J. Pharmacol 153(2), 299–308 (2008).h 51. Miller, A. M. & Stella, N. CB2 receptor-mediated migration of immune cells: it can go either way. BR. J. Ph (2008).h ( ) 52. Thelen, M. Dancing to the tune of chemokines. Nat. Immunol. 2(2), 129–134 (2001). h g ( ) ( ) 53. Fung, S., Cherry, A. E., Xu, C. & Stella, N. Alkylindole-sensitive receptors modulate microglial cell migration and proliferation. Glia 63, 1797–1808 (2015).t 4. Franklin, A., Parmentier-Batteur, S., Walter, L., Greenberg, D. A. & Stella, N. Palmitoylethanolamide increases after focal cerebra ischemia and potentiates microglial cell motility. J Neurosci. 23(21), 7767–75 (2003). 55. Walter, L. et al. Nonpsychotropic cannabinoid receptors regulate microglial cell migration. J. Neurosci. 23(4), 1398–405 (2003). Walter, L. et al. Nonpsychotropic cannabinoid receptors regulate m , p y p p g g g J ( ), ( 56. Sacerdote, P. et al. Author Contributions Conceived and designed the experiments: F.G., L.L., V.D., S.M., Performed the experiments: S.B., M.E.G., R.R., I.M., R.I., F.A.I., E.D.A., F.P., F.R., A.F., Analyzed the data. Contributed materials/analysis tools: G.B., A.R., L.C., V.d.N., Wrote the paper: F.G., S.M., V.D., All authors read and approved the final manuscript. g We would to thank Shimos Alameen and Carmela Belardo for technical support. g We would to thank Shimos Alameen and Carmela Belardo for technical support. www.nature.com/scientificreports/ The nonpsychoactive component of marijuana cannabidiol modulates chemotaxis and IL-10 and production of murine macrophages both in vivo and in vitro. J. Neuroimmunol. 159(1–2), 97–105 (2005). 57. Gonsiorek, W. et al. Endocannabinoid 2-arachidonyl glycerol is a full agonist through human type 2 cannabinoid rece antagonism by anandamide. Mol Pharmacol. 57(5), 1045–50 (2000). 58. Sugiura, T. et al. Evidence that 2-arachidonoylglycerol but not N-palmitoylethanolamine or anandamide is the physiological ligand for the cannabinoid CB2 receptor. Comparison of the agonistic activities of various cannabinoid receptor ligands in HL-60 cells. J Biol Chem 275(1), 605–12 (2000). 59. Martin-Moreno, A. M. et al. Cannabidiol and other cannabinoids reduce microglial activation in vitro and in vivo: relevance to Alzheimer’s disease. Mol. Pharmacol. 79, 964–973 (2011). 0. Di Marzo, V. et al. Palmitoylethanolamide inhibits the expression of fatty acid amide hydrolase and enhances the anti-proliferative effect of anandamide in human breast cancer cells. Biochem. J. 358, 249–255 (2001). f 61. Clark, A., Wodarski, R., Guida, F., Sasso, O. & Malcangio, M. Cathepsin S release from primary cultured microglia is regulat the P2X7 receptor. Glia 58(14), 1710–1726 (2010).h p 62. Luongo, L. et al. The A1 adenosine receptor as a new player in microglia physiology. Glia 62, 122–132 (2014).if 62. Luongo, L. et al. The A1 adenosine receptor as a new player in m h 63. Iannotti, F. A., Barrese, V., Formisano, L., Miceli, F. & Taglialatela, M. Specification of skeletal muscle differentiation by repr element-1 silencing transcription factor (REST)-regulated Kv7.4 potassium channels. Mol. Biol. Cell. 24, 274–84 (2013). 64. Matias, I., Gonthier, M. P., Petrosino, S., Docimo, L., Capasso, R., Hoareau, L., Monteleone, P., Roche, R., Izzo, A. A. & Di Marzo, V. Role and regulation of acylethanolamides in energy balance: focus on adipocytes and beta-cells. Br J Pharmacol 152(5), 676–90 (2007). 65. Piscitelli, F., Carta, G., Bisogno, T., Murru, E., Cordeddu, L., Berge, K., Tandy, S., Cohn, J. S., Griinari, M., Banni, S. & Di Marzo, V. Effect of dietary krill oil supplementation on the endocannabinoidome of metabolically relevant tissues from high-fat-fed mice. Nutr Metab 8(1), 51 (2011). www.nature.com/scientificreports/ 40 Li Y C l G d i f h h b i kid 293 li i ll bi l i l i 40. Lin, Y. C. et al. Genome dynamics of the human embryonic kidney 293 lineage in response to cell biology manipulations. Nat. Commun. 5, 4767 (2014).f 0. Lin, Y. C. et al. Genome dynamics of the human embryonic kidney 293 lineage in response to cell biology manipulations. Nat Commun. 5, 4767 (2014). 1 L i F l S d f h l i f h d bi id i i d l f l i l l i l h i ff 41. Loria, F. et al. Study of the regulation of the endocannabinoid system in a virus model of multiple sclerosis reveals a therapeutic effect of palmitoylethanolamide. Eur. J. Neurosci 28, 633–41 (2008). 1. Loria, F. et al. Study of the regulation of the endocannabinoid system in a virus model of multiple sclerosis reveals a therapeutic effec of palmitoylethanolamide. Eur. J. Neurosci 28, 633–41 (2008). p y 2. Ribes, S. et al. Toll-like receptor stimulation enhances phagocytosis and intracellular killing of nonencapsulated and encapsulated Streptococcus pneumoniae by murine microglia. Infect. Immun. 78(2), 865–71 (2010).ffi 43. Nau, R., Ribes, S., Djukic, M. & Eiffert, H. Strategies to increase the activity of microglia as efficient protectors of the brain against infections. Front. Cell. Neurosci 8, 138 (2014). Scientific Reports \| 7: 375 \| DOI:10.1038/s41598-017-00342-1 10 www.nature.com/scientificreports/ Scientific Reports \| 7: 375 \| DOI:10.1038/s41598-017-00342-1 Acknowledgements g We would to thank Shimos Alameen and Carmela Belardo for technical support. Additional Information Supplementary information accompanies this paper at doi:10.1038/s41598-017-00342-1 Supplementary information accompanies this paper at doi:10.1038/s41598-017-00342-1 Competing Interests: The authors declare that they have no competing interests. Competing Interests: The authors declare that they have no competing interests. ublisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and nstitutional affiliations. This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/ © The Author(s) 2017 Scientific Reports \| 7: 375 \| DOI:10.1038/s41598-017-00342-1 11
https://openalex.org/W4388611291	https://www.qeios.com/read/45UUKH/pdf	English	null	Review of: "Prevalence of Buruli Ulcer Among Residents in Jasikan Municipality: A Cross-Sectional Study"	null	2,023	cc-by	251	Qeios, CC-BY 4.0 · Review, November 12, 2023 Review of: "Prevalence of Buruli Ulcer Among Residents in Jasikan Municipality: A Cross-Sectional Study" Cheryl M E McCrindle1 1 University of Pretoria Potential competing interests: No potential competing interests to declare. https://doi.org/10.32388/UXQ4X5 Review of: Prevalence of Buruli Ulcer Among Residents in Jasikan Municipality: A Cross- Sectional Study The reviewer(s) rated it 1/5 Declarations Methods: The study employed a descriptive cross-sectional study design. A simple random sampling method was used to select respondents. A sample of 56 cases of Buruli ulcer was identified from the population.. COMMENT No inclusion or exclusion criteria. No diagnostic criteria included. It was NOT a Descriptive Cross-Sectional Study. It was NOT Qeios ID: 45UUKH · https://doi.org/10.32388/45UUKH Qeios ID: 45UUKH · https://doi.org/10.32388/45UUKH 1/2 Qeios, CC-BY 4.0 · Review, November 12, 2023 Qeios ID: 45UUKH · https://doi.org/10.32388/45UUKH randomized randomized Results: The prevalence of the disease was determined to be 8.0 per 10,000 population. COMMENT: Total population not indicated. If sample size is 56, pop size is presumably 70 000, which is far less than actual population. How was the sample randomized? it is a PURPOSIVE selection of diagnosed cases . No details on where it was diagnosed , diagnostic criteria. Conclusions: Buruli ulcer was found to be high among the population. COMMENT : High?????? 8 per 10 000????. We have HIV in RSA with a prevalence of 20% ( 20 out of every 100 adults, 200 in 1000, 2000 in 10 000. That is HIGH) Qeios ID: 45UUKH · https://doi.org/10.32388/45UUKH 2/2
https://openalex.org/W2102420764	https://www.ocl-journal.org/articles/ocl/pdf/2011/06/ocl2011186p307.pdf	English	null	The role of omega-3 fatty acids in child development	Oilseeds and fats, crops and lipids/OCL. Oilseeds & fats crops and lipids	2,011	cc-by	5,485	Saskia JM OSENDARP Saskia JM OSENDARP Saskia JM OSENDARP Unilever R&D Vlaardingen, The Netherlands & Division of Human Nutrition, Wageningen University, The Netherlands <saskia.osendarp@unilever.com> Abstract: Omega-3 long chain polyunsaturated fatty acids (n-3 LCPUFA) are important constituents of the maturing brain and therefore considered crucial for brain development in utero and in early infancy. However, it is uncertain whether n-3 LCPUFA supplementation during pregnancy and lactation can have beneﬁcial, sustainable effects on visual or cognitive development. Beneﬁcial effects on child cognitive function after supplementation with EPA and DHA during pregnancy and lactation were observed at 4 years of age, but not at 3, 6 months or 7 years. In term infants LCPUFA when given in relative high dosages, seems to improve visual acuity, but not cognitive function. Evidence for an effect of LCPUFA supplementation of preterm infants remains inconclusive. In children older than 2 years of age, epidemiological evidence suggests an association between psychiatric or neurodevelopmental disorders and omega-3 fatty acid deﬁciencies. However, the evidence from randomized controlled trials exploring the impact of omega-3 fatty acids on cognitive performance or brain function in school-aged children is not conclusive. In conclusion, n-3 LCPUFA are highly present in the maturing brain and are important for normal brain functioning and development. When provided in relative high dosages, n-3 LCPUFA may improve visual acuity in term infants. However, it remains unclear whether supplementation with n-3 LCPUFA during pregnancy, early infancy, and childhood can improve cognitive function. Unilever R&D Vlaardingen, The Netherlands & Division of Human Nutrition, Wageningen University, The Netherlands <saskia.osendarp@unilever.com> Key words: child development, omega-3 fatty acids, neurodevelopmental disorders psychiatric disorders developmental outcomes of the off- spring (> 2 years) were reported from both observational studies and rando- mized controlled trials (RCTs) (ﬁgure 1). brane ﬂuidity, membrane activity-bound enzymes, number and afﬁnity of recep- tors, function of neuronal membrane ionic channels, and production of neuro- transmitters and brain peptides (Yehuda, 2003). Although DHA and AA are the major structural components of the central nervous system, there is currently no consensus whether dietary supple- mentation of LCPUFA has beneﬁts for visual and cognitive development of infants. There isconsiderableinterest inthe roleof certain long chain polyunsaturated fatty acids (LCPUFA), in visual and cognitive developmentthroughout childhood.The n-3 fatty acid docosahexaenoic acid (DHA) and the n-6 fatty acid arachidonic acid (AA) are the major LCPUFA in the brain (Martinez, 1992). Saskia JM OSENDARP DHA and AA are rapidly incorporated in the nervous tissue of retina and brain during the brain’s growth spurt, which mainly takes place from the last trimester of pregnancy up to 2yearsofage(DobbingandSands, 1973; Clandinin et al., 1980; Martinez, 1992). Beyond development of the central nervous system, n-3 and n-6 fatty acids may inﬂuence brain function throughout life by modiﬁcations of neuronal mem- Three RCTs from high-income countries supplemented withn-3 fatty acids during pregnancy. Two of these trials reported improvements in problem solving at 9 months of age (Judge et al., 2007) or hand-eye coordination at 30 months of age (Dunstan et al., 2008). However, as a lot of outcome measures were tested in these RCTs without adjusting p-values accordingly, positive ﬁndings may have been due to chance. In contrast, one recent large study with high-DHA ﬁshoil in Australia did not observe beneﬁts in cognitive and language scores at 18 months of age after maternal supplementation during pregnancy (Makrides et al., 2010). JOURNEES CHEVREUL 2011 LIPIDS AND BRAIN JOURNEES CHEVREUL 2011 LIPIDS AND BRAIN To cite this article: Osendarp SJM. The role of omega-3 fatty acids in child development. OCL 2011 ; 18(6) : 307-313. doi : 10.1684/ ocl.2011.0417 Evidence from term and pre-term infants Evidence on the effects of LCPUFA during infancy have been well-documented and summarised in recent meta-analysis. Overall, there is no good evidence for a beneﬁcial effect of LCPUFA supplemen- tation on mental development in infants (< 2 yr). A recent meta-analysis on the effects of LCPUFA supplementation of pre-term infants on neurodevelopment up to 18 months of age reported that four out of seven studies did not show a beneﬁt; the three trials which showed improvements on cognitive develop- ment scores all used the newer version of the Bayles Scales of Infant Develop- ment, suggesting that the effects may have been too subtle to be detected with other methods (Schulzke et al., 2011). No beneﬁcial effects of LCPUFA supple- mentation on visual development were seen either (Schulzke et al., 2011). A meta-analysis on the effects of LCPUFA supplementation in term infants did not show an effect on either mental or psychomotor development (Simmer et al., 2008). Outcomes on visual acuity were inconsistent, with six out of nine studies not showing a beneﬁcial effect. Beneﬁcial effects on visual acuity in term infants were seen with higher doses of DHA (0.35% or 0.36% in formula) and when measured by electrophysiological tests. Figure 1. Brain development in childhood. Adapted from Thompson, RA, Nelson CA. Develop- ment. Am Psychol 2001; 56: 5-15. Animation: Gogtay et al., 2007 feeding mothers for improving child growth and development conducted a pooled analysis of outcomes on ﬁve clustered areas of neurodevelopment, i.e. language development, intelligence/ problem solving ability, psychomotor development, motor development and child attention. Overall, no signiﬁcant effects were found, except for attention at 5 years which was due to one single study (Jensen et al., 2010). It was concluded that based on the limited evidence, LCPUFA supplementation dur- ing the lactation period did not appear to improve children’s neurodevelopment. been conducted in low-income coun- tries: in Bangladesh and Mexico. Both studies did not observe differences betweeninfants fromtreated andcontrol mothers in development scores at 10 or 18 months of age were observed (Tofail et al., 2006; Ramakrishan et al., 2010). One study reported on the effects of maternal supplementation with 803 mg EPA and 1183 mg DHA from 18 weeks gestation until 3 months postpartum. Beneﬁcial effects were observed on mental processing index at 4 years (Helland et al., 2003; Helland et al., 2008) but not any more at 7 years of age. Evidence from pregnant and lactating mothers There is suggestive evidence for a beneﬁcial effect of early exposure to n-3 fatty acids on children’s cognitive development. Repeatedly, positive ﬁn- dings on a beneﬁcial effect of n-3 fatty acid supplementation or ﬁsh intake during pregnancy and/or lactation on *Lecture from the symposium ‘‘Lipids and Brain II’’ held by The French Society for the Study of Lipids (Sfel) - Paris 28-30 March 2011, cf. OCL 18-4 and 18-5. doi: 10.1684/ocl.2011.0417 Two intervention trials on prenatal supplementation with n-3 LCPUFA have of omega-3 fatty acids in child development. OCL 2011 ; 18(6) : 307-313. doi : 10.1684/ OCL VOL. 18 N8 6 novembre-de´cembre 2011 307 Article disponible sur le site http://www.ocl-journal.org ou http://dx.doi.org/10.1051/ocl.2011.0417 intake during pregnancy and/or lacta- tion on developmental outcomes of the offspring up to 14 years were reported from observational studies. The small number of RCTs, however, reported only few positive ﬁndings on single outcome measures, which may have been due to chance. Overall, there is no consistency in the beneﬁts observed, with improvements being reported for general IQ, memory, verbal, motor performance, attention, hyperactivity and social behaviour. Birth -8 -6 -4 -2 0 2 4 6 8 10 12 2 4 6 8 10 12 14 16 18 20 Months Years Human brain development Conception Myelination Neuralisation Cell proliferation and migration Sensorimotor cortex Parietal and temporal association cortex Prefontal cortex TRENDS in Cognitive Sciences Figure 1. Brain development in childhood. Adapted from Thompson, RA, Nelson CA. Develop- ment. Am Psychol 2001; 56: 5-15. Animation: Gogtay et al., 2007 Human brain development Evidence from term and pre-term infants Infant, maternal or cord blood DHA status was repeatedly found to be positively correlated to cognitive out- comes (Helland 2003 et al.;, Helland et al., 2008; Dunstan et al., 2008). Six (Lederman et al., 2008; Oken et al., 2008; Mendez et al., 2008; Gale et al., 2008; Budtz-Jorgensen et al., 2007; Hibbeln et al., 2007) out of seven (Strain et al., 2008) observational studies report a signiﬁcant beneﬁcial association of maternal ﬁsh intake during pregnancy and children’s development up to 14 years of age on one or more sub-scales (PDI/motor abilities; hyperactivity & social behaviour; verbal und full IQ) in combination with non-signiﬁcant trends for other outcome measures suggesting beneﬁcial effects of ﬁsh intake. Overall, there is no consistency with regard to the areas of mental develop- ment that might beneﬁt from maternal n-3 fatty acid supplementation; impro- vements were reported for general IQ, memory, verbal, motor performance, attention, hyperactivity and social beha- viour. Evidence from healthy children Studies in healthy children (> 2 yr) do not show consistent beneﬁcial effects of a DHA or EPA supplementation. In total, seven out of ten randomised controlled intervention studies in 4-14 year old children did not ﬁnd a beneﬁcial effect A recent Cochrane review (Delgado- Noguera et al., 2010) on the effects of supplementation with LCPUFA to breast- In summary, repeatedly, positive ﬁn- dings on a beneﬁcial effect of n-3 or ﬁsh 308 308 OCL VOL. 18 N8 6 novembre-de´cembre 2011 2009). The positive effects found cannot for sure be contributed to EPA & DHA as the ﬁsh ﬂour also contains vitamins and minerals which might have had an effect. An fMRI study in 33 UK boys found signiﬁcantly increased cortical activation during a sustained attention task upon eight weeks of supplementation with 400 or 1200 mg/d DHA (McNamara et al., 2010), but no differences between groups in behavioural test outcomes. Another study in UK children found signiﬁcant effects of a three month intervention with EPA (558 mg/d), DHA (174 mg/d) and GLA (60 mg/d) onreadingageand working memory but not on behaviour. However, this study had some quality limitations and is still not published in a peer-reviewed paper (Portwood, 2011). 2009). The positive effects found cannot for sure be contributed to EPA & DHA as the ﬁsh ﬂour also contains vitamins and minerals which might have had an effect. An fMRI study in 33 UK boys found signiﬁcantly increased cortical activation during a sustained attention task upon eight weeks of supplementation with 400 or 1200 mg/d DHA (McNamara et al., 2010), but no differences between groups in behavioural test outcomes. Another study in UK children found signiﬁcant effects of a three month intervention with EPA (558 mg/d), DHA (174 mg/d) and GLA (60 mg/d) onreadingageand working memory but not on behaviour. However, this study had some quality limitations and is still not published in a peer-reviewed paper (Portwood, 2011). LCPUFA on cognitive outcomes (Kirby et al., 2010; Kennedy et al., 2009; Muthayya et al., 2009; Ryan et al., 2008; Osendarp et al., 2007; de Jong etal.,2010).The dosages usedinthistrial variedsigniﬁcantlyandrangedfrom100- 1,200 mg DHA/EPA per day. Evidence from diseased populations LCPUFA and DHA in particular are abundantly present in the human deve- loping brain. However, the evidence base for a beneﬁcial effect of dietary LCPUFA on child’s development is thin and has inconsistent ﬁndings. Supportive evidence for a beneﬁcial role of n-3 fatty acids in cognitive functioning is mainly coming from studies in speciﬁc patient groups (PKU) or children with develop- mental disorders (ADHD, neurophysiolo- gic diseases, n-3 deﬁciency). It is questionable however, how far this evi- dence can be extrapolated to the general population. There is some evidence from studies in children with phenylkenonuria (PKU) on the role of n-3 fatty acids in neurotrans- mission and behavioural outcomes. Chil- dren with PKU have very low natural intakes of EPA and DHA due to dietary restrictions (but a good supply of ALA). Two studies (one of them an open label study) in treated PKU children showed that these children had prolonged visual evoked potentials compared to healthy controls. After supplementation with DHA, EPA (+ALA and AA) wave latency of visual evoked potentials decreased in both studies, indicating more rapid cen- tral nervous system information process- ing (Beblo et al., 2001). The open-label study also showed signiﬁcant improve- mentsof EPA+DHAsupplementation on a test of motor function and coordination (Beblo et al., 2007). When the subjects of the controlled trial were assessed again three years later after the end of the treatment to see whether improvements had lasted in the longer term, initial improvements had returned to baseline (Agostini et al., 2003). Because the conversion of ALA to DHA is very limited, many health authorities currently advice to ensure a minimum intake of DHA and EPA from the diet, despite the fact that in principle the body should be able to make these fatty acids. EPA and DHA are for that reason considered ‘‘conditionally essential’’ (Uauy and Dangour, 2009). Recommendations for the intake of EPA and DHA for adults are mainly based on primary and secondary prevention of cardiovascular disease (CVD; 250- 500 mg/d EPA+DHA or 1-2 portions of fatty ﬁsh per week). For children from 2- 10 years the FAO recently set recom- mendations for an adequate EPA and DHAintake,i.e.100 to 250 mg(depend- ing on the age) (FAO, 2010) in line with adults aiming at the early prevention of chronic disease (Koletzko et al., 2010). In children and adults, PUFA intakes are generally below the recommendations. Evidence from healthy children Evidence from studies in children with ADHD show some positive ﬁndings on self-reported behaviour. Four (Richardson and Puri, 2002; Stevens et al., 2003; Sinn, 2007; Johnsonetal., 2009) out ofﬁve (Razetal., 2009) randomized controlled trials sup- plementing a mix of n-3 fatty acids (120- 730 mg/d) and n-6 fatty acids (60 to 135 mg/d) showed improvements on self-reported ADHD symptoms. Six more studies of varying quality, two of them were open-label studies, supplemented n-3 fatty acids alone (Voigt et al., 2001; Hiramaya et al., 2004; Joshi et al., 2006; Sorgi et al., 2007; Vaisman et al., 2008; Gustafsson et al., 2010). Only the open- label studies showed a signiﬁcant effect of ALA (400 mg/d) or very high doses of EPA/DHA (16 g/d) on behavioural out- comes (Joshi et al., 2006; Sorgi et al., 2007). Three of the four randomized controlled trials supplementing DHA (+EPA) point into the same direction (Voigt et al., 2001; Vaisman et al., 2008; Gustafsson et al., 2010) (ﬁgure 2). lopmental disorders. Evidence from studies in children with ADHD show some positive ﬁndings on self-reported behaviour. Four (Richardson and Puri, 2002; Stevens et al., 2003; Sinn, 2007; Johnsonetal., 2009) out ofﬁve (Razetal., 2009) randomized controlled trials sup- plementing a mix of n-3 fatty acids (120- 730 mg/d) and n-6 fatty acids (60 to 135 mg/d) showed improvements on self-reported ADHD symptoms. Six more studies of varying quality, two of them were open-label studies, supplemented n-3 fatty acids alone (Voigt et al., 2001; Hiramaya et al., 2004; Joshi et al., 2006; Sorgi et al., 2007; Vaisman et al., 2008; Gustafsson et al., 2010). Only the open- label studies showed a signiﬁcant effect of ALA (400 mg/d) or very high doses of EPA/DHA (16 g/d) on behavioural out- comes (Joshi et al., 2006; Sorgi et al., 2007). Three of the four randomized controlled trials supplementing DHA (+EPA) point into the same direction (Voigt et al., 2001; Vaisman et al., 2008; Gustafsson et al., 2010) (ﬁgure 2). cognitive outcome measures. Other studies did not show an association between DHA status at birth or intake during the ﬁrst six months with cognitive performance at 4 or 7 yr of age (Bakker et al., 2003; Ghys et al., 2002), but a positive relationship with improved motor performance at 5 (Rask-Nissila et al., 2002) and 7 years of age (Bakker et al., 2009) (ﬁgure 2). Evidence from healthy children The question whether additional dietary supply of DHA, is required for cognitive functioning after the age of 2 years, remains therefore unanswered. Rapoport et al. (2007) studied the turnover of DHA in the human brain by measuring the daily incorporation of radio-labelled DHA into the brain. He stated that the incorporation rate of DHA equal the rate of loss because DHA cannot be synthesized de novo in the brain, neither does its precursor (ALA) contribute signiﬁcantly (<1%) to brain DHA. He found that the adult brain uses and replaces 4.6 mg of DHA per day. This suggests that DHA in the brain might need to be replenished to main- tain certain levels. However, in a rat study Rapoport et al. also showed that the rate of liver synthesis of DHA from ALA was more than sufﬁcient to main- tain brain DHA (Rapoport et al., 2007). In summary, there is suggestive evidence – from observational studies – of a positive relationship between n-3 FA status at birth or ﬁsh intake during childhood and developmental outcomes in healthy chil- dren up to the age of 18 years. Yet the results might be confounded by other inﬂuencing factors and don’t prove a cause-effect relationship. The potential beneﬁcial effects are not conﬁrmed in intervention trials. The majority (i.e. 7 out of 10) of randomized controlled inter- vention trials in healthy children does not show an effect of n-3 fatty acid sup- plementation on cognitive outcomes. As several studies in speciﬁc patient groups (ADHD, PKU) showed that EPA/ DHA supplementation had a beneﬁcial effect on functional cognitive outcomes (primarily behaviour related) one can hypothesise that dietary fatty acids do arrive at the target organ and that they are indeed used in their role in cognitive functioning. However, this would still need to be conﬁrmed. Evidence from healthy children In addition, all of these RCTs had some quality limitations, like a low intervention dose, a too short intervention period, use of non-validated tests, tests in non-native language, ceiling effect, multiple asses- sors, inappropriate control product, a small sample size, which might explain the lack of positive ﬁndings. relationship between n-3 FA or ﬁsh intake and cognitive outcomes. Five cross-sectional studies assessing 4- 16 yr old children found a signiﬁcant positive association between ﬁsh, n-3, or total PUFA intake and at least some cognitive outcomes (Kirby et al., 2010, Theodore et al., 2009; Kim et al., 2009; Zhang et al., 2005, Freire et al., 2010). Two other studies did not ﬁnd any signiﬁcant associations (Mendez et al., 2008, Eilander et al., 2010). Five longitudinal studies, followed chil- dren form birth to childhood (Bakker et al., 2003, 2009; Ghys et al., 2002; Gale et al., 2010) or throughout childhood (Rask-Nissila et al., 2002; Aberg et al., 2009); two studies showed signiﬁcant beneﬁcial effects of DHA fortiﬁed formula during the ﬁrst 6 months (Gale et al., 2010) and ﬁsh intake during adolescence (Aberg et al., 2009) on children’s overall Three RCTs did ﬁnd some beneﬁcial effects (Dalton et al., 2009; McNamara et al., 2010; Portwood, 2011). One RCT in South African children of low SES with a habitual low ﬁsh intake found signi- ﬁcant effects of a spread fortiﬁed with ﬁsh ﬂour on verbal recognition, discri- mination and spelling (Dalton et al., Observational (cross-sectional and lon- gitudinal) studies in healthy children suggest that there might be a positive Socio-cultural factors Poverty Psycho-social risk factors Biological risk factors (Nutrition) Brain development and function Mental development Sensori- motor Social- emotional Cognitive- language Figure 2. Conceptual framework on role of nutrition in mental development. Adapted from Walker et al., Lancet 2007 Socio-cultural factors Poverty Psycho-social risk factors Biological risk factors (Nutrition) Brain development and function Mental development Sensori- motor Social- emotional Cognitive- language gure 2. Conceptual framework on role of nutrition in mental development. Adapted from Walker et al., Lancet 2007 Socio-cultural factors Poverty Psycho-social risk factors Biological risk factors (Nutrition) Brain development and function Mental development Sensori- motor Social- emotional Cognitive- language Psycho-social risk factors Conceptual framework on role of nutrition in mental development. Adapted from Walker et al., Lancet 2007 OCL VOL. 18 N8 6 novembre-de´cembre 2011 309 OCL VOL. 18 N8 6 novembre-de´cembre 2011 309 lopmental disorders. 310 OCL VOL. 18 N8 6 novembre-de´cembre 2011 Evidence from diseased populations Also EPA and DHA intakes are lower than recommended; however, data are very limited (Harika et al., 2011). Despite the on-going debate on the exact role of LCPUFA in cognitive, efforts to increase It is unfortunate that most of the intervention studies reported to date suffer from methodological limitations. Failure to ﬁnd positive effects of LCPUFA on child development might have been due to intervention dosages which were too low, too short interventions, limita- tions related to the use of psychological tests, inappropriate control products and small sample sizes. On the other hand, trials which did demonstrate positive effects, may in fact have reported false- positive ﬁndings particularly due to a high number of test outcomes included and the possibility of chance ﬁndings. In addition, there is some evidence from EPA/DHA intervention studies on cogni- tive functioning in children with deve- 310 310 OCL VOL. 18 N8 6 novembre-de´cembre 2011 12000 10000 8000 6000 4000 2000 0 -2000 20 40 60 80 100 120 140 160 www.medscape.com Martinez M. J Pediatr. 1992 : 120 : S129-S138. DHA accumulates in the brain beginning In utero through toddlerhood DHA accumulation in the brain, mmol/forebrain Postmenstrual age, weeks Third trimester Figure 3. Brain DHA during gestation and early childhood. Adapted from Martinez M. J Pediatr 1992, 120: S129-s138 DHA accumulates in the brain beginning In utero through toddlerhood Figure 3. Brain DHA during gestation and early childhood. Adapted from Martinez M. J Pediatr 1992, 120: S129-s138 Figure 3. Brain DHA during gestation and early childhood. Adapted from Martinez M. J Pediatr 1992, 120: S129-s138 consumption of these fatty acids in children should therefore be supported (ﬁgure 3). supplementation improves visual evoked potentials in children with phenylketonuria. Neurology 2001; 57: 1488-91. Delgado-Noguera MF, Calvache JA, Bonﬁll Cosp X. Supplementation with long chain polyunsaturated fatty acids (LCPUFA) to breastfeeding mothers for improving child growth and development. Cochrane Data- base Syst Rev 2010; (12): CD007901. Delgado-Noguera MF, Calvache JA, Bonﬁll Cosp X. Supplementation with long chain polyunsaturated fatty acids (LCPUFA) to breastfeeding mothers for improving child growth and development. Cochrane Data- base Syst Rev 2010; (12): CD007901. Delgado-Noguera MF, Calvache JA, Bonﬁll Cosp X. Supplementation with long chain polyunsaturated fatty acids (LCPUFA) to breastfeeding mothers for improving child growth and development. Cochrane Data- base Syst Rev 2010; (12): CD007901. Beblo S, Reinhardt H, Demmelmair H, Muntau AC, Koletzko B. Evidence from diseased populations Effect of ﬁsh oil supplementa- tionon fatty acid status, coordination, and ﬁne motor skills in children with phenylketonuria. J Pediatr 2007; 150: 479-84. REFERENCES Acta Paediatrica 2010; 99: 72-7. Gustafsson PA, Birberg-Thornberg U, Duchen K, et al. EPA supplementation improves teacher-rated behaviour and oppositional symptoms in children with ADHD. Acta Paediatr 2010; 99: 1540-9. Kirby, et al. A double-blind, placebo-control- led study investigating the effects of omega-3 supplementation in children aged 8-10 years from a mainstream school population. Res Dev Disabil 2010a; 31: 718-30. Rapoport SI, Rao JS, Igarashi M. Brain metabolism of nutritionally essential poly- unsaturated fatty acids depends on both the diet and the liver. Prostaglandins Leukot Essent Fatty Acids 2007; 77: 251-61. Helland, et al. Maternal supplementation with very-long-chain n-3 fatty acids during pre- gnancy and lactation augments children’s IQ at 4 years of age. Pediatrics 2003; 111: e39-44. Kirby, et al. Childrens’ learning behaviour and the association with cheek cell poly- unsaturated fatty acid levels. Res Dev Disabil 2010b; 31: 731-42. Rask-Nissila L, Jokinen E, Terho P, et al. Effects of diet on the neurologic development of children at 5 years of age: the STRIP project. J Pediatr 2002; 140: 328-33. Helland, et al. Effect of supplementing pregnant and lactating mothers with n-3 very-long-chain fatty acids on children’s IQ and body mass index at 7 years of age. Pediatrics 2008; 122: e472-9. Koletzko B, Uauy R, Palou A, et al. Dietary intake of eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) in children - a workshop report. Br J Nutr 2010; 103: 923-8. Raz R, Carasso RL, Yehuda S. The inﬂuence of short-chain essential fatty acids on children with attention-deﬁcit/hyperactivity disorder: a double-blind placebo-controlled study. J Child Adolesc Psychopharmacol 2009; 19: 167-77. Hibbeln, et al. Maternal seafood consump- tion in pregnancy and neurodevelopmental outcomes in childhood (ALSPAC study): an observational cohort study. Lancet 2007; 369: 578-85. Lederman et al., Relation between Cord Blood Mercury Levels and Early Child Deve- lopment in a World Trade Center Cohort. Environ Health Perspect 2008; 116: 1085-91. Richardson AJ, Puri BK. A randomized double- blind, placebo-controlled study of the effects of supplementation with highly unsaturated fatty acids on ADHD-related symptoms in children with speciﬁc learning difﬁculties. Prog Neuropsychopharmacol Bio Psychiatry 2002; 26: 233-9. Hiramaya S, Hamazaki T, Terasawa K. Effect of docosahexaenoic acid-containing food administration on symptoms of attention- deﬁcit/hyperactivity disorder – a placebo- controlled double-blind study. Eur J Clin Nutr 2004; 58: 467-73. Makrides M, Collins CT, Gibson RA. Impact of fatty acid status on growth and neurobeha- vioral development in humans. REFERENCES FAO. Fats and fatty acids in human nutrition. Report of an expert consultation. Food and Nutrition paper 91. Rome 2010. Judge MP, Harel O, Lammi-Keefe CJ. Mater- nal consumption of a docosahexanoic acid- containing functional food during pre- gnancy: beneﬁt for infant performance on problem-solving but not on recognition memory tasks at 9 mo of age. Am J Clin Nutr 2007; 85: 1572-7. Freire C, Ramos R, Lopez-Espinosa MJ, et al. Hair mercury levels, ﬁsh consumption, and cognitive development in preschool children from Granada, Spain. Environ Res 2010; 110: 96-104. Oken et al., Maternal Fish Intake during Pregnancy, Blood Mercury Levels, and Child Cognition at Age 3 Years in a US Cohort. Am J Epidemiol 2008; 167: 1171-81. Harika RK, Cosgrove MC, Osendarp SJM, Verhoef P, Zock PL. Fatty acid intakes of children and adolescents around the world are lower than recommended for future cardiovascular health: a review of dietary intake data. B J Nutr 2011(in press); doi: 10.1017/S0007114511001528. Gale, et al. Oily ﬁsh intake during pregnancy – association with lower hyperactivity but not with higher full-scale IQ in offspring. J Child Psychol Psychiatry 2008; 49: 1061-8. Osendarp SJ, Baqhurst K, Bryan J, et al.; The NEMO study group. Effect of a 12-mo micro- nutrient intervention on learning and memory in well-nourished and marginally nourished school-aged children: 2 parallel, randomized, placebo-controlled studies in Australia and Indonesia. Am J Clin Nutr 2007; 86: 1082-93. Gale, et al. Breastfeeding, the use of doco- sahexaenoic acid-fortiﬁed formulas in infancy and neuropsychological function in child- hood. Arch Dis Child 2010; 95: 174-9. Kennedy et al. Cognitive and mood effects of 8 weeks’ supplementation with 400 mg or 1000 mg of the omega-3 essential fatty acid docosahexaenoic acid (DHA) in healthy children aged 10-12 years. Nutr Neurosci 2009; 12:48-56. Portwood N. The Durham trial. http://www. durhamtrial.org/primary%20main.htm. (accessed at 18 November 2011). Ghys et al., Red blood cell and plasma phospholipid arachidonic and docosahexae- noic acid levels at birth and cognitive development at 4 years of age. Early Hum Dev 2002; 69: 83-90. Ramakrishan U, Stein AD, Parra-Cabrera S, et al. Effects of docosahexanoic acid supple- mentation during pregnancy on gestational age and size at birth: randomized, double- blind, placebo-controlled trial in Mexico. Food Nutrition Bulletin 2010l 31: S108-16. Kim J, Winkvist A, Aberg MAI, et al. Fish consumption and school grades in Swedish Adolescents: a study of the large general population. REFERENCES de Jong, et al. The Groningen LCPUFA study: no effect of postnatal long-chain polyunsa- turated fatty acids in healthy term infants on neurological condition at 9 years. Br J Nutr 2010; 104: 566-72. Aberg, et al. Fish intake of Swedish male adolescents is a predictor of cognitive per- formance. Acta Paediatr 2009; 98: 555-60. Budtz-Jorgensen, et al. Separation of Risks and Beneﬁts of Seafood Intake. Environ Health Perspect 2007; 115: 323-7. Agostini C, Verduci E, Massetto N, et al. Long term effects of long chain polyunsaturated fats in hyperphenylalaninemic children. Arch Dis Child 2003; 88: 582-3. Dobbing J, Sands J. Quantitative growth and development of human brain. Arch Dis Child 1973; 48: 757-67. Clandinin MT, Chappell JE, Leong S, Heim T, Swyer PR, Chance GW. Extrauterine fatty acid accretion in infant brain: implications for fatty acid requirements. Early Hum Dev 1980; 4: 131-8. Bakker, et al. Long-chain polyunsaturated fatty acids at birth and cognitive function at 7 y of age. Eur J Clin Nutr 2003; 57: 89-95. Dunstan, et al. Cognitive assessment of children at age 2(1/2) years after maternal ﬁsh oil supplementation in pregnancy: a randomised controlled trial. Arch Dis Child Fetal Neonatal Ed 2008; 93: F45-50. Bakker, et al. Relationship between long- chain polyunsaturated fatty acids at birth and motor function at 7 years of age. Eur J Clin Nutr 2009; 63: 499-504. Dalton, et al. A randomised control trial in schoolchildren showed improvement in cognitive function after consuming a bread spread, containing ﬁsh ﬂour from a marine source. Prostaglandins Leukot Essent Fatty Acids 2009; 80 :143-9. Dalton, et al. A randomised control trial in schoolchildren showed improvement in cognitive function after consuming a bread spread, containing ﬁsh ﬂour from a marine source. Prostaglandins Leukot Essent Fatty Acids 2009; 80 :143-9. Eilander, et al. Undernutrition, fatty acid and micronutrient status in relation to cognitive performance in Indian school children: a Beblo S, Reinhardt H, Muntau AC, Mueller- Felber W, Roscher AA, Koletzko B. Fish oil 311 OCL VOL. 18 N8 6 novembre-de´cembre 2011 311 outcome of Attention deﬁcit Hyperactivity Disorder (ADHD). Prostaglandins Leukot Essent Fatty Acids 2006; 74: 17-21. cross-sectional study. Br J Nutr 2010; 103: 1056-64. Muthayya et al., Effect of fortiﬁcation with multiple micronutrients and n-3 fatty acids on growth and cognitive performance in Indian schoolchildren: the CHAMPION (Chil- dren’s Health and Mental Performance Inﬂuenced by Optimal Nutrition) Study. Am J Clin Nutr 2009; 89: 1766-75. REFERENCES Maternal Child Nutr 2011; 7 (Suppl .2): 80-8. Jensen, et al. Effects of early maternal docosahexaenoic acid intake on neuropsy- chological status and visual acuity at ﬁve years of age of breast-fed term infants. J Pediatr 2010; 157: 900-5. Martinez M, Tissue levels of polyunsaturated fatty acids during early human development. J Pediatr 1992; 120: S129-38. Ryan et al., Assessing the effect of docosa- hexaenoic acid on cognitive functions in healthy, preschool children: a randomized, placebo-controlled, double-blind study. Clin Pediatr (Phila) 2008; 47: 355-62. McNamara, et al. Docosahexaenoic acid supplementation increases prefrontal cortex activation during sustained attention in heal- thy boys: a placebo-controlled, dose-ranging, functional magnetic resonance imaging study. Am J Clin Nutr 2010; 91: 1060-7. Johnson M, Ostlund s, Fransson G, Kadesj€o B, Gillberg C. Omega-3/omega-6 fatty acids for attention deﬁcit hyperactivity disorder: a randomized placebo-controlled trial in chil- dren and adolescents. J Atten Disord 2009; 12: 394-401. Schulzke SM, Patole SK, Simmer K. Long- chain polyunsaturated fatty acid supplemen- tation in preterm infants. Cochrane Database of Systematic Reviews 2011, Issue 2. Mendez et al., Maternal ﬁsh and other seafood intakes during pregnancy and child neurodevelopment at age 4 years. Public Health Nutr 2008 : 1-9. Simmer K, Patole S, Rao SC. Longchain polyunsaturated fatty acid supplementation in infants born at term. Cochrane Database of Systematic Reviews 2008, Issue 1. Joshi K, Lad S, Kale M, et al. Supplementation with ﬂax oil and vitamin C improves the Joshi K, Lad S, Kale M, et al. Supplementation with ﬂax oil and vitamin C improves the 312 312 OCL VOL. 18 N8 6 novembre-de´cembre 2011 Sinn N. Physical fatty acid deﬁciency signs in children with ADHD symptoms. Prostaglan- dinsLeukotEssentFatty Acids2007;77: 109-15. Development Nutrition Study. Neurotoxico- logy 2008; 29: 776-82. tion performance in children with inatten- tion: effect of dietary n-3 fatty acids containing phospholipids. Am J Clin Nutr 2008; 87: 1170-80. tion performance in children with inatten- tion: effect of dietary n-3 fatty acids containing phospholipids. Am J Clin Nutr 2008; 87: 1170-80. Theodore, et al. Dietary patterns and intel- ligence in early and middle childhood. Intelligence, 2009; 37: 506-13. Sorgi PJ, Hallowell EM, Hutchins HL, Sears B. Effects of an open-label pilot study with high- dose EPA/DHA concentrates on plasma phospholipids and behaviour in children with attention deﬁcit hyperactivity disorder. Nutr J 2007; 6: 16. Voigt RG., Llorente AM, Jensen CL, Fraley JK, Berretta MC, Heird WC. OCL VOL. 18 N8 6 novembre-de´cembre 2011 313 REFERENCES A randomized, double-blind, placebo-controlled trial of docosahexaenoic acid supplementation in children with attention-deﬁcit/hyperactivity disorder. J Pediatr 2001; 139: 189-96. Tofail F, Kabir I, Hamadani JD, et al. Supplementation of ﬁsh-oil and soy-oil dur- ing pregnancy and psychomotor develop- ment of infants. J of Health, Population and Nutrition 2006; 24: 48-56. Stevens L, Zhan W, Peck L, et al. EFA supplementation in children with inatten- tion, hyperactivity, and other disruptive behaviors. Lipids 2003; 38: 1007-21. Yehuda S, Omega-6/omega-3 ratio and brain-related functions. World Rev Nutr Diet 2003; 92: 37-56. Uauy R, Dangour AD. Dietary fat quality for optimal health and well-being: overview of recommendations. Ann Nutr Metab 2009; 54 (Suppl. 1): 2-7. Strain JJ, Davidson PW, Bonham MP, et al. Associations of maternal long-chain polyun- saturated fatty acids, methyl mercury, and infant development in the Seychelles Child Zhang, et al. Dietary fat intake is associated with psychosocial and cognitive functioning of school-aged children in the United States. J Nutr 2005; 135: 1967-73. Vaisman N, Kaysar N, Zaruk-Adasha Y, et al. Correlation between changes in blood fatty acid composition and visual sustained atten-
https://openalex.org/W2944134194	https://acp.copernicus.org/articles/19/14195/2019/acp-19-14195-2019.pdf	English	null	Single-particle experiments measuring humidity and inorganic salt effects on gas-particle partitioning of butenedial	Atmospheric chemistry and physics	2,019	cc-by	13,521	Single-particle experiments measuring humidity and inorganic sal effects on gas-particle partitioning of butenedial Even under dry (RH < 5 %) conditions, the evaporation rate of butenedial is orders of magnitude lower than what would be expected if butenedial existed purely as a dialdehyde in the particle, implying an equilibrium strongly favoring hydrated forms and the strong preference of certain dialdehyde com- pounds to remain in a hydrated form even under lower water content conditions. Butenedial exhibits a salting-out effect in the presence of sodium chloride and sodium sulfate, in con- trast to glyoxal. The outcomes of these experiments are also helpful in guiding the design of future EDB-MS experiments. Even under dry (RH < 5 %) conditions, the evaporation rate of butenedial is orders of magnitude lower than what would be expected if butenedial existed purely as a dialdehyde in the particle, implying an equilibrium strongly favoring hydrated forms and the strong preference of certain dialdehyde com- pounds to remain in a hydrated form even under lower water content conditions. Butenedial exhibits a salting-out effect in the presence of sodium chloride and sodium sulfate, in con- trast to glyoxal. The outcomes of these experiments are also helpful in guiding the design of future EDB-MS experiments. 1 Introduction Atmospheric aerosol particles contribute signiﬁcantly to health and climate effects (Boucher et al., 2013; Cohen et al., 2017). In order to understand and predict the extent and com- position of aerosol particles across all environments, models of atmospheric aerosol require well-constrained experimen- tally derived parameters describing aerosol behavior over their entire multiday lifetime in the atmosphere. One key process is gas–particle partitioning, meaning a quantitative measure of what fraction of a given atmospheric chemical species exists in the gas compared to the particle phase under a given set of atmospheric conditions. For ex- ample, accurate understanding of gas–particle partitioning is necessary to understand to what extent a given compound contributes to aerosol particle loading in a particular envi- Single-particle experiments measuring humidity and inorganic sal effects on gas-particle partitioning of butenedial Adam W. Birdsall1, Jack C. Hensley2, Paige S. Kotowitz3,a, Andrew J. Huisman3,b, and Frank N. Keutsch1,2,4 1Department of Chemistry and Chemical Biology, Harvard University, Cambridge, MA, USA 2School of Engineering and Applied Sciences, Harvard University, Cambridge, MA, USA 3Chemistry Department, Union College, Schenectady, NY, USA 4Department of Earth and Planetary Sciences, Harvard University, Cambridge, MA, USA anow at: nanoComposix, San Diego, CA, USA bnow at: Gentex Corporation, Zeeland, MI, USA orrespondence: Adam Birdsall (abirdsall@g.harvard.edu) and Frank Keutsch (keutsch@seas.harva Correspondence: Adam Birdsall (abirdsall@g.harvard.edu) and Frank Keutsch (keutsch@seas.harvard.edu) Received: 2 May 2019 – Discussion started: 7 May 2019 Received: 2 May 2019 – Discussion started: 7 May 2019 Revised: 7 October 2019 – Accepted: 19 October 2019 – Published: 26 November 2019 Abstract. An improved understanding of the fate and prop- erties of atmospheric aerosol particles requires a detailed process-level understanding of fundamental factors inﬂuenc- ing the aerosol, including partitioning of aerosol compo- nents between the gas and particle phases. Laboratory exper- iments with levitated particles provide a way to study fun- damental aerosol processes over timescales relevant to the multiday lifetime of atmospheric aerosol particles, in a con- trolled environment in which various characteristics relevant to atmospheric aerosol can be prepared (e.g., high surface- to-volume ratio, highly concentrated or supersaturated solu- tions, changes to relative humidity). In this study, the four- carbon unsaturated compound butenedial, a dialdehyde pro- duced by oxidation of aromatic compounds that undergoes hydration in the presence of water, was used as a model or- ganic aerosol component to investigate different factors af- fecting gas–particle partitioning, including the role of lower- volatility “reservoir” species such as hydrates, timescales in- volved in equilibration between higher- and lower-volatility forms, and the effect of inorganic salts. The experimental approach was to use a laboratory system coupling particle levitation in an electrodynamic balance (EDB) with parti- cle composition measurement via mass spectrometry (MS). In particular, by ﬁtting measured evaporation rates to a ki- netic model, the effective vapor pressure was determined for butenedial and compared under different experimental con- ditions, including as a function of ambient relative humidity and the presence of high concentrations of inorganic salts. Atmos. Chem. Phys., 19, 14195–14209, 2019 https://doi.org/10.5194/acp-19-14195-2019 © Author(s) 2019. This work is distributed under the Creative Commons Attribution 4.0 License. Atmos. Chem. Phys., 19, 14195–14209, 2019 https://doi.org/10.5194/acp-19-14195-2019 © Author(s) 2019. This work is distributed under the Creative Commons Attribution 4.0 License. A. W. Birdsall et al.: Gas-particle partitioning of butenedial Schematic illustrating the meaning of effective vapor pressure, Pvap,eff, and effective Henry’s law constant, KH,eff. A compound partitioning between the gas and particle phase (here, butenedial, BD) can exist in the gas phase (BDgas). In the particle phase the compound can exist in its standard form (BDpart) but also in fast equilibrium with a “reservoir” species via reversible chemical equilibrium (here, hydrated butenedial, BD-hydpart). The reservoir species can also partition to the gas phase (BD-hydgas), but with a lower vapor pressure. The formal vapor pressure and Henry’s law constant of butenedial are related to the equilibrium between BDgas and BDpart alone. In contrast, Pvap,eff and KH,eff are related to the equilibrium between the gas phase and the combined particle-phase abundance of both BD and reservoir species BD-hyd. Additionally, Pvap,eff and KH,eff can be modulated by the presence of inorganic ions in the particle phase, shown here as generic cations X+ and generic anions Y −, which may have a salting-in or salting-out ef- fect. Researchers have used global-scale chemical transport models to investigate the effect of representation of KH,eff on modeled secondary organic aerosol (SOA) concentrations. In particular, KH,eff has been used as a parameter describing the loss of gas-phase oxidized semivolatile organic compounds (SVOCs) via wet deposition, as a pathway competing with SOA formation. In these modeling studies, one approach has been to use a single uniform KH,eff of 103 to 105 M atm−1 for SVOCs for which KH,eff is otherwise unknown (e.g., Pye and Seinfeld, 2010). An alternate approach has been to ob- tain a model-derived parameterized dependence of KH,eff on volatility that distinguishes between anthropogenic and bio- genic precursors (Hodzic et al., 2014). Overall, calculated secondary organic aerosol loadings have been shown to de- pend on the representation of KH,eff (Hodzic et al., 2016; Knote et al., 2015), underlining the importance of an im- proved fundamental understanding of how KH,eff depends on the chemical behavior of a compound within the complex matrix of atmospheric aerosol particles. particle partitioning equilibrium as well as on the equilibrium with the reservoir species (Ervens and Volkamer, 2010). The prototypical molecule for this process is the two-carbon di- aldehyde glyoxal, which is known to exist largely in hydrated form in the presence of condensed-phase water, leading to a large KH,eff (Ip et al., 2009). Published by Copernicus Publications on behalf of the European Geosciences Union. Published by Copernicus Publications on behalf of the European Geosciences Union. A. W. Birdsall et al.: Gas-particle partitioning of butenedial 14196 Figure 1. Schematic illustrating the meaning of effective vapor pressure, Pvap,eff, and effective Henry’s law constant, KH,eff. A compound partitioning between the gas and particle phase (here, butenedial, BD) can exist in the gas phase (BDgas). In the particle phase the compound can exist in its standard form (BDpart) but also in fast equilibrium with a “reservoir” species via reversible chemical equilibrium (here, hydrated butenedial, BD-hydpart). The reservoir species can also partition to the gas phase (BD-hydgas), but with a lower vapor pressure. The formal vapor pressure and Henry’s law constant of butenedial are related to the equilibrium between BDgas and BDpart alone. In contrast, Pvap,eff and KH,eff are related to the equilibrium between the gas phase and the combined particle-phase abundance of both BD and reservoir species BD-hyd. Additionally, Pvap,eff and KH,eff can be modulated by the presence of inorganic ions in the particle phase, shown here as generic cations X+ and generic anions Y −, which may have a salting-in or salting-out ef- fect. ronment. Despite seemingly being a straightforward question to answer, reconciling observations of gas–particle partition- ing with scientists’ best understanding of the involved com- pounds remains an area of active research (Bilde et al., 2015). p The vapor pressure of a compound is an important param- eter affecting its gas–particle partitioning behavior, but is not necessarily sufﬁcient to completely describe the partitioning. Strictly speaking, a vapor pressure describes the equilibrium between pure condensed- and gas-phase forms of a partic- ular compound. However, in the complex matrix of atmo- spheric aerosol particles, a compound can instead be thought of as exhibiting an “effective vapor pressure”, Pvap,eff, mean- ing the vapor pressure apparently demonstrated by the com- pound when at equilibrium in a system consisting of a gas phase and single condensed phase, if the role of effects be- yond Raoult’s Law (i.e., a mole fraction dependence), were ignored. In particular, the effective vapor pressure describes the vapor pressure that would be calculated if the activity co- efﬁcient were unity and condensed-phase chemical equilib- ria of the compound of interest with other “reservoir” forms were disregarded (see below). Similarly, while the Henry’s law constant, KH, is used to describe gas–particle partition- ing of a compound in a dilute aqueous solution, the behavior of a compound in atmospheric aerosol can be described using an “effective Henry’s law constant”, KH,eff. Figure 1. www.atmos-chem-phys.net/19/14195/2019/ A. W. Birdsall et al.: Gas-particle partitioning of butenedial 14197 Figure 2. Butenedial (a) in equilibrium with its singly (b) and doubly (c) hydrated forms. The cis isomers of butenedial and its hydrates are shown here but no assumptions are made about the geometric isomer studied in this work. Figure 2. Butenedial (a) in equilibrium with its singly (b) and doubly (c) hydrated forms. The cis isomers of butenedial and its hydrates are shown here but no assumptions are made about the geometric isomer studied in this work. vations made in levitated particle experiments can, with care, be extrapolated to ambient atmospheric conditions, for ex- ample, by accounting for the difference in size (and hence surface-area-to-volume ratio) between laboratory and atmo- spheric particles. work we have studied the four-carbon unsaturated dialde- hyde butenedial. Butenedial has been observed in the at- mosphere and can be produced as a ﬁrst-generation oxida- tion product of aromatic compound precursors (Birdsall and Elrod, 2011; Dumdei and O’Brien, 1984; Shepson et al., 1984). Furthermore we are interested in butenedial in terms of what its behavior reveals about the behavior of aldehy- dic compounds present in the atmosphere. One character- istic of butenedial is that the aldehyde functional groups are expected to readily hydrate under aqueous conditions, such as in an aqueous aerosol phase whose water content is governed by the relative humidity (RH) of the surround- ing gas phase (Fig. 2). The properties of butenedial, such as Pvap,eff or KH,eff, are expected to vary greatly depending on whether butenedial primarily exists in a hydrated or non- hydrated form, as has been observed for glyoxal. Another dicarbonyl compound, the three-carbon compound methyl- glyoxal, has also received appreciable study (e.g., Curry et al., 2018), though one notable structural difference is it con- tains one aldehydic group and one ketone. Unlike for gly- oxal and methylglyoxal, Pvap,eff or KH,eff for butenedial pre- viously has not been measured experimentally, to the best of our knowledge. Measurement of the gas–particle partitioning of butenedial provides further insight into the behavior of di- carbonyl compounds larger than glyoxal and methylglyoxal. Previous work in our laboratory has developed a technique termed electrodynamic balance–mass spectrometry (EDB- MS) to levitate individual charged droplets with diameter on the order of 10 µm in an electrodynamic balance (EDB) and then measure the droplet’s composition with mass spectrom- etry (MS) (Birdsall et al., 2018). Other single levitated par- ticle MS methods have been reported (Jacobs et al., 2017). A. W. Birdsall et al.: Gas-particle partitioning of butenedial Other groups have developed methods to measure vapor pressures of organic compounds using highly precise parti- cle diameter measurements using optical sizing techniques (Krieger et al., 2012). One difference between MS and op- tical sizing approaches is that with MS, vapor pressures of individual chemical components can be readily extracted in a multicomponent system. In this research article, we report experiments in which we used our EDB-MS instrumentation to measure Pvap,eff of butenedial under different experimental conditions, by measuring butenedial’s evaporation rate and ﬁtting the data to a kinetic model in which Pvap,eff is the free parameter. KH,eff can also be calculated based on a calcu- lated particle water content, using measured relative humid- ity and a thermodynamic model. First we measured Pvap,eff of butenedial in purely organic aerosol particles under two relative humidity conditions. We then measured Pvap,eff of butenedial in mixed organic–inorganic aerosol particles, con- taining either sodium chloride or sodium sulfate under humid conditions. The results demonstrate the possible role of dif- ferent factors on the effective vapor pressures and Henry’s law constants of organic aerosol components as well as help inform future EDB-MS experiments. One approach to address these questions is via labora- tory experiments studying individual particles levitated in a chamber surrounded by a bath gas. The advantages of this ap- proach include the ability to isolate and quantify individual processes of interest under controlled conditions. An addi- tional strength of levitated particle experiments is the abil- ity to create a system that exhibits certain key characteris- tics of atmospheric aerosol: a relatively high surface-area- to-volume ratio enabling investigation of the interaction be- tween bulk and multiphase processes, a condensed phase that can achieve highly concentrated or supersaturated condi- tions that are often present in atmospheric aerosol particles, and the ability to levitate particles over timescales consis- tent with their multiday atmospheric lifetimes so that “slow” processes can be studied. Furthermore, the ability to ﬂow a pure bath gas continually through the levitation chamber sim- pliﬁes the system compared to analogous experiments per- formed in an environmental chamber, removing the need for continual gas-phase measurements and reducing the set of multiphase processes that need to be accounted for, such as the effects of a more complex gas-phase matrix. The obser- A. W. Birdsall et al.: Gas-particle partitioning of butenedial The process here of reversible formation of a reservoir species is in contrast to the nonre- versible formation of a product that takes place in the process of reactive uptake, for example by the isoprene epoxydiols (Lin et al., 2012). A second effect could be if nonideal mixing, within a single phase, between the compound of interest and other aerosol components causes the activity coefﬁcient of the compound, γ , to differ strongly from unity. Here Pvap,eff or KH,eff differs from the value in a corresponding ideal mixture by a factor of γ . When the nonideality is due to interactions with inorganic compounds, which can be highly concentrated in ambient aerosol particles, the effect is called a “salting-in” (γ < 1) or “salting-out” (γ > 1) interaction, depending on whether the interaction favors or disfavors, respectively, the presence of the compound in the phase in question. The effect of salting in or salting out on chemical systems with atmo- spheric relevance has been assessed with both experimental and theoretical approaches (Toivola et al., 2017; Wang et al., 2014; Waxman et al., 2015; Yu et al., 2011). To this end, this research article focuses on two possible effects on Pvap,eff and KH,eff: formation of condensed phase “reservoir” species and nonideality in the condensed phase due to interactions with inorganic ions (Fig. 1). First, the compound may experience a second equilibrium in the condensed phase with a “reservoir” species with a lower vapor pressure. This reservoir species may be formed by chemical reactions such as hydration or oligomerization. If a large proportion of the compound exists in the form of the reservoir species, the result would be a lower Pvap,eff or higher KH,eff, whose value can depend on both the gas– The gas–particle partitioning behavior of a given com- pound will depend on its chemical properties. In the present Atmos. Chem. Phys., 19, 14195–14209, 2019 www.atmos-chem-phys.net/19/14195/2019/ 2.1 Butenedial synthesis Butenedial was synthesized in our laboratory following literature procedure (Avenati and Vogel, 1982). In brief, 2.4 M 2,5-dihydro-2,5-dimethoxyfuran (DMDF, TCI Amer- ica, 98 %) was hydrolyzed at 25 ◦C over ∼10 d in an aque- ous solution containing acetic acid (VWR, 99.7 %) diluted to 3.4 M in deionized water. Reaction progress was mon- www.atmos-chem-phys.net/19/14195/2019/ A. W. Birdsall et al.: Gas-particle partitioning of butenedial 14198 Figure 3. NMR spectrum of synthesized butenedial in deuterated water solvent. Includes peaks associated with hydrated butenedial (hyd. BD), diethylmalonic acid (DMA), used as an internal standard (int. std.), and solvent (HDO). Figure 4. Sample mass spectrum of droplet trapped in EDB and measured with MS, using the BD + PEG-6 + NaCl (XNaCl = 0.140) experimental solution. Both the raw and background- subtracted spectra are shown. Butenedial is observed at its parent ion of 85 m/z. PEG-6 is observed at its parent ion of 283 m/z, l ith t k f t / l b l d ith ∗ Figure 3. NMR spectrum of synthesized butenedial in deuterated water solvent. Includes peaks associated with hydrated butenedial (hyd. BD), diethylmalonic acid (DMA), used as an internal standard (int. std.), and solvent (HDO). Figure 3. NMR spectrum of synthesized butenedial in deuterated water solvent. Includes peaks associated with hydrated butenedial (hyd. BD), diethylmalonic acid (DMA), used as an internal standard (int. std.), and solvent (HDO). Figure 4. Sample mass spectrum of droplet trapped in EDB and measured with MS, using the BD + PEG-6 + NaCl (XNaCl = 0.140) experimental solution. Both the raw and background- subtracted spectra are shown. Butenedial is observed at its parent ion of 85 m/z. PEG-6 is observed at its parent ion of 283 m/z, along with at known fragment m/z labeled with ∗. itored via NMR. Rotary evaporation was used to remove acetic acid, residual DMDF, and excess water. The mass frac- tion of the solution after rotary evaporation determined to be butenedial was quantiﬁed via quantitative addition of di- ethylmalonic acid (Sigma Aldrich, 98 %) as an internal stan- dard, and found to be 0.75 ± 0.02, with the lack of major unidentiﬁed peaks in the spectrum implying the balance of the composition to consist of water. No further puriﬁcation was deemed necessary based on the collected product NMR spectrum (Fig. 3). on a heated plate (220 ◦C) and the resulting vapors are ion- ized via a corona discharge emanating from a charged nee- dle and drawn into the inlet of a commercial time-of-ﬂight mass spectrometer, operated at unit mass resolution (JEOL AccuTOF). A sample mass spectrum of a single particle con- taining butenedial, hexaethylene glycol (PEG-6, an internal standard), and sodium chloride is shown in Fig. 4. A. W. Birdsall et al.: Gas-particle partitioning of butenedial Mass spectra were collected using commercial software (JEOL MassCenter), resulting in a brief (few second) signal “pulse” above the background on m/z channels correspond- ing to components of the vaporized droplet (Fig. 5). As previ- ously, the intensity at each m/z channel was quantiﬁed as the height of the pulse above the background. The peak height calculation was automated using an algorithm that detected the peak at each mass channel of interest and subtracted an average background value from before the peak onset, and it was checked by eye for correctness. www.atmos-chem-phys.net/19/14195/2019/ Atmos. Chem. Phys., 19, 14195–14209, 2019 A. W. Birdsall et al.: Gas-particle partitioning of butenedial Figure 4. Sample mass spectrum of droplet trapped in EDB and measured with MS, using the BD + PEG-6 + NaCl (XNaCl = 0.140) experimental solution. Both the raw and background- subtracted spectra are shown. Butenedial is observed at its parent ion of 85 m/z. PEG-6 is observed at its parent ion of 283 m/z, along with at known fragment m/z labeled with ∗. A. W. Birdsall et al.: Gas-particle partitioning of butenedial www.atmos-chem-phys.net/19/14195/2019/ 2.3 Experimental systems We used two types of experiments to probe the effect of RH and inorganic content on the evaporation rate of butenedial and on extracted values of Pvap,eff and KH,eff. The exper- imental solutions used in the current study are reported in Table 1. Overall two types of experiments were performed. We also veriﬁed our observations were not affected by reactive chemistry between PEG-6 and butenedial. Because butenedial contains aldehyde groups and PEG-6 contains al- cohol groups, conceivably a hemiacetal could be formed by their reaction, which would lower Pvap,eff of butenedial. We repeated the humid evaporation measurements with diethyl- malonic acid as the internal standard rather than PEG-6, for which no chemical reaction with butenedial is expected. We observed no difference in evaporation rate, within experi- mental uncertainty, of butenedial between the PEG-6 and diethylmalonic acid data, implying the presence of PEG-6 did not measurably inﬂuence the evaporation rate. Further- more, we observed no peaks in the mass spectrum at m/z ratios consistent with a PEG-6–butenedial hemiacetal. Re- cent work has shown that an model hemiacetal oligomer was detected intact via mass spectrometry when quickly (mil- lisecond timescale) vaporized from impact onto a heated rod at ∼160 ◦C (Claﬂin and Ziemann, 2019), which suggests under our similar analytical conditions we would also ex- pect to observe the intact hemiacetal, if present. Nor did we observe a peak in a 2-D NMR heteronuclear multiple One type of experiment was designed to investigate the effect of RH on the evaporation rate (and hence Pvap,eff) of butenedial. These experiments used droplets from a solution containing a mixture of butenedial and hexaethylene glycol (PEG-6, i.e., the polyethylene glycol hexamer, 99 %, Sigma Aldrich), the latter of which served as an internal standard for reasons discussed in Sect. 2.3.1. Droplets were exposed to two different RH levels: “dry” experiments for which the bath gas was dry nitrogen gas with RH in the EDB measured to be < 5 %, and “humid” experiments for which the bath gas was humidiﬁed nitrogen gas created by ﬂowing nitrogen gas through a water bubbler with RH in the EDB measured to be 75 ± 5 %. A second type of experiment was designed to probe the effect of the presence of inorganic species on Pvap,eff of butenedial. As for the RH experiments, droplets contained butenedial along with PEG-6 acting as an internal standard. A. W. Birdsall et al.: Gas-particle partitioning of butenedial 14199 Figure 5. Sample extracted ion time series from measurement of a single droplet, whose mass spectrum is shown in Fig. 4. The peak intensities in the 85 and 283 m/z mass channels above the surround- ing background are used to quantify the abundance of butenedial and PEG-6 in the droplet, respectively. of 0.071 and 0.140; one Na2SO4 solution was studied with XNa2SO4 of 0.105. These experiments were performed with a humidiﬁed levitation chamber at RH 75 ± 5 %. Separate trials in a similarly designed EDB equipped with additional spectroscopic instrumentation (described in Sect. 3.2 of Krieger et al. 2018) were performed to con- ﬁrm the particles remained deliquesced at our experimen- tal humidity for the inorganic particles. The observed two- dimensional angular optical scattering pattern remained vis- ible at the experimental humidity, implying a spherical del- iquesced particle (Braun and Krieger, 2001). No change in fringe pattern was observed when drying a particle contain- ing butenedial and PEG-6 without an inorganic salt, mean- ing that no phase separation or efﬂorescence occurred in the salt-free system. In contrast, when dry nitrogen was intro- duced as the bath gas for salt-containing particles, the fringe pattern was observed to change, indicating that efﬂorescence had occurred due to the presence of the inorganic salt. Figure 5. Sample extracted ion time series from measurement of a single droplet, whose mass spectrum is shown in Fig. 4. The peak intensities in the 85 and 283 m/z mass channels above the surround- ing background are used to quantify the abundance of butenedial and PEG-6 in the droplet, respectively. 2.3.1 PEG-6 as internal standard We deemed PEG-6 to be an appropriate internal standard and major condensed-phase component for these experiments for a number of reasons. PEG-6 was used as an internal standard to account for droplet-to-droplet variability in total amount of signal measured with the mass spectrometer because it shows a clear peak at 283 m/z in the mass spectrum (MH+) and its vapor pressure is low enough for its evaporation to be neg- ligible over the timescale of the experiments. We also used PEG-6 as an internal standard in a previous set of experi- ments (Birdsall et al., 2018). Based on those previous results, the viscosity of a PEG-6 matrix is sufﬁciently low to expect no condensed-phase diffusion effects limiting the evapora- tion rate of butenedial. same calibration data using 18 µm diameter polymethyl- methacrylate (PMML) spheres as previously. The absolute accuracy of this technique is estimated at 20 %. 2.2 Droplet levitation and measurement Microdroplets were generated, levitated, and measured via mass spectrometry using the same EDB-MS apparatus as previously described (Birdsall et al., 2018). In brief, a droplet-on-demand particle generator injects a ∼140 pL droplet from an aqueous solution into the EDB levitation chamber, through a charged coil that gives the droplet charge. A combination of AC and DC electrodes, in a “dual-ring” ge- ometry, establishes an electric ﬁeld that conﬁnes the charged droplet in the center of the levitation chamber. The water con- tent from the aqueous solution rapidly equilibrates with the relative humidity of the levitation chamber (∼1 s). A sin- gle droplet is allowed to reside in the center of the levitation chamber for a deﬁned period of time while a purge ﬂow of dry or humidiﬁed nitrogen gas ﬂows through the levitation chamber (95 sccm), maintaining a ﬁxed RH in the levitation chamber and preventing the accumulation of gases evaporat- ing from the droplet. At the end of the residence time in the droplet levitation chamber, the gas ﬂow and electric ﬁeld are manipulated to eject the droplet out of the bottom of the lev- itation chamber, through a straight length of 1/4 in. (outer diameter) stainless steel tubing to the ionization region. In the ionization region, which like the levitation chamber is at ambient pressure, the droplet impacts a glass slide mounted One change from the previously described setup was to sample the mass spectrometer at a faster rate of 3 Hz, com- pared to previous sampling at 1 Hz. Each mass spectrum had sufﬁcient signal, with better temporal sampling of the droplet signal peak. With the current set of experiments, we observed less particle-to-particle variability in the normalized analyte signal compared to previously (Birdsall et al., 2018). This could potentially be explained by our previous technique un- dersampling the peak shape. As previously described (Birdsall et al., 2018), droplet sizes were characterized after initial trapping and before the ejection procedure using a spring point technique, which relies on a quantitative relationship of the visible onset of droplet instability (the “spring point”) in the EDB with the voltage conditions and the droplet diameter (Davis, 1985). The droplet diameters determined for this study used the Atmos. Chem. Phys., 19, 14195–14209, 2019 www.atmos-chem-phys.net/19/14195/2019/ A. W. Birdsall et al.: Gas-particle partitioning of butenedial (1): model numerically integrates a differential equation describ- ing evaporation of droplet components via Maxwellian ﬂux (Seinfeld and Pandis, 2006), as in Eq. (1): bond correlation (HMBC) experiment of the precursor PEG- 6–butenedial aqueous solution consistent with coupling be- tween the butenedial and PEG-6 moieties of a putative hemi- acetal, as would be expected. dni dt = 4πrDgi(c∞i −csi), (1) (1) To check whether salting out of PEG-6 would affect the extracted butenedial Pvap,eff in the inorganic experiments, additional evaporation model runs (see Sect. 2.4) were per- formed in which the model representation of PEG-6 had a vapor pressure 10 times larger than the literature value. This was used as a conservative upper bound of a salting-out ef- fect. With this higher vapor pressure of PEG-6, evaporation of PEG-6 still proceeded slowly enough over experimental timescales such that there was no discernible difference in the extracted butenedial Pvap,eff (< 1 % change). in which r is the particle radius, Dgi is the gas-phase diffu- sion constant of species i, c∞i is the gas-phase concentra- tion of species i at inﬁnite distance from the particle surface (taken to be zero in the EDB), and csi is the gas-phase sur- face concentration of species i. We assume the particle is an ideal mixture and in equilibrium at its surface with the gas phase, so the gas-phase surface concentration is then given by Eq. (2): in which r is the particle radius, Dgi is the gas-phase diffu- sion constant of species i, c∞i is the gas-phase concentra- tion of species i at inﬁnite distance from the particle surface (taken to be zero in the EDB), and csi is the gas-phase sur- face concentration of species i. We assume the particle is an ideal mixture and in equilibrium at its surface with the gas phase, so the gas-phase surface concentration is then given by Eq. (2): csi = Xi Pvapi,eff kT , (2) 2.4 Evaporation model and determination of Pvap,eff and KH,eff (2) (2) where Xi is the particle-phase mole fraction of species i, Pvapi,eff is the pure component vapor pressure of species i at temperature T inside the EDB, and k is the Boltzmann constant. Because the droplet surface is curved, the vapor pressure of a compound above a charged levitated droplet could conceivably be elevated due to the Kelvin effect. The particle charge also could lower the vapor pressure of a com- pound with a signiﬁcant dipole moment. However, follow- ing Sect. C1 of Huisman et al. (2013) to calculate the com- bined Kelvin and charge stabilization effects, we conclude the effects are negligible considering the particle diameter and charge for these experiments. Pvap,eff of butenedial under each experimental condition was determined by ﬁtting a kinetic model that describes the changing composition of a droplet in time to observations. To determine the uncertainty in the extracted Pvap,eff, the anal- ysis considered the uncertainty in the model input parame- ters and the uncertainty introduced by shot-to-shot noise in collected data, as detailed below. The data consist of a set of individual observations for each trapped and measured droplet, corresponding to a normalized abundance of remain- ing butenedial (relative to PEG-6) after butenedial evapora- tion has proceeded for a certain amount of time. Plotting a set of these data points for a single type of experiment shows a decay over time in the normalized butenedial signal. The in- dividual data points are binned by time spent in the EDB, and bootstrapped realizations of the binned data are compared to the kinetic model. Due to the fast equilibration of water, its evaporation is not explicitly represented. Instead, the water content is calcu- lated at each time step to preserve a ﬁxed mole fraction of wa- ter in the particle. The evaporation of PEG-6 is explicitly rep- resented using experimental physical parameters available in a review of polyethylene glycol vapor pressures (Krieger et al., 2018). However, over the experimental timescales in- volved the evaporation of PEG-6 was calculated to be negli- gible, making for our purposes PEG-6 effectively involatile. The model representations of sodium chloride and sodium 2.3 Experimental systems Additionally the droplets contained one of two inorganic compounds, either sodium chloride (NaCl) or sodium sul- fate (Na2SO4). Two NaCl solutions were studied with XNaCl Atmos. Chem. Phys., 19, 14195–14209, 2019 www.atmos-chem-phys.net/19/14195/2019/ 14200 A. W. Birdsall et al.: Gas-particle partitioning of butenedial Table 1. Parameters describing the experimental conditions used in each experiment type. These parameters and associated uncertainties were used as model inputs. Relative molar quantities of butenedial, PEG-6, and the inorganic salts were determined via the mass composition of the prepared precursor solution. The overall mole fraction of water was calculated from the experimental relative humidity (75 ± 5 % for the humid experiments and 2.5 ± 2.5 % for the dry experiments) and the solution composition via the thermodynamic model AIOMFAC (see body text). Radius measurements were obtained via a spring point technique (see body text). Provided uncertainties are 1σ values and incorporated into the model calculations as explained in the main text. Composition RH XBD XPEG−6 XNaCl XNa2SO4 XH2O r (µm) T (K) N BD + PEG-6 < 5 % 0.457 ± 0.013 0.518 0 0 0.025 ± 0.013 12.9 ± 1.5 300.10 ± 0.75 19 BD + PEG-6 75 ± 5 % 0.0960 ± 0.0028 0.109 0 0 0.795 ± 0.013 14.4 ± 1.7a 300.10 ± 0.75 9 BD + PEG-6 + NaCl (no. 1) 75 ± 5 % 0.0976 ± 0.0026 0.0514 0.071 0 0.780 ± 0.020 13.8 ± 2.1b 300.10 ± 0.75 16 BD + PEG-6 + NaCl (no. 2) 75 ± 5 % 0.0255 ± 0.0007 0.0145 0.140 0 0.820 ± 0.015 13.8 ± 2.1b 300.10 ± 0.75 17 BD + PEG-6 + Na2SO4 75 ± 5 % 0.0621 ± 0.0027 0.0528 0 0.105 0.795 ± 0.018 13.8 ± 2.1b 300.10 ± 0.75 8 a Radius distribution imputed from dry BD + PEG-6, adjusted by factor of 1.12 to account for the presence of water. b Radius distribution imputed from NaCl (no. 1) experiment. BD + PEG-6 + Na2SO4 75 ± 5 % 0.0621 ± 0.0027 0.0528 0 0.105 0.795 ± 0.018 13.8 ± 2.1b 300.10 ± 0.75 8 a Radius distribution imputed from dry BD + PEG-6, adjusted by factor of 1.12 to account for the presence of water. b Radius distribution imputed from NaCl (no. 1) experiment. a Radius distribution imputed from dry BD + PEG-6, adjusted by factor of 1.12 to account for the presence of water. b Radius distribution imputed from NaCl (no. 1) experim model numerically integrates a differential equation describ- ing evaporation of droplet components via Maxwellian ﬂux (Seinfeld and Pandis, 2006), as in Eq. 2.4.2 Parameter inputs and uncertainty All parameters aside from Pvap,eff of butenedial are con- strained by knowledge of the experimental system. These pa- rameters include the initial mole fraction of butenedial, tem- perature, RH, the gas-phase diffusivity of butenedial, and the initial droplet radius. The droplet diameter is provided by the spring-point measurements. The temperature and RH is pro- vided by a RH and temperature probe (Sensirion SHT31). Because of the different molar sensitivities of the mass spectrometer to butenedial and PEG-6, a scaling factor needed to be applied to scale the measured ratios of butene- dial and PEG-6 signal intensities to the molar ratio. The scal- ing factor was determined independently for each type of ex- periment from the data points for the droplets that had resided in the EDB for 5 min or less. Assuming negligible butene- dial evaporation over this timescale, the molar ratio for these droplets was assumed to be equal to the known starting mo- lar ratio of the precursor aqueous solutions. A bootstrapping procedure was used to estimate the mean and standard devi- ation in normalized peak intensity for the droplets that had resided in the EDB for 5 min or less. From that, along with the known molar composition of the starting precursor aque- ous solution, a mean and standard deviation in the scaling factor was determined. Physical parameters used to represent each compound are given in Table 2. Other model input parameters are given in Table 1. The gas-phase diffusivity of butenedial (along with the dihydrate) in air was estimated using the Fuller–Schettler– Giddings equation, which is based on the molecular weights and volumes of the components of a binary gas mixture. The molecular volume of butenedial was calculated based on its molecular structure using literature parameters (Welty et al., 1984). The density of butenedial is assumed to be 1.06 g m−3 based on literature densities of aldehydes butane- dial and glutaraldehyde (CRC Handbook of Chemistry and Physics, 2008). The initial molar ratio of butenedial and other non-water components of the droplet (e.g., PEG-6, sodium chloride, sodium sulfate) is known from the relative concentrations of the compounds in the precursor aqueous solution, which was prepared quantitatively. However, the overall mole fraction of butenedial in the droplet depends on the droplet’s water content. For the dry experiment the RH is measured to be 5 % or lower. A. W. Birdsall et al.: Gas-particle partitioning of butenedial the activities of water in solution and in the gas phase (i.e., RH) are equal. sulfate had vapor pressures of 0, preventing any modeled evaporation. Sodium chloride and sodium sulfate are repre- sented as being fully dissociated into two and three ions, re- spectively. The model does not include any representation of acid–base equilibrium. To calculate the mole fraction of water corresponding to the RH of the humid experiments we used the thermo- dynamic model AIOMFAC via its publicly accessible on- line interface (https://aiomfac.lab.mcgill.ca/, last access: 1 May 2019), using the known starting composition of the non-water particle components. AIOMFAC uses a functional group approach to calculate the equilibrium thermodynamic activities of a mixed inorganic–organic solution assumed to exist in a single deliquesced phase (Zuend et al., 2011, 2008). In particular, butenedial was deﬁned in terms of its functional groups in its hydrated form because our results imply that is its dominant form in our experiments (see below). The mole fraction of water in the particle, XH2O, was assumed to be ﬁxed over the entire course of the experiment, which is true if the activity coefﬁcient of water does not change appreciably as butenedial evaporates from the particle. We checked the validity of the assumption by using AIOMFAC to compare the change in calculated XH2O for the humid, inorganic-free experiment, between its initial composition and composition after all butenedial evaporated. We found the change in cal- culated XH2O to be approximately 0.005, implying the effect of this assumption on our results is negligible within other sources of uncertainty. To make the ﬁtting procedure less computationally ex- pensive, rather than repeatedly numerically integrating the differential equation we developed an analytical expres- sion that approximates a solution to Maxwellian ﬂux under certain simplifying assumptions. In particular, we assumed the droplet radius and mole fraction of water were time- invariant, and all non-butenedial, non-water droplet compo- nents did not evaporate over the experimental timescale. A description of the resulting analytical solution and ﬁtting pro- cedure, along with validation of the technique, is given in Sects. S1 and S2 in the Supplement. 2.4.1 Maxwell ﬂux description of particle evaporation The kinetic model describing evaporation was implemented in pyvap, an open-source Python package that has been pre- viously described (Birdsall et al., 2018). In brief, the kinetic Atmos. Chem. Phys., 19, 14195–14209, 2019 www.atmos-chem-phys.net/19/14195/2019/ A. W. Birdsall et al.: Gas-particle partitioning of butenedial 14201 A. W. Birdsall et al.: Gas-particle partitioning of butenedial 14202 compound used in the model. A gas phase diffusivity is not applicable (n/a) to the inorganic salts. Table 2. Physical parameters for each compound used in the model. A gas phase diffusivity is not applicable (n/a) to the inorganic salts. Table 2. Physical parameters for each compound used in the model. A gas phase diffusivity is not applicable Table 2. Physical parameters for each compound used in the model. A gas phase diffusivity is not applicable (n/a) to the inorganic salts. Compound M (kg mol−1) ρ (kg m−3) P (298 K) (Pa) Dgas (10−6 m2 s−1) Butenedial dihydrate 0.1201a 1060 See results 8.56 ± 0.86 PEG-6 0.2823 1180 3.05 × 10−5 4.26 NaCl 0.05844 2160 0b n/a Na2SO4 0.142 2660 0b n/a a Note that because all observations of butenedial in the condensed phase were consistent with existing in its dihydrated form, the model used physical parameters of butenedial dihydrate. b Because inorganic salts sodium chloride and sodium sulfate are essentially involatile, their vapor pressures were assumed to be effectively 0. a Note that because all observations of butenedial in the condensed phase were consistent with existing in its dihydrated form, the model used physical parameters of butenedial dihydrate. b Because inorganic salts sodium chloride and sodium sulfate are essentially involatile, their vapor pressures were assumed to be effectively 0. where MWavg is the weighted average of molecular weights of the chemical components of the particle and ρavg is the weighted average of densities of the chemical components of the particle. tions were assumed to match that of the NaCl no. 1 experi- ment. The resulting diameter distributions used are given in Table 1. We used a combined Monte Carlo and bootstrapping ap- proach to estimate the uncertainty in the retrieved butenedial Pvap,eff due to uncertainties in the other model input param- eters. The implementation of the uncertainty analysis is de- scribed in Sect. S3. The analysis incorporates the uncertainty associated with model input parameters, including droplet di- ameter, temperature, scaling factor, and gas-phase diffusivity of butenedial, as reported in Table 1. Using this approach, the per-experiment uncertainties due to uncertainty in model input parameters and shot-to-shot noise in the data vary by experiment and are reported in Table 3 as 95 % conﬁdence in- tervals. Compared to other uncertainties in our approach, this is likely a dominant source of uncertainty. A. W. Birdsall et al.: Gas-particle partitioning of butenedial For comparison, we previously described an uncertainty estimation technique in which we simply considered limiting cases of temperature and diameter parameters to give upper and lower bounds of Pvap,eff of various polyethylene glycols (Birdsall et al., 2018). Using that less detailed uncertainty treatment, which did not consider uncertainty due to measurement variability, the pre- vious estimated uncertainty range was between ±15 % and ±25 %. Because the dominant source of uncertainty in Eq. (4) is Pvap,eff, we derive the 95 % conﬁdence interval in KH,eff by using the same 95 % conﬁdence interval as in the correspond- ing Pvap,eff. 3.1 Humidity dependence Figure 6 shows data and the corresponding best model ﬁt for the humidity dependence butenedial evaporation exper- iments. The extracted values for the extracted Pvap,eff of butenedial under the two RH conditions are 28.1 mPa (95 % CI 13.1, 47.8 mPa) and 34.2 mPa (95 % CI 18.8, 54.9 mPa) for the dry and humid conditions, respectively. (The uncer- tainty values correspond to a 95 % conﬁdence interval de- rived from the combined Monte Carlo sampling and boot- strapping approach described in the SI and reﬂects the uncer- tainty due to uncertainty in the model input parameters and the shot-to-shot noise in the data.) We interpret these results to imply Pvap,eff under the dry conditions and Pvap,eff under the humid conditions are indistinguishable, within the uncer- tainties in our measurements and ﬁtting procedure. Note the extracted Pvap,eff values are similar despite the differences in evaporation timescales. This is a consequence of the evap- oration rate scaling with the concentration of butenedial in the particle in Eq. (2). Because the mole fraction of butene- dial decreases as water content increases, under humid con- ditions butenedial evaporation is expected to proceed more slowly even if Pvap,eff is unchanged. This previously has been referred to as the “Raoult’s Law effect” (Prisle et al., 2010). 2.4.2 Parameter inputs and uncertainty Though the possible presence of water in these low amounts are accounted for in our uncertainty analysis (see below), we ﬁnd under these conditions the presence of water has a negligible effect on accurately predicting the mole frac- tion of butenedial. In contrast for the humid experiments the water content is a non-negligible contribution to the overall molar composition. When equilibrated with the surrounding gas phase, the droplet water concentration will be such that Spring-point-derived radii were available for the dry ex- periment (16 radii, mean 12.9 µm, and standard deviation 1.5 µm) and NaCl no. 1 experiment (16 radii, mean 13.8 µm, and standard deviation 2.1 µm). Spring-point-derived sizes were not available for the humid (inorganic-free), NaCl no. 2, and Na2SO4 experiments. However, due to the similarity in droplet behavior (e.g., response to ﬂow and voltage condi- tions during particle transfer) the distribution of droplet di- ameters for those experiments could be imputed from the available measurements. The humid experiment diameter distribution was obtained by scaling the dry experiment di- ameter distribution by a factor of 1.115 to account for the calculated increase in diameter due to the presence of water. The NaCl no. 2 and Na2SO4 experiment diameter distribu- www.atmos-chem-phys.net/19/14195/2019/ Atmos. Chem. Phys., 19, 14195–14209, 2019 www.atmos-chem-phys.net/19/14195/2019/ A. W. Birdsall et al.: Gas-particle partitioning of butenedial 14203 Table 3. Extracted effective vapor pressures Pvap,eff for butenedial (BD) with 95 % conﬁdence interval bounds (given within parentheses) obtained from uncertainties in model input parameters described in the main text, along with effective Henry’s law constants KH,eff and Setschenow constants KS, where applicable. Uncertainties in KH,eff and KS derived from propagating uncertainties in associated Pvap,eff values. Values not applicable to particular compositions are marked n/a. Table 3. Extracted effective vapor pressures Pvap,eff for butenedial (BD) with 95 % conﬁdence interval bounds (given within parentheses) obtained from uncertainties in model input parameters described in the main text, along with effective Henry’s law constants KH,eff and Setschenow constants KS, where applicable. Uncertainties in KH,eff and KS derived from propagating uncertainties in associated Pvap,eff values. Values not applicable to particular compositions are marked n/a. Composition RH Ionic Pvap,eff KH,eff KS strength (M) (BD, 300 K) (mPa) (107 M atm−1) (m−1) BD + PEG-6 < 5 % n/a 28.1 (13.1, 47.8) n/a n/a BD + PEG-6 75 ± 5 % n/a 34.2 (18.8, 54.9) 6.0 (3.7, 11) n/a BD + PEG-6 + NaCl (no. 1) 75 ± 5 % 5.3 66 (36, 105) 3.1 (2.0, 5.7) +0.056 (0.012, 0.16) BD + PEG-6 + NaCl (no. 2) 75 ± 5 % 9.6 169 (71, 301) 1.2 (0.68, 2.9) +0.074 (0.047, 0.15) BD + PEG-6 + Na2SO4 75 ± 5 % 21.0 177 (64, 376) 1.2 (0.55, 3.2) +0.096 (0.056, 0.21) nolle et al., 2011) (accessed through UManSysProp; Top- ping et al., 2016) predicts a vapor pressure for butenedial of 350 Pa at 298 K, and the Nannoolal et al. method (Nan- noolal et al., 2008) (also accessed through UManSysProp) predicts an even higher vapor pressure of 1.5 kPa at 298 K. If butenedial were primarily in the form of the dialdehyde, we would expect to observe a much faster rate of evaporation than was actually observed. In contrast, using the dihydrated form of butenedial, EVAPORATION predicts a vapor pres- sure of 2.2 mPa at 298 K, which is almost within the same order of magnitude as the measured butenedial Pvap,eff. The Nannoolal et al. method predicts a very low vapor pressure of 2.5 × 10−4 mPa at 298 K for the butenedial dihydrate. dial completely decompose or fragment during the vaporiza- tion and ionization process into butenedial monomers that are detected at the butenedial ion parent ion value of 85 m/z. A. W. Birdsall et al.: Gas-particle partitioning of butenedial The likelihood of complete oligomer decomposition appears low. A recent study of different classes of model oligomers reported ether oligomers were detected intact via mass spec- trometry after thermal desorption at ∼160 ◦C (Claﬂin and Ziemann, 2019). Because we have no observational evidence for oligomer formation, in contrast to our observation of ex- tensive butenedial hydration in NMR spectra, we conclude hydration of butenedial is the more likely explanation for the observation of a relatively low Pvap,eff. Finally, as elucidated in Sect. 2.3.1, we observe no evidence for the formation of cross-products of butenedial reacting with PEG-6 that would reduce the Pvap,eff of butenedial. y Both EVAPORATION and the Nannoolal et al. method use a group contribution approach; the extremely low value of the Nannoolal et al. method prediction may reﬂect the limitations of its group contribution parameterization for this chemical structure. Additionally, in comparison with experi- mental vapor pressures, EVAPORATION has been reported to show a trend of underestimating vapor pressures at low vapor pressure (O’Meara et al., 2014). This trend could help account for the discrepancy between our measured Pvap,eff and the EVAPORATION model estimate, though we cannot rule out the possibility that the discrepancy between mea- sured Pvap,eff and predicted butenedial dihydrate vapor pres- sure could possibly reﬂect a contribution to Pvap,eff by the evaporation of butenedial monohydrate present in the parti- cle. EVAPORATION predicts the monohydrate to have a va- por pressure of 2.3 Pa, whereas the Nannoolal et al. method predicts a vapor pressure of 220 mPa. However, the lack of any signiﬁcant aldehyde peak in the NMR spectrum suggests equilibrium strongly favors the dihydrate form of butenedial under humid conditions. p Furthermore the fact that the two Pvap,eff are indistin- guishable, at least within the estimated uncertainty, implies butenedial primarily exists in a hydrated form not only un- der conditions with a high water content, but also under con- ditions with lower water content. This observation is also consistent with the collected NMR spectra of butenedial, in which only hydrated butenedial peaks are observed, even un- der conditions with a lower residual water content. A. W. Birdsall et al.: Gas-particle partitioning of butenedial There may be to some degree a shift in equilibrium between the hydrated and non-hydrated forms of butenedial under the different RH conditions, which could lead to a change in Pvap,eff, but our results imply any change would correspond to a change in Pvap,eff less than our uncertainty in extracted Pvap,eff. p A possible mechanistic explanation for this observation could be due to slow kinetics of dehydration under dry con- ditions. For all experiments, regardless of RH, the particle was initially generated from an aqueous solution in which equilibrium appears to strongly favor butenedial dihydrate. When the droplet is injected into the EDB under dry condi- tions, the water content of the particle quickly (∼1 s) equili- brates. Within the dry particle, butenedial may be kinetically frozen over experimental timescales in the dihydrate form, even if the dialdehyde becomes more thermodynamically fa- The low observed Pvap,eff of butenedial could also con- ceivably reﬂect a contribution from formation of butenedial oligomers, which would similarly lead to a lower Pvap,eff. However, we see no evidence for oligomer formation in our mass spectra or NMR spectrum. For oligomer formation to be consistent with our observations, it would have to be the case that ether oligomers hypothetically formed from butene- Atmos. Chem. Phys., 19, 14195–14209, 2019 2.4.3 Effective Henry’s law constant Using the extracted Pvap,eff along with the calculated molar- ity of butenedial, we estimated an effective Henry’s law con- stant KH,eff for all humid experiments. The butenedial mo- larity was calculated using the water content calculated with the thermodynamic model AIOMFAC (Sect. 2.4.2) and as- suming the overall molecular weight and density of the par- ticle was the weighted average of the molecular weights and densities of its chemical components, respectively. The full expression is then given by Eq. (3): KH,eff = XBDMW−1 avgρavg XBDPvap,eff (3) One implication of the measured effective vapor pressures is that considering the hydration of butenedial is crucial for a correct prediction of its Pvap,eff and hence its gas–particle partitioning behavior. If the butenedial in the particle primar- ily consisted of butenedial in its dialdehyde form, the ex- pected Pvap,eff would be expected to be orders of magnitude larger. For example, the EVAPORATION model (Comper- (3) or after simpliﬁcation KH,eff = MW−1 avgρavg Pvap,eff , (4) KH,eff = MW−1 avgρavg Pvap,eff , (4) Atmos. Chem. Phys., 19, 14195–14209, 2019 Atmos. Chem. Phys., 19, 14195–14209, 2019 www.atmos-chem-phys.net/19/14195/2019/ Atmos. Chem. Phys., 19, 14195–14209, 2019 A. W. Birdsall et al.: Gas-particle partitioning of butenedial A. W. Birdsall et al.: Gas-particle partitioning of butenedial 14204 A. W. Birdsall et al.: Gas-particle partitioning of butenedial Figure 6. Experiments used to determine the effective vapor pressure of butenedial (BD) in a droplet also containing hexaethylene glycol (PEG-6), under dry (RH < 5 %) and humid (RH 75 ± 5 %) conditions. Small pink points are observations of individual droplets and the line is the best model ﬁt, as described in the text. Large points are the mean binned values from the bootstrapping procedure, with x error bars showing the width of the bins and y error bars representing a 95 % conﬁdence interval of the mean value in the bin, over all bootstrapped realizations. The plotted model ﬁt is for the mean vapor pressure obtained by averaging over all model ﬁts to all realizations. The vapor pressure is reported as a 95 % conﬁdence interval over repeating the ﬁtting procedure over 10 000 bootstrapped realizations. Figure 6. Experiments used to determine the effective vapor pressure of butenedial (BD) in a droplet also containing hexaethylene glycol (PEG-6), under dry (RH < 5 %) and humid (RH 75 ± 5 %) conditions. Small pink points are observations of individual droplets and the line is the best model ﬁt, as described in the text. Large points are the mean binned values from the bootstrapping procedure, with x error bars showing the width of the bins and y error bars representing a 95 % conﬁdence interval of the mean value in the bin, over all bootstrapped realizations. The plotted model ﬁt is for the mean vapor pressure obtained by averaging over all model ﬁts to all realizations. The vapor pressure is reported as a 95 % conﬁdence interval over repeating the ﬁtting procedure over 10 000 bootstrapped realizations. vorable. Additional experiments would be necessary to test this explanation. gests the direction of salting in and salting out for a dicar- bonyl compound may be inﬂuenced by the extent to which it contains hydrophobic regions, represented in butenedial by the alkenyl group and in methylglyoxal by the methyl group. Using Eq. (4) we calculated KH,eff of butenedial to be 6.0 (95 % CI 3.7, 11) ×107 M atm−1 in the humid, inorganic- free experiment. (Uncertainties in KH,eff arise from propagat- ing uncertainties in Pvap,eff, given the relationship in Eq. 4). A. W. Birdsall et al.: Gas-particle partitioning of butenedial For comparison, KH,eff of glyoxal has been previously mea- sured to be 4.19×105 M atm−1 in an inorganic-free aqueous phase (Ip et al., 2009). The magnitude of the measured KH,eff for butenedial compared to glyoxal suggests butenedial may have a strong tendency to partition into an available aqueous phase, ignoring the effect of inorganic compounds. Calculated KH,eff values for the three inorganic salt- containing experiments, again using Eq. (4), are given in Ta- ble 3. Furthermore the magnitude of the salting-out effect can be quantiﬁed using the Setschenow coefﬁcient KS, using the formulation of Waxman et al. (2015) in Eq. (5), where KS = 1 cslog KH,w KH,s (5) KS = 1 cslog KH,w KH,s (5) and cs is the condensed-phase salt concentration (using mo- lality, m), KH,w is the effective Henry’s law constant in the absence of the inorganic salt, and KH,s is the effective Henry’s law constant in the presence of the inorganic salt. The calculated values of KS using Eq. (5) for the three in- organic experiments are given in Table 3, with uncertainties derived from propagating the uncertainties in the effective Henry’s law constants. The KS values for the two NaCl ex- periments (+0.056 (95 % CI 0.012, 0.16), +0.074 (95 % CI 0.047, 0.15)) have similar values within our uncertainties. KS for sodium sulfate (+0.096 (95 % CI 0.056, 0.21)) is some- what larger than the KS for sodium chloride in our exper- iments, though the 95 % conﬁdence intervals largely over- lap. Interestingly, this is a similar trend to that previously reported for methylglyoxal in the presence of the same two inorganic salts, of 0.06 and 0.16 m−1, respectively (Waxman et al., 2015). www.atmos-chem-phys.net/19/14195/2019/ 14204 3.2 Inorganic salt dependence The plotted model ﬁt is for the mean vapor pressure obtained by averaging over all model ﬁts to all realizations. The vapor pressure is reported as a 95 % conﬁdence interval over repeating the ﬁtting procedure over 10 000 bootstrapped realizations. Figure 7. Experiments used to determine the effective vapor pressure of butenedial (BD) in a droplet also containing hexaethylene glycol (PEG-6) and either sodium chloride (NaCl) or sodium sulfate (Na2SO4), under humid (RH 75 ± 5 %) conditions. Small pink points are observations of individual droplets and the line is the best model ﬁt, as described in the text. Large points are the mean binned values from the bootstrapping procedure, with x-axis error bars showing the width of the bins and y-axis error bars representing a 95 % conﬁdence interval of the mean value in the bin, over all bootstrapped realizations. The plotted model ﬁt is for the mean vapor pressure obtained by averaging over all model ﬁts to all realizations. The vapor pressure is reported as a 95 % conﬁdence interval over repeating the ﬁtting procedure over 10 000 bootstrapped realizations. its higher mole fraction concentration. However, thermody- namic calculations performed in AIOMFAC suggest this be- havior is unlikely. Though a more extensive set of calcula- tions would be necessary to rigorously check for the pres- ence of partial phase separation, calculations performed us- ing AIOMFAC assuming a single well-mixed phase under the experimental conditions show the activities of the organic components remain well below 1 without extremely large (i.e., ≫1) activity coefﬁcients. These calculations are con- sistent with the droplets consisting of a single mixed phase under these experimental conditions. ulation mean (i.e., droplets measured using the spring point technique to be smaller than average did not have their ana- lyte molecule evaporate more quickly) (Birdsall et al., 2018). This suggests that though there is considerable uncertainty in our spring point measurements, it would not be helpful to constrain the model based on the measured diameter of each individual droplet. As a secondary effect, uncertainty in water content of the droplets as a calculated parameter may also contribute uncer- tainty to our Pvap,eff determination. 3.2 Inorganic salt dependence Figure 7 shows data for inorganic salt dependence experi- ments. For both the sodium chloride and sodium sulfate ex- periments, the extracted butenedial Pvap,eff is larger than the Pvap,eff measured in the organic-only cases, with values of 66 mPa (95 % CI 36, 105 mPa) and 169 mPa (95 % CI 71, 301 mPa) for the XNaCl of 0.071 and 0.140 sodium chlo- ride experiments, respectively, and 177 mPa (95 % CI 64, 376 mPa) for the sodium sulfate experiment. The fact that the Pvap,eff for butenedial becomes higher in solutions con- taining both inorganic salts, by up to a factor of 5 under our experimental conditions, implies the inorganic salts in this case have a salting-out effect. Interestingly, previous work with a different dialdehyde, glyoxal, demonstrated that the presence of both sodium chloride and sodium sulfate led to a shift in the hydration equilibrium favoring the formation of the hydrate, implying a salting-in effect (Waxman et al., 2015; Yu et al., 2011). In contrast, the three-carbon dicar- bonyl methylglyoxal has been measured to exhibit a salting- out effect for both sodium chloride and sodium sulfate (Wax- man et al., 2015). The fact that salting-out effects have now been observed for both methylglyoxal and butenedial sug- A competing hypothesis to potentially explain the mea- sured higher Pvap,eff in the mixed organic–inorganic droplets could be phase separation. If the presence of the inorganic components led to the formation of two different condensed phases with butenedial predominantly in an organic-rich phase, the Pvap,eff of butenedial would be higher due to Atmos. Chem. Phys., 19, 14195–14209, 2019 www.atmos-chem-phys.net/19/14195/2019/ A. W. Birdsall et al.: Gas-particle partitioning of butenedial 14205 A. W. Birdsall et al.: Gas-particle partitioning of butenedial 14205 Figure 7. Experiments used to determine the effective vapor pressure of butenedial (BD) in a droplet also containing hexaethylene glycol (PEG-6) and either sodium chloride (NaCl) or sodium sulfate (Na2SO4), under humid (RH 75 ± 5 %) conditions. Small pink points are observations of individual droplets and the line is the best model ﬁt, as described in the text. Large points are the mean binned values from the bootstrapping procedure, with x-axis error bars showing the width of the bins and y-axis error bars representing a 95 % conﬁdence interval of the mean value in the bin, over all bootstrapped realizations. 3.2 Inorganic salt dependence Our representation of the uncertainty in the mole fraction of water reﬂects solely un- certainty in the relative humidity measurements upon which the model representation is based; we do not include a repre- sentation of the accuracy of the AIOMFAC model calculation used to determine the corresponding mole fraction of water itself. The water content is not used to constrain the droplet diameter (which is constrained by the spring point method, above), but it does have a signiﬁcant effect on determining the overall mole fraction of butenedial in the droplet. For ex- ample, the mean measured RH for the humid experiments was 75 %, but based on AIOMFAC model calculations the mean mole fraction of water can be up to 0.82, depending on the hygroscopicity of the solution. Because the butene- dial is a ﬁxed mole fraction of the non-water portion of the solution, for the same set of measurements a difference in assumed mole fraction of water between 0.75 and 0.82 cor- responds to a difference in extracted butenedial Pvap,eff by a factor of (0.18 −0.25)/0.25, or almost 30 %. Because the true uncertainty in the AIOMFAC activity correction is likely smaller than the difference between the ideal and calculated nonideal cases, 30 % likely represents a highly conservative upper bound on uncertainty arising from the AIOMFAC wa- ter mole fraction calculation. 3.3 Sources of uncertainty The dominant source of uncertainty in our Pvap,eff extraction procedure is uncertainty in the droplet diameter. The uncer- tainty in the measured diameter for a population of particles is estimated to be up to a factor of 50 % about the mean, which has a particularly large effect because vapor pressure scales with surface area. Using the spring point technique to measure the starting diameter of each trapped droplet has an inherent uncertainty. Additionally there is some degree in droplet-to-droplet variability in starting diameter for each population of droplets. Our modeling approach treats the di- ameter of a population of droplets for a single experiment type as a single mean value, with an associated standard de- viation. In part we treated the data this way because spring point-derived droplet diameters were not available for every droplet included in this analysis, though consistency in col- lected data across the data set implied each set of droplets could be described in terms of a single distribution of diame- ters. Additionally, previous analysis of a data set using spring point diameter measurements did not observe a correlation between residuals of the remaining analyte and the mea- sured starting diameter for that particle, relative to the pop- Atmos. Chem. Phys., 19, 14195–14209, 2019 www.atmos-chem-phys.net/19/14195/2019/ Atmos. Chem. Phys., 19, 14195–14209, 2019 4 Conclusions being kinetically frozen as a hydrate over the experimen- tal timescales of tens of minutes to hours. If this is indeed the case, the ability of our EDB-MS approach to determine butenedial’s Pvap,eff is a consequence of butenedial hydrate evaporating over a timescale during which the dehydration reaction is negligible. This analysis points to the care that needs to be taken when considering the behavior of atmo- spheric aerosol. If using an experimental approach with ap- preciably different conditions from those reported here (e.g., temperature or butenedial concentration), the rates of the competing evaporation and dehydration pathways may ex- hibit different scaling dependencies and hence lead to a dif- ferent conclusion about butenedial’s behavior. One attractive quality of EDB-MS is it allows monitoring of the evolution of a particle’s composition over the multiday lifetime of an at- mospheric aerosol particle. It is not necessary to use extreme experimental conditions to simulate some aspect of multiple days of atmospheric lifetime in a shorter period of time. A set of laboratory experiments have used an electrodynamic balance–mass spectrometry (EDB-MS) technique to measure the evaporation of butenedial from organic-only and mixed organic–inorganic levitated droplets under dry and humid conditions. With this setup the speciﬁc process of interest, gas–particle partitioning, was studied in isolation in an envi- ronment in which a single particle of known composition is exposed to a continually refreshed bath gas of pure dry or hu- midiﬁed nitrogen. This approach simpliﬁes the measurement compared to performing an experiment in an environmental chamber, which will have a more complex gas-phase matrix, because no gas-phase measurement is required and no pro- cess other than evaporation needs to be modeled. We measured the effective vapor pressure (Pvap,eff) of butenedial, under both low (RH < 5 %) and higher (RH 70 %) humidity conditions, to be approximately 30 mPa, which is 4 orders of magnitude lower than the expected va- por pressure of a four-carbon dialdehyde. This result implies butenedial exists primarily in a hydrated form, across a wide range of RH conditions, and the gas–particle partitioning of butenedial in ambient particles favors the particle phase more strongly due to butenedial’s hydration. The importance of hy- dration reactions in affecting gas–particle partitioning is con- sistent with previous work studying atmospherically relevant aldehydes, most notably glyoxal. 4 Conclusions Observation of a higher Pvap,eff of butenedial in the pres- ence of high concentrations of inorganic salts sodium chlo- ride or sodium sulfate imply a salting-out effect that in- creased Pvap,eff of butenedial by up to a factor of approxi- mately 6 to 8 under the most concentrated inorganic condi- tions used in this experiment. In ambient particles the magni- tude of the effect is expected to depend on the inorganic ion concentration, as described by the Setschenow coefﬁcients. In general, the magnitude of the observed Setschenow coef- ﬁcients is such that only with inorganic ion concentrations on the scale of > 1 M is a signiﬁcant effect predicted. The salting-out effect is predicted to be negligible with inorganic ion concentrations on the order of 100 µM to 1 mM. Conse- quently, the measured salting-out effect is predicted to have an inﬂuence on aqueous aerosol particles but not on cloud water. The magnitude and sign of the salting-in or salting- out effect is also known to depend on the identity of the ion, so further study would be necessary to investigate the effect of other inorganic ions. In general, better understanding of factors inﬂuencing gas–particle partitioning with this level of chemical detail improves predictions of the composition and fate of organic aerosol and its chemical constituents. These results emphasize the importance of considering the gas–particle partitioning of atmospheric compounds based on their actual chemical form in the condensed phase, rather than their pure form in isolation. For the case of butenedial the current results suggest butenedial’s Pvap,eff can be repre- sented by a single value across all ambient RH. The formation of intramolecular hydrogen bonds by hy- drated butenedial may have a substantive effect on its volatil- ity. Based on the structure of butenedial hydrate, the role of cis-trans isomerism is expected to play a role, with the cis form of butenedial dihydrate more likely to be able to form intramolecular hydrogen bonds and therefore demonstrate a high vapor pressure, compared to the trans form. Though the isomeric form of the precursor (cis) combined with the syn- thetic mechanism suggest synthesis of purely cis butenedial, evidence from our NMR spectra does not support this con- clusion, instead suggesting our experiments were performed with a mixture of cis and trans isomers. www.atmos-chem-phys.net/19/14195/2019/ 14206 A. W. Birdsall et al.: Gas-particle partitioning of butenedial Atmos. Chem. Phys., 19, 14195–14209, 2019 Supplement. The supplement related to this article is available on- line at: https://doi.org/10.5194/acp-19-14195-2019-supplement. Boucher, O., Randall, D., Artaxo, P., Bretherton, C., Feingold, G., Forster, P., Kerminen, V.-M., Kondo, Y., Liao, H., Lohmann, U., Rasch, P., Satheesh, S. K., Sherwood, S., Stevens, B., and Zhang, X. Y.: Clouds and aerosols, in: Climate Change 2013: The Phys- ical Science Basis, Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change, Cambridge University Press, Cambridge, United King- dom, 1552 pp., 2013. Author contributions. AWB, JCH, and FNK designed the experi- ments. AWB and JCH performed the laboratory experiments and analyzed the data. PK and AJH developed the electrodynamic bal- ance used to assess droplet efﬂorescence behavior. JCH developed the analytical solution approximating droplet evaporation and ﬁt- ting procedure. AWB prepared the paper with contributions from all coauthors. Rasch, P., Satheesh, S. K., Sherwood, S., Stevens, B., and Zhang, Braun, C. and Krieger, U. K.: Two-dimensional angular light- scattering in aqueous NaCl single aerosol particles during deliquescence and efﬂorescence, Opt. Express, 8, 314–321, https://doi.org/10.1364/OE.8.000314, 2001. Competing interests. The authors declare that they have no conﬂict of interest. Claﬂin, M. S. and Ziemann, P. J.: Thermal desorption behavior of hemiacetal, acetal, ether, and ester oligomers, Aerosol Sci. Tech., 53, 473–484, https://doi.org/10.1080/02786826.2019.1576853, 2019. Acknowledgements. The authors thank Ulrich Krieger and Steven Wofsy for useful discussions. Cohen, A. J., Brauer, M., Burnett, R., Anderson, H. R., Frostad, J., Estep, K., Balakrishnan, K., Brunekreef, B., Dandona, L., Dan- dona, R., Feigin, V., Freedman, G., Hubbell, B., Jobling, A., Kan, H., Knibbs, L., Liu, Y., Martin, R., Morawska, L., Pope, C. A., Shin, H., Straif, K., Shaddick, G., Thomas, M., van Dingenen, R., van Donkelaar, A., Vos, T., Murray, C. J. L., and Forouzanfar, M. H.: Estimates and 25-year trends of the global burden of disease attributable to ambient air pollution: An analysis of data from the Global Burden of Diseases Study 2015, Lancet, 389, 1907–1918, https://doi.org/10.1016/S0140-6736(17)30505-6, 2017. Financial support. This research has been supported by the National Science Foundation, Division of Chemistry (grant no. 1808084), the National Science Foundation, Division of Grad- uate Education (grant nos. 1144152 and 1745303), and the Harvard University Faculty of Arts and Sciences Dean’s Competitive Fund for Promising Scholarship. Compernolle, S., Ceulemans, K., and Müller, J.-F.: EVAPO- RATION: a new vapour pressure estimation methodfor or- ganic molecules including non-additivity and intramolec- ular interactions, Atmos. Chem. Phys., 11, 9431–9450, https://doi.org/10.5194/acp-11-9431-2011, 2011. Review statement. Supplement. The supplement related to this article is available on- line at: https://doi.org/10.5194/acp-19-14195-2019-supplement. This paper was edited by Ryan Sullivan and re- viewed by two anonymous referees. CRC Handbook of Chemistry and Physics: A Ready-Reference of Chemical and Physical Data, 89th ed., edited by: Lide, D. R., National Institute of Standards and Technology, CRC Press LLC: Boca Raton, 2736 pp, ISBN 9781420066791, 2008. References Avenati, M. and Vogel, P.: Face selectivity of the Diels-Alder additions of 2-substituted 5,6-bis((E)- chloromethylidene)Bicyclo[2.2.2]Octanes, Helv. Chim. Acta., 65, 204–216, https://doi.org/10.1002/hlca.19820650119, 1982. Curry, L. A., Tsui, W. G., and McNeill, V. F.: Technical note: Up- dated parameterization of the reactive uptake of glyoxal and methylglyoxal by atmospheric aerosols and cloud droplets, At- mos. Chem. Phys., 18, 9823–9830, https://doi.org/10.5194/acp- 18-9823-2018, 2018. Bilde, M., Barsanti, K., Booth, M., Cappa, C. D., Donahue, N. M., Emanuelsson, E. U., McFiggans, G., Krieger, U. K., Marcolli, C., Topping, D., Ziemann, P., Barley, M., Clegg, S., Dennis-Smither, B., Hallquist, M., Hallquist, Å. M., Khlystov, A., Kulmala, M., Mogensen, D., Percival, C. J., Pope, F., Reid, J. P., Silva, M. A. V. R., Rosenoern, T., Salo, K., Soonsin, V. P., Yli-Juuti, T., Prisle, N. L., Pagels, J., Rarey, J., Zardini, A. A., and Riipinen, I.: Saturation vapor pressures and transition enthalpies of low- volatility organic molecules of atmospheric relevance: From di- Davis, E. J.: Electrodynamic balance stability characteristics and applications to the study of aerocolloidal particles, Langmuir, 1, 379–387, https://doi.org/10.1021/la00063a021, 1985. Dumdei, B. E. and O’Brien, R. J.: Toluene degradation prod- ucts in simulated atmospheric conditions, Nature, 311, 248–250, https://doi.org/10.1038/311248a0, 1984. Atmos. Chem. Phys., 19, 14195–14209, 2019 A. W. Birdsall et al.: Gas-particle partitioning of butenedial 14207 A. W. Birdsall et al.: Gas-particle partitioning of butenedial carboxylic acids to complex mixtures, Chem. Rev., 115, 4115– 4156, https://doi.org/10.1021/cr5005502, 2015. carboxylic acids to complex mixtures, Chem. Rev., 115, 4115– 4156, https://doi.org/10.1021/cr5005502, 2015. damental factors inﬂuencing gas–particle partitioning in at- mospheric aerosol. Birdsall, A. W. and Elrod, M. J.: Comprehensive NO-dependent study of the products of the oxidation of atmospherically rele- vant aromatic compounds, J. Phys. Chem. A, 115, 5397–5407, https://doi.org/10.1021/jp2010327, 2011. Code and data availability. The Python package pyvap used as the kinetic model of droplet evaporation is available at https://github. com/awbirdsall/pyvap (last access: 6 October 2019, Birdsall, 2019). Data used to generate paper ﬁgures are available upon request. Birdsall, A. W., Krieger, U. K., and Keutsch, F. N.: Electrodynamic balance–mass spectrometry of single particles as a new platform for atmospheric chemistry research, Atmos. Meas. Tech., 11, 33– 47, https://doi.org/10.5194/amt-11-33-2018, 2018. Birdsall, A. W.: pyvap: Kinetic model of particle evaporation, avail- able at: https://github.com/awbirdsall/pyvap, last access: 6 Octo- ber 2019. Supplement. The supplement related to this article is available on- line at: https://doi.org/10.5194/acp-19-14195-2019-supplement. 4 Conclusions However, we do not observe a double exponential shape to our evaporation data, which could imply either the vapor pressures of the cis or trans isomers do not have appreciably different vapor pres- sures, or the difference in evaporation rates is obscured by the noise in our data. Future studies on a wider set of com- pounds may help illuminate the effect. This work also helps inform the design of future exper- imental work using EDB-MS instrumentation. A measure- ment of particle diameter with lower uncertainty than the spring point technique would meaningfully reduce the un- certainty in extracted effective vapor pressures, particularly with a continuous diameter measurement. Other research with EDBs has demonstrated the utility of optical sizing tech- niques for performing this measurement (Gorkowski et al., 2018; Haddrell et al., 2012; Hargreaves et al., 2010; Marsh et al., 2017; Steimer et al., 2015; Zardini et al., 2006). In sum, this work uses an EDB-MS laboratory technique to better understand the dependence of a model organic com- pound’s effective vapor pressure on particle composition, and hence help contribute to an improved understanding of fun- The fact that Pvap,eff is invariant between the two RH conditions may imply in these experiments butenedial re- mained in its hydrated form even under dry conditions with little condensed-phase water. This may be due to butenedial www.atmos-chem-phys.net/19/14195/2019/ Atmos. Chem. Phys., 19, 14195–14209, 2019 A. W. Birdsall et al.: Gas-particle partitioning of butenedial J., Bilde, M., and Donahue, N. M.: Humidity inﬂuence on gas-particle phase partitioning of α- pinene + O3 secondary organic aerosol, Geophys. Res. Lett., 37, L01802, https://doi.org/10.1029/2009GL041402, 2010. Hodzic, A., Kasibhatla, P. S., Jo, D. S., Cappa, C. D., Jimenez, J. L., Madronich, S., and Park, R. J.: Rethinking the global sec- ondary organic aerosol (SOA) budget: stronger production, faster removal, shorter lifetime, Atmos. Chem. Phys., 16, 7917–7941, https://doi.org/10.5194/acp-16-7917-2016, 2016. Pye, H. O. T. and Seinfeld, J. H.: A global perspective on aerosol from low-volatility organic compounds, Atmos. Chem. Phys., 10, 4377–4401, https://doi.org/10.5194/acp-10-4377-2010, 2010. Huisman, A. J., Krieger, U. K., Zuend, A., Marcolli, C., and Peter, T.: Vapor pressures of substituted polycarboxylic acids are much lower than previously reported, Atmos. Chem. Phys., 13, 6647– 6662, https://doi.org/10.5194/acp-13-6647-2013, 2013. Seinfeld, J. H. and Pandis, S. N.: Mass transfer aspects of atmo- spheric chemistry, in: Atmospheric Chemistry and Physics, 2nd ed., John Wiley & Sons, Inc., Hoboken, USA, 537–587, 2006. Shepson, P. B., Edney, E. O., and Corse, E. W.: Ring fragmentation reactions in the photooxidations of toluene and o-xylene, J. Phys. Chem., 88, 4122–4126, https://doi.org/10.1021/j150662a053, 1984. Ip, H. S. S., Huang, X. H. H., and Yu, J. Z.: Effective Henry’s law constants of glyoxal, glyoxylic acid, and glycolic acid, Geophys. Res. Lett., 36, L01802, https://doi.org/10.1029/2008GL036212, 2009. Steimer, S. S., Krieger, U. K., Te, Y.-F., Lienhard, D. M., Huisman, A. J., Luo, B. P., Ammann, M., and Peter, T.: Electrodynamic balance measurements of thermodynamic, kinetic, and optical aerosol properties inaccessible to bulk methods, Atmos. Meas. Tech., 8, 2397–2408, https://doi.org/10.5194/amt-8-2397-2015, 2015. Jacobs, M. I., Davies, J. F., Lee, L., Davis, R. D., Houle, F., and Wilson, K. R.: Exploring chemistry in microcom- partments using guided droplet collisions in a branched quadrupole trap coupled to a single droplet, paper spray mass spectrometer, Anal. Chem., 89, 12511–12519, https://doi.org/10.1021/acs.analchem.7b03704, 2017. Toivola, M., Prisle, N. L., Elm, J., Waxman, E. M., Volka- mer, R., and Kurtén, T.: Can COSMOTherm predict a salting in effect?, J. Phys. Chem. A, 121, 6288–6295, https://doi.org/10.1021/acs.jpca.7b04847, 2017. Knote, C., Hodzic, A., and Jimenez, J. L.: The effect of dry and wet deposition of condensable vapors on secondary organic aerosols concentrations over the continental US, Atmos. Chem. Phys., 15, 1–18, https://doi.org/10.5194/acp-15-1-2015, 2015. Topping, D., Barley, M., Bane, M. K., Higham, N., Aumont, B., Dingle, N., and McFiggans, G.: UManSysProp v1.0: an online and open-source facility for molecular property prediction and atmospheric aerosol calculations, Geosci. A. W. Birdsall et al.: Gas-particle partitioning of butenedial 14208 ries of polyethylene glycols, Atmos. Meas. Tech., 11, 49–63, https://doi.org/10.5194/amt-11-49-2018, 2018. ries of polyethylene glycols, Atmos. Meas. Tech., 11, 49–63, https://doi.org/10.5194/amt-11-49-2018, 2018. Ervens, B. and Volkamer, R.: Glyoxal processing by aerosol mul- tiphase chemistry: towards a kinetic modeling framework of secondary organic aerosol formation in aqueous particles, At- mos. Chem. Phys., 10, 8219–8244, https://doi.org/10.5194/acp- 10-8219-2010, 2010. Lin, Y.-H., Zhang, Z., Docherty, K. S., Zhang, H., Budisulis- tiorini, S. H., Rubitschun, C. L., Shaw, S. L., Knipping, E. M., Edgerton, E. S., Kleindienst, T. E., Gold, A., and Sur- ratt, J. D.: Isoprene epoxydiols as precursors to secondary or- ganic aerosol formation: Acid-catalyzed reactive uptake studies with authentic compounds, Environ. Sci. Technol., 46, 250–258, https://doi.org/10.1021/es202554c, 2012. Gorkowski, K., Donahue, N. M. and Sullivan, R. C.: Emerging in- vestigator series: Determination of biphasic core–shell droplet properties using aerosol optical tweezers, Environ. Sci.-Proc. Imp., 20, 1512–1523, https://doi.org/10.1039/C8EM00166A, 2018. Marsh, A., Rovelli, G., Song, Y.-C., Pereira, K. L., Willoughby, R. E., Bzdek, B. R., Hamilton, J. F., Orr-Ewing, A. J., Topping, D. O., and Reid, J. P.: Accurate representations of the physic- ochemical properties of atmospheric aerosols: When are labora- tory measurements of value?, Faraday Discuss., 200, 639–661, https://doi.org/10.1039/C7FD00008A, 2017. Haddrell, A. E., Davies, J. F., Yabushita, A., and Reid, J. P.: Accounting for Changes in Particle Charge, Dry Mass and Composition Occurring During Studies of Sin- gle Levitated Particles, J. Phys. Chem. A, 116, 9941–9953, https://doi.org/10.1021/jp304920x, 2012. Nannoolal, Y., Rarey, J., and Ramjugernath, D.: Estimation of pure component properties: Part 3. Estimation of the vapor pressure of non-electrolyte organic compounds via group contributions and group interactions, Fluid Phase Equilibr., 269, 117–133, https://doi.org/10.1016/j.ﬂuid.2008.04.020, 2008. Hargreaves, G., Kwamena, N.-O. A., Zhang, Y. H., Butler, J. R., Rushworth, S., Clegg, S. L., and Reid, J. P.: Measurements of the Equilibrium Size of Supersaturated Aqueous Sodium Chlo- ride Droplets at Low Relative Humidity Using Aerosol Optical Tweezers and an Electrodynamic Balance, J. Phys. Chem. A, 114, 1806–1815, https://doi.org/10.1021/jp9095985, 2010. O’Meara, S., Booth, A. M., Barley, M. H., Topping, D., and McFiggans, G.: An assessment of vapour pressure estima- tion methods, Phys. Chem. Chem. Phys., 16, 19453–19469, https://doi.org/10.1039/C4CP00857J, 2014. Hodzic, A., Aumont, B., Knote, C., Lee-Taylor, J., Madronich, S., and Tyndall, G.: Volatility dependence of Henry’s law con- stants of condensable organics: Application to estimate deposi- tional loss of secondary organic aerosols, Geophys. Res. Lett., 41, 4795–4804, https://doi.org/10.1002/2014GL060649, 2014. Prisle, N. L., Engelhart, G. A. W. Birdsall et al.: Gas-particle partitioning of butenedial Model Dev., 9, 899– 914, https://doi.org/10.5194/gmd-9-899-2016, 2016. Krieger, U. K., Marcolli, C., and Reid, J. P.: Exploring the complexity of aerosol particle properties and processes using single particle techniques, Chem. Soc. Rev., 41, 6631–6662, https://doi.org/10.1039/c2cs35082c, 2012. Wang, C., Lei, Y. D., Endo, S., and Wania, F.: Measur- ing and modeling the salting-out effect in ammonium sul- fate solutions, Environ. Sci. Technol., 48, 13238–13245, https://doi.org/10.1021/es5035602, 2014. Krieger, U. K., Siegrist, F., Marcolli, C., Emanuelsson, E. U., Gø- bel, F. M., Bilde, M., Marsh, A., Reid, J. P., Huisman, A. J., Riipinen, I., Hyttinen, N., Myllys, N., Kurtén, T., Bannan, T., Percival, C. J., and Topping, D.: A reference data set for vali- dating vapor pressure measurement techniques: homologous se- Waxman, E. M., Elm, J., Kurtén, T., Mikkelsen, K. V., Ziemann, P. J., and Volkamer, R.: Glyoxal and methylglyoxal Setschenow Atmos. Chem. Phys., 19, 14195–14209, 2019 www.atmos-chem-phys.net/19/14195/2019/ A. W. Birdsall et al.: Gas-particle partitioning of butenedial 14209 salting constants in sulfate, nitrate, and chloride solutions: Measurements and Gibbs energies, Environ. Sci. Technol., 49, 11500–11508, https://doi.org/10.1021/acs.est.5b02782, 2015. salting constants in sulfate, nitrate, and chloride solutions: Measurements and Gibbs energies, Environ. Sci. Technol., 49, 11500–11508, https://doi.org/10.1021/acs.est.5b02782, 2015. Zuend, A., Marcolli, C., Luo, B. P., and Peter, T.: A thermo- dynamic model of mixed organic-inorganic aerosols to pre- dict activity coefﬁcients, Atmos. Chem. Phys., 8, 4559–4593, https://doi.org/10.5194/acp-8-4559-2008, 2008. Welty, J. R., Wicks, C. E., and Wilson, R. E.: Fundamentals of mo- mentum, heat, and mass transfer, 3rd ed., Wiley, New York, 832 pp., 1984. Zuend, A., Marcolli, C., Booth, A. M., Lienhard, D. M., Soon- sin, V., Krieger, U. K., Topping, D. O., McFiggans, G., Pe- ter, T., and Seinfeld, J. H.: New and extended parameterization of the thermodynamic model AIOMFAC: calculation of activ- ity coefﬁcients for organic-inorganic mixtures containing car- boxyl, hydroxyl, carbonyl, ether, ester, alkenyl, alkyl, and aro- matic functional groups, Atmos. Chem. Phys., 11, 9155–9206, https://doi.org/10.5194/acp-11-9155-2011, 2011. Yu, G., Bayer, A. R., Galloway, M. M., Korshavn, K. J., Fry, C. G., and Keutsch, F. N.: Glyoxal in aqueous ammonium sulfate so- lutions: Products, kinetics and hydration effects, Environ. Sci. Technol., 45, 6336–6342, https://doi.org/10.1021/es200989n, 2011. Zardini, A. A., Krieger, U. K., and Marcolli, C.: White light Mie res- onance spectroscopy used to measure very low vapor pressures of substances in aqueous solution aerosol particles, Opt. Express, 14, 6951–6962, https://doi.org/10.1364/OE.14.006951, 2006. Atmos. Chem. Phys., 19, 14195–14209, 2019 www.atmos-chem-phys.net/19/14195/2019/ www.atmos-chem-phys.net/19/14195/2019/ www.atmos-chem-phys.net/19/14195/2019/
https://openalex.org/W2162198091	https://europepmc.org/articles/pmc4216368?pdf=render	English	null	Risk factors for depth of infiltration in the differentiated depressed early gastric carcinoma: a preliminary analysis	Diagnostic pathology	2,014	cc-by	5,113	RESEARCH Open Access © 2014 Yang et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Yang et al. Diagnostic Pathology 2014, 19:206 http://www.diagnosticpathology.org/content/19/1/206 Yang et al. Diagnostic Pathology 2014, 19:206 http://www.diagnosticpathology.org/content/19/1/206 Keywords: Differentiated depressed early gastric cancer, Depth of infiltration, Clinicopathologic Keywords: Differentiated depressed early gastric cancer, Depth of infiltration, Clinicopathologic factor Risk factors for depth of infiltration in the differentiated depressed early gastric carcinoma: a preliminary analysis Shu-dong Yang1†, Zhi-xin Qian2†, Qiang Zhan3, Qun-yan Zhou3 and Guo-min Lu3 Correspondence: zhanq33@163.com †Equal contributors 3Department of Gastroenterology, Wuxi People’s Hospital Affiliated with Nanjing Medical University, Wuxi 214023, China Full list of author information is available at the end of the article Abstract Background: To analyze the clinicopathologic factors associated with mucosal and submucosal infiltration in differentiated depressed early gastric cancer, and screening factors that can predict depth of infiltration before endoscopic treatment. Methods: The study included 35 cases of mucosal carcinomas and 66 cases of submucosal carcinomas according to the pathological diagnosis. The relevant clinicopathologic factors were investigated by univariate and multivariate analysis. Results: The average depth of the depressed lesions for the submucosal group was significantly more than that for the mucosal group. The proportion of the lesions with rough bottom surface and abnormal surrounding folds was significantly higher in the submucosal group compared to that in the mucosal group. Logistic regression analysis indicated that the above-mentioned three factors were independent risk factors that could be used to predict mucosal and submucosal infiltration. Area under the curve (AUC) of receiver operating characteristic (ROC) of the ordinal above-mentioned three factors for predicting submucosal infiltration was 0.716, 0.663, 0.704, respectively. Stratified analysis showed that the 100% cases with lesion depth ≥2.5 mm and rough bottom surface developed submucosal infiltration regardless of the morphological changes of the folds. Conclusion: The study identified independent risk factors for predicting mucosal and submucosal infiltration in depressed differentiated early gastric cancer, which may evaluate the degree of penetration before endoscopic treatment. Virtual Slides: The virtual slide(s) for this article can be found here: http://www.diagnosticpathology.diagnomx.eu/ vs/13000_2014_206 Virtual Slides: The virtual slide(s) for this article can be found here: http://www.diagnosticpathology.diagnomx.eu vs/13000_2014_206 Keywords: Differentiated depressed early gastric cancer, Depth of infiltration, Clinicopathologic factor Background early gastric cancer who received gastrectomy and lymph node dissection, and assessed the clinicopathological factors for their possible association with lymph node metastasis. They eventually developed evaluation cri- teria for EMR and ESD treatment of early gastric cancer before surgery. Endoscopic mucosal resection (EMR) and endoscopic sub- mucosal dissection (ESD) have become one of the standard treatment methods for early-stage gastric cancer in Japan. The development of these two endoscopic technologies has benefited from research on two screening criteria for operability for endoscopic resection. Yamao et al. [1] and Gotoda et al [2], investigated thousands of patients with EMR and ESD are therapeutic endoscopic techniques that offer the advantage of acquiring specimens from the lesion for pathological analysis. These have become the most accurate approaches for the diagnosis of early gastric cancer. For instance, results of the pathological examin- ation can confirm the cure or suggest additional treat- ment. The disease-specific 5- and 10-year survival rates * Correspondence: zhanq33@163.com †Equal contributors 3Department of Gastroenterology, Wuxi People’s Hospital Affiliated with Nanjing Medical University, Wuxi 214023, China Full list of author information is available at the end of the article Yang et al. Diagnostic Pathology 2014, 19:206 http://www.diagnosticpathology.org/content/19/1/206 Page 2 of 7 Page 2 of 7 (depressed). Type II can be sub-classified into: type IIa (surface elevated), type IIb (flat) and type IIc (surface depressed). Both type III and type IIc were considered as depressed lesions. can reach 99% for patients who meet the above evaluation criteria for EMR treatment of early gastric cancer before surgery [3]. Early gastric ESD offers the advantage of high percent- age of enbloc resection rates. According to a published report, enbloc resection rates of 92.7% and curative re- section rates of 73.6% were achieved [4]. Local recur- rence of early gastric cancer after ESD treatment was less than 2% [5]. It is worth noting that the healing ef- fects similar to those obtained with traditional surgical treatment could be achieved only if the lesions met the criteria for curative resection, highlighting the import- ance of accurate pre and post-operative assessment. Subjects and inclusion/exclusion criteria Patients who had been diagnosed for depressed early gastric cancer at Wuxi People’s Hospital Affiliated with Nanjing Medical University between January 2005 and December 2011 were recruited for this study. The subjects received surgical resection and lymphadenectomy and the patho- logical analysis confirmed diagnosis of differentiated type early gastric cancer. Cases with undifferentiated type early gastric cancer were excluded due to the controversy sur- rounding the use of endoscopy for such types of gastric cancer. All case diagnoses were made by a pathologist with 16 years clinical experience and confirmed by an in- dependent pathologist with over 40 years clinical experi- ence. Informed consents were obtained from all subjects, and the study was approved by the medical ethics com- mittee of People’s Hospital of Wuxi, an affiliated hospital of Nanjing Medical University. Results A total of 101 cases with depressed lesions were evaluated (35 cases in the mucosal group, 34.7%, Figure 1; 66 cases in the submucosal group, 65.3%, Figure 2). General patient information and clinicopathologic data are shown in Tables 1 and 2. Background The mistakes in preoperative evaluation do occur (error rate was 7% in the EMR group and 16% in the ESD group) and are confirmed by postoperative pathological examin- ation.4 According to preoperative indications defined by Japanese Gastric Cancer Association (JGCA) [6,7], endo- scopic examination and endoscopic biopsy are sufficient to evaluate the diameter of the lesion, formation of ulcer and the pathological type of early gastric cancer. How- ever, practical and putative criteria to evaluate the depth of mucosal or submucosal infiltration are still lacking. The following clinical data were acquired: (1) gender; (2) age; (3) location of the lesions (stomach was divided into upper 1/3, middle 1/3 and lower 1/3 based on JGCA guidelines;7 (4) size of the lesion (largest diameter of the bottom surface of depressed area); (5) presence of ulcer (obvious ulcer or scarring evidence of a previous ulcer); (6) depth of the lesion (measured as the distance between planes of the specimen mounted on a pathology slide from the highest point of the lesion edge to the lowest point of the lesion basilar portion); (7) bottom of the lesion: smooth or rough (surface is not flat with particle-like bulge or small mucosal island); (8) folds surrounding the lesions: normal folds were marked as negative and broken folds or enlarged/fusion folds were marked as positive; (9) degree of gastric cancer differenti- ation: highly differentiated was defined as grade I cancer and moderately differentiated was defined as grade II can- cer; (10) depth of infiltration: the lesions were classified as mucosal gastric cancer or submucosal gastric cancer. Given the difficulty in determining the depth of infil- tration in the depressed early gastric cancer, we followed the method of Yamao T and Gotoda T [1,2], and retro- spectively reviewed the clinicopathological characteris- tics of mucosal and submucosal gastric cancer in order to identify the factors for preoperative evaluation in gas- tric cancer. Statistical analysis Th d The data were processed using SPSS 17.0 software (SPSS Inc., Chicago, IL). All the samples were classified into the mucosal group and the submucosal group based on the depth of infiltration. T test or Wilcoxon rank test were used to process quantitative data, and categorical data were processed by chi-square test. Logistic regres- sion multiple factor analysis was used to evaluate inde- pendent factors that influenced the mucosal/submucosal infiltration of early gastric cancer. Receiver Operating Characteristic (ROC) curves of independent factors were plotted based on the above results of Logistic regression analysis to reflect morphological factors, which may affect the depth of infiltration. The area under the curve (AUC) was calculated and cut-off values were estimated. Stratified analysis was performed for statistically signifi- cant morphological factors to evaluate the distribution of mucosal and submucosal groups in various condi- tions. P < 0.05 was defined as statistically significant. Univariate analysis of the depressed lesions According to the classification of gastric cancer based on the guidelines laid out by JGCA [7], the lesions were grouped as: type I (elevated), type II and type III Univariate analysis was applied to the data of acquired characteristics. T test or Wilcoxon rank test were used Yang et al. Diagnostic Pathology 2014, 19:206 http://www.diagnosticpathology.org/content/19/1/206 Page 3 of 7 Table 1 General patient information Figure 1 A representative moderately differentiated adenocarcinoma with invasion of the muscular layer of mucosa. HE staining, low power (40×) microscopic view. Table 1 General patient information N =101 Patient data Males, N = 71 Females, N = 30 Yes No Yes No Upper abdominal pain 50 21 19 11 Abdominal distention 41 30 19 11 Decreased food appetite 16 55 5 25 Nausea 25 46 11 19 Vomiting 3 68 3 27 Acid regurgitation 5 66 3 27 Belching 16 55 14 16 Haematemesis 5 66 0 30 Black stool 14 57 0 30 Weight loss 9 62 3 27 Fatigue 7 64 3 27 History of hypertension 27 44 16 14 History of coronary heart disease 5 66 0 30 History of diabetes mellitus 11 60 8 22 History of hepatitis 5 66 0 30 History of cirrhosis 3 68 0 30 Long-term* history of smoking 13 58 0 30 Long-term history of drinking alcohol 3 68 0 30 Long-term* oral administration of aspirin 3 68 0 30 Special occupation 3**** 68 0 30 Family history 5 66 3 27 15-30 years; 7-15 years; 3-5 years; **All cases were dust-exposure. Figure 1 A representative moderately differentiated adenocarcinoma with invasion of the muscular layer of mucosa. HE staining, low power (40×) microscopic view. to process quantitative data, and categorical data were processed by chi-square test. Results showed that the le- sion depth of the submucosal group (average 3.9 ± 2.5 mm) was more than that of the mucosal group (aver- age 2.3 ± 1.5 mm). The analysis of the bottom of the le- sion showed that the mucosal group had 30 (85.7%) cases with smooth bottom and 5 (14.3%) cases with rough bottom, while the submucosal group had 35 (53.0%) cases of smooth bottom and 31 (47.0%) cases of rough bottom. Multivariate analysis of depressed lesions According to the results of univariate analysis, logistic binary regression model was used to evaluate the size, depth, bottom surface, and surrounding folds of the le- sions (P < 0.1). Regression model was established using forward stepwise regression, and the results obtained from the three-step regression analysis showed that depth, surface roughness and surrounding folds of the lesions are independent predictors of tumor filtration (Table 4). Figure 2 A representative moderately differentiated adenocarcinoma with invasion of the submucosa. HE staining, low power (40×) microscopic view. Univariate analysis of the depressed lesions The analysis of folds surrounding the le- sions showed that the mucosal group had 27 (77.1%) cases with normal folds and 8 (22.9%) cases with abnor- mal folds, while the submucosal group had 24 (36.4%) cases with normal folds and 42 (63.6%) cases with abnor- mal folds. The submucosal group showed higher rates of rough bottom surface and abnormal folds compared to the mucosal group. The collective data is presented in Table 3. to the mucosal group. The collective data is presented in Table 3. Stratified analysis Based on the cut-off value of 2.5 mm, the lesions were divided into two groups: <2.5 mm group and ≥2.5 mm group. Stratified analysis suggested that 100% of the lesions in early, depressed gastric cancer with depth ≥2.5 mm, rough bottom surface and normal/abnormal surrounding folds developed submucosal infiltration (Table 6). submucosal infiltration We further plotted ROC curves to estimate the diagnos- tic performance of the three independent predictors: depth, surface roughness and surrounding folds of the depressed lesions in gastric cancer (Figure 3). AUCs of depth, surface roughness and surrounding folds of the Figure 2 A representative moderately differentiated adenocarcinoma with invasion of the submucosa. HE staining, low power (40×) microscopic view. Yang et al. Diagnostic Pathology 2014, 19:206 http://www.diagnosticpathology.org/content/19/1/206 Page 4 of 7 Table 2 Clinical and pathological characteristics of the patients Characteristic N Percentage (%) Sex Male 71 70.3 Female 30 29.7 Age in years 61.2 ± 9.8 Location of lesions Upper 1/3 24 23.8 Middel 1/3 30 29.7 Lower 1/3 47 46.5 Size of the lesions in mm 15.5 ± 12.3 Presence of ulcer No 53 52.5 Yes 48 47.5 Depth of depressed lesions in mm 3.4 ± 2.4 Roughness of the bottom Smooth 65 64.4 Rough 36 35.6 Surrounding folds Normal 51 50.5 Abnormal 50 49.5 Degree of differentiation Highly differentiated 10 9.9 Moderately differentiated 91 91.1 Data are presented as N or mean ± SD. Table 2 Clinical and pathological characteristics of the patients lesions were 0.716 (P = 0.000, 95% CI 0.613-0.819), 0.663 (P = 0.007, 95% CI 0.556-0.771) and 0.704 (P = 0.001, 95% CI 0.598-0.810), respectively. The optimal cut-off value of depth of depressed lesions for predicting sub- mucosal infiltration was 2.5 mm based on Youden’s index (Table 5). lesions were 0.716 (P = 0.000, 95% CI 0.613-0.819), 0.663 (P = 0.007, 95% CI 0.556-0.771) and 0.704 (P = 0.001, 95% CI 0.598-0.810), respectively. The optimal cut-off value of depth of depressed lesions for predicting sub- mucosal infiltration was 2.5 mm based on Youden’s index (Table 5). lesions were 0.716 (P = 0.000, 95% CI 0.613-0.819), 0.663 (P = 0.007, 95% CI 0.556-0.771) and 0.704 (P = 0.001, 95% CI 0.598-0.810), respectively. The optimal cut-off value of depth of depressed lesions for predicting sub- mucosal infiltration was 2.5 mm based on Youden’s index (Table 5). Data are presented as N (%) or mean ± SD. Discussion Endoscopic mucosal resection (EMR) and endoscopic submucosal dissection (ESD) are widely accepted endo- scopic techniques for treating early gastric cancer. The indications for applying endoscopic therapy are: differen- tiated type early gastric cancer; no evidence of lymph node metastasis; infiltration is confined to mucosal or minute submucosal infiltration [8]. The rate of lymph node metastasis is less than 3% when the gastric cancer is confined to mucosal infiltration [9]. However, sub- mucosal infiltration leads to a higher risk of metastasis [10]. Due to frequent submucosal infiltration and lymph node metastasis [11], evaluation of the depth of infiltra- tion of depressed type of early gastric cancer is crucial for decisions regarding the therapeutic regimen [8,12]. Table 3 Univariate analysis of differentiated depressed early gastric carcinoma Factors Values Statistical value p value Mucosal gastric cancer Submucosal gastric cancer Sex Male 24 (66.2) 47 (71.2) χ2 = 0.076 0.782 Female 11 (33.8) 19 (28.8) Age in years 62.3 ± 9.2 60.6 ± 10.2 t = 0.402 0.843 Lesion location Upper 1/3 6 (17.1) 18 (27.3) χ2 = 1.994 0.369 Middle 1/3 13 (37.2) 17 (25.7) Lower 1/3 16 (45.7) 31 (47.0) Size of the lesion in mm 12.4 ± 10.8 17.2 ± 12.8 U = 894.0 0.061 Ulcer Absence 22 (62.9) 31 (47.0) χ2 = 2.315 0.128 Presence 13 (37.1) 35 (53.0) Depth in mm 2.3 ± 1.5 3.9 ± 2.5 U = 656.5 0.000 Roughness of the bottom Smooth 30 (85.7) 35 (53.0) χ2 = 10.651 0.001 Rough 5 (14.3) 31 (47.0) Surrounding folds Normal 27 (77.1) 24 (36.4) χ2 = 15.215 0.000 Abnormal 8 (22.9) 42 (63.6) Degree of differentiation Highly differentiated 4 (11.4) 6 (9.1) χ2 = 0.140 0.708 Moderately differentiated 31 (88.6) 60 (90.9) Data are presented as N (%) or mean ± SD. Table 3 Univariate analysis of differentiated depressed early gastric carcinoma Yang et al. Diagnostic Pathology 2014, 19:206 http://www.diagnosticpathology.org/content/19/1/206 Page 5 of 7 Table 4 Multivariate analysis of differentiated depressed early gastric carcinoma B S.E. Wals df Sig. Exp (B) Exp (B) 95% CI Lower limit Upper limit Lesion depth 0.425 0.151 7.949 1 0.005 1.530 1.138 2.056 Bottom roughness 1.348 0.608 4.918 1 0.027 3.851 1.170 12.681 Surrounding folds 1.448 0.520 7.746 1 0.005 4.254 1.535 11.794 Abbreviations: B regression coefficient, S.E. Discussion standard error, Wals the square of the ratio of the regression coefficient and standard error, df degrees of freedom, Sig probability, Exp (B) the exponential function, and is the result of regression coefficient B value index calculation, CI confidence interval. Table 4 Multivariate analysis of differentiated depressed early gastric carcinoma breviations: B regression coefficient, S.E. standard error, Wals the square of the ratio of the regression coefficient and standard error, df probability, Exp (B) the exponential function, and is the result of regression coefficient B value index calculation, CI confidence interval analysis to these factors to predict the depth of filtration. It was reported that the diameter of the lesion is associ- ated with the depth of infiltration of gastric cancer [19], but our results indicated that there was no significant difference between mucosal and submucosal groups. Be- sides, only seven cases of depressed lesions among 37 cases of gastric cancer were included [19], which limits the analysis performed in that study. The parameters, such as the location of the lesion, presence of ulcer and histology type are irrelevant to the depth of filtration, which is consistent with a previously published study [20]. ROC curve analysis further illustrates the value of the three factors for the diagnosis of lesion infiltration. Among the three factors, the depth of the lesion is super- ior to the lesion bottom roughness and surrounding folds in term of its value for making a diagnosis. Our findings that the presence of rough bottom surface and irregular folds also aid in the diagnosis of submucosal gastric cancer are in agreement with a pervious study [21]. There is no consensus on the preoperative diagnosis of the depth of infiltration. Computed tomography (CT) scan does not assess the depth of tumor invasion [13]. The use of endoscopic ultrasonography is also contro- versial in determining the depth of infiltration [14]. A summary of 18 studies suggested that sensitivity varied from 18.2% to 100% (average 87.8%) and specificity var- ied from 34.7% to 100% (average 80.2%) when using endoscopic ultrasound for the diagnosis of submucosal invasion of early gastric cancer [15]. It was concluded that it is not a mature method for differentiating be- tween mucosal and submucosal infiltration. Tradition- ally, gastroscopy has been used for early gastric cancer staging with an accuracy of 79.0%, which is similar to that achieved by endoscopic ultrasound [16-18]. aYouden’s index = (sensitivity + specificity)-1; bthe optimal cut-off value. Discussion Thus, ordinary endoscopy could provide sufficient information to determine the best treatment options for the resection of early gastric cancer. Our study identified independent predictive factors for infiltration based on the analysis of multiple clinical and pathomorphological data. We further applied stratified It should be noted that in this study, the converging folds were considered as ulcer but not the abnormal changes of the surrounding folds. Only the broken and enlarged folds were defined as abnormal folds. The for- mation of ulcers and subsequent scarring and conver- ging fold normally indicate tissue fibrosis but do not necessarily suggest penetration of tumor, although some gastric cancers with converging folds are confined to the Figure 3 ROC curve to predict infiltration depth based on lesion depth, bottom roughness and surrounding folds. Table 5 ROC curve and Youden’s index based on the depth of the lesions Lesion depth Sensitivity Specificity 1—specificity Youden’s indexa 0.000 1.000 0.000 1.000 0.000 1.250 0.864 0.429 0.571 0.293 1.750 0.848 0.429 0.571 0.277 2.500b 0.621 0.714 0.286 0.315 3.500 0.439 0.771 0.229 0.210 4.500 0.409 0.829 0.171 0.238 5.500 0.197 1.000 0.000 0.197 6.000 0.167 1.000 0.000 0.167 7.500 0.121 1.000 0.000 0.121 9.000 0.061 1.000 0.000 0.061 11.000 0.000 1.000 0.000 0.000 aYouden’s index = (sensitivity + specificity)-1; bthe optimal cut-off value. Table 5 ROC curve and Youden’s index based on the depth of the lesions Yang et al. Diagnostic Pathology 2014, 19:206 http://www.diagnosticpathology.org/content/19/1/206 Page 6 of 7 Table 6 Stratified analysis of the differentiated depressed early gastric carcinoma Lesion depth Bottom roughness Grouping based on penetration depth Sum Mucosal surface Submucosal surface < 2.5 mm Smooth Surrounding folds No Count 16 7 23 Percentage 69.5% 31.5% 100.0% Yes Count 4 6 10 Percentage 40.0% 60.0% 100.0% Rough Surrounding folds No Count 4 4 8 Percentage 50.0% 50.0% 100.0% Yes Count 1 8 9 Percentage 11.1% 88.9% 100.0% ≥2.5 mm Smooth Surrounding folds No Count 7 11 16 Percentage 43.8% 56.2% 100.0% Yes Count 3 11 14 Percentage 21.4% 79.6% 100.0% Rough Surrounding folds No Count 0 2 2 Percentage 0.0% 100.0% 100.0% Yes Count 0 17 17 Percentage 0.0% 100.0% 100.0% Table 6 Stratified analysis of the differentiated depressed early gastric carcinoma mucosal layer [22]. However, the folds may break or en- large when the top of the folds are infiltrated by the tumor. Discussion If annular dike appears, the tumor may have in- vaded the muscular layer. Thus, abnormal folds but not the presence or absence of ulcers may provide useful in- formation for evaluating the depth of penetration in dif- ferentiated gastric cancer. for evaluating mucosal or submucosal infiltration in de- pressed gastric cancer based on retrospective analysis of the pathology samples. The results indicate that further studies from multiple centers and with larger sample sizes are required. References 1. Yamao T, Shirao K, Ono H, Kondo H, Saito D, Yamaguchi H, Sasako M, Sano T, Ochiai A, Yoshida S: Risk factors for lymph node metastasis from intramucosal gastric carcinoma. Cancer 1996, 77:602–606. 2. Gotoda T, Yanagisawa A, Sasako M, Ono H, Nakanishi Y, Shimoda T, Kato Y: Incidence of lymph node metastasis from early gastric cancer: 1. Yamao T, Shirao K, Ono H, Kondo H, Saito D, Yamaguchi H, Sasako M, Sano T, Ochiai A, Yoshida S: Risk factors for lymph node metastasis from intramucosal gastric carcinoma. Cancer 1996, 77:602–606. 2. Gotoda T, Yanagisawa A, Sasako M, Ono H, Nakanishi Y, Shimoda T, Kato Y: Incidence of lymph node metastasis from early gastric cancer: 1. Yamao T, Shirao K, Ono H, Kondo H, Saito D, Yamaguchi H, Sasako M, Sano T, Ochiai A, Yoshida S: Risk factors for lymph node metastasis from intramucosal gastric carcinoma. Cancer 1996, 77:602–606. Authors’ contributions SY and ZQ participated in the design of the study, performed the statistical analysis and draft the manuscript. QZ conceived of the study, and participated in its design and coordination and helped to draft the manuscript. QZ and GL collected, analyzed and organized cases data. All authors read and approved the final manuscript. Competing interests h h ( ) d l d p g The author(s) declared no potential competing interest in the manuscript, including financial, consultant, institutional and other relationships. Based on the Youden’s index derived from the ROC curve, we defined 2.5 mm as the cut-off value to divide the lesions into two groups (lesion depth <2.5 mm or ≥2.5 mm). Indeed, cases with lesion depth ≥2.5 mm with rough bottom surface had 100% submucosal penetration, re- gardless of the status of the surrounding folds. How- ever, 69.5% cases with lesion depth <2.5 mm, smooth bottom and normal surrounding folds were confined to the mucosa. However, we measured the depth using specimens mounted on glass slides for pathological analysis; as such, the distance is expected to have been affected (shrunk by 10-20%) by the formalin fixation and dehydration processes that are required for slide preparation. Our study aimed to reflect the conditions found under endoscopy and before any biopsy is taken. Therefore, the cut-off value of 2.5 mm may actually be 3 mm in living stomach. It is important that any future re- search studies, either from our lab or others, should de- sign and use measurement methods that will reduce this kind of error. Author details 1 1Department of Pathology, Wuxi People’s Hospital Affiliated with Nanjing Medical University, Wuxi 214023, China. 2Department of Internal Medicine, Jiangsu Taizhou People’s Hospital, Taizhou 225300, China. 3Department of Gastroenterology, Wuxi People’s Hospital Affiliated with Nanjing Medical University, Wuxi 214023, China. Received: 26 May 2014 Accepted: 8 October 2014 Acknowledgements W h k M dj d B We thank Medjaden Bioscience Limited for assisting in the preparation of this manuscript. We thank Dr. Rong-chao Sun for conducting the independent diagnostic review of all cases. Conclusions Sano T, Ochiai A, Yoshida S: Risk factors for lymph node metastasis from intramucosal gastric carcinoma. Cancer 1996, 77:602–606. 2. Gotoda T, Yanagisawa A, Sasako M, Ono H, Nakanishi Y, Shimoda T, Kato Y: Incidence of lymph node metastasis from early gastric cancer: Overall, we identified lesion depth, bottom roughness and surrounding folds as independent predictive factors g 2. Gotoda T, Yanagisawa A, Sasako M, Ono H, Nakanishi Y, Shimoda T, Kato Y: Incidence of lymph node metastasis from early gastric cancer: Page 7 of 7 Yang et al. Diagnostic Pathology 2014, 19:206 http://www.diagnosticpathology.org/content/19/1/206 21. Choi J, Kim SG, Im JP, Kim JS, Jung HC, Song IS: Endoscopic prediction of tumor invasion depth in early gastric cancer. Gastrointest Endosc 2011, 73:917–927. estimation with a large number of cases at two large centers. Gastric Cancer 2000, 3:219–225. estimation with a large number of cases at two large centers. Gastric Cancer 2000, 3:219–225. 3. Uedo N, Iishi H, Tatsuta M, Ishihara R, Higashino K, Takeuchi Y, Imanaka K, Yamada T, Yamamoto S, Yamamoto S, Tsukuma H, Ishiguro S: Longterm outcomes after endoscopic mucosal resection for early gastric cancer. Gastric Cancer 2006, 9:88–92. 22. Sano T, Okuyama Y, Kobori O, Shimizu T, Morioka Y: Early gastric cancer. Endoscopic diagnosis of depth of invasion. Dig Dis Sci 1990, 35:1340–1344. 22. Sano T, Okuyama Y, Kobori O, Shimizu T, Morioka Y: Early gastric cancer. Endoscopic diagnosis of depth of invasion. Dig Dis Sci 1990, 35:1340–1344. 4. Oda I, Saito D, Tada M, Iishi H, Tanabe S, Oyama T, Doi T, Otani Y, Fujisaki J Ajioka Y, Hamada T, Inoue H, Gotoda T, Yoshida S: A multicenter retrospective study of endoscopic resection for early gastric cancer. Gastric Cancer 2006, 9:262–270. doi:10.1186/s13000-014-0206-8 Cite this article as: Yang et al.: Risk factors for depth of infiltration in the differentiated depressed early gastric carcinoma: a preliminary analysis. Diagnostic Pathology 2014 19:206. 5. Chaves DM, Maluf Filho F, De Moura EG, Santos ME, Arrais LR, Kawaguti F, Sakai P: Endoscopic submucosal dissection for the treatment of early esophageal and gastric cancer–initial experience of a western center. Clinics (Sao Paulo) 2010, 65:377–382. 6. Ono S, Fujishiro M, Niimi K, Goto O, Kodashima S, Yamamichi N, Omata M: Technical feasibility of endoscopic submucosal dissection for early gastric cancer in patients taking anti-coagulants or anti-platelet agents. Dig Liver Dis 2009, 41:725–728. 7. Japanese Gastric Cancer Association: Japanese classification of gastric carcinoma. Conclusions 2nd English Edition. Gastric Cancer 1998, 1:11. 8. Kim SG: Estimation by gross findings in early gastric cancer. Clin Endosc 2012, 45:245–247. 9. Sano T, Kobori O, Muto T: Lymph node metastasis from early gastric cancer: endoscopic resection of tumour. Br J Surg 1992, 79:241–244. 10. Li Y, Zhao Q, Fan LQ, Tan BB, Zhang ZD, Liu Y: Analysis of Lymph Node Dissection Range-Related Factors for Early Gastric Cancer Operation. Hepatogastroenterology 2013, 60:971–974. 11. Nakamura T, Yao T, Kakeji Y, Anai H, Morita M, Oda Y, Maehara Y: Depressed type of intramucosal differentiated-type gastric cancer has high cell proliferation and reduced apoptosis compared with the elevated type. Gastric Cancer 2013, 16:94–99. 12. Noh H, Park JJ, Yun JW, Kwon M, Yoon DW, Chang WJ, Oh HY, Joo MK, Lee BJ, Kim JH, Yeon JE, Kim JS, Byun KS, Bak YT: Clinicopathologic characteristics of patients who underwent curative additional gastrectomy after endoscopic submucosal dissection for early gastric cancer or adenoma. Korean J Gastroenterol 2012, 59:289–295. 13. Lim JS, Yun MJ, Kim MJ, Hyung WJ, Park MS, Choi JY, Kim TS, Lee JD, Noh SH, Kim KW: CT and PET in stomach cancer: preoperative staging and monitoring of response to therapy. Radiographics 2006, 26:143–156. 14. Okada K, Fujisaki J, Kasuga A, Omae M, Yoshimoto K, Hirasawa T, Ishiyama A, Yamamoto Y, Tsuchida T, Hoshino E, Igarashi M, Takahashi H: Endoscopic ultrasonography is valuable for identifying early gastric cancers meeting expanded-indication criteria for endoscopic submucosal dissection. Surg Endosc 2011, 25:841–848. 15. Kwee RM, Kwee TC: The accuracy of endoscopic ultrasonography in differentiating mucosal from deeper gastric cancer. Am J Gastroenterol 2008, 103:1801–1809. 16. Ahn HS, Lee HJ, Yoo MW, Kim SG, Im JP, Kim SH, Kim WH, Lee KU, Yang HK: Diagnostic accuracy of T and N stages with endoscopy, stomach protocol CT, and endoscopic ultrasonography in early gastric cancer. J Surg Oncol 2009, 99:20–27. 17. Choi J, Kim SG, Im JP, Kim JS, Jung HC, Song IS: Comparison of endoscopic ultrasonography and conventional endoscopy for prediction of depth of tumor invasion in early gastric cancer. Endoscopy 2010, 42:705–713. Conclusions Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color ﬁgure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color ﬁgure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color ﬁgure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit Submit your next manuscript to BioMed Central and take full advantage of: Submit your next manuscript to BioMed Central and take full advantage of: 18. Choi J, Kim SG, Im JP, Kim JS, Jung HC, Song IS: Is endoscopic ultrasonography indispensable in patients with early gastric cancer prior to endoscopic resection? Surg Endosc 2010, 24:3177–3185. 19. Guo M, Liu H: Pathological char acter of early gastric cancer. Chin J Curr Clin Med 2008, 6:300–301. • Convenient online submission 20. Abe S, Oda I, Shimazu T, Kinjo T, Tada K, Sakamoto T, Kusano C, Gotoda T: Depth-predicting score for differentiated early gastric cancer. Gastric Cancer 2011, 14:35–40.
https://openalex.org/W2795467017	https://journal.niidi.ru/jofin/article/download/693/601	Russian	null	DIAGNOSTIC MEANING OF DETERMINATION OF HOMOCYSTEIN IN THE BLOOD SERUM IN CHILDREN WITH NEUROINFECTIONS	Žurnal infektologii	2,018	cc-by	3,269	Резюме Резюме The aim of the study was to determine the serum level of homocysteine in children with neuroinfections. Цель: определить уровень гомоцистеина в сыворот- ке крови у детей при нейроинфекциях. Mаterials and methods: A blood serum assay of 60 chil- dren with viral and bacterial neuroinfections was performed using the enzyme immunoassay to evaluate homocystein- emia as a marker of endothelial dysfunction. Материалы и методы: методом иммуноферментно- го анализа проведено исследование сыворотки крови 60 детей с вирусными и бактериальными нейроинфекция- ми для оценки гомоцистеинемии как маркера эндотели- альной дисфункции. Results: In interpreting the results, the average level of hyperhomocysteinemia in patients with bacterial and viral neuroinfections was established. The highest homocysteine levels in serum were observed in bacterial neuroinfections, especially in the presence of signs of meningoencephalitis and pneumococcal etiology of the disease. Dependence of changes in the level of homocysteinemia from the sex of sick children was not revealed. Результаты: при интерпретации результатов установлен средний уровень гипергомоцистеинемии у пациентов при бактериальных и вирусных нейроинфек- циях. Наибольшие показатели гомоцистеина в сыворот- ке крови отмечены при бактериальных нейроинфекциях, особенно при наличии признаков менингоэнцефалита и пневмококковой этиологии заболевания. Зависимости изменения уровня гомоцистеинемии от пола больных детей не выявлено. Conclusion: Neuroinfections in children, regardless of the cause, are accompanied by hyperhomocysteinemia, indi- cating the presence of endothelial activation in this patholo- gy, more pronounced in the case of a bacterial etiology of the infectious process and the combined lesion of the soft dura mater and the brain substance. Заключение: нейроинфекции у детей, независимо от причины, сопровождаются гипергомоцистеинемией, свидетельствуя о наличии активации эндотелия при данной патологии, более выраженной в случае бактери- альной этиологии инфекционного процесса и сочетан- ном поражении оболочек и вещества головного мозга. Key words: homocysteine, endothelial dysfunction, neu- roinfections, children. Ключевые слова: гомоцистеин, дисфункция эндоте- лия, нейроинфекции, дети. Оригинальное исследование Оригинальное исследование DOI: 10.22625/2072-6732-2018-10-1-42-46 Diagnostic meaning of determination of homocystein in the blood serum in children with neuroinfections T S Berezovskaya N A Miromanova Chita State Medical Academy, Chita, Russia Chita State Medical Academy, Chita, Russia Diagnostic meaning of determination of homocystein in the blood serum in children with neuroinfections T.S. Berezovskaya, N.A. Miromanova Chita State Medical Academy Chita Russia Diagnostic meaning of determination of homocystein in the blood serum in children with neuroinfectio T.S. Berezovskaya, N.A. Miromanova Chit St t M di l A d Chit R i цах. Однако исследование данного показателя при нейроинфекциях у детей остается изученным не- достаточно. Цель исследования – определить уровень го- моцистеина в сыворотке крови у детей при нейро- инфекциях. Введение прежде всего сосудистого эндотелия, также играет ведущую роль в формировании защитных и патоло- гических процессов при нейроинфекциях. прежде всего сосудистого эндотелия, также играет ведущую роль в формировании защитных и патоло- гических процессов при нейроинфекциях. До настоящего времени сохраняется актуаль- ность углубленного изучения нейроинфекций, что связано с высоким риском развития тяжелых и осложненных форм, большим процентом инвали- дизации пациентов и возможным летальным исхо- дом [1, 2]. Тяжесть нейроинфекций зависит как от этиологии заболевания, так и от возраста пациен- та. Наиболее тяжело нейроинфекции протекают у детей раннего возраста ввиду быстрого развития у них осложнений, таких как отек головного мозга и судорожный синдром. На течение и исход данного заболевания в детском возрасте существенное вли- яние оказывает выраженность интратекального воспаления, обусловливающего гемоликвороцир- куляторные и ишемические нарушения мозга [3]. Состояние сосудов центральной нервной системы, В настоящее время эндотелий рассматривается как активная метаболическая система, поддержи- вающая сосудистый гомеостаз путем осуществле- ния разнообразных функций (модуляция сосу- дистого тонуса, регуляция хемотаксических, ре- паративных и воспалительных процессов в ответ на локальное повреждение, формирование вне- клеточного матрикса) [4]. Принимая во внимание широкий спектр причин, приводящих к измене- нию нормального функционирования сосудистого эндотелия, на сегодняшний момент удалось уста- новить, что эндотелиальная дисфункция является неизменным патогенетическим звеном широкого круга заболеваний и состояний [5–7]. Том 10, № 1, 2018 ЖУРНАЛ ИНФЕКТОЛОГИИ 42 Оригинальное исследование Одним из маркеров дисфункции эндотелия является изменение уровня гомоцистеина, пред- ставляющий собой заменимую цитотоксичную и нейротоксичную аминокислоту, которая при- сутствует во всех клетках организма и плазме крови в незначительном количестве. Гомоци- стеин – продукт реакции трансметилирования; в плазме крови приблизительно 80% его находит- ся в связанном с альбумином состоянии, около 1% в свободной форме, остальная часть присутствует в виде дисульфидов с цистеином, гомоцистеином (гомоцистин) и другими соединениями. Все гомо- цистеин-содержащие производные определяют- ся понятием «общий гомоцистеин» плазмы крови [8]. Уровень содержания гомоцистеина в плазме крови человека зависит от возраста, пола, питания и наследственных факторов. Известно, что у муж- чин показатели данной аминокислоты выше, чем у женщин; во время беременности уровень гомоци- стеина снижается в два раза и восстанавливается до обычного уровня только после родов. У детей содержание гомоцистеина в плазме крови значи- тельно ниже, чем у взрослых: так, у детей до 12 лет, независимо от пола, физиологической нормой считается уровень до 5 мкмоль/л. В период поло- вого созревания уровень гомоцистеинемии со- ставляет у девочек до 6 мкмоль/л, у мальчиков до 7 мкмоль/л. цах. Однако исследование данного показателя при нейроинфекциях у детей остается изученным не- достаточно. цах. Материалы и методы Исследовано 60 образцов сыворотки крови детей с нейроинфекциями вирусной и бактери- альной этиологии (по 30 человек в каждой груп- пе), получавших лечение в Краевой клинической инфекционной больнице (г. Чита, Забайкальский край) в период с 2008 по 2014 г. Контрольную груп- пу составили 20 здоровых детей аналогичного воз- раста (табл. 1). Характеристика группы с бактериальными нейроинфекциями. Поражение нервной систе- мы чаще протекало с изолированным воспале- нием оболочек головного мозга, признаки ме- нингоэнцефалита наблюдались в 16,6% случаев (5 больных). Этиологические причины бактери- альных нейроинфекций: Neisseria meningitidis (N.meningitidis) – 56,7% (17 пациентов), Streptococcus pneumoniae (S. рneumoniae) – 6,7%, Haemophilus influenzae b (Hib) – 6,7%, сочетание N. meningitidis и S. рneumoniae – 3,3%. В 26,6% (8 случаев) этиологию бактериальной нейроин- фекции установить не удалось. При менингокок- ковой инфекции наблюдалось сочетание пораже- ния нервной системы с септицемией (15 больных). При пневмококковой, гемофильной и в случае не верифицированных нейроинфекций наблюдалось только изолированное поражение оболочек голов- ного мозга и/или вещества головного мозга. Изме- нения цереброспинальной жидкости характери- зовались выраженным цитозом (1115±693 клетки в мкл), нейтрофильного характера (80±10,9%), уровень белка составил 0,29±0,06 г/л. Для оценки рисков развития патологических состояний принято выделять величины гомоцисте- инемии: низкий уровень – содержание гомоци- стеина в плазме крови ≤ 7 мкмоль/л, средний – от 8 до 11 мкмоль/л, высокий – от 12 до 16 мкмоль/л и очень высокий – более 17 мкмоль/л [8]. Нару- шения утилизации данной аминокислоты при воз- действии патологических факторов приводят к на- коплению в организме продуктов окисления гомо- цистеина и развитию патологических эффектов, прежде всего к быстрой травматизации клеток эндотелия с развитием их дисфункции [9]. Роль го- моцистеинемии доказана при гипертонии, атеро- склерозе, патологии плода и плаценты [10], а так- же панкреатите и панкреонекрозе [11, 12]. Таким образом, современные данные свидетельствуют о том, что уровень гомоцистеинемии может служить самостоятельным маркером патогенетических из- менений при некоторых нозологических едини- Характеристика группы с вирусными нейро- инфекциями. Этиологическими агентами вирус- ных нейроинфекций являлись неполиомиелит- ные энтеровирусы – (93,3%): Коксаки А9 – 7,1%, ЕСНО 6 – 21,4%, Коксаки В4 – 10,7%, Коксаки с комбинацией серотипов В2 и В4 – 3,6%, в осталь- ных случаях типирования энтеровирусов не про- Клиническая характеристика групп исследуемых, Me±SD ЖУРНАЛ ИНФЕКТОЛОГИИ Том 10, № 1, 2018 43 Клиническая характеристика групп исследуемых, Me±SD Группа Количество, n Средний возраст, мес. Распределение по полу (абс. Введение Однако исследование данного показателя при нейроинфекциях у детей остается изученным не- достаточно. Материалы и методы ,%) Девочки Мальчики Бактериальные нейроинфекции 30 16±7,8 17 (56,7%) 13 (43,3%) Вирусные нейроинфекции 30 72±17,9 11 (36,7%) 19 (63,3%) Контроль 20 60±15,5 10 (50%) 10 (50%) Клиническая характеристика групп исследуемых, Me±SD 43 Оригинальное исследование водилось. Поражение нервной системы в этой группе протекало в виде поражения оболочек го- ловного мозга. В половине случаев, помимо сероз- ного менингита, у больных наблюдалось пораже- ние слизистой оболочки ротоглотки (герпетиче- ская ангина, фарингит). Изменения цереброспи- нальной жидкости характеризовались невысоким цитозом (94,5±33,2 клетки в 1 мкл), нейтрофиль- ным характером (68±18,9%) и нормальным содер- жанием белка 0,165±0,05 г/л. вался на более высоких значениях по сравнению с группой больных при вирусных нейроинфекци- ях (9,1±0,66 мкмоль/л, р ˂0,01). вался на более высоких значениях по сравнению с группой больных при вирусных нейроинфекци- ях (9,1±0,66 мкмоль/л, р ˂0,01). Интересным фактом нашего исследования явилась констатация повышения уровня гомо- цистеина и в группе здоровых детей, составив 5,14±0,71 мкмоль/л. Данный показатель по обще- принятой классификации гипергомоцистеине- мии находится на пограничных значениях нормы и повышения, что требует дополнительных ис- следований для уточнения причин данного состо- яния у здоровых детей. При анализе показателей уровня гомоцистеина в сыворотке крови здоро- вых детей в зависимости от пола выявлено его повышение у мальчиков (5,8±1,6 мкмоль/л) отно- сительно показателя девочек, который составил 4,2±1,14 мкмоль/л, р<0,01 (табл. 3). Для определения концентрации гомоцистеина (мкмоль/л) в крови пациентов проводилось иссле- дование методом ИФА с использованием наборов фирмы «Axis-Shield» (Норвегия). Забор материала осуществлялся однократно в 1–2-е сут госпитали- зации. Исследования проводились в лаборатории экспериментальной и клинической биохимии и иммунологии НИИ молекулярной медицины Чи- тинской государственной медицинской академии. Полученные результаты статистически обработа- ны в соответствии с рекомендациями по обработке данных медико-биологических исследований [13]. Количественные показатели выражены в виде среднего значения (Ме) и стандартного отклоне- ния (SD). Достоверность различий между двумя независимыми совокупностями оценивалась с по- мощью критерия Манна – Уитни. Различия пока- зателей считали достоверными при p≤0,05. Уровень гомоцистеинемии у мальчиков при бактериальных нейроинфекциях не отличался от показателей девочек (см. табл. 3). Учитывая разнообразие этиологических при- чин бактериальных нейроинфекций, особый интерес представляло изучение содержания го- моцистеина в зависимости от вида возбудителя и клинических проявлений. Благодаря этому уда- лось выявить, что при изолированном менинго- кокковом менингите содержание гомоцистеина составило 10,19±0,62 мкмоль/л, что отличалось от аналогичных показателей у детей в случае раз- вития комбинированных форм менингококко- вой инфекции (8,91±0,64 мкмоль/л, р≤0,05). При Hib-ассоциированном поражении центральной нервной системы гомоцистеинемия составила 9,34±1,3 мкмоль/л, гнойных менингитах неуточ- ненной этиологии – 8,31±0,98 мкмоль/л. Более Результаты и обсуждение Установлено, что уровень гомоцистеина в сыво- ротке крови повышался относительно контроля во всех группах исследуемых (табл. 2). Однако уровень гомоцистеина у пациентов при бактериальных нейроинфекциях регистриро- Содержание гомоцистеина в сыворотке крови у детей при нейроинфекциях, Me±SD Показатель Бактериальные нейроинфекции, n=30 Вирусные нейроинфекции, n=30 Контроль, n=20 Гомоцистеин, мкмоль/л 9,1±0,66 8,42±0,53 5,14±0,71 р ˂0,0001 ˂0,0001 р1 ˂0,01 р – статистическая значимость различий с группой контроля; р1 – статистическая значимость различий между группами с вирусными и бактериальными нейроинфекциями. Содержание гомоцистеина в сыворотке крови у детей при нейроинфекциях, Me±SD Таблица 3 Том 10, № 1, 2018 ЖУРНАЛ ИНФЕКТОЛОГИИ 44 Уровень гомоцистеинемии в исследуемых группах в зависимости от пола, Me±SD Группа Гомоцистеин, мкмоль/л Статистическая значимость, р Девочки Мальчики Бактериальные нейроинфекции, n=30 9,04±1,42 8,99±0,54 ˃0,05 Вирусные нейроинфекции, n=30 8,5±0,52 8,42±0,96 ˃0,05 Контроль, n=20 4,2±1,14 5,8±1,6 ˂0,01 Уровень гомоцистеинемии в исследуемых группах в зависимости от пола, Me±SD 44 Оригинальное исследование высокие показатели гомоцистеина наблюдались при пневмококковой инфекции: так, его уровень в среднем составил 16,05±6,7 мкмоль/л, значитель- но отличаясь не только от группы контроля, но и от показателей пациентов с иной этиологией ней- роинфекции (р<0,001). Максимальные значения гипергомоцистеинемии выявлены у пациента с клиникой менингоэнцефалита пневмококковой этиологии (22,79 мкмоль/л). Кроме того, содер- жание гомоцистеина в случае развития бактери- ального менингоэнцефалита (17,2±3,9 мкмоль/л) превышало аналогичные показатели пациентов с признаками изолированного бактериального ме- нингита в 2 раза (р˂0,001), что может свидетель- ствовать о более интенсивном вовлечении в пато- логический процесс церебральных сосудов. Полученные результаты свидетельствуют, что уровень гомоцистеина в сыворотке крови детей при нейроинфекциях может служить одним из ди- агностических критериев развития эндотелиаль- ной дисфункции, в том числе для оценки тяжести течения инфекционного процесса. Выводы 1. При бактериальных и вирусных нейроин- фекциях повышенный уровень гомоцистеина в сыворотке крови у детей свидетельствует об акти- вации эндотелия. 2. Наибольший уровень гомоцистеинемии реги- стрировался у детей при бактериальных нейроин- фекциях, особенно ассоциированных с Streptococcus pneumoniae, а также в случае сочетанного пораже- ния оболочек и вещества головного мозга. логический процесс церебральных сосудов. Уровень гомоцистеинемии у пациентов с изо- лированным поражением оболочек головного мозга, ассоциированного с вирусными агентами, составил 8,44±0,56 мкмоль/л и не отличался от такового при комбинации серозного менингита с другими проявлениями энтеровирусной инфек- ции (8,32±0,36 мкмоль/л, p≥0,05). Известно, что повреждение сосудистого русла при нейроинфек- ционных заболеваниях является обязательным патогенетическим звеном в связи с преимуще- ственно гематологическим путем распростране- ния возбудителей [14]. Однако проникновение микроорганизмов через гематоэнцефалический барьер и их гематогенная интратекальная цирку- ляция могут обусловливать прямое или опосредо- ванное поражение церебральных вен и артерий, приводя к развитию васкулопатий или васкулитов с признаками сосудистой дисфункции, играющей ключевую роль в характере течения и исходов па- тологического процесса. Выявленное нами повы- шение уровня гомоцистеина у детей при нейро- инфекциях, даже при изолированном поражении вещества и оболочек головного мозга, подтверж- дают имеющиеся данные об активации эндотелия, вероятно, как результат развития церебральных васкулитов или васкулопатий [15]. Особую роль при этом играют агенты бактериального проис- хождения, так как имеют значительные факторы агрессии, благодаря которым они реализуют свои патогенные свойства и вызывают бурную ответ- ную реакцию организма, в том числе и активацию эндотелия. В результате этого нарушается утилиза- ция гомоцистеина и его производных, которые на- капливаются в сосудистом русле и способствуют еще большему поражению сосудистого эндотелия. Поврежденные эндотелиальные клетки начинают потенцировать процессы тромбообразования, воз- никает тромбогенная активация всех компонентов системы гемостаза (сосудистой стенки, тромбоци- тов, плазменно-коагуляционных звеньев), усугу- бляющие в конченом итоге ишемию тканей. 3. Гипергомоцистеинемия у детей при бактериаль- ных и вирусных нейроинфекциях не зависит от пола. ЖУРНАЛ ИНФЕКТОЛОГИИ Том 10, № 1, 2018 Литература: 1. Нейроинфекции у детей / под ред. Н. В. Скрипченко. – СПб.: Тактик-Студио, 2015. – 856 с. 2. Скрипченко, Н.В. Нейроинфекции у детей: тенден- ции и перспективы / Н.В. Скрипченко [и др.] // Российский вестник перинаталогии. – 2016. – 4. – С.9-22. 3. Сорокина, М.Н. Вирусные энцефалиты и менингиты у детей / М.Н. Сорокина, Н.В. Скрипченко. – М.: Медицина, 2004. – 346 с. 4. Марков, Х.М. Молекулярные механизмы дисфунк- ции сосудистого эндотелия / Х.М. Марков // Кардиология. – 2005. – Т. 45, № 12. – С. 62–72. 5. Малеев, В.В. Система гемостаза и состояние эндоте- лия при инфекционной патологии / В.В. Малеев [и др.] // Инфекционные болезни. – 2009. – T.7, № 1. – С. 11–15. 5. Малеев, В.В. Система гемостаза и состояние эндоте- лия при инфекционной патологии / В.В. Малеев [и др.] // Инфекционные болезни. – 2009. – T.7, № 1. – С. 11–15. 6. Азаренок, А.А. Роль вируса гриппа и его поверхност- ных белков в развитии дисфункции эндотелия : автореф. дис. канд. биол. Наук / А.А. Азаренок. – СПб., 2014. – 29 с. 7. Мироманова, Н.А. Маркеры дисфункции эндотелия при осложненных и неосложненных формах гриппа у де- тей / Н.А. Мироманова, А.М. Мироманов // Журнал ин- фектологии. – 2016. – Т. 8, № 4 – С. 66–71. 8. Наумов, А.В. Гомоцистеин. Медико-биологические проблемы: монография / А.В. Наумов. – Минск: Профес- сиональные издания, 2013. – 311 с. 9. Наумов, А.В. Гомоцистеин в патогенезе микроцир- куляторных и тромботических осложнений / А.В. Наумов, Т.Н. Гриневич, В.М. Найдина // Тромбоз, гемостаз и реоло- гия. – 2012. – Т. 49, № 4 – С. 9–19. 10. Клинические аспекты гиперцистеинемии: моногра- фия / под ред. В.А. Снежицкого, В.М. Пырочкина [и др.]. – ГрГМУ, 2011. – 292 с. 11. Working Group IAP/APA Acute Pancreatitis Guidelines. IAP/APA evidence-based guidelines for the management of acute pancreatitis / М. Besselink [et al.] // Pancreatology. – 2013. – 13. – Р. 1-15 12. Дунаевская, С.С. Значение гомоцистеинемии в оценке тяжести острого панкреатита / С.С. Дунаевская, Ю.С. Винник, Д.А. Антюфриева // ЭНИ Забайкальский ме- дицинский вестник. – 2015. – №3. – С. 13–16. 13. Юнкеров, В.И. Математико-статистическая обра- ботка данных медицинских исследований / В.И. Юнкеров, С.Г. Григорьев. – СПб.: ВМедА, 2005. – 292 с. 45 ЖУРНАЛ ИНФЕКТОЛОГИИ Том 10, № 1, 2018 ЖУРНАЛ ИНФЕКТОЛОГИИ Том 10, № 1, 2018 Оригинальное исследование / N.А. Miromanova, A.M. Mironov // Journal of Infectology. – 2016.- T. 8, №4 – P. Литература: 66 – 71 (in Russian). 14. Скрипченко, Н.В. Клинико-лучевая диагностика це- ребральных васкулитов при нейроинфекциях у детей / Н.В. Скрипченко [и др.] // Российский вестник перинаталогии и педиатрии. – 2010. – №1. – С.101–106. 8. Homocysteine. Medico – biological problems: mono- graph / A.V. Naumov. – Minsk: Professional publications, 2013. – 311 s (in Russian). 15. Скрипченко, Н.В. Инфекционные васкулиты: их роль в органной патологии / Н.В. Скрипченко, Т.Н. Трофи- мова, Е.С. Егорова // Журнал инфектологии. – 2010. – Т. 2, № 1. – С. 7–17. 9. Naumov, A.V. Homocysteine in the pathogenesis of mi- crocirculatory and thrombotic complications / A.V. Naumov, TN Grinevich, V.M. Naidina // Thrombosis, hemostasis and rheology. – 2012. – Vol. 49, No. 4 – P. 9-19 (in Russian). 10. Clinical aspects of hypercysteinemia: monograph / ed. V.A. Snezhitsky, V.M. Pyrochkina [and others] – GrGMU, 2011. – 292p (in Russian). Березовская Татьяна Сергеевна – ассистент кафедры детских инфекций Читинской государственной медицинской академии; тел.: 8(3022)35-43-24, факс 8(3022)32-30-58, е-mail: baranchygovats@mail.ru Мироманова Наталья Анатольевна – заведующая кафедрой детских инфекций Читинской государственной медицинской академии, д.м.н., доцент; тел.: 8(3022)35-43-24, факс 8(3022)32-30-58, е-mail: detinf-chita@mail.ru Том 10, № 1, 2018 ЖУРНАЛ ИНФЕКТОЛОГИИ References Petersburg, 2014. – 29 p (in Russian). 15. Skripchenko, N.V. Infectious vasculitis: their role in or- gan pathogens / N.V. Skripchenko, TN. Trofimova, E.S. Egoro- va // Journal of Infectology. – 2010. – Т.2, №1. – P. 7 – 17 (in Russian). 7. Miromanova, N.A. Markers of endothelial dysfunction in complicated and uncomplicated forms of influenza in children 7. Miromanova, N.A. Markers of endothelial dysfunction in complicated and uncomplicated forms of influenza in children Авторский коллектив: References 1. Neuroinfections in children / Ed. N. V. Skripchenko. – SPb .: Tactic-Studio, 2015.- 856 p (in Russian). 1. Neuroinfections in children / Ed. N. V. Skripchenko. – SPb .: Tactic-Studio, 2015.- 856 p (in Russian). 11. Working Group IAP / APA Acute Pancreatitis Guide- lines. IAP / APA evidence-based guidelines for the manage- ment of acute pancreatitis / M. Besselink [et al.] // Pancreatol- ogy. – 2013. – 13. – R. 1-15 2. Skripchenko, N.V. Neuroinfections in children: trends and perspectives. N.V. Skripchenko, M.V. Ivanova, A.A. Vilnitz, E.Yu.Skripchenko // Russian herald of perinatology. – 2016. – 4. – C.9-22 (in Russian). 12. Dunaevskaya, S.S. The importance of homocysteinemia in assessing the severity of acute pancreatitis / SS. Dunaevska- ya, Yu.S. Vinnik, D.A. Antyufrieva // ENI Transbaikal Medical Herald. – 2015. – № 3. – P. 13 – 16 (in Russian). 3. Sorokina, M.N. Viral encephalitis and meningitis in chil- dren / M.N. Sorokina, N.V. Skripchenko. – Moscow: Medi- cine, 2004. – 346 p (in Russian). 4. Markov, H.M. Molecular mechanisms of vascular endo- thelial dysfunction / Kh.M. Markov // Cardiology. – 2005. – Т. 45, №12. – P. 62 – 72 (in Russian). 13. Yunkerov, V.I. Mathematico – statistical processing of medical research data. Yunkerov, S.G. Grigoryev. – SPb .: VM- EDA, 2005. – 292 p (in Russian). 5. Maleev, V.V. The system of hemostasis and the state of the endothelium in infectious pathology / V.V. Maleev [and others] // Infectious diseases. – 2009. – T.7, No. 1. – P. 11-15 (in Russian). 5. Maleev, V.V. The system of hemostasis and the state of the endothelium in infectious pathology / V.V. Maleev [and others] // Infectious diseases. – 2009. – T.7, No. 1. – P. 11-15 (in Russian). 14. Skripchenko, N.V. Clinical and radiation diagnosis of cerebral vasculitis in neuroinfections in children / N.V. Skrip- chenko, TN. Trofimova, E.S. Egorova, N.V. Morgatsky, E.A. Kosmacheva // Russian herald of perinatology and pediatrics. – 2010. – №1. – P.101-106 (in Russian). 6. Azarenok, A.A. The role of the influenza virus and its sur- face proteins in the development of endothelial dysfunction: author›s abstract. dis. Cand. Biol. Sciences / A.A. Azarenok. – St. Petersburg, 2014. – 29 p (in Russian). 6. Azarenok, A.A. The role of the influenza virus and its sur- face proteins in the development of endothelial dysfunction: author›s abstract. dis. Cand. Biol. Sciences / A.A. Azarenok. – St. Березовская Татьяна Сергеевна – ассистент кафедры детских инфекций Читинской государственной медицинской академии; тел.: 8(3022)35-43-24, факс 8(3022)32-30-58, е-mail: baranchygovats@mail.ru Мироманова Наталья Анатольевна – заведующая кафедрой детских инфекций Читинской государственной Авторский коллектив: Том 10, № 1, 2018 ЖУРНАЛ ИНФЕКТОЛОГИИ 46
https://openalex.org/W2015339207	https://ped-rheum.biomedcentral.com/counter/pdf/10.1186/1546-0096-7-20	English	null	The incidence of juvenile rheumatoid arthritis in Quebec: a population data-based study	Pediatric rheumatology online journal	2,009	cc-by	3,694	BioMed Centra Pediatric Rheumatology Open Acces Short Report The incidence of juvenile rheumatoid arthritis in Quebec: a population data-based study Debbie Ehrmann Feldman1, Sasha Bernatsky2 and Michelle Houde3 Address: 1Université de Montréal, Interdisciplinary Research Centre in Rehabilitation, and Public Health Department of Montreal, Pavillon 7077 du Parc, CP 6128, Succ Centre-ville, Montréal, Qc H3C 3J7, Canada, 2McGill University, Montreal, Canada Division of Clinical Epidemiology, Research Institute of the McGill University Health Centre, Royal Victoria Hospital, V Building, 687 Pine Avenue West, Montreal, QC H3A 1A1, Canada and 3Public Health Department of Montreal, Agence de la santé et des services sociaux de Montréal, Direction de Santé Publique, 1301 rue Sherbrooke Est, Montréal (Qc) H2L 1M3, Canada Email: Debbie Ehrmann Feldman - debbie.feldman@umontreal.ca; Sasha Bernatsky - sasha.bernatsky@mail.mcgill.ca; Michelle Houde - mhoude@santepub-mtl.qc.ca * Corresponding author Abstract Objective: To determine the population incidence of juvenile rheumatoid arthritis (JRA) in Published: 19 November 2009 Pediatric Rheumatology 2009, 7:20 doi:10.1186/1546-0096-7-20 Received: 21 July 2009 Accepted: 19 November 2009 This article is available from: http://www.ped-rheum.com/content/7/1/20 © 2009 Feldman et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. BioMed Centra Pediatric Rheumatology Open Acces Short Report The incidence of juvenile rheumatoid arthritis in Quebec: a population data-based study Debbie Ehrmann Feldman1, Sasha Bernatsky2 and Michelle Houde3 Address: 1Université de Montréal, Interdisciplinary Research Centre in Rehabilitation, and Public Health Department of Montreal, Pavillon 7077 du Parc, CP 6128, Succ Centre-ville, Montréal, Qc H3C 3J7, Canada, 2McGill University, Montreal, Canada Division of Clinical Epidemiology, Research Institute of the McGill University Health Centre, Royal Victoria Hospital, V Building, 687 Pine Avenue West, Montreal, QC H3A 1A1, Canada and 3Public Health Department of Montreal, Agence de la santé et des services sociaux de Montréal, Direction de Santé Publique, 1301 rue Sherbrooke Est, Montréal (Qc) H2L 1M3, Canada Email: Debbie Ehrmann Feldman - debbie.feldman@umontreal.ca; Sasha Bernatsky - sasha.bernatsky@mail.mcgill.ca; Michelle Houde - mhoude@santepub-mtl.qc.ca * Corresponding author Abstract Objective: To determine the population incidence of juvenile rheumatoid arthritis (JRA) in Published: 19 November 2009 Pediatric Rheumatology 2009, 7:20 doi:10.1186/1546-0096-7-20 Received: 21 July 2009 Accepted: 19 November 2009 This article is available from: http://www.ped-rheum.com/content/7/1/20 © 2009 Feldman et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. BioMed Central Short Report The incidence of juvenile rheumatoid arthritis in Quebec: a population data-based study y e Ehrmann Feldman1, Sasha Bernatsky2 and Michelle Houde3 Address: 1Université de Montréal, Interdisciplinary Research Centre in Rehabilitation, and Public Health Department of Montreal, Pavillon 7077 du Parc, CP 6128, Succ Centre-ville, Montréal, Qc H3C 3J7, Canada, 2McGill University, Montreal, Canada Division of Clinical Epidemiology, Research Institute of the McGill University Health Centre, Royal Victoria Hospital, V Building, 687 Pine Avenue West, Montreal, QC H3A 1A1, Canada and 3Public Health Department of Montreal, Agence de la santé et des services sociaux de Montréal, Direction de Santé Publique, 1301 rue Sherbrooke Est, Montréal (Qc) H2L 1M3, Canada mail: Debbie Ehrmann Feldman - debbie.feldman@umontreal.ca; Sasha Bernatsky - sasha.bernatsky@mail.mcgill Michelle Houde - mhoude@santepub-mtl.qc.ca * Corresponding author Received: 21 July 2009 Accepted: 19 November 2009 Published: 19 November 2009 Published: 19 November 2009 diatric Rheumatology 2009, 7:20 doi:10.1186/1546-0096-7-20 This article is available from: http://www.ped-rheum.com/content/7/1/20 © 2009 Feldman et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. ; This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Abstract Objective: To determine the population incidence of juvenile rheumatoid arthritis (JRA) in Quebec. Methods: We obtained data from Quebec's physician claims database. Incident cases were defined as having a visit for JRA in 2000, no visit in the previous 3 years, a confirmed diagnosis by an arthritis specialist, or having ≥ 2 visits to any physician for JRA, ≥ 2 months apart but within 2 years. Results: Cumulative incidence of JRA was 17.8/100,000. Mean age at diagnosis was 9.8 ± 4.6 years, 68% were female and more persons were diagnosed in winter. Subjects had a median of 10 medical visits over the first year. Conclusion: Our population based incidence estimate was similar to others. Children and adolescents with JRA are heavy users of medical care. Additional study of environmental or climate- related triggers may be warranted. ment appear promising and early drug therapy combined with rehabilitation can optimize outcomes [3]. ment appear promising and early drug therapy combined with rehabilitation can optimize outcomes [3]. O Short Report The incidence of juvenile rheumatoid arthritis in Quebec: a population data-based study Debbie Ehrmann Feldman*1, Sasha Bernatsky2 and Michelle Houde3 Open Access Page 1 of 4 (page number not for citation purposes) Methods b We obtained data from a physician claims administrative database covering all residents of the province of Québec (the Régie de l'assurance maladie du Québec: RAMQ), after receiving permission from the Commission d'accès à l'information du Québec (Quebec Access to Information Commission). Our study population consisted of all chil- dren, 16 years and under, who had a visit to a physician for JRA (International Classification of Diseases-9 code 714) in the year 2000. Data were available for these chil- dren and adolescents for the period between January 1997 and June 2003. We report the cumulative incidence of juvenile arthritis for the year 2000. In addition, we describe incidence by sex, season, and socio-economic status. Socioeconomic status was based on a validated indicator that utilizes postal code to estimate neighborhood socioeconomic sta- tus and provides an ecological index of material and social deprivation [11]. We dichotomized socioeconomic status at the top two quintiles versus the lower three. We also describe medical service use for the incident JRA cases. Among the 842 children and adolescents with at least one physician visit coded for JRA in the year 2000 (which we considered as suspected cases of JRA), we established that 523 had had no prior physician visits for JRA in the pre- ceding 3 years. For a more specific case-definition of inci- dent JRA among these 523 subjects with suspected JRA, we http://www.ped-rheum.com/content/7/1/20 http://www.ped-rheum.com/content/7/1/20 Pediatric Rheumatology 2009, 7:20 ince of Quebec, Canada using a physician reimbursement database that covers the entire population. We report inci- dence for JRA (as opposed to JIA) since the database con- tains a well-defined code for this diagnosis. included only incident cases that had either been diag- nosed by a pediatric arthritis specialist, or who had ≥ 2 vis- its for JRA at least two months apart but within a two year span [10]. We included as pediatric arthritis specialists all rheumatologists, internists or pediatric immunologists who had provided care for JRA patients in a pediatric hos- pital, and all pediatricians who over the study period had recorded over 50 JRA patient visits. Background g Juvenile idiopathic arthritis (JIA) is a group of inflamma- tory arthropathies that includes three main groups: juve- nile rheumatoid arthritis (JRA), juvenile spondyloarthropathies (or enthesitis-related arthritides) and juvenile psoriatic arthritis [1]. The etiology is unknown and the pathophysiology is poorly understood. Due to the effects of chronic inflammation on joints, affected children often suffer permanent disability, impaired functional status, and poor quality of life in their adult lives [2]. However, recent advances in treat- Epidemiologic studies have noted a relatively wide inci- dence of juvenile arthritis ranging between approximately 10 and 23 per 100,000 persons, with a known female pre- dominance [4-7]. Variations in estimates may be due to differences in diagnostic criteria, case ascertainment, or to regional and time differences [8,9]. Most of the incidence studies to date have relied only on physician reporting. We sought to determine the incidence of JRA in the prov- Page 1 of 4 (page number not for citation purposes) Page 1 of 4 (page number not for citation purposes) Page 2 of 4 (page number not for citation purposes) Discussion Our study produced the first population-based cumula- tive incidence for JRA in Quebec. This was estimated at 17.8 cases per 100,000 Quebec residents aged ≤ 16, a fig- ure within the range of other incidence studies of juvenile arthritis to date. We noted a clear female preponderance of JIA which is in keeping with known demograph- ics,[5,13]. and mean age at diagnosis was also consistent [5,7]. Children and adolescents who are newly diagnosed with JRA appear to be heavy users of medical care, on average experiencing 17.4 physician visits in the first year post- diagnosis. In comparison, 2004 National Ambulatory Medical Care Survey data suggested an average of 2.4 phy- sician visits per year among children aged ≤ 15 years [16]. Brunner et al., in a sample of prevalent JRA patients, reported similar figures to ours in terms of average rheu- matology visits (between 4.4 and 5.8 visits per year) and ophthalmology visits(between1.6 to 2.0) [17]. A British study indicated that children with asthma had, on aver- age, 3.4 to 4.4 annual visits for asthma to out-patient gen- eral practice [18]. Although these figures are comparable to ours with respect to visits for JRA to an arthritis special- ist, total medical visits and pediatric or family physician visits are higher in our cohort, possibly indicative of higher medical care use by children with JRA. Our results are similar to the JRA incidence rate of 19.5/ 100,000 in Finland [5], JIA incidence rate of 21.7/100,000 in Estonia,[7] and 15/100,000 in the Nordic countries [6]. Although we have not validated our algorithm with patient clinical data, the coherence of these rates with other studies that use clinical data suggests that adminis- trative data may be useful to study the epidemiology of JRA. Further, our study covered the entire population of the province of Québec including those living in rural and outlying regions. Our results are similar to the JRA incidence rate of 19.5/ 100,000 in Finland [5], JIA incidence rate of 21.7/100,000 in Estonia,[7] and 15/100,000 in the Nordic countries [6]. Although we have not validated our algorithm with patient clinical data, the coherence of these rates with other studies that use clinical data suggests that adminis- trative data may be useful to study the epidemiology of JRA. Further, our study covered the entire population of the province of Québec including those living in rural and outlying regions. http://www.ped-rheum.com/content/7/1/20 Table 1: Description of Incident Cases of N = 267 Girls High socioeconomic status Season of diagnosis Winter (December 21-March 20) Spring (March 21-June 20) Summer (June21-September 20) Fall (September21-December 20) Descriptive statistics are recorded in Table 1. There was a trend (P < 0.01) suggesting more incident cases in the winter season (December 21-March 20)[12]. In the first year post diagnosis, the median number of total doctor visits was 10 (inter-quartile range (IQR) 6- 17). Medical visits are described in Table 2. Results We identified 267 new cases of JRA in 2000, in the prov- ince of Quebec - 203 were diagnosed by an arthritis spe- cialist and 64 by a pediatrician, family physician or another specialist (Figure 1). Based on census data, there Definition of incident JRA cases in Quebec Figure 1 Definition of incident JRA cases in Quebec. Definition of incident JRA cases in Quebec Figure 1 Definition of incident JRA cases in Quebec. J g Definition of incident JRA cases in Quebec. Page 2 of 4 (page number not for citation purposes) Page 2 of 4 (page number not for citation purposes) http://www.ped-rheum.com/content/7/1/20 http://www.ped-rheum.com/content/7/1/20 Pediatric Rheumatology 2009, 7:20 Table 2: Medical visits for incident cases of JRA (n = 267), over the first 12 months following initial diagnosis 12 months post diagnosis Mean (SD) Median (IQR) All medical visits 17.4 (28.4) 10.0 (6.0-17.0) Arthritis specialist 4.3 (5.6) 3.0 (1.0-5.0) Pediatrician or Family Physician 6.3 (9.6) 3.0 (2.0-7.0) Ophthalmologist 1.8 (1.7) 1.0 (0.0-2.0) Radiologist 0.8 (1.9) 0.0 (0.0-1.0) Other specialist 4.9 (15.4) 1.0 (0.0-4.0) Table 2: Medical visits for incident cases of JRA (n = 267), over the first 12 months following initial diagnosis were approximately 1.5 million Quebec residents aged ≤ 16 in 2001, so we estimate the incidence of JRA in 2000 as 17.8 cases per 100,000 (95% confidence interval: 16.05/ 100,000, 19.55/100,000). The mean age at diagnosis was 9.8 years (standard deviation 4.56, median 10.0). were approximately 1.5 million Quebec residents aged ≤ 16 in 2001, so we estimate the incidence of JRA in 2000 as 17.8 cases per 100,000 (95% confidence interval: 16.05/ 100,000, 19.55/100,000). The mean age at diagnosis was 9.8 years (standard deviation 4.56, median 10.0). 16 in 2001, so we estimate the incide 17.8 cases per 100,000 (95% confid 100,000, 19.55/100,000). The mean 9.8 years (standard deviation 4.56, m Descriptive statistics are recorded in trend (P < 0.01) suggesting more winter season (December 21-March In the first year post diagnosis, the total doctor visits was 10 (inter-qu 17). Medical visits are described in T Discussion Our study produced the first popu tive incidence for JRA in Quebec. T 17.8 cases per 100,000 Quebec resid ure within the range of other inciden arthritis to date. We noted a clear fe of JIA which is in keeping with ics,[5,13]. and mean age at diagnos [5,7]. Our results are similar to the JRA in 100,000 in Finland [5], JIA incidence in Estonia,[7] and 15/100,000 in the Although we have not validated patient clinical data, the coherence other studies that use clinical data s trative data may be useful to study JRA. Further, our study covered the the province of Québec including th outlying regions. According to our data, there was a tre nosis during the winter months. One suggested subsets of JRA may be mo and spring, [14] although this was Canadian study [15]. Unfortunately guish between specific JRA sub-types tive data. Competing interests The authors declare that they have no competing interests. J 11. Pampalon R, Duncan C, Subramanian SV, Jones K: Geographies of health perception in Quebec: a multilevel perspective. Soc Sci Med 1999, 48:1483-1490. Discussion According to our data, there was a trend towards first diag- nosis during the winter months. One study from Israel has suggested subsets of JRA may be more common in winter and spring, [14] although this was not seen in a large Canadian study [15]. Unfortunately, we could not distin- guish between specific JRA sub-types from our administra- tive data. Since our study relies on data from an administrative data- base, there are important potential limitations. There is no indication of severity or type of arthritis- factors which influence medical visits and costs [19]. Nevertheless, our study does cover the entire population of the province of Quebec and provides estimates of JRA incidence and med- ical care use on a population level. Another limitation is that the incidence may have been over-estimated if chil- dren and adolescents with previously diagnosed disease moved to Quebec and were included as incident cases in our study. Table 1: Description of Incident Cases of JRA N = 267 n % Girls 181 67.8 High socioeconomic status 115 43.9 Season of diagnosis Winter (December 21-March 20) 79 29.6 Spring (March 21-June 20) 60 22.5 Summer (June21-September 20) 63 23.6 Fall (September21-December 20) 65 24.3 Table 1: Description of Incident Cases of JRA In conclusion, our population-based study of children and adolescents in Quebec estimated 17.8 incident JRA cases per 100,000 which is similar to physician-reported studies. We confirmed that these individuals are heavy users of medical care, which emphasizes the impact of arthritis in the young. As well, the suggestion of seasonal Page 3 of 4 (page number not for citation purposes) Page 3 of 4 (page number not for citation purposes) Acknowledgements This study was funded by the Canadian Institutes for Health Research. Dr. Ehrmann Feldman was supported by the Arthritis Society, is currently sup- ported by the fonds de la recherche en santé du Québec (FRSQ) and Uni- versité de Montréal, School of Rehabilitation and Dr. Sasha Bernatsky is a Canadian Arthritis Network Scholar and is supported by the Canadian Institutes of Health Research, the Fonds de la recherche en santé du Québec (FRSQ), and the McGill University Health Centre Research Insti- tute and Department of Medicine. The authors would like to acknowledge the support of the Agence de la santé et des services sociaux de Montréal - Direction de la santé publique. 18. Thomas M, Kocevar VS, Zhang Q, Yin DD, Price D: Asthma- related health care resource use among asthmatic children with and without concomitant allergic rhinitis. Pediatrics 2005, 115:129-134. 19. Bernatsky S, Duffy C, Malleson P, Feldman DE, St Pierre Y, Clarke AE: Economic impact of juvenile idiopathic arthritis. Arthritis Rheum 2007, 57:44-48. Authors' contributions 12. Arsham H: Test for seasonality of a time series. 2009 [http:home.ubalt.edu/ntsbarsh/Business-stat/otherapplets/TestSea son.htm]. Ref Type: Electronic Citation DEF was responsible for carrying out the study, involved in conception and design, acquisition of data, analysis, interpretation and writing of the manuscript. 13. Andersson GB: Juvenile arthritis--who gets it, where and when? A review of current data on incidence and prevalence. Clin Exp Rheumatol 1999, 17:367-374. SB helped with conception and design, acquisition of data, analysis, interpretation and critical revision of the manuscript. 14. Uziel Y, Pomeranz A, Brik R, Navon P, Mukamel M, Press J, Barash J, Tauber T, Harel L, Virgilis D, Bibi H, Heldenberg D, Wolach B: Sea- sonal variation in systemic onset juvenile rheumatoid arthri- tis in Israel. J Rheumatol 1999, 26:1187-1189. J 15. Feldman BM, Birdi N, Boone JE, Dent PB, Duffy CM, Ellsworth JE, Lang BA, Laxer RM, Lewkonia RM, Malleson PN, Oen KG, Paquin JD, Rosenberg AM, Schneider R, Silverman ED: Seasonal onset of sys- temic-onset juvenile rheumatoid arthritis. J Pediatr 1996, 129:513-518. MH was involved in data analysis, interpretation, and crit- ical revision of the manuscript. 16. Hing E, Cherry DK, Woodwell DA: National Ambulatory Medi- cal Care Survey: 2004 summary. Adv Data 2006:1-33. 17 B HI T l J B MT C MS K SL M l All authors approved the final submitted version of the manuscript. 17. Brunner HI, Taylor J, Britto MT, Corcoran MS, Kramer SL, Melson PG, Kotagal UR, Graham TB, Passo MH: Differences in disease outcomes between medicaid and privately insured children: possible health disparities in juvenile rheumatoid arthritis. Arthritis Rheum 2006, 55:378-384. List of abbreviations 8. Doran MF, Pond GR, Crowson CS, O'Fallon WM, Gabriel SE: Trends in incidence and mortality in rheumatoid arthritis in Rochester, Minnesota, over a forty-year period. Arthritis Rheum 2002, 46:625-631. JIA: Juvenile idiopathic arthritis; JRA: Juvenile rheumatoid arthritis; SD: Standard deviation; IQR: Inter-quartile range. 9. Manners PJ, Bower C: Worldwide prevalence of juvenile arthri- tis why does it vary so much? J Rheumatol 2002, 29:1520-1530. 10. MacLean CH, Louie R, Leake B, McCaffrey DF, Paulus HE, Brook RH, Shekelle PG: Quality of care for patients with rheumatoid arthritis. JAMA 2000, 284:984-992. http://www.ped-rheum.com/content/7/1/20 Pediatric Rheumatology 2009, 7:20 trends may mean that additional study of environmental triggers in JRA is warranted. trends may mean that additional study of environmental triggers in JRA is warranted. 7. Pruunsild C, Uibo K, Liivamagi H, Tarraste S, Talvik T, Pelkonen P: Incidence of juvenile idiopathic arthritis in children in Esto- nia: a prospective population-based study. Scand J Rheumatol 2007, 36:7-13. Publish with BioMed Central and every scientist can read your work free of charge p 3. Wallace CA: Current management of juvenile idiopathic arthritis. Best Pract Res Clin Rheumatol 2006, 20:279-300. p 3. Wallace CA: Current management of juvenile idiopathic arthritis. Best Pract Res Clin Rheumatol 2006, 20:279-300. 4. Symmons DP, Jones M, Osborne J, Sills J, Southwood TR, Woo P: Pediatric rheumatology in the United Kingdom: data from the British Pediatric Rheumatology Group National Diag- nostic Register. J Rheumatol 1996, 23:1975-1980. g J 5. Kaipiainen-Seppanen O, Savolainen A: Changes in the incidence of juvenile rheumatoid arthritis in Finland. Rheumatology (Oxford) 2001, 40:928-932. 6. Berntson L, Andersson GB, Fasth A, Herlin T, Kristinsson J, Lahdenne P, Marhaug G, Nielsen S, Pelkonen P, Rygg M, Nordic Study Group: Incidence of juvenile idiopathic arthritis in the Nordic coun- tries. A population based study with special reference to the validity of the ILAR and EULAR criteria. J Rheumatol 2003, 30:2275-2282. 6. Berntson L, Andersson GB, Fasth A, Herlin T, Kristinsson J, Lahdenne P, Marhaug G, Nielsen S, Pelkonen P, Rygg M, Nordic Study Group: Incidence of juvenile idiopathic arthritis in the Nordic coun- tries. A population based study with special reference to the validity of the ILAR and EULAR criteria. J Rheumatol 2003, 30:2275-2282. References 1. Petty RE, Southwood TR, Baum J, Bhettay E, Glass DN, Manners P, Maldonado-Cocco J, Suarez-Almazor M, Orozco-Alcala J, Prieur AM: Revision of the proposed classification criteria for juvenile idiopathic arthritis, Durban 1997. J Rheumatol 1998, 25:1991-1994. 1. Petty RE, Southwood TR, Baum J, Bhettay E, Glass DN, Manners P, Maldonado-Cocco J, Suarez-Almazor M, Orozco-Alcala J, Prieur AM: Revision of the proposed classification criteria for juvenile idiopathic arthritis, Durban 1997. J Rheumatol 1998, 25:1991-1994. 2. Minden K, Niewerth M, Listing J, Biedermann T, Bollow M, Schontube M, Zink A: Long-term outcome in patients with juvenile idio- pathic arthritis. Arthritis Rheum 2002, 46:2392-2401. 2. Minden K, Niewerth M, Listing J, Biedermann T, Bollow M, Schontube M, Zink A: Long-term outcome in patients with juvenile idio- pathic arthritis. Arthritis Rheum 2002, 46:2392-2401. Publish with BioMed Central and every scientist can read your work free of charge "BioMed Central will be the most significant development for disseminating the results of biomedical research in our lifetime." Sir Paul Nurse, Cancer Research UK Your research papers will be: available free of charge to the entire biomedical community peer reviewed and published immediately upon acceptance cited in PubMed and archived on PubMed Central yours — you keep the copyright Submit your manuscript here: http://www.biomedcentral.com/info/publishing_adv.asp BioMedcentral Page 4 of 4 (page number not for citation purposes) Publish with BioMed Central and every scientist can read your work free of charge
https://openalex.org/W4361264729	https://figshare.com/articles/journal_contribution/Supplementary_Figure_2_from_Repression_of_RUNX1_Activity_by_EVI1_A_New_Role_of_EVI1_in_Leukemogenesis/22368512/1/files/39813581.pdf	English	null	Supplementary Figure 3 from Repression of RUNX1 Activity by EVI1: A New Role of EVI1 in Leukemogenesis	null	2,023	cc-by	218	EVI1 Detection EVI1 Refe- rence EVI1 K562 cells, lysate K562 cells, IP: αRUNX1 IP: αRUNX1 WB 1 2 3 K562 cells, lysate K562 cells, IP: αRUNX1 RUNX1 RUNX1 Detection IgG IP: αRUNX1 WB 1 2 EVI1 Detection EVI1 Refe- rence EVI1 K562 cells, lysate K562 cells, IP: αRUNX1 IP: αRUNX1 WB 1 2 3 K562 cells, lysate K562 cells, IP: αRUNX1 RUNX1 RUNX1 Detection IgG IP: αRUNX1 WB 1 2 EVI1 Detection EVI1 Refe- rence EVI1 K562 cells, lysate K562 cells, IP: αRUNX1 IP: αRUNX1 WB 1 2 3 EVI1 Detection EVI1 Refe- rence IP: αRUNX1 WB 1 2 3 K562 cells, lysate K562 cells, IP: αRUNX1 RUNX1 RUNX1 Detection IgG IP: αRUNX1 WB 1 2 Supplemental Figure 2S. Endogenous EVI1 and RUNX1 interact. K562 cell proteins (lane 2) were co-IP with anti-RUNX1 antibody (lane 1) and analyzed by Western blot. Lane 3 represents a lysate from K562 cells transiently transfected with an EVI1-expressing plasmid as a size marker. K562 cells, lysate K562 cells, IP: αRUNX1 RUNX1 RUNX1 Detection IgG IP: αRUNX1 WB 1 2 Supplemental Figure 2S. Endogenous EVI1 and RUNX1 interact. K562 cell proteins (lane 2) were co-IP with anti-RUNX1 antibody (lane 1) and analyzed by Western blot. Lane 3 represents a lysate from K562 cells transiently transfected with an EVI1-expressing plasmid as a size marker.
https://openalex.org/W3201969414	https://europepmc.org/articles/pmc8541623?pdf=render	English	null	The Role of IL-6 and ET-1 in the Diagnosis of Coronary MicroVascular Disease in Women	Journal of personalized medicine	2,021	cc-by	11,576	Citation: Gurz˘au, D.; Sitar-T˘aut, A.; Caloian, B.; Gu¸setu, G.; Com¸sa, H.; Frîngu, F.; Zdrenghea, D.; Pop, D. The Role of IL-6 and ET-1 in the Diagnosis of Coronary MicroVascular Disease in Women. J. Pers. Med. 2021, 11, 965. https://doi.org/10.3390/jpm11100965 Keywords: ischemic heart disease in women; microvascular angina; IL6; ET1 Article The Role of IL-6 and ET-1 in the Diagnosis of Coronary MicroVascular Disease in Women u, Adela Sitar-Tăut , Bogdan Caloian , Gabriel Gu¸setu, Hora¸tiu Com¸sa, Florina Frîngu, enghea and Dana Pop Diana Gurzău, Adela Sitar-Tăut , Bogdan Caloian , Gabriel Gu¸setu, Hora¸tiu Com¸sa, Florina Frîngu Dumitru Zdrenghea and Dana Pop Cardiology Department, Clinical Rehabilitation Hospital, “Iuliu Hat,ieganu” University of Medicine and Pharmacy, 400012 Cluj-Napoca, Romania; gurzaudiana@yahoo.com (D.G.); adelasitar@yahoo.com (A.S.-T.); gusetu@gmail.com (G.G.); dh.comsa@gmail.com (H.C.); ﬂorina.fringu@yahoo.com (F.F.); dzdrenghea@yahoo.com (D.Z.); pop67dana@gmail.com (D.P.) * Correspondence: bogdan912@yahoo.com; Tel.: +40-2642-07021 Abstract: Background: Microvascular angina is a common clinical entity, with about a three-fold higher frequency in women. The pathogenesis of microvascular angina has not been much studied, but inﬂammation and endothelial dysfunction have been incriminated as the main mechanisms of this disease. Methoss: Our purpose was to analyze whether certain inﬂammatory markers, i.e., interleukin 6 (IL-6) and endothelin 1 (ET-1), can play a role in the diagnosis of microvascular angina in women. Results: Ninety women with ischemic heart disease were divided into two groups, based on their afﬂiction with either microvascular or macrovascular disease. In general, the levels of IL6 and ET1 were similar between the two groups. Analyzing these marker levels according to the number of coronary lesions, we obtained an increased IL6 value that was similar for patients with microvascular angina, one-vessel, and two-vessel coronary disease, but signiﬁcantly lower than in women with three-vessel coronary lesions. Also, in microvascular angina, IL6 level was correlated with the NYHA IV functional class. Unexpectedly, the level of ET1 was correlated with left ventricular systolic dysfunction. Conclusions: In women with an increased suspicion of microvascular angina, in whom microvascular dysfunction cannot be tested invasively, IL-6 level, unlike the ET-1 level, might be considered a diagnostic marker of this disease. Journal of Personalized Medicine Journal of Personalized Medicine Journal of Personalized Medicine Citation: Gurz˘au, D.; Sitar-T˘aut, A.; Caloian, B.; Gu¸setu, G.; Com¸sa, H.; Frîngu, F.; Zdrenghea, D.; Pop, D. The Role of IL-6 and ET-1 in the Diagnosis of Coronary MicroVascular Disease in Women. J. Pers. Med. 2021, 11, 965. https://doi.org/10.3390/jpm11100965 1. Introduction Ischemic heart disease through microvascular involvement is a common clinical entity, with about a three-fold higher frequency in as women compared with men [1]. The pathogenesis of coronary microvascular disease has not been much studied, as it is a somewhat newer clinical entity. Coronary microvascular involvement refers to coronary microcirculation that is not visible angiographically and comprises all vessels (prearterioles, arterioles, capillaries) with a diameter smaller than 0.4–0.5 mm, circulation that represents about 95% of coronary vascularization and is, in fact, the main site of blood ﬂow self- regulation as well as vascular resistance [2]. These self-regulation mechanisms ensure myocardial blood ﬂow, including in situations of obstruction or signiﬁcant stenosis of the epicardial arteries, through the compensatory vasodilation of microvascular circulation [2]. When this compensatory vasodilation process is inefﬁcient or overcome, it results in a reduction of coronary ﬂow reserve, with the development of ischemia. Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional afﬁl- iations. Copyright: © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). y p The causes considered responsible for deﬁcient vasodilation would be anatomical microcirculation abnormalities, vascular spasm at this level, or the presence of endothelial dysfunction [3,4]. From a practical point of view, the diagnosis of microvascular angina is most frequently established by coronarography, where following visualization of normal epicardial arteries, the coronary microcirculation physiology is assessed by determining coronary ﬂow, ﬂow reserve, as well as vascular resistance. Invasive testing of microvascular J. Pers. Med. 2021, 11, 965. https://doi.org/10.3390/jpm11100965 https://www.mdpi.com/journal/jpm J. Pers. Med. 2021, 11, 965 2 of 16 dysfunction is a procedure that is not routinely performed. This procedure is technically extremely laborious, requiring special software and, of course, an experienced doctor; it is also time consuming, increasing the duration of the procedure by up to 30 min and is much more expensive. Due to this fact, we need other clinical and paraclinical parameters to help us establish the diagnosis of microvascular coronary dysfunction [5]. p g y y The presence of endothelial dysfunction plays a primary role in the development of this particular form of ischemic heart disease. This involves an imbalance of vasodilator and vasoconstrictor peptides, in favor of peptides with a vasoconstrictive effect, of which ET-1(endothelin-1) is the most important. 1. Introduction ET-1 is the most powerful endogenous vaso- constrictor produced by vascular endothelium, involved in the occurrence of endothelial dysfunction and inﬂammation, but also in the process of vascular remodeling [6,7]. ET-1 acts through receptors that are bound to two G proteins, namely endothelin A-ETA receptor and endothelin B-ETB receptor. The effects of ET-1 coupling with the two receptors differ, so by binding to ETA receptors, located mainly in smooth muscle vascular cells, vascular vasoconstriction is mediated, and coupling to ETB receptors, located mainly in endothelial cells, mediates vascular vasodilation [7,8] However, the coupling with the ETA receptor has a greater clinical importance, importance as this coupling is involved in the occurrence of hypertension, atherosclerotic disease, consequently cardiovascular diseases; more, it seems that it has a role in the pathogenesis of chronic kidney disease and in diabetes, especially its vascular complications [7,8]. Its role in the diagnosis and prognosis of patients with microvascular angina has been very little studied. Most of the studies evaluating and conﬁrming the role of endothelin as a mediator of microvascular dysfunction have focused on microvascular dysfunction secondary to percutaneous revascularization in patients with acute coronary syndromes [9], but there are extremely few studies that assess this role in patients with microvascular angina alone. p g Alongside endothelial dysfunction, inﬂammation also plays an extremely important role in coronary microcirculation, with pertinent consequences (there is a close interaction between the two processes) [6]. Inﬂammation, through the presence of cytokines such as IL-6 (interleukin 6), C reactive protein, can also underlie coronary microvascular endothe- lial dysfunction in patients with microvascular angina. IL-6 is a cytokine derived from T lymphocytes, macrophages, and fat cells, having a major role in triggering and maintaining the inﬂammatory cascade by stimulating ﬁbrinogenesis and stimulating C-reactive protein, with consequences for initiating and accelerating the atherosclerotic process [10,11]. Inter- leukin practically stimulates the hepatic production of acute phase proteins, such as those mentioned above, proteins that are known to be strongly involved in the pathogenesis of ischemic coronary heart disease. Also, another link in the pathogenetic chain induced by IL6 that underlies the occurrence of ischemic heart disease is represented by the increase of blood viscosity, with the activation and proliferation of platelets. 1. Introduction Subsequently, at the level of the vascular wall, by activating the monocytes by IL6, the ﬁbrinogen is deposited, after which by inhibition also induced by IL6, the activity and the plasma level of lipase lipopro- teins decrease, resulting in the increase of macrophage lipid absorption. Finally, interleukin 6 will activate the hypothalamic-pituitary-adrenal axis, which will contribute to the onset and maintenance of atherosclerosis, hypertension, obesity, and insulin resistance [11,12]. So far, there are no studies analyzing the role of interleukin 6, consequently inﬂammation, in the diagnosis and prognosis of patients with ischemic heart disease through microvascular involvement, and even less so in female patients. In this context, the current study aims to analyze whether certain inﬂammatory markers, i.e., interleukin 6 (IL-6) and endothelin 1 (ET-1), can play a role in the diagnosis of ischemic heart disease through microvascular involvement in women. 2. Materials and Methods The study included 90 patients diagnosed with ischemic heart disease, admitted to the Clinical Rehabilitation Hospital Cluj-Napoca, Romania, with a mean age of 69.02 ± 9.34 years. J. Pers. Med. 2021, 11, 965 3 of 16 3 of 16 Ischemic heart disease is deﬁned according to the 2019 recommendations of the diag- nostic and treatment guidelines for chronic coronary syndromes of the European Society of Cardiology, based on clinical symptoms, rest ECG, exercise testing, echocardiography and coronarography [13]. Thus, for coronary macrovascular disease, coronary lesions were considered signiﬁcant for stenosis > 70% or >50% in the case of the main trunk [13]. The diagnosis of microvascular angina was made based on typical angina symptoms, correlated with electrocardiographic rest changes, positive exercise test, and coronarographically normal epicardial arteries with delayed coronary ﬂow of the contrast substance. We men- tion that the invasive testing of microvascular dysfunction, by evaluating vasoreactivity, coronary blood ﬂow and coronary ﬂow reserve is not performed routinely, being a very ex- pensive and time-consuming procedure. Precisely for these reasons, we wanted to evaluate other clinical and paraclinical parameters to help us diagnose coronary heart disease. Cardiovascular risk factors and echocardiographic parameters were evaluated for all patients. Also, all women underwent coronarography. The patients included in the study were divided into two groups: the ﬁrst group with coronary microvascular disease (50%) or microvascular angina, and the second group was represented by patients with macrovascular atherosclerotic coronary lesions (50%). The venous blood samples required for the measurement of serum IL-6 and ET-1 concentrations were taken on the day of admission, after obtaining the patients’ informed consent regarding biological sample collection for research purposes and the conﬁden- tiality of the results obtained. By centrifugation at 1500× g, at a temperature of 4 ◦C for 10 min, the serum was separated from the rest of the blood composition, and subsequently this was transferred to 1 mL cryotubes and stored at a temperature of −70 ◦C. Both IL-6 and ET-1 concentrations were measured using the ELISA method. Test calibration was performed according to the manufacturer’s recommendations. The reference interval for ET-1 was between (1–3) pmol/L, and a value considered above the normal limit of IL-6 was ≥6.5 pg/mL. For CRP (C-reactive protein), normal value was considered to be less than 0.6 mg/dL. 2. Materials and Methods Regarding the plasma concentration of IL-6, a single extreme value of 500 pg/mL was recorded, this being found in a patient with a very important atheroscle- rotic load, respectively the patient present with triple vessel coronary heart disease (LAD, CX, RCA), with percutaneous revascularization but with repeated stent thrombosis, with endarterectomy on the right internal carotid artery, with repetitive ischemic stroke, grade III arterial hypertension, major valvulopathies, congestive heart failure, obesity and gas- troesophageal reﬂux disease. Echocardiographically, the telediastolic, telesystolic diameters and the left atrial diam- eter in the parasternal long axis view were measured, the LV ejection fraction (LVEF%) was calculated using the biplane Simpson’s method, which was considered normal at a value higher than 50%, diastolic dysfunction was assessed, and the presence of parietal kinetic disorders was also recorded. The exclusion criteria were represented by the presence of acute or chronic infections, systemic or autoimmune diseases, ongoing or previous neoplastic pathology and recent surgical procedures Statistical analysis was performed in Writer, Presentation and Spreadsheets (WPS) Ofﬁce 2019. For all statistical analyses and subsequent diagrams, R 4.0.0 was used. We considered that p < 0.05 was statistically signiﬁcant and p < 0.10 only showed a tendency to statistical signiﬁcance. g The selected patients were informed about the study protocol and signed an informed consent form. The present clinical study was approved by the local ethics committee (approval number 2606/4.04.2018) and was performed in accordance with the ethical standards established by the 1964 Declaration of Helsinki and its later amendments. 3. Results The general characteristics of the patients included in the study are represented in Table 1. As the table shows, there were no signiﬁcant differences between the two groups J. Pers. Med. 2021, 11, 965 4 of 16 regarding age, associated risk factors, symptoms, except for palpitations which were frequently found in women with coronary microvascular disease. In fact, atrial ﬁbrillation was present in a signiﬁcantly higher proportion in these women compared to those with macrovascular involvement (53.3 vs. 66.66%, p = 0.0201). On the other hand, women with coronary macrovascular disease more frequently had a history of acute coronary syndromes and heart failure. Table 1. Characteristics of the patients included in the study. Table 1. Characteristics of the patients included in the study. 3. Results Parameters Details Global Group 1 Microvascular Disease (Non-Obstructive Coronary Artery Disease) Group 2 Macrovascular Disease (Obstructive Coronary Artery Disease) p Value Number of patients 45 (50%) 45 (50%) Age (Mean ± SD) 69.02 ± 9.34 67.37 ± 9.03 70.66 ± 9.46 p = ns Smoking (%) Yes 30 (33.3) 13 (28.88) 17 (37.77) p = ns No 60 (67.6) 32 (71.11) 28 (62.22) Symptoms (%-yes) Typical angina 54 (60) 23 (51.11) 31 (68.88) p = ns Atypical pain 19 (21.1%) 13 (28.88) 6 (13.33) p = ns Palpitations 37 (41.1%) 25 (55.55) 12 (26.66) p = 0.0101 Dyspnea 67 (74.4%) 33 (73.33) 34 (75.55) p = ns Decreased exercise tolerance 74 (82.2%) 35 (77.77) 39 (86.66) p = ns No of macrovascular coronary lesions (%) One-vessel 15 (33.3) - 15 (33.3) p = ns Two-vessel 15 (33.3) - 15 (33.3) Three-vessel 15 (33.3) - 15 (33.3) Total cholesterol (mg/dL) Mean ± SD 171.9 ±46.55 174.24 ±50.18 169.55 ± 43.05 p = ns LDL cholesterol (mg/dL) Mean ± SD 101.18 ±42.15 105.80 ±43.15 96.57 ±41.09 p = ns HDL cholesterol (mg/dL) Mean ± SD 43.46 ±10.48 43.60 ± 11.85 43.33 ± 9.04 p = ns Triglycerides (mg/dL) Mean ± SD 140.14 ±69.56 (119.5) 125.55 ± 50.88 (117) 154.73 ±82.23 (137) p = ns Uric acid (mg/dL) Mean ± SD 6.76 ±2.08 6.61 ±1.96 6.91 ± 2.2 p = ns CRP (mg/dL) Mean ± SD 1.13 ± 1.26 1.24 ± 1.67 1.01 ± 0.65 p = ns IL-6 * (pg/mL) Mean ± SD 19.66 ± 55.09 (7.7) 12.36 ± 16.36 (7.5) 26.95± 75.9 (8) p = ns Log IL-6 Mean ± SD 0.8528 ± 0.5809 0.7760 ± 0.5664 0.9297 ± 0.5913 p = ns ET-1 * (pmol/L) Mean ± SD 1.67 ± 0.5 (1.7) 1.63 ± 0.42 (1.7) 1.7 ± 0.57 (1.7) p = ns Log ET-1 Mean ± SD 0.20 ± 0.12 () 0.19 ± 0.11 (0.23) 0.21 ± 0.13 (0.23) p = ns Diabetes mellitus (%) Yes 41 (45.6%) 18 (40) 23 (51.11) p = ns No 49 (54.4%) 27 (60) 22 (48.88) Table 1. Characteristics of the patients included in the study. 5 of 16 J. Pers. Med. 2021, 11, 965 Table 1. Cont. 3. Results Parameters Details Global Group 1 Microvascular Disease (Non-Obstructive Coronary Artery Disease) Group 2 Macrovascular Disease (Obstructive Coronary Artery Disease) p Value Obesity (%) Yes 64 (71.1%) 32 (71.11) 32 (71.11) p = ns No 26 (28.9%) 13 (28.88) 13 (28.88) PPH of ACS (%) Yes 33 (36.7%) 5 (11.11) 28 (62.22) p < 0.0001 No 57 (63.3%) 40 (88.88) 17 (37.77) STEMI 15 (45.5%) 2 (4.44) 13(28.88) p < 0.0001 NSTEMI 14 (42.4%) 2(4.44) 12 (26.66) UA 4 (12.1%) 1 (2.22) 3 (6.66) Without 57 (63.3%) 40 (88.88) 17 (37.77) PAD (%) Yes 14 (15.6%) 5 (11.11) 9 (20) p = ns No 76 (84.4%) 40 (88.88) 36 (80) CHF/LVF (%) Yes 68 (75.6%) 28 (62.22) 40(88.88) p = 0.0070 No 22 (24.4%) 17 (37.77) 5(11.11) AFi (%) Yes 48 (53.3%) 30 (66.66) 18 (40) p = 0.0201 No 42 (46.7%) 15 (33.33) 27 (60) Permanent 14 (15.6%) 9 (20) 5 (11.11) p = ns Persistent 10 (11.1%) 6 (13.33) 4 (8.88) Paroxysmal 24 (26.7%) 15 (33.33) 9 (20) Without 42 (46.7%) 15(33.33) 27(60) Ischemic stroke (%) Yes 26 (28.9%) 11 (24.44) 15 (33.33) p = ns No 64 (71.1%) 34 (75.55) 30(66.66) CKD (%) Yes 22 (24.4%) 9 (25) 13 (28.88) p = ns No 68 (75.6%) 36 (75) 32 (71.11) Anxiety-depressive disorder (%) Yes 34 (37.8%) 18 (40) 16 (35.55) p = ns No 56 (62.2%) 27(60) 29 (64.44) IL-6—interleukin 6; ET-1—endothelin 1; ACS—acute coronary syndrome; PPH—personal pathological history; UA—unstable angina; PAD—chronic peripheral artery disease; CHF—congestive heart failure; LVF—left ventricular failure; AFi—atrial ﬁbrillation; CKD—chronic kidney disease; ns—not signiﬁcant. * The normality conditions were not met. Table 1. Cont. IL-6—interleukin 6; ET-1—endothelin 1; ACS—acute coronary syndrome; PPH—personal pathological history; UA—unstable angina; PAD—chronic peripheral artery disease; CHF—congestive heart failure; LVF—left ventricular failure; AFi—atrial ﬁbrillation; CKD—chronic kidney disease; ns—not signiﬁcant. * The normality conditions were not met. When analyzing the correlation between the serum levels of inﬂammatory biomarkers, depending on the degree of coronary involvement, expressed by the number of coronary lesions, we observed a statistically signiﬁcant difference in the group of patients with multivascular coronary lesions (Table 2). The IL-6 value was signiﬁcantly higher in this group compared to patients with one-vessel or two-vessel coronary involvement, as well as to patients with microvascular angina. 3. Results The IL-6 value was similar in the group of patients with microvascular angina and in the group of patients with macrovascular involvement, with one-vessel and two-vessel coronary lesions. There were no differences in ET-1 levels between the two groups or among patients with macrovascular involvement, regardless of the number of affected vessels. These details are presented in Table 2, Figures 1 and 2. J. Pers. Med. 2021, 11, 965 6 of 16 f Table 2. Correlation between serum IL-6 and ET-1 levels and the degree of coronary involvement, explained by the number of coronary lesions. Parameter Number of Coronary Lesions p Value without One-Vessel Two-Vessel Three-Vessel IL-6 (pg/mL) Mean ± SD 12.36 ± 16.36 (7.5) 10.06 ± 10.41 (7.7) 12.71 ± 19.17 (6.6) 58.08 ± 126.76 (12.10) p = 0.030 Log IL-6 Mean ± SD 0.7760 ± 0.5664 0.7752 ± 0.4852 0.8335 ±0.4597 1.1803 ± 0.7412 p = 0.018 SD-standard deviation, p value obtained in the Kruskal–Wallis test and Mann–Whitney tests. ET-1 (pmol/L) Mean ± SD 1.63 ± 0.42 (1.7) 1.78 ± 0.7 (1.7) 1.57 ± 0.41 (1.6) 1.76 ± 0.58 (1.7) p = 0.7316 Log ET-1 Mean ± SD 0.19 ± 0.11 (0.23) 0.22 ± 0.14 (0.23) 0.18 ± 0.12 (0.20) 0.22 ± 0.14 (0.23) p = 0.7316 Table 2. Correlation between serum IL-6 and ET-1 levels and the degree of coronary involvement, explained by the num- ber of coronary lesions. Parameter Number of Coronary Lesions p Value without One-Vessel Two-Vessel Three-Vessel IL-6 (pg/mL) Mean ± SD 12.36 ± 16.36 (7.5) 10.06 ± 10.41 (7.7) 12.71 ± 19.17 (6.6) 58.08 ± 126.76 (12.10) p = 0.030 Log IL-6 Mean ± SD 0.7760 ± 0.5664 0.7752 ± 0.4852 0.8335 ±0.4597 1.1803 ± 0.7412 p = 0.018 SD-standard deviation, p value obtained in the Kruskal–Wallis test and Mann–Whitney tests. ET-1 (pmol/L) Mean ± SD 1.63 ± 0.42 (1.7) 1.78 ± 0.7 (1.7) 1.57 ± 0.41 (1.6) 1.76 ± 0.58 (1.7) p = 0.7316 Log ET-1 Mean ± SD 0.19 ± 0.11 (0.23) 0.22 ± 0.14 (0.23) 0.18 ± 0.12 (0.20) 0.22 ± 0.14 (0.23) p = 0.7316 Figure 1. Relationship between the serum IL-6 value and the number of affected coronary vessels. IL-6 levels were similar in the group of patients with microvascular angina, one and two-vessel coronary artery disease, but statistically significantly higher among patients with three-vessel coro- nary lesions (p = 0.030). 3. Results The plasma concentration of ET-1 did not show signiﬁcant differences either in terms of the type of coronary heart disease, microvascular versus macrovascular, either in terms of the number of coronary arteries involved for patients with obstructive coronary heart disease. 0—microvascular angina; 1—one-vessel coronary disease; 2—two-vessel coronary disease; 3—three-vessel coronary disease. Figure 2. The relationship between plasma ET-1 concentration and the number of coronary lesions. The plasma concentration of ET-1 did not show significant differences either in terms of the type of Figure 2. The relationship between plasma ET-1 concentration and the number of coronary lesions. The plasma concentration of ET-1 did not show signiﬁcant differences either in terms of the type of coronary heart disease, microvascular versus macrovascular, either in terms of the number of coronary arteries involved for patients with obstructive coronary heart disease. 0—microvascular angina; 1—one-vessel coronary disease; 2—two-vessel coronary disease; 3—three-vessel coronary disease. Figure 2. The relationship between plasma ET-1 concentration and the number of coronary lesions. The plasma concentration of ET-1 did not show significant differences either in terms of the type of Figure 2. The relationship between plasma ET-1 concentration and the number of coronary lesions. The plasma concentration of ET-1 did not show signiﬁcant differences either in terms of the type of coronary heart disease, microvascular versus macrovascular, either in terms of the number of coronary arteries involved for patients with obstructive coronary heart disease. 0—microvascular angina; 1—one-vessel coronary disease; 2—two-vessel coronary disease; 3—three-vessel coronary disease. Table 3. The relationship between IL-6 and ET-1 values and patients’ age. Table 3. The relationship between IL-6 and ET-1 values and patients’ age. Global Group 1—Microvascular Disease Group 2—Macrovascular Disease IL-6–age Rho 0.283 p = 0.0075 Rho 0.301 p = 0.0459 Rho 0.230 p = 0.1277 Mean Age 67.37 ± 9.03 70.66 ± 9.46 ET-1–age Rho 0.0963 p = 0.3635 Rho 0.0827 p = 0.5835 Rho 0.0944 p = 0.5313 By analyzing the inﬂuence of IL-6 and ET-1 markers on systolic ventricular function, assessed through the left ventricular ejection fraction value, and on diastolic function, we obtained statistically signiﬁcant correlations only in the group of patients with microvas- cular angina, where the ET-1 level was statistically signiﬁcantly associated with the left ventricular ejection fraction value (Rho −0.440, p = 0.0035). All statistical data are presented in Table 4, Figure 3a,b. Table 4. The correlation between the IL-6/ET-1 value and left ventricular function. 3. Results 0—microvascular angina; 1—one-vessel coronary disease; 2—two-vessel coronary disease; 3—three-vessel coronary disease. Figure 1. Relationship between the serum IL-6 value and the number of affected coronary vessels. IL- 6 levels were similar in the group of patients with microvascular angina, one and two-vessel coronary artery disease, but statistically signiﬁcantly higher among patients with three-vessel coronary lesions (p = 0.030). 0—microvascular angina; 1—one-vessel coronary disease; 2—two-vessel coronary disease; 3—three-vessel coronary disease. Figure 1. Relationship between the serum IL-6 value and the number of affected coronary vessels. IL-6 levels were similar in the group of patients with microvascular angina, one and two-vessel coronary artery disease, but statistically significantly higher among patients with three-vessel coro- nary lesions (p = 0.030). 0—microvascular angina; 1—one-vessel coronary disease; 2—two-vessel coronary disease; 3—three-vessel coronary disease. Figure 1. Relationship between the serum IL-6 value and the number of affected coronary vessels. IL- 6 levels were similar in the group of patients with microvascular angina, one and two-vessel coronary artery disease, but statistically signiﬁcantly higher among patients with three-vessel coronary lesions (p = 0.030). 0—microvascular angina; 1—one-vessel coronary disease; 2—two-vessel coronary disease; 3—three-vessel coronary disease. IL-6 and ET-1 values were not correlated with the LDL cholesterol level recorded in the two groups of patients, for patients in the group with microvascular involvement Rho −0.185, p = 0.283 versus Rho −0.122, p = 0.4181 for patients with macrovascular involvement. When analyzing the relationship between IL-6 level and age, a statistical correlation was observed between these two variables, but when evaluating this interrelation com- paratively between the two groups of patients, the IL-6 value was correlated with age only in the ﬁrst group, of patients with microvascular involvement (Rho 0.301, p = 0.0459), not in the second group (Rho 0.230, p = 0.1277). Regarding the ET-1 value, this was not correlated with patients’ age in the general population of the study or with coronary in- volvement. Table 3 shows data regarding the correlation between the values of biomarkers and patients’ age. 7 of 16 o vessel sel coro- o vessel J. Pers. Med. 2021, 11, 965 Figure 2. The relationship between plasma ET-1 concentration and the number of coronary lesions. The plasma concentration of ET-1 did not show significant differences either in terms of the type of Figure 2. The relationship between plasma ET-1 concentration and the number of coronary lesions. 3. Results Serum IL-6 concentration was not correlated with LVEF%, regardless of the type of coronary heart disease. LVEF%-left ventricular ejection fraction. In contrast, inflammation assessed based on the IL-6 and ET-1 value was not corre- lated with the presence of atrial fibrillation in the group of patients with microvascular involvement or in the group of patients with macrovascular coronary lesions (Table 5). In contrast, inﬂammation assessed based on the IL-6 and ET-1 value was not corre- lated with the presence of atrial ﬁbrillation in the group of patients with microvascular involvement or in the group of patients with macrovascular coronary lesions (Table 5). Table 5. Interrelation between IL-6/ET-1 and the development of atrial fibrillation comparatively between the two groups. Group 1-Microvascular Disease Group 2-Macrovascular Disease with AFi without AFi with AFi without AFi IL-6-ET-1 30 patients- rho 0.193 p = 0.2980 15 patients- rho 0.418 p = 0.1181 18 patients- rho 0.161 p = 0.505 27 patients- rho 0.147 p = 0.4542 Spearman correlation coefficient (R), and p > 0.05 was considered statistically significant. AFi—atrial fibrillation; IL-6—interleukin 6; ET-1—endothelin 1. Table 5. Interrelation between IL-6/ET-1 and the development of atrial ﬁbrillation comparatively between the two groups. Group 1-Microvascular Disease Group 2-Macrovascular Disease with AFi without AFi with AFi without AFi IL-6-ET-1 30 patients- rho 0.193 p = 0.2980 15 patients- rho 0.418 p = 0.1181 18 patients- rho 0.161 p = 0.505 27 patients- rho 0.147 p = 0.4542 Spearman correlation coefﬁcient (R), and p > 0.05 was considered statistically signiﬁcant. AFi—atrial ﬁbrillation; IL-6—interleukin 6; ET-1—endothelin 1. Table 5. Interrelation between IL-6/ET-1 and the development of atrial fibrillation comparatively between Table 5. Interrelation between IL-6/ET-1 and the development of atrial ﬁbrillation comparatively between ation between IL-6/ET-1 and the development of atrial fibrillation comparatively between the two groups. ation between IL-6/ET-1 and the development of atrial ﬁbrillation comparatively between the two groups. Tables 6 and 7 illustrates the correlations between the studied biomarkers and the presence of associated comorbidities depending on the type of coronary involvement. Tables 6 and 7 illustrates the correlations between the studied biomarkers and the presence of associated comorbidities depending on the type of coronary involvement. Table 6. The correlation between the serum levels of IL-6/Log IL-6 and the presence of associated comorbidities in the patients included in the study. 3. Results Global Group 1—Microvascular Disease Group 2—Macrovascular Disease IL-6-LVEF% Rho −0.186 p = 0.0789 Rho −0.263 p = 0.0813 Rho −0.121 p = 0.4225 ET-1-LVEF% Rho −0.203 p = 0.0561 Rho −0.440 p = 0.0035 Rho 0.0261 p = 0.8625 IL-6–diastolic dysfunction Rho 0.128 p = 0.2255 Rho 0.0637 p = 0.6728 Rho 0.197 p = 0.1905 ET-1–diastolic dysfunction Rho 0.0915 p = 0.3881 Rho 0.265 p = 0.0783 Rho −0.123 p = 0.4159 LVEF%—left ventricular ejection fraction. Table 4. The correlation between the IL-6/ET-1 value and left ventricular function. 8 of 16 8 of 15 J. Pers. Med. 2021, 11, 965 J Pers Med 2021 11 x FO Figure 3. (a) Correlation between the ET-1 level and the ejection fraction value in the two groups. ET-1 levels were statis- tically significantly correlated with LVEF% values in patients with microvascular disease, but not in patients with obstruc- tive coronary heart disease. (b) Graphic representation of the correlation between the IL-6 value and the ejection fraction value. Serum IL-6 concentration was not correlated with LVEF%, regardless of the type of coronary heart disease. LVEF%- left ventricular ejection fraction. Figure 3. (a) Correlation between the ET-1 level and the ejection fraction value in the two groups. ET-1 levels were statistically signiﬁcantly correlated with LVEF% values in patients with microvascular disease, but not in patients with obstructive coronary heart disease. (b) Graphic representation of the correlation between the IL-6 value and the ejection fraction value. Serum IL-6 concentration was not correlated with LVEF%, regardless of the type of coronary heart disease. LVEF%-left ventricular ejection fraction. Figure 3. (a) Correlation between the ET-1 level and the ejection fraction value in the two groups. ET-1 levels were statis- tically significantly correlated with LVEF% values in patients with microvascular disease, but not in patients with obstruc- tive coronary heart disease. (b) Graphic representation of the correlation between the IL-6 value and the ejection fraction value. Serum IL-6 concentration was not correlated with LVEF%, regardless of the type of coronary heart disease. LVEF%- left ventricular ejection fraction. Figure 3. (a) Correlation between the ET-1 level and the ejection fraction value in the two groups. ET-1 levels were statistically signiﬁcantly correlated with LVEF% values in patients with microvascular disease, but not in patients with obstructive coronary heart disease. (b) Graphic representation of the correlation between the IL-6 value and the ejection fraction value. 3. Results ET-1 Log ET-1 Mean Standard Deviation Median Mean Standard Deviation Median Diabetes mellitus yes 1.61 0.29 1.60 p = 0.3190 0.20 0.08 0.20 p = 0.31 no 1.72 0.63 1.70 0.20 0.15 0.23 Obesity yes 1.71 0.54 1.70 p = 0.5658 0.21 0.12 0.23 p = 0.56 no 1.59 0.41 1.67 0.18 0.11 0.22 PAD yes 1.85 0.53 1.75 p = 0.0994 0.25 0.11 0.24 p = 0.09 no 1.64 0.50 1.60 0.19 0.12 0.20 CKD yes 1.76 0.62 1.70 p = 0.4442 0.22 0.13 0.23 p = 0.44 no 1.64 0.46 1.67 0.19 0.12 0.22 Ischemic stroke yes 1.71 0.55 1.70 p = 0.3319 0.21 0.14 0.23 p = 0.33 no 1.66 0.49 1.60 0.20 0.11 0.20 yes—with the mentioned comorbidity; no—without the mentioned comorbidity; PAD—chronic peripheral artery disease; CKD—chronic kidney disease. Due to the fact that ET-1 does not comply with normal conditions, logs were calculated. There was a statistically signiﬁcant correlation between the IL-6 level and the NYHA functional class only in the group of patients with microvascular involvement (Figure 4a). Table 6. Cont. IL-6 Log IL-6 Mean Standard Deviation Median Mean Standard Deviation Median CKD yes 14.30 22.53 7.35 p = 0.3312 0.7832 0.5703 0.8662 p = 0.52 no 21.39 62.14 8.10 0.8754 0.5866 0.9085 Ischemic stroke yes 39.69 97.12 9.55 p = 0.0497 1.0791 0.6541 0.9793 p = 0.0176 no 11.52 17.68 7.55 yes—with the mentioned comorbidity; no—without the mentioned comorbidity; PAD—chronic peripheral artery disease; CKD—chronic kidney disease. Due to the fact that IL-6 does not comply with normal conditions, logs were calculated. Table 6. Cont. Table 6. Cont. yes—with the mentioned comorbidity; no—without the mentioned comorbidity; PAD—chronic peripheral artery disease; CKD—chronic kidney disease. Due to the fact that IL-6 does not comply with normal conditions, logs were calculated. Table 7. The correlation between the serum levels of ET-1/Log ET-1 and the presence of associated comorbidities in the patients included in the study. 3. Results IL-6 Log IL-6 Mean Standard Deviation Median Mean Standard Deviation Median Diabetes mellitus yes 14.61 22.94 8.00 p = 0.8174 0.8216 0.5667 0.9031 p = 0.64 no 23.88 71.77 7.50 0.8790 0.5970 0.8751 Obesity yes 24.51 64.66 8.10 p = 0.0353 0.9377 0.5886 0.9085 p = 0.0288 no 7.72 7.90 6.35 0.6439 0.5140 0.8024 PAD yes 20.82 34.63 7.55 p = 0.6047 0.7749 0.7610 0.8779 p = 0.58 no 19.45 58.26 7.75 0.8672 0.5465 0.8893 CKD yes 14.30 22.53 7.35 p = 0.3312 0.7832 0.5703 0.8662 p = 0.52 no 21.39 62.14 8.10 0.8754 0.5866 0.9085 Table 6. The correlation between the serum levels of IL-6/Log IL-6 and the presence of associated comorbidities in the patients included in the study. IL-6 Log IL-6 Mean Standard Deviation Median Mean Standard Deviation Median Diabetes mellitus yes 14.61 22.94 8.00 p = 0.8174 0.8216 0.5667 0.9031 p = 0.64 no 23.88 71.77 7.50 0.8790 0.5970 0.8751 Obesity yes 24.51 64.66 8.10 p = 0.0353 0.9377 0.5886 0.9085 p = 0.0288 no 7.72 7.90 6.35 0.6439 0.5140 0.8024 PAD yes 20.82 34.63 7.55 p = 0.6047 0.7749 0.7610 0.8779 p = 0.58 no 19.45 58.26 7.75 0.8672 0.5465 0.8893 J. Pers. Med. 2021, 11, 965 9 of 16 Table 6. Cont. IL-6 Log IL-6 Mean Standard Deviation Median Mean Standard Deviation Median CKD yes 14.30 22.53 7.35 p = 0.3312 0.7832 0.5703 0.8662 p = 0.52 no 21.39 62.14 8.10 0.8754 0.5866 0.9085 Ischemic stroke yes 39.69 97.12 9.55 p = 0.0497 1.0791 0.6541 0.9793 p = 0.0176 no 11.52 17.68 7.55 yes—with the mentioned comorbidity; no—without the mentioned comorbidity; PAD—chronic peripheral artery disease; CKD—chronic kidney disease. Due to the fact that IL-6 does not comply with normal conditions, logs were calculated. Table 7. The correlation between the serum levels of ET-1/Log ET-1 and the presence of associated comorbidities in the patients included in the study. 3. Results ET-1 Log ET-1 Mean Standard Deviation Median Mean Standard Deviation Median Diabetes mellitus yes 1.61 0.29 1.60 p = 0.3190 0.20 0.08 0.20 p = 0.31 no 1.72 0.63 1.70 0.20 0.15 0.23 Obesity yes 1.71 0.54 1.70 p = 0.5658 0.21 0.12 0.23 p = 0.56 no 1.59 0.41 1.67 0.18 0.11 0.22 PAD yes 1.85 0.53 1.75 p = 0.0994 0.25 0.11 0.24 p = 0.09 no 1.64 0.50 1.60 0.19 0.12 0.20 CKD yes 1.76 0.62 1.70 p = 0.4442 0.22 0.13 0.23 p = 0.44 no 1.64 0.46 1.67 0.19 0.12 0.22 Ischemic stroke yes 1.71 0.55 1.70 p = 0.3319 0.21 0.14 0.23 p = 0.33 no 1.66 0.49 1.60 0.20 0.11 0.20 yes—with the mentioned comorbidity; no—without the mentioned comorbidity; PAD—chronic peripheral artery disease; CKD—chronic kidney disease. Due to the fact that ET-1 does not comply with normal conditions, logs were calculated. Table 7. The correlation between the serum levels of ET-1/Log ET-1 and the presence of associated comorbidities in the patients included in the study. yes—with the mentioned comorbidity; no—without the mentioned comorbidity; PAD—chronic peripheral artery disease; CKD—chronic kidney disease. Due to the fact that ET-1 does not comply with normal conditions, logs were calculated. There was a statistically signiﬁcant correlation between the IL-6 level and the NYHA functional class only in the group of patients with microvascular involvement (Figure 4a). Regarding the relationship between the ET-1 value and the NYHA functional class, no statistical correlation was found regardless of the type of coronary involvement (Figure 4b). 10 of 16 g ) class, no (Fi 10 of 16 g ) class, no (Fi J. Pers. Med. 2021, 11, 965 Figure 4. (a). Relationship between IL-6 value and NYHA functional class comparatively between the two groups. The increased value of IL6 was directly associated with NYHA functional class, but only for patients with microvascular angina, not for those with coronary heart disease. (b). Relationship between ET-1 level and NYHA functional class comparatively between the two groups. The ET1 value was not correlated with the NYHA functional class, regardless of the type of coronary heart disease, microvascular or macrovascular. NYHA class—New York Heart Association, Functional Classification. Figure 4. (a). Relationship between IL-6 value and NYHA functional class comparatively between the two groups. 4. Discussion 4. Discussion It is well known that coronary disease with normal epicardial arteries has been diag- nosed in an extremely low proportion in specialized studies. Thus, in the CRUSADE reg- istry, only 8.6% of patients had myocardial infarction without ST elevation, with normal It is well known that coronary disease with normal epicardial arteries has been di- agnosed in an extremely low proportion in specialized studies. Thus, in the CRUSADE registry, only 8.6% of patients had myocardial infarction without ST elevation, with normal coronary arteries [14]. Also, myocardial infarction with normal epicardial arteries is found in a much higher proportion in women than in men. The latest ﬁndings presented in the GUSTO IIb trial show that myocardial infarction without obstructive coronary lesions was detected in 22% of women versus only 10% of men, which can mainly be explained by the presence of endothelial dysfunction [15]. In the current study, we attempted to evaluate the role of some markers of inﬂammation (IL-6 and ET-1) in women diagnosed with microvascular angina compared to those diagnosed with macrovascular angina. g p g g Thus, we found in the ﬁrst place that between the two forms of coronary involvement, there were no differences regarding the values of lipid fractions, uric acid, CRP, IL-6, or ET-1. This result is extremely important, practically demonstrating the not-so-benign nature of ischemic heart disease through microvascular involvement, as it was believed, until recently, to be. Dyslipidemia plays a central role in the pathogenesis of atherosclerotic disease, the control of dyslipidemia being one of the main pillars of the therapeutic strategy in ischemic coronary macrovascular disease. However, when speaking about the role of dyslipidemia in the pathogenesis of microvascular angina, things are not so clear. So far, there are no studies describing the direct implication of dyslipidemia in the pathophysiology of microvascular angina. An indirect role can be attributed to dyslipidemia through secondary inﬂammation, which is known to play an important role in the development of microvascular angina. This aspect also explains the importance of treatment with statins in patients with microvascular angina [16]. 3. Results The increased value of IL6 was directly associated with NYHA functional class, but only for patients with microvascular angina, not for those with coronary heart disease. (b). Relationship between ET-1 level and NYHA functional class comparatively between the two groups. The ET1 value was not correlated with the NYHA functional class, regardless of the type of coronary heart disease, microvascular or macrovascular. NYHA class—New York Heart Association, Functional Classiﬁcation. Figure 4. (a). Relationship between IL-6 value and NYHA functional class comparatively between the two groups. The increased value of IL6 was directly associated with NYHA functional class, but only for patients with microvascular angina, not for those with coronary heart disease. (b). Relationship between ET-1 level and NYHA functional class comparatively between the two groups. The ET1 value was not correlated with the NYHA functional class, regardless of the type of coronary heart disease, microvascular or macrovascular. NYHA class—New York Heart Association, Functional Classification. Figure 4. (a). Relationship between IL-6 value and NYHA functional class comparatively between the two groups. The increased value of IL6 was directly associated with NYHA functional class, but only for patients with microvascular angina, not for those with coronary heart disease. (b). Relationship between ET-1 level and NYHA functional class comparatively between the two groups. The ET1 value was not correlated with the NYHA functional class, regardless of the type of coronary heart disease, microvascular or macrovascular. NYHA class—New York Heart Association, Functional Classiﬁcation. 4. Discussion 4. Discussion The relationship between uric acid and ischemic coronary macrovascular disease is relatively well known, hyperuricemia increasing the risk of cardiovascular diseases through endothelial dysfunction secondary to oxidative stress, with a reduction of the nitric oxide level, blood platelet activation and an increase in the release of circulating cytokines, these processes underlying the initiation and perpetuation of the atherosclerotic process [17]. At the same time, women with ischemic coronary disease with associated hyperuricemia have adverse cardiovascular events and higher mortality rates [18]. Also, hyperuricemia, per se, represents an inﬂammatory state, so that uric acid may further contribute to the development of microvascular angina. J. Pers. Med. 2021, 11, 965 11 of 16 11 of 16 Literature studies have conﬁrmed the interrelation between high IL-6 levels and coronary disease, as well as the role of IL-6 assessment for the early detection of patients with extensive coronary lesions. It has also been demonstrated that, including at val- ues > 1 pg/mL, in patients with intermediate cardiovascular risk, IL-6 can be an important predictor for ischemic coronary disease [19]. All these aspects are more studied in relation to coronary macrovascular disease compared to the implications of IL-6 level in the patho- genesis of coronary microvascular disease. So far, there are extremely few studies analyzing the direct implications of IL-6 in the development of microvascular angina, but it seems that in women with microvascular angina, there are eight inﬂammatory biomarkers that are associated with microvascular angina, and the majority of biomarkers have followed the pro-inﬂammatory CRP, TNF-α—IL-6 pathway [20]. Consequently, the results of our study conﬁrm the previously mentioned statements. Women in the ﬁrst group, similarly to those of the group with macrovascular involvement, had increased IL-6 values, much above the normal limit. Given the high similarity of the values between the two groups, it could be said that in microvascular angina, endothelial dysfunction expressed by the ET-1 level is as important as in atherosclerotic coronary disease. There are even studies that have identiﬁed a mutation/dysregulation of ET-1 in patients with microvascular angina and support the possibility of using gene therapy with the site of action at the level of the ETA receptor in these patients. p p At the same time, although the plasma concentration of CRP was increased in both groups, the mean values were similar, with no statistically signiﬁcant difference. 4. Discussion 4. Discussion This result conﬁrms that inﬂammation is equally present in both forms of coronary heart disease. Much is known about the relationship between CRP and obstructive coronary heart disease, CRP is not only an excellent biomarker for detecting inﬂammation, but is also directly involved in the pathogenesis of atherosclerosis. Elevated CRP levels are associated with an increased risk of myocardial infarction, peripheral artery disease, sudden cardiac death and ischemic stroke, and, compared to other inﬂammatory biomarkers, it does not show diurnal variations, remaining stable for long periods of time. More importantly, the increased plasma concentration of CRP demonstrated a speciﬁcity in predicting the risk of cardiovascular disease [21]. However, regarding the implications of CRP in microvascular dysfunction, although the information in the literature is not so numerous compared to obstructive coronary disease, it seems that CRP is an important predictor of microvascular dysfunction, with inﬂammation playing a central role in the pathophysiology and complications of microvas- cular heart disease [22]. At the same time, reducing the plasma concentration of CRP by treatment with statins and/or antiplatelet agents seems to improve microvascular dysfunction [23]. By analyzing the correlation of IL-6 with the degree of coronary involvement, we observed that the IL-6 level, therefore a more signiﬁcant inﬂammatory process, was found only in patients with multivascular atherosclerotic lesions. In contrast, the IL-6 level was similar in the group of patients with microvascular angina, with one-vessel and two- vessel coronary lesions. This aspect suggests, on the one hand, marked inﬂammation in microvascular angina, comparable to one-vessel and two-vessel atherosclerotic coronary disease, and so a less benign nature of microvascular angina. This is the ﬁrst study that compares the degree of inﬂammation represented by the IL-6 level between the two types of ischemic coronary disease. On the other hand, the results suggest that a signiﬁcantly increased IL-6 level is a predictor of multivascular coronary disease, which conﬁrms the literature data, namely that a high IL-6 level is associated with a poorer prognosis, a higher risk of infarction and higher cardiovascular mortality [20]. However, ET-1 levels were not associated with the type of ischemic heart disease or with the degree of atherosclerotic coronary involvement. 4. Discussion 4. Discussion This result is somewhat surprising, in the ﬁrst place because we did not expect the mean ET-1 value to be similar between the two groups but, rather, to be higher in the group of patients with atherosclerotic disease, and, in the second place, we expected ET-1 in the second group to be directly proportional J. Pers. Med. 2021, 11, 965 12 of 16 to the number of coronary lesions given the direct implication of ET-1 in the atherosclerotic process. The results obtained in our study contradict the literature data, which demonstrate that the ET-1 level is associated with an advanced atherosclerotic process and progressive ischemic coronary disease [24]. Also, the ET-1 level is a predictor of mortality in patients with trivascular coronary disease [24]. Although we have no explanation for the absence of the differences in plasma ET-1 levels between the two groups, we can conclude that endothelial dysfunction among patients with microvascular angina is as expressed as in the case of coronary macrovascular disease. Consequently, these patients show, in time, an increased risk of developing an atherosclerotic process and macrovascular coronary lesions. p g p y It is well known that, with aging, the risk of cardiovascular diseases increases. In women with menopause the pro-inﬂammatory status becomes more pronounced and the risk of ischemic heart disease increasing signiﬁcantly. However, when analyzing the correlation of inﬂammatory markers with age, the correlation was found only among patients in the ﬁrst group, with microvascular angina. Therefore, in women with microvas- cular angina, the slightly higher IL-6 level obtained was also due to the direct correlation with age and with the consequences of aging. In contrast, in the presence of coronary macrovascular disease, inﬂammation expressed by the IL-6 level is not inﬂuenced by age, being independent of it. Regarding the relationship between the plasma ET-1 level and age, studies demonstrate a direct causality between this and the development of cardiovascular diseases [25]. However, it is not yet clear whether the increase in ET-1 concentration is secondary to pathologies that develop with aging or if ET-1 concentration secondary to aging is causal for disease progression. It is certain that in addition to being involved in the aging process and in chronic diseases, ET-1 plays an extremely important role in the physi- ological processes of the organism [26]. 4. Discussion 4. Discussion 2021, 11, 965 13 of 16 13 of 16 hand, obesity is also associated with OSA (obstructive sleep apnea syndrome) these pa- tients having an increased incidence of cardiovascular disease, especially ischemic coronary heart disease. The pathophysiological mechanisms of OSA involved in ischemic coronary heart disease are represented by chronic activation of the sympathetic system, oxidative stress, endothelial dysfunction, and chronic inﬂammation. [31,32]. Therefore, the levels of markers, such as CRP, TNF-alpha, IL-6, IL-8, but also that of ET1, are increased, and all being strongly involved in the pathogenesis of atherosclerotic disease. A recently published study, which evaluated the correlation between serum levels of ET-1 and LOX-1(lectin-like oxidized low-density lipoprotein receptor-1), both being involved in the atherosclerotic process, in patients with sleep apnea syndrome and the risk of cardiovascular adverse events, demonstrated a direct relationship between the presence of sleep apnea syndrome and the level of ET-1 and, consequently, a higher cardiovascular risk [32]. Indeed, we did not analyze the prevalence of OSA among the obese patients included in the study, but there is a high probability that they also hide this respiratory pathology, which accentuates inﬂammation and endothelial dysfunction and implicitly the atherosclerotic process. This may explain the higher value of IL-6 in the group of patients with obstructive coronary heart disease. IL-6 is also implicated in the development of ischemic stroke, particularly in the occlusion of small brain vessels. Also, the plasma IL-6 level is a predictor of recurrences as well as of the prognosis of patients with ischemic stroke [33]. It has been proven that about one third of patients with ischemic stroke, but without known coronary disease, have in fact coronary lesions in a proportion of over 50%, and 3% of these are at a risk of developing acute myocardial infarction at one year [34]. In its turn, ischemic coronary disease is associated with an increased risk of stroke, which is one of the most redoubtable extracardiac complications of myocardial infarction [35]. All these statements are valid in the case of ischemic heart disease through macrovascular involvement. The interrelation between microvascular angina and stroke is not currently very well known, but taken independently, studies demonstrate the implication of microvascular cerebral dysfunction as a mechanism in the development of ischemic stroke [36]. 4. Discussion 4. Discussion In our study, the ET-1 level was not correlated with patients’ age; consequently, the plasma ET-1 concentration was most probably inﬂuenced by the associated ischemic pathology. We found in our study that patients with microvascular angina had atrial ﬁbrillation in a statistically signiﬁcantly higher proportion. So far, there are no studies that compare the incidence of atrial ﬁbrillation depending on ischemic coronary disease, microvascular and macrovascular, as it is well known that both forms are important etiologies of atrial ﬁbrillation. Neither the IL-6 nor the ET-1 levels were correlated with the presence of atrial ﬁbrillation in either of the two groups. This result could be explained by the small number of cases with associated atrial ﬁbrillation, more precisely 53.3%, as well as by the lower level of the biomarkers, particularly ET-1, while IL-6 has proved to be an important mediator in the pathophysiological process of atrial ﬁbrillation [27]. In its turn, the ET-1 level increases in patients with myocardial ischemia, which results in myocyte hypertrophy, myocardial ﬁbrosis, and atrial dilatation. All this leads to an increase in atrial wall stress and to an alteration of atrial geometry, with consequences for electrical stability and the development of atrial ﬁbrillation. Thus, the increased plasma ET-1 level is a strong and independent predictor of atrial ﬁbrillation, predominantly among patients with ischemic coronary disease and heart failure [28]. An analysis of other comorbidities, in our study, showed that IL-6 was correlated only with the presence of obesity and ischemic stroke, and only in the group of patients with macrovascular coronary lesions. Obesity is also characterised by a pro-inﬂammatory status of more reduced intensity, but chronic, termed metabolic inﬂammation, in which the activated monocytes inﬁltrate the adipose tissue, and differentiate into resident adi- pose macrophages which in turn secrete inﬂammatory cytokines, among which IL-6 [29]. Consequently, obese patients have increased IL-6 values, which results in an increased risk of cardiovascular diseases, insulin resistance, and diabetes mellitus [30]. These data are conﬁrmed by the results of our study, where the IL-6 level in obese patients was indeed cor- related with coronary macrovascular disease. Why was it correlated only with this and not with microvascular angina as well? Things are not very clear, but a plausible explanation would be the fact that obesity through its consequences is more frequently associated with the atherosclerotic process, therefore with coronary macrovascular disease. On the other J. Pers. Med. 4. Discussion 4. Discussion The absence of a correlation between the level of inﬂammation and the occurrence of ischemic stroke among patients with microvascular angina might also be explained by the lower level of inﬂammation among these patients. g p Regarding the inﬂuence of inﬂammatory biomarkers on cardiac functions, on systolic and diastolic function, respectively, we observed no signiﬁcant correlation between these functions and plasma IL-6 level. This result is again surprising, and contradicts the data obtained in previous studies, which practically supported the hypothesis that, including in patients who are asymptomatic and without associated cardiovascular pathology, IL-6 value is indirectly correlated with the presence of regional left ventricular systolic dysfunction, thus representing a pathogenic link to the development of heart failure [37]. This result could be explained in the ﬁrst place by the small number of patients included in the study, and in the second place, possibly by the need for a higher IL-6 level. In contrast, ET-1 is correlated with systolic dysfunction expressed by the ejection fraction value only among patients with microvascular angina, not in the second group. The ET-1 value was similar in the two groups, but this correlation might be explained by the more diffuse location of endothelial dysfunction in patients with microvascular involvement, compared to those with atherosclerotic lesions, where the alterations are more localized and are limited to a certain coronary territory, and a correlation with systolic dysfunction in these patients would require a more extensive endothelial dysfunction, therefore a much higher ET-1 level. Thus, it can be stated that the ET-1 level can be a marker of systolic dysfunction in patients with coronary microvascular involvement. p y By further analyzing the presence of heart failure, an important complication of ischemic heart disease, it should be emphasized that one of the most important cytokines with effects both regarding the inﬂammatory and immune response, which is involved in the pathophysiology of heart failure, is IL-6 [38]. It has been proved that the IL-6 level J. Pers. Med. 2021, 11, 965 14 of 16 continues to increase progressively in patients with heart failure, with the aggravation of the NYHA functional class, and also, the high level is independently associated with the risk of mortality and hospitalization. In our study, increased IL-6 level was correlated with the NYHA functional class only in patients with microvascular angina, the NYHA IV class being diagnosed only in these patients. 4. Discussion 4. Discussion In contrast, although ET-1 is also a biomarker implicated in the pathophysiological processes of heart failure and its plasma concentrations are generally increased both in heart failure with reduced ejection fraction and in heart failure with preserved ejection fraction and have a negative prognostic role, in our study, ET-1 level was not increased and was not associated with the stage of heart failure. This lack of correlation between ET-1 level and the NYHA functional class is explained precisely by this normal level of ET-1 obtained in our study. Finally, we want to emphasize that establishing the role of the two biomarkers in the pathophysiology of ischemic heart disease, and of its complications, requires further studies. 5. Conclusions In patients with an increased suspicion of microvascular angina, in whom microvas- cular dysfunction cannot be invasively assessed, IL-6 levels, unlike the ET-1 levels, in association with other clinical and paraclinical parameters, may be a helpful marker for the diagnosis of microvascular coronary heart disease. Author Contributions: Conceptualization, D.G. and D.P.; Data curation, D.G. and A.S.-T.; Formal analysis, D.G. and G.G.; Investigation, D.G.; Methodology, D.G., A.S.-T.; Resources, D.G. and H.C.; Software, D.G., A.S.-T., F.F.; Supervision, D.Z. and D.P.; Validation, B.C., D.Z. and D.P.; Visualization, D.G. and D.P.; Writing–original draft, D.G. and D.P.; Writing–review and editing, D.G. All authors have read and agreed to the published version of the manuscript. Funding: The study was funded by the institutional research grant “Iuliu Hatieganu” University of Medicine and Pharmacy, Cluj-Napoca, Romania (PCD 2018- No. 1680/45/19.01.2018). Funding: The study was funded by the institutional research grant “Iuliu Hatieganu” University of Medicine and Pharmacy, Cluj-Napoca, Romania (PCD 2018- No. 1680/45/19.01.2018). Institutional Review Board Statement: The present clinical study was approved by the local Ethics Committee of the Rehabilitation Hospital, Cluj-Napoca, Romania (approval number 2606/4.04.2018) and was performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments. Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. Acknowledgments: This paper was published under the frame of European Social Found, Human Capital Operational Programme 2014–2020, project no. POCU/380/6/13/125171. Conﬂicts of Interest: The authors declare no conﬂict of interest. Conﬂicts of Interest: The authors declare no conﬂict of interest. Conﬂicts of Interest: The authors declare no conﬂict of interest. p p p y 7. Sutton, G.; Pugh, D.; Dhaun, N. Developments in the Role of Endothelin-1 in Atherosclerosis: A Potential Therapeutic Target? Am. J. Hypertens. 2019, 32, 813–815. [CrossRef] [PubMed] 6. Yanagisawa, M.; Kurihara, H.; Kimura, S.; Tomobe, Y.; Mitsui, Y.; Yazaki, Y.; Goto, K.; Masaki, T. A novel potent vasoconstrictor peptide produced by vascular endothelial cells. Nature 1988, 332, 411–415. [CrossRef] 5. Ford, T.J.; Ong, P.; Sechtem, U.; Beltrame, J.; Camici, P.G.; Crea, F.; Kaski, J.-C.; Merz, C.N.B.; Pepine, C.J.; Shimokawa, H.; et al. Assessment of Vascular Dysfunction in Patients Without Obstructive Coronary Artery Disease: Why, How, and When. JACC Cardiovasc. Interv. 2020, 13, 1847–1864. [CrossRef] [PubMed] References 1. Ong, P.; Athanasiadis, A.; Borgulya, G.; Vokshi, I.; Bastiaenen, R.; Kubik, S.; Hill, S.; Schäufele, T.; Mahrholdt, H.; Kaski, J.C.; et al. Clinical usefulness, angiographic characteristics, and safety evaluation of intracoronary acetylcholine provocation testing among 921 consecutive white patients with unobstructed coronary arteries. Circulation 2014, 129, 1723–1730. [CrossRef] 2 Lanza G A ; Crea F Primary coronary microvascular dysfunction: Clinical presentation pathophysiology and management 1. Ong, P.; Athanasiadis, A.; Borgulya, G.; Vokshi, I.; Bastiaenen, R.; Kubik, S.; Hill, S.; Schäufele, T.; Mahrholdt, H.; Kaski, J.C.; et al. Clinical usefulness, angiographic characteristics, and safety evaluation of intracoronary acetylcholine provocation testing among 921 consecutive white patients with unobstructed coronary arteries. Circulation 2014, 129, 1723–1730. [CrossRef] 1. Ong, P.; Athanasiadis, A.; Borgulya, G.; Vokshi, I.; Bastiaenen, R.; Kubik, S.; Hill, S.; Schäufele, T.; Mahrholdt, H.; Kaski, J.C.; et al. Clinical usefulness, angiographic characteristics, and safety evaluation of intracoronary acetylcholine provocation testing among 921 consecutive white patients with unobstructed coronary arteries. Circulation 2014, 129, 1723–1730. [CrossRef] p y , , [ ] 2. Lanza, G.A.; Crea, F. Primary coronary microvascular dysfunction: Clinical presentation, pathophysiology, and management. Circulation 2010, 121, 2317–2325. [CrossRef] [PubMed] [ ] [ ] 3. Cocco, G.; Jerie, P. Angina pectoris in patients without ﬂow-limiting coronary artery disease (cardiac syndrome X). A forest of a variety of trees. Cardiol. J. 2015, 22, 605–612. [CrossRef] [PubMed] y 4. Della Rocca, D.G.; Pepine, C.J. Some thoughts on the continuing dilemma of angina pectoris. Eur. Heart J. 2014, 35, 1361–1364. [CrossRef] [PubMed] 5. Ford, T.J.; Ong, P.; Sechtem, U.; Beltrame, J.; Camici, P.G.; Crea, F.; Kaski, J.-C.; Merz, C.N.B.; Pepine, C.J.; Shimokawa, H.; et al. Assessment of Vascular Dysfunction in Patients Without Obstructive Coronary Artery Disease: Why, How, and When. JACC Cardiovasc. Interv. 2020, 13, 1847–1864. [CrossRef] [PubMed] 5. Ford, T.J.; Ong, P.; Sechtem, U.; Beltrame, J.; Camici, P.G.; Crea, F.; Kaski, J.-C.; Merz, C.N.B.; Pepine, C.J.; Shimokawa, H.; et al. Assessment of Vascular Dysfunction in Patients Without Obstructive Coronary Artery Disease: Why, How, and When. JACC Cardiovasc. Interv. 2020, 13, 1847–1864. [CrossRef] [PubMed] 6. Yanagisawa, M.; Kurihara, H.; Kimura, S.; Tomobe, Y.; Mitsui, Y.; Yazaki, Y.; Goto, K.; Masaki, T. A novel potent vasoconstrictor peptide produced by vascular endothelial cells. Nature 1988, 332, 411–415. [CrossRef] 6. Yanagisawa, M.; Kurihara, H.; Kimura, S.; Tomobe, Y.; Mitsui, Y.; Yazaki, Y.; Goto, K.; Masaki, T. A novel potent vasoconstrictor peptide produced by vascular endothelial cells. y 4. Della Rocca, D.G.; Pepine, C.J. Some thoughts on the continuing dilemma of angina pectoris. Eur. Heart J. 2014, 35, 1361–1364. [CrossRef] [PubMed] References Nature 1988, 332, 411–415. [CrossRef] p p p y 7. Sutton, G.; Pugh, D.; Dhaun, N. Developments in the Role of Endothelin-1 in Atherosclerosis: A Potential Therapeutic Target? Am. J. Hypertens. 2019, 32, 813–815. [CrossRef] [PubMed] 15 of 16 J. Pers. Med. 2021, 11, 965 8. Dhaun, N.; Webb, D.J. Endothelins in cardiovascular biology and therapeutics. Nat. Rev. Cardiol. 2019, 16, 491–502. [CrossRef] 9. Guddeti, R.R.; Prasad, A.; Matsuzawa, Y.; Aoki, T.; Rihal, C.; Holmes, D.; Best, P.J.; Lennon, R.; Lerman, L.O.; Lerman, A. Role of endothelin in microvascular dysfunction following percutaneous coronary intervention for non-ST elevation acute coronary syndromes: A single-centre randomised controlled trial. Open Heart 2016, 3, e000428. [CrossRef] 8. Dhaun, N.; Webb, D.J. Endothelins in cardiovascular biology and therapeutics. Nat. Rev. Cardiol. 2019, 16, 491–502. [CrossRef] 9. Guddeti, R.R.; Prasad, A.; Matsuzawa, Y.; Aoki, T.; Rihal, C.; Holmes, D.; Best, P.J.; Lennon, R.; Lerman, L.O.; Lerman, A. Role of endothelin in microvascular dysfunction following percutaneous coronary intervention for non-ST elevation acute coronary syndromes: A single-centre randomised controlled trial. Open Heart 2016, 3, e000428. [CrossRef] y g p [ ] 10. Davies, R.; Choy, E. Clinical experience of IL-6 blockade in rheumatic diseases—Implications on IL-6 biology and disease pathogenesis. Semin. Immunol. 2014, 26, 97–104. [CrossRef] p g 11. Bacchiega, B.C.; Bacchiega, A.B.; Usnayo, M.J.G.; Bedirian, R.; Singh, G.; Pinheiro, G. Interleukin 6 Inhibition and Coronary Artery Disease in a High-Risk Population: A Prospective Community-Based Clinical Study. J. Am. Heart Assoc. 2017, 6, e005038. [CrossRef] 12. Su, D.; Li, Z.; Li, X.; Chen, Y.; Zhang, Y.; Ding, D.; Xia, M.; Qiu, J.; Ling, W. Association between serum interleukin-6 concentration and mortality in patients with coronary artery disease. Med. Inﬂamm. 2013, 2013, 726178. [CrossRef] h S l G bb l k ll l 13. Lazzerini, P.E.; Laghi-Pasini, F.; Acampa, M.; Srivastava, U.; Bertolozzi, I.; Giabbani, B.; Finizola, F.; Vanni, F.; Dokollari, A.; Natale, M.; et al. Systemic Inﬂammation Rapidly Induces Reversible Atrial Electrical Remodeling: The Role of Interleukin-6-Mediated Changes in Connexin Expression. J. Am. Heart Assoc. 2019, 8, e011006. [CrossRef] [PubMed] 14. Knuuti, J.; Wijns, W.; Saraste, A.; Capodanno, D.; Barbato, E.; Funck-Brentano, C.; Prescott, E.; Storey, R.F.; Deaton, C.; Cuisset, T.; et al. ESC Scientiﬁc Document Group, 2019 ESC Guidelines for the diagnosis and management of chronic coronary syndromes: The Task Force for the diagnosis and management of chronic coronary syndromes of the European Society of Cardiology (ESC). Eur. References Heart J. 2020, 41, 407–477. [CrossRef] [PubMed] 15. Patel, M.R.; Chen, A.Y.; Peterson, E.D.; Newby, L.K.; Pollack Jr, V.C.; Brindis, R.G.; Gibson, C.M.; Kleiman, N.S.; Saucedo, J.F.; Bhatt, D.L.; et al. Prevalence, predictors, and outcomes of patients with non-ST-segment elevation myocardial infarction and insigniﬁcant coronary artery disease: Results from the Can Rapid risk stratiﬁcation of Unstable angina patients Suppress Adverse outcomes with Early implementation of the ACC/AHA Guidelines (CRUSADE) initiative. Am. Heart J. 2006, 152, 641–647. [PubMed] 16. Lanza, G.A.; Crea, F. Acute coronary syndromes without obstructive coronary atherosclerosis: The tiles of a complex puzzle. Circ. Cardiovasc. Interv. 2014, 7, 278–281. [CrossRef] [PubMed] 17. Recio-Mayoral, A.; Rimoldi, O.E.; Camici, P.G.; Kaski, J.C. Inﬂammation and microvascular dysfunction in cardiac syndrome X patients without conventional risk factors for coronary artery disease. JACC Cardiovasc. Imaging 2013, 6, 660–667. [CrossRef] 18 Kei A ; Koutsouka F ; Makri A ; Elisaf M Uric acid and cardiovascular risk: What genes can say Int J Clin Pract 2018 17. Recio-Mayoral, A.; Rimoldi, O.E.; Camici, P.G.; Kaski, J.C. Inﬂammation and microvascular dysfunction in cardiac syndrome X patients without conventional risk factors for coronary artery disease. JACC Cardiovasc. Imaging 2013, 6, 660–667. [CrossRef] 18. Kei, A.; Koutsouka, F.; Makri, A.; Elisaf, M. Uric acid and cardiovascular risk: What genes can say. Int. J. Clin. Pract. 2018, 72, e13048. [CrossRef] 19. Wainstein, M.V.; Mossmann, M.; Araujo, G.N.; Gonçalves, S.C.; Gravina, G.L.; Sangalli, M.; Veadrigo, F.; Matte, R.; Reich, R.; Costa, F.G.; et al. Elevated serum interleukin-6 is predictive of coronary artery disease in intermediate risk overweight patients referred for coronary angiography. Diabetol. Metab. Syndr. 2017, 9, 67. [CrossRef] 20. Schroder, J.; Mygind, N.D.; Frestad, D.; Michelsen, M.; Suhrs, H.E.; Bove, K.B.; Gustafsson, I.; Kastrup, J.; Prescott, E. Pro- inﬂammatory biomarkers in women with non-obstructive angina pectoris and coronary microvascular dysfunction. Int. J. Cardiol. Heart Vasc. 2019, 24, 100370. [CrossRef] 21. Shrivastava, A.K.; Singh, H.V.; Raizada, A.; Singh, S.K. C-reactive protein, inﬂammation and coronary heart disease. Egypt. Heart J. 2015, 67, 89–97. [CrossRef] 22. Tong, D.C.; Whitbourn, R.; MacIsaac, A.; Wilson, A.; Burns, A.; Palmer, S.; Layland, J. High-Sensitivity C-Reactive Protein Is a Predictor of Coronary Microvascular Dysfunction in Patients with Ischemic Heart Disease. Front. Cardiovasc. Med. 2018, 4, 81. [CrossRef] 23. Tomai, F. C reactive protein and microvascular function. Heart 2004, 90, 727–728. [CrossRef] 24. References Zhang, C.; Tian, J.; Jiang, L.; Xu, L.; Liu, J.; Zhao, X.; Feng, X.; Wang, D.; Zhang, Y.; Sun, K.; et al. Prognostic Value of Plasma Big Endothelin-1 Level among Patients with Three-Vessel Disease: A Cohort Study. J. Atheroscler. Thromb. 2019, 26, 959–969. [CrossRef] [PubMed] 25. Donato, A.J.; Gano, L.B.; Eskurza, I.; Silver, A.E.; Gates, P.E.; Jablonski, K.; Seals, D.R. Vascular endothelial dysfunction with aging: Endothelin-1 and endothelial nitric oxide synthase. Am. J. Physiol. Heart Circ. Physiol. 2009, 297, H425–H432. [CrossRef] aging: Endothelin-1 and endothelial nitric oxide synthase. Am. J. Physiol. Heart Circ. Physiol. 2009, 297, H425–H432. [CrossRef] 26. Barton, M. Aging and endothelin: Determinants of disease. Life Sci. 2014, 118, 97–109. [CrossRef] [PubMed] 26. Barton, M. Aging and endothelin: Determinants of disease. Life Sci. 2014, 118, 97–109. [CrossRef] [PubMed] g g f 27. Marcus, G.M.; Whooley, M.A.; Glidden, D.V.; Pawlikowska, L.; Zaroff, J.G.; Olgin, J.E. Interleukin-6 and atrial ﬁbrillation in patients with coronary artery disease: Data from the Heart and Soul Study. Am. Heart J. 2008, 155, 303–309. [CrossRef] 28 M F Ni b M Z i k A B d J Gilli A M Ch M K W D R V A i ti f l ft t i l 27. Marcus, G.M.; Whooley, M.A.; Glidden, D.V.; Pawlikowska, L.; Zaroff, J.G.; Olgin, J.E. Interleukin-6 patients with coronary artery disease: Data from the Heart and Soul Study. Am. Heart J. 2008, 155, 303– 28. Mayyas, F.; Niebauer, M.; Zurick, A.; Barnard, J.; Gillinov, A.M.; Chung, M.K.; Wagoner, D.R.V. Association of left atrial endothelin-1 with atrial rhythm, size, and ﬁbrosis in patients with structural heart disease. Circ. Arrhythm. Electrophysiol. 2010, 3, 369–379. [CrossRef] 29. Sindhu, S.; Thomas, R.; Shihab, P.; Sriraman, D.; Behbehani, K.; Ahmad, R. Obesity Is a Positive Modulator of IL-6R and IL-6 Expression in the Subcutaneous Adipose Tissue: Signiﬁcance for Metabolic Inﬂammation. PLoS ONE 2015, 10, e0133494. [CrossRef] 30. El-Mikkawy, D.M.E.; EL-Sadek, M.A.; EL-Badawy, M.A.; Samaha, D. Circulating level of interleukin-6 in relation to body mass indices and lipid proﬁle in Egyptian adults with overweight and obesity. Egypt. Rheumatol. Rehabil. 2020, 47, 7. [CrossRef] 31. Solarz, D.E.; Mullington, J.M.; Meier-Ewert, H.K. Sleep, Inﬂammation and cardiovascular disease. Front. Biosci. 2012, 4, 2490–2501. [CrossRef] 30. El-Mikkawy, D.M.E.; EL-Sadek, M.A.; EL-Badawy, M.A.; Samaha, D. Circulating level of interleukin-6 in relation to body mass indices and lipid proﬁle in Egyptian adults with overweight and obesity. Egypt. Rheumatol. Rehabil. 2020, 47, 7. 32. Kosacka, M.; Brzecka, A. Endothelin-1 and LOX-1 as Markers of Endothelial Dysfunction in Obstructive Sleep Apnea Patients. Int. J. Environ. Res. Public Health 2021, 18, 1319. [CrossRef] 37. Yan, A.T.; Yan, R.T.; Cushman, M.; Redheuil, A.; Tracy, R.P.; Arnett, D.K.; Rosen, B.D.; McClelland, R.L.; Bluemke, D.A.; Lima, J.A.C. Relationship of interleukin-6 with regional and global left-ventricular function in asymptomatic individuals without clinical cardiovascular disease: Insights from the Multi-Ethnic Study of Atherosclerosis. Eur. Heart J. 2010, 31, 875–882. [CrossRef] 38. Gullestad, L.; Ueland, T.; Vinge, L.E.; Finsen, A.; Yndestad, A.; Aukrust, P. Inﬂammatory cytokines in heart failure: Mediators and markers. Cardiology 2012, 122, 23–35. [CrossRef] 37. Yan, A.T.; Yan, R.T.; Cushman, M.; Redheuil, A.; Tracy, R.P.; Arnett, D.K.; Rosen, B.D.; McClelland, R.L.; Bluemke, D.A.; Lima, J.A.C. Relationship of interleukin-6 with regional and global left-ventricular function in asymptomatic individuals without clinical cardiovascular disease: Insights from the Multi-Ethnic Study of Atherosclerosis. Eur. Heart J. 2010, 31, 875–882. [CrossRef] 38 Gullestad L ; Ueland T ; Vinge L E ; Finsen A ; Yndestad A ; Aukrust P Inﬂammatory cytokines in heart failure: Mediators 36. Koton, S.; Schneider, A.L.C.; Windham, B.G.; Mosley, T.H.; Gottesman, R.F.; Coresh, J. Microvascular Brain Disease Progression and Risk of Stroke: The ARIC Study. Stroke 2020, 51, 3264–3270. [CrossRef] [PubMed] p g J [ ] 34. Gunnoo, T.; Hasan, N.; Khan, M.S.; Slark, J.; Bentley, P.; Sharma, P. Quantifying the risk of heart disease following acute ischaemic stroke: A meta-analysis of over 50,000 participants. BMJ Open 2016, 6, e009535. [CrossRef] 35. Putaala, J.; Nieminen, T. Stroke Risk Period After Acute Myocardial Infarction Revised. J. Am. Heart Assoc. 2018, 7, e011200. [CrossRef] [PubMed] J , , [ ] 33. Shaaﬁ, S.; Shariﬁpour, E.; Rahmanifar, R.; Hejazi, S.; Andalib, S.; Nikanfar, M.; Baradarn, B.; Mehdizadeh, R. Interleukin-6, a l bl f f h k l [ b d] c ca ca d ovascu a d sease: s g ts o t e Mu t t c Study o t e osc e os s. ur. eart J. 0 0, 3 , 8 5 88 . [C oss e ] 38. Gullestad, L.; Ueland, T.; Vinge, L.E.; Finsen, A.; Yndestad, A.; Aukrust, P. Inﬂammatory cytokines in heart failure: Mediators and markers. Cardiology 2012, 122, 23–35. [CrossRef] J , , [ ] 33. Shaaﬁ, S.; Shariﬁpour, E.; Rahmanifar, R.; Hejazi, S.; Andalib, S.; Nikanfar, M.; Baradarn, B.; Mehdizadeh, R. Interleukin-6, a reliable prognostic factor for ischemic stroke. Iran. J. Neurol. 2014, 13, 70–76. [PubMed] , , [ ] ﬁpour, E.; Rahmanifar, R.; Hejazi, S.; Andalib, S.; Nikanfar, M.; Baradarn, B.; Mehdizadeh, R. Interleukin- References [CrossRef] p p gyp g y 31. Solarz, D.E.; Mullington, J.M.; Meier-Ewert, H.K. Sleep, Inﬂammation and cardiovascular disease. Front. Biosci. 2012, 4, 2490–2501. [CrossRef] 16 of 16 16 of 16 J. Pers. Med. 2021, 11, 965 32. Kosacka, M.; Brzecka, A. Endothelin-1 and LOX-1 as Markers of Endothelial Dysfunction in Obstructive Sleep Apnea Patients. Int. J. Environ. Res. Public Health 2021, 18, 1319. [CrossRef] J [ ] 33. Shaaﬁ, S.; Shariﬁpour, E.; Rahmanifar, R.; Hejazi, S.; Andalib, S.; Nikanfar, M.; Baradarn, B.; Mehdizadeh, R. Interleukin-6, a reliable prognostic factor for ischemic stroke. Iran. J. Neurol. 2014, 13, 70–76. [PubMed] 33. Shaaﬁ, S.; Shariﬁpour, E.; Rahmanifar, R.; Hejazi, S.; Andalib, S.; Nikanfar, M.; Baradarn, B.; Meh reliable prognostic factor for ischemic stroke. Iran. J. Neurol. 2014, 13, 70–76. [PubMed] p g 34. Gunnoo, T.; Hasan, N.; Khan, M.S.; Slark, J.; Bentley, P.; Sharma, P. Quantifying the risk of heart disease following acute ischaemic stroke: A meta-analysis of over 50,000 participants. BMJ Open 2016, 6, e009535. [CrossRef] 34. Gunnoo, T.; Hasan, N.; Khan, M.S.; Slark, J.; Bentley, P.; Sharma, P. Quantifying the risk of hea stroke: A meta-analysis of over 50,000 participants. BMJ Open 2016, 6, e009535. [CrossRef] 35. Putaala, J.; Nieminen, T. Stroke Risk Period After Acute Myocardial Infarction Revised. J. Am. Heart Assoc. 2018, 7, e011200. [CrossRef] [PubMed] 36. Koton, S.; Schneider, A.L.C.; Windham, B.G.; Mosley, T.H.; Gottesman, R.F.; Coresh, J. Microvascular Brain Disease Progression and Risk of Stroke: The ARIC Study. Stroke 2020, 51, 3264–3270. [CrossRef] [PubMed] 37. Yan, A.T.; Yan, R.T.; Cushman, M.; Redheuil, A.; Tracy, R.P.; Arnett, D.K.; Rosen, B.D.; McClelland, R.L.; Bluemke, D.A.; Lima, J.A.C. Relationship of interleukin-6 with regional and global left-ventricular function in asymptomatic individuals without clinical cardiovascular disease: Insights from the Multi-Ethnic Study of Atherosclerosis. Eur. Heart J. 2010, 31, 875–882. [CrossRef] 38. Gullestad, L.; Ueland, T.; Vinge, L.E.; Finsen, A.; Yndestad, A.; Aukrust, P. Inﬂammatory cytokines in heart failure: Mediators and markers. Cardiology 2012, 122, 23–35. [CrossRef]
https://openalex.org/W4200082658	https://www.frontiersin.org/articles/10.3389/fmed.2021.794299/pdf	English	null	Sustained Release of Gas6 via mPEG-PLGA Nanoparticles Enhances the Therapeutic Effects of MERTK Gene Therapy in RCS Rats	Frontiers in medicine	2,021	cc-by	9,164	ORIGINAL RESEARCH published: 14 December 2021 doi: 10.3389/fmed.2021.794299 ORIGINAL RESEARCH published: 14 December 2021 doi: 10.3389/fmed.2021.794299 Sustained Release of Gas6 via mPEG-PLGA Nanoparticles Enhances the Therapeutic Effects of MERTK Gene Therapy in RCS Rats Shen Wu 1,2†, Yingyan Mao 1,3†, Qian Liu 1,2, Xuejing Yan 1,2, Jingxue Zhang 1,2,3* and Ningli Wang 1,2,3* 1 Beijing Institute of Ophthalmology, Beijing Tongren Eye Center, Beijing Tongren Hospital, Capital Medical University, Beijing Ophthalmology & Visual Sciences Key Laboratory, Beijing, China, 2 Collaborative Innovation Center for Brain Disorders, Beijing Institute of Brain Disorders, Capital Medical University, Beijing, China, 3 Beijing Advanced Innovation Center for Big Data-Based Precision Medicine, Beijing Tongren Hospital, Beihang University, Capital Medical University, Beijing, China Keywords: retinitis pigmentosa, Gas6 nanoparticles, sustained release, phagocytosis, gene therapy Edited by: Shaochong Zhang, Sun Yat-sen University, China Edited by: Shaochong Zhang, Sun Yat-sen University, China Edited by: Shaochong Zhang, Sun Yat-sen University, China Reviewed by: Songtao Yuan, Nanjing Medical University, China Shipeng Wen, Beijing University of Chemical Technology, China Correspondence: Ningli Wang wningli@vip.163.com Jingxue Zhang jingxuezh@ccmu.edu.cn †These authors have contributed equally to this work Previous researches utilizing MER proto-oncogene tyrosine kinase (MERTK) gene therapy in Royal College of Surgeons (RCS) rats evidenced its effectiveness in treating MERTK-associated retinitis pigmentosa (RP). Speciﬁc ligands for receptor tyrosine kinases, such as growth arrest-speciﬁc 6 (Gas6), may enhance retinal phagocytosis via the MERTK receptor, and consequently, enhance the therapeutic effects of gene therapy. In order to overcome the short life effect of the injected Gas6 protein, we constructed a Gas6 loaded methoxy-poly (ethylene glyeol)-poly (lactic-co-glycolic acid) (mPEG-PLGA) nanoparticles (Gas6 NPs) system which allowed for localized and sustained Gas6 protein release, and therefore, a prolonged biological effect. Our data demonstrated that Gas6 protein release from Gas6 NPs preserved the bioactivity and promoted retinal pigment epithelium (RPE) phagocytosis in vitro. In vivo studies showed that RCS rats in the hMERTK/Gas6 NPs group exhibiting the highest electroretinogram responses and more complete retinal structure than that in other groups, further demonstrating that the co-administration of AAV2-BEST1-hMERTK and Gas6 NPs could protect photoreceptors from degeneration. These ﬁndings strongly suggest that Gas6 NPs are a promising method to enable the sustained release of Gas6 protein and could therefore enhance the therapeutic effects of gene therapy for MERTK-associated RP. Reviewed by: Songtao Yuan, Nanjing Medical University, China Shipeng Wen, Beijing University of Chemical Technology, China Correspondence: Ningli Wang wningli@vip.163.com Jingxue Zhang jingxuezh@ccmu.edu.cn †These authors have contributed equally to this work Specialty section: This article was submitted to Ophthalmology, a section of the journal Frontiers in Medicine Specialty section: This article was submitted to Ophthalmology, a section of the journal Frontiers in Medicine Received: 13 October 2021 Accepted: 24 November 2021 Published: 14 December 2021 Received: 13 October 2021 Accepted: 24 November 2021 Published: 14 December 2021 Characterization and in vitro Protein Release of Gas6 NPs Gas6 NPs were prepared using the double emulsion (w/o/w) technique as illustrated in Figure 1A. The encapsulation of Gas6 into mPEG-PLGA nanoparticles was conﬁrmed by FTIR spectroscopy (Figure 2A). In the FTIR spectrum of Gas6, the peaks at 1,633 and 3,339 cm−1 are attributed to N-H deformation and C=O stretching vibrations, respectively. Compared to the FTIR spectrum of pure mPEG-PLGA nanoparticles (blank NPs), the peaks at 1,633 and 3,339 cm−1 corresponding to the signal of Gas6 appeared in the FTIR spectrum of Gas6 NPs, conﬁrming the presence of Gas6 in Gas6 NPs. Numerous studies have reported on the eﬀectiveness of gene replacement therapy for MERTK-associated RP (7, 8). For example, by transplanting an RPE-speciﬁc AAV vector, AAV-VMD2-hMerTK, into subretinal space or vitreous cavity could provide long-term photoreceptor rescue in the RCS rats (MERTK-associated retinal dystrophy model) (9, 10). However, although pre-clinical animal models and initial clinical trials suggested a beneﬁcial eﬀect, double- blinded clinical trials with large cohorts of patients failed to show eﬃcacy (11). These disappointing results were attributed, at least partially, long-term MerTK mutations can lead to destruction of retina microenvironment (12). Clearly, it is important to explore more eﬀective therapies to rescue retinal function and morphology in individuals with MERTK-associated RP, especially after the initiation of retinal degeneration. The encapsulation percentage of the resultant Gas6 NPs was 75%. A TEM image of the resultant Gas6 NPs is presented in Figure 2B. It is observed that Gas6 NPs have a well-deﬁned spherical morphology and high uniformity. The average hydrodynamic diameter of Gas6 NPs determined by DLS was 175.3 nm (Figure 2C). DLS data also evidenced low polydispersity and asymmetric size distributions. As observed by DLS, the size of Gas6 NPs in room temperature water can be maintained at ∼180 nm for at least 7 days (Figure 2D). Combining the results of FTIR spectroscopy, TEM observations, and DLS characterizations, it can be concluded that Gas6 was successfully encapsulated into Gas6 NPs. Speciﬁc ligands for receptor tyrosine kinases, such as growth arrest-speciﬁc 6 (Gas6), may enhance retinal phagocytosis via the MERTK receptor (13). Modulating local environmental factors by Gas6 to provide conditions that are more conducive for functional rescue and repair may maximize the therapeutic eﬀect for RP due to phagocytic dysfunction. RESULTS dystrophies revealed ∼3% of RP cases are attributable to MERTK mutations (4, 5). Such mutations result in defective phagocytosis, which causes the retinal pigment epithelium (RPE) failing to shed photoreceptor outer segments (6). INTRODUCTION Wu S, Mao Y, Liu Q, Yan X, Zhang J and Wang N (2021) Sustained Release of Gas6 via mPEG-PLGA Nanoparticles Enhances the Therapeutic Effects of MERTK Gene Therapy in RCS Rats. Front. Med. 8:794299. doi: 10.3389/fmed.2021.794299 Retinitis pigmentosa (RP), a group of progressive, hereditary diseases that causes irreversible vision loss, is responsible for blindness in more than 2 million people worldwide (1, 2). Mutations in more than 70 genes have been associated with RP (1). One of the mutation genes is the MER proto-oncogene tyrosine kinase (MERTK), which encodes a transmembrane receptor tyrosine kinase, have been identiﬁed to cause RP in patients (3). Two large-scale molecular surveys of retinal Retinitis pigmentosa (RP), a group of progressive, hereditary diseases that causes irreversible vision loss, is responsible for blindness in more than 2 million people worldwide (1, 2). Mutations in more than 70 genes have been associated with RP (1). One of the mutation genes is the MER proto-oncogene tyrosine kinase (MERTK), which encodes a transmembrane receptor tyrosine kinase, have been identiﬁed to cause RP in patients (3). Two large-scale molecular surveys of retinal December 2021 \| Volume 8 \| Article 794299 Frontiers in Medicine \| www.frontiersin.org Gas6 NPs Enhances Retinal Phagocytosis Wu et al. Characterization and in vitro Protein Release of Gas6 NPs However, using Gas6 protein as a drug has many disadvantages, such as short half- life and chemical instability in vivo, and may necessitate frequent intraocular injections to maintain long-term eﬀects. This can lead to many complications, such as inﬂammation, bleeding, and patient compliance issues, which will greatly limit its widespread practical application. To overcome these drawbacks, sustained- release formulations that deliver the protein continuously, thus maintaining the concentration within the therapeutic window for an extended period, have been explored (14). Encapsulating proteins into injectable microspheres or nanoparticles comprised of biodegradable polymers ensures that it maintains its properties and activities (15, 16). Polymers derived from D,L-lactic and glycolic acids, like poly(lactide-co-glycolide) (PLGA), are widely employed with the latter aim in mind (17, 18). PLGA has been approved for use in drug and protein delivery systems by the United States Food and Drug Administration (U.S. FDA) due to its controlled and sustained- release properties, low toxicity, and biocompatibility with tissue and cells. Gas6 protein release kinetics were determined by the Human Gas6 Elisa Kit (19). Based on the standard curve established in advance, the cumulative amount of Gas6 released at each time point was calculated. The Gas6 release proﬁle exhibited a gradual sustained release pattern throughout the experimental period (Figure 2E) and indicated that Gas6 can be continuously released for more than 2 weeks. Frontiers in Medicine \| www.frontiersin.org In vitro Biocompatibility and Bioactivity of Gas6 NPs In order to be clinically applicable, a material must have excellent biocompatibility. We therefore investigated the eﬀect of Gas6 NPs on the growth and proliferation of hfRPE cells to determine its biocompatibility. As depicted in Figure 3A, the hfRPE cells co-cultured with Gas6 NPs exhibited similar cell morphology and proliferation as the control group. Moreover, in vitro biocompatibility of Gas6 NPs was assessed using CCK8 (Figure 3B). The results demonstrated that the absorbance of CCK8 in the group treated with Gas6 NPs was almost the same as that of the control group before 4 d, but was slightly lower than that of the control group from 5 d to 7 d. We then calculated the relative proliferation rate, which represents cell viability. As indicated in Figure 3C, the cell viability of Gas6 NPs decreases from day 1 to 6 and stabilizes on day 7. The cell viability is 87.61 ± 1.23% on day 7 in the group treated with Gas6 NPs. According to the evaluation criterion in the International Standard ISO10993, the cytotoxicity of Gas6 NPs can be given a ranking of 1 and qualiﬁed. This result demonstrated that the Gas6 NPs system has good biocompatibility. In this study, Gas6 protein was encapsulated into methoxy- poly (ethylene glyeol)-poly (lactic-co-glycolic acid) (mPEG- PLGA) nanoparticles (Gas6 NPs) using the double emulsion technique (Figure 1A). We investigated the bioactivity of Gas6 protein released from Gas6 NPs in vitro. To assess whether Gas6 NPs enhance the therapeutic eﬀects of gene therapy for MERTK-associated RP in vivo, we co-transplanted AAV2- BEST1-hMERTK and Gas6 NPs (hMERTK/Gas6 NPs) into RCS rats to demonstrate its therapeutic potential in terms of visual function (Figure 1B). In addition, we examined whether the AAV2-BEST1-hMERTK/Gas6 NPs system is eﬀective in rescuing photoreceptors from degeneration in RCS rats. Thereafter, we examined the eﬀects of Gas6 and Gas6 NPs on the phagocytic function of hfRPE cells by phagocytosis December 2021 \| Volume 8 \| Article 794299 Frontiers in Medicine \| www.frontiersin.org 2 Wu et al. Gas6 NPs Enhances Retinal Phagocytosis FIGURE 1 \| Schematic illustration of (A) the preparation of Gas6 NPs and (B) Evaluation of the protect effects by co-administration of AAV2-BEST1-hMERTK and Gas6 NPs in RCS rats. dpi: days post injection. FIGURE 1 \| Schematic illustration of (A) the preparation of Gas6 NPs and (B) Evaluation of the protect effects by co-administration of AAV2-BEST1-hMERTK and Gas6 NPs in RCS rats. In vitro Biocompatibility and Bioactivity of Gas6 NPs dpi: days post injection. FIGURE 1 \| Schematic illustration of (A) the preparation of Gas6 NPs and (B) Evaluation of the protect effects by co-administration of AAV2-BEST1-hMERTK and Gas6 NPs in RCS rats. dpi: days post injection. FIGURE 1 \| Schematic illustration of (A) the preparation of Gas6 NPs and (B) Evaluation of the protect effects by co-administration of AAV2-BEST1-hMERTK and Gas6 NPs in RCS rats. dpi: days post injection. FIGURE 1 \| Schematic illustration of (A) the preparation of Gas6 NPs and (B) Evaluation of the protect effects by co-administration of AAV2-BEST1-hMERTK and Gas6 NPs in RCS rats. dpi: days post injection. FIGURE 2 \| Characterization of Gas6 NPs. (A) FTIR spectra of Gas6, NPs, and Gas6 NPs. (B) TEM image of Gas6 NPs. Scale bar=200 nm. (C) DLS characterization of Gas6 NPs. (D) Stability of Gas6 NPs sizes during 7 days characterized by DLS (n = 3). (E) Cumulative Gas6 release of Gas6 NPs (n = 3) at 37◦C. FIGURE 2 \| Characterization of Gas6 NPs. (A) FTIR spectra of Gas6, NPs, and Gas6 NPs. (B) TEM image of Gas6 NPs. Scale bar=200 nm. (C) DLS characterization of Gas6 NPs. (D) Stability of Gas6 NPs sizes during 7 days characterized by DLS (n = 3). (E) Cumulative Gas6 release of Gas6 NPs (n = 3) at 37◦C. December 2021 \| Volume 8 \| Article 794299 3 Frontiers in Medicine \| www.frontiersin.org Wu et al. Gas6 NPs Enhances Retinal Phagocytosis FIGURE 3 \| Results of the in vitro biocompatibility and cytotoxicity. (A) Microscope images of hfRPE cells cultured for 1, 3, and 7 d. (B) CCK-8 OD450 of hfRPE cells co-cultured with/without Gas6 NPs during 7 days. (C) Cell viability of hfRPE cells co-cultured with Gas6 NPs during 7 days. The data are expressed as mean ± SD (All tests were performed in triplicate, bar = 50 µm). P < 0.05. FIGURE 3 \| Results of the in vitro biocompatibility and cytotoxicity. (A) Microscope images of hfRPE cells cultured for 1, 3, and 7 d. (B) CCK-8 OD450 of hfRPE cells co-cultured with/without Gas6 NPs during 7 days. (C) Cell viability of hfRPE cells co-cultured with Gas6 NPs during 7 days. The data are expressed as mean ± SD (All tests were performed in triplicate, bar = 50 µm). P < 0.05. blotting (Figure 5A). Frontiers in Medicine \| www.frontiersin.org In vitro Biocompatibility and Bioactivity of Gas6 NPs The fold changes of pFAKY861, GTP-Rac1, and pAKT473 in hfRPE cells pretreated with Gas6 (4.911 ± 3.262, 2.744 ± 1.181, and 7.061 ± 3.547) and Gas6 NPs (5.591± 2.412, 2.784 ± 1.159, and 9.680± 6.285) were higher than that of the control group 1.5 h after the beads were added (Figure 5B). When the incubation time was extended to 3 h, the expression of active proteins remained higher than that of the control group and fold changes were computed as 2.300 ± 0.985 (pFAK861), 1.939 ± 0.232 (GTP-Rac1), and 2.352 ± 0.513 (pAKT473) pretreated with Gas6 and 2.667 ± 1.250 (pFAK861), 1.957 ± 0.348 (GTP-Rac1), and 2.266 ± 0.845 (pAKT473) pretreated with Gas6 NPs (Figure 5B). Collectively, these results suggest that Gas6 NPs provide a safe and eﬀective mean for delivery of Gas6 in vitro. assay. When pretreating with Gas6 protein and Gas6 NPs, the location of the ﬂuorescent beads engulfed by hfRPE cells was observed by laser confocal microscope (Figure 4A). The number of ﬂuorescent beads engulfed by hfRPE cells in both Gas6 and Gas6/NPs groups were signiﬁcantly higher than that of the control group at 1.5 h (8.95 ± 2.30 and 9.15 ± 1.51, respectively, compared to 4.43 ± 1.13 engulfed beads/cell) and 3 h (14.88 ± 5.54 and 15.08 ± 5.51, respectively, compared to 7.33 ± 1.47 engulfed beads/cell) (Figure 4B). The Gas6 protein is the ligand of the MERTK receptor and can enhance phagocytosis by RPE cells via the MERTK- FakY861-Rac1 signaling pathway (13). Therefore, we detected key proteins involved in Gas6-induced phagocytosis by western December 2021 \| Volume 8 \| Article 794299 4 Wu et al. Gas6 NPs Enhances Retinal Phagocytosis FIGURE 4 \| Phagocytosis Assay of human fRPE cells pretreating with Gas6 and Gas6 NPs at 1.5 and 3 h. (A) Laser Confocal Microscope images of hfRPE cells pretreating with Gas6 and Gas6 NPs at 1.5 and 3 h. (B) Average number of ﬂuorescent beads engulfed by single hfRPE cell. The data are expressed as mean ± SD (tests were performed in triplicate and 40 cells were counted in all, bar = 25 µm). P < 0.05. In vivo Rescue of MERTK-Associated R i l D i b G 6 NP RCS rats (Figure 6) following the subretinal injection of AAV2- h FIGURE 4 \| Phagocytosis Assay of human fRPE cells pretreating with Gas6 and Gas6 NPs at 1.5 and 3 h. In vitro Biocompatibility and Bioactivity of Gas6 NPs hMERTK/Gas6 group (49.35 ± 5.78 µm and 25.77 ± 4.95 µm,). The rats were ﬁnally sacriﬁced and the eyeballs were taken out. The H&E staining images of retina were shown in Figure 9A. The results evidenced that the ONL nuclear layers in control group (3 ± 0.71) was signiﬁcantly less than that of the hMERTK group (6.11 ± 1.23), hMERTK/Gas6 group (6.56 ± 1.24), and hMERTK/Gas6 NPs group (10.89 ± 1.36) (Figure 9B). A statistically signiﬁcant diﬀerence was observed between the hMERTK/Gas6 NP group and the other 2 treatment groups. However, there was no signiﬁcant diﬀerence between the hMERTK and hMERTK/Gas6 group. And the result of the ONL thickness is consistent with the result of ONL nuclear layers (Supplementary Figure 2). Altogether, these results suggest that the combined treatment of AAV2-BEST1-hMERTK and Gas6 NPs had a greater protective eﬀect on photoreceptors. We further evaluated the architecture in the retina of the RCS rats via OCT. Representative OCT images of the retina were taken horizontally across the optic nerve head (ONH), and the imaging location was marked on the fundus image with a black line (Figure 8A). Then we quantitatively measure the thickness of the outer nuclear layer (ONL) at 100, 200, 400-µm of temporal retina and 100, 200, 400, 600-µm of nasal retina from the ONH (Figure 8B). We can clearly note that the ONL thickness Frontiers in Medicine \| www.frontiersin.org In vitro Biocompatibility and Bioactivity of Gas6 NPs FIGURE 6 \| Detection of MERTK expression in RCS rats’ retinas following delivery of AAV2-BEST1-hMERTK. hMERTK (labeled as red) expression by immunoﬂuorescence (bar = 50 µm). f MERTK expression in RCS rats’ retinas following delivery of AAV2-BEST1-hMERTK. hMERTK (labeled as red) expression by ar = 50 µm). FIGURE 6 \| Detection of MERTK expression in RCS rats’ retinas following delivery of AAV2-BEST1-hMERTK. hMERTK (labeled as red) expression by immunoﬂuorescence (bar = 50 µm). NPs intervention) and 19 dpi (12 days after Gas6 and Gas6 NPs intervention). The RCS rats in all groups demonstrated typical ERG responses with a and b-waves (Figure 7 and Supplementary Figure 1). At 12 dpi, the hMERTK/Gas6 NPs group exhibited signiﬁcantly higher b-wave (295.533 ± 61.598 µV) in dark-adapted ERG responses at 0.01 cd.s/m2 than other groups. The b-wave amplitudes of the hMERTK group (199.022 ± 28.187 µV) and hMERTK/Gas6 group (214.114 ± 50.677 µV) were also higher than that of the control group (146.179 ± 29.720 µV) at 0.01 cd.s/m2; there is no signiﬁcant diﬀerence between the hMERTK group and hMERTK/Gas6 group (Figure 7A).To evaluate the sustained protective eﬀect of the combined treatment, we also recorded ERGs at 19 dpi. As illustrated in Figure 7B, the dark-adapted ERG responses in the hMERTK, hMERTK/Gas6 and hMERTK/Gas6 NPs groups were all signiﬁcantly higher than that of the control group at low light intensity (0.01 cd.s/m2). Moreover, the hMERTK/Gas6 NPs group had the largest b-wave amplitude among the 3 treatment groups, similar to that recorded at 12 dpi. There was no signiﬁcant diﬀerence between the hMERTK and hMERTK/Gas6 groups at 19 dpi too. NPs intervention) and 19 dpi (12 days after Gas6 and Gas6 NPs intervention). The RCS rats in all groups demonstrated typical ERG responses with a and b-waves (Figure 7 and Supplementary Figure 1). At 12 dpi, the hMERTK/Gas6 NPs group exhibited signiﬁcantly higher b-wave (295.533 ± 61.598 µV) in dark-adapted ERG responses at 0.01 cd.s/m2 than other groups. The b-wave amplitudes of the hMERTK group (199.022 ± 28.187 µV) and hMERTK/Gas6 group (214.114 ± 50.677 µV) were also higher than that of the control group (146.179 ± 29.720 µV) at 0.01 cd.s/m2; there is no signiﬁcant diﬀerence between the hMERTK group and hMERTK/Gas6 group (Figure 7A).To evaluate the sustained protective eﬀect of the combined treatment, we also recorded ERGs at 19 dpi. In vitro Biocompatibility and Bioactivity of Gas6 NPs (A) Laser Confocal Microscope images of hfRPE cells pretreating with Gas6 and Gas6 NPs at 1.5 and 3 h. (B) Average number of ﬂuorescent beads engulfed by single hfRPE cell. The data are expressed as mean ± SD (tests were performed in triplicate and 40 cells were counted in all, bar = 25 µm). P < 0.05. FIGURE 4 \| Phagocytosis Assay of human fRPE cells pretreating with Gas6 and Gas6 NPs at 1.5 and 3 h. (A) Laser Confocal Microscope images of hfRPE cells pretreating with Gas6 and Gas6 NPs at 1.5 and 3 h. (B) Average number of ﬂuorescent beads engulfed by single hfRPE cell. The data are expressed as mean ± SD (tests were performed in triplicate and 40 cells were counted in all, bar = 25 µm). P < 0.05. In vivo Rescue of MERTK-Associated Retinal Degeneration by Gas6 NPs RCS rats (Figure 6) following the subretinal injection of AAV2- BEST1-hMERTK. Additionally, in order to assess whether Gas6 NPs enhanced the therapeutic eﬀects of gene therapy in RCS rats, we recorded and analyzed the ERG response and OCT results of the rats at 12 days post injection (12 dpi, 5 days after Gas6 and Gas6 RCS rats with inherited retinal degeneration caused by a deletion in the MERTK gene were used to evaluate the protective eﬀect of the combination strategy. Firstly, we demonstrated that the hMERTK protein could be sexpressed in the RPE cell layer of December 2021 \| Volume 8 \| Article 794299 Frontiers in Medicine \| www.frontiersin.org Wu et al. Gas6 NPs Enhances Retinal Phagocytosis GURE 5 \| Western blot analysis of the key proteins involved in Gas6 induced phagocytosis. (A) Relative optical density determined by densitometry using ImageJ tware. (B) Protein relative expression in control, Gas6 and Gas6 NPs groups. The data are expressed as mean ± SD. FIGURE 5 \| Western blot analysis of the key proteins involved in Gas6 induced phagocytosis. (A) Relative optical density determined by densitometry using ImageJ software. (B) Protein relative expression in control, Gas6 and Gas6 NPs groups. The data are expressed as mean ± SD. December 2021 \| Volume 8 \| Article 794299 Frontiers in Medicine \| www.frontiersin.org 6 Wu et al. Gas6 NPs Enhances Retinal Phagocytosis FIGURE 6 \| Detection of MERTK expression in RCS rats’ retinas following delivery of AAV2-BEST1-hMERTK. hMERTK (labeled as red) expression by immunoﬂuorescence (bar = 50 µm). In vitro Biocompatibility and Bioactivity of Gas6 NPs As illustrated in Figure 7B, the dark-adapted ERG responses in the hMERTK, hMERTK/Gas6 and hMERTK/Gas6 NPs groups were all signiﬁcantly higher than that of the control group at low light intensity (0.01 cd.s/m2). Moreover, the hMERTK/Gas6 NPs group had the largest b-wave amplitude among the 3 treatment groups, similar to that recorded at 12 dpi. There was no signiﬁcant diﬀerence between the hMERTK and hMERTK/Gas6 groups at 19 dpi too. curve of the nasal side was clearly separated (Figure 8B). The average ONL thickness of the nasal retina was calculated and shown in Figure 8C. After co-administration of AAV2-BEST1- hMERTK and Gas6 NPs, the thickness of nasal ONL was 54.70 ± 7.53 µm and 31.09 ± 5.27 µm respectively at 12 dpi and 19 dpi. The thickness of the ONL was signiﬁcantly higher than that of the Control group (39.00 ±7.50 µm and 19.54 ± 2.73 µm), hMERTK group (48.45 ± 6.98 µm and 24.51 ± 2.52 µm,) and hMERTK/Gas6 group (49.35 ± 5.78 µm and 25.77 ± 4.95 µm,). curve of the nasal side was clearly separated (Figure 8B). The average ONL thickness of the nasal retina was calculated and shown in Figure 8C. After co-administration of AAV2-BEST1- hMERTK and Gas6 NPs, the thickness of nasal ONL was 54.70 ± 7.53 µm and 31.09 ± 5.27 µm respectively at 12 dpi and 19 dpi. The thickness of the ONL was signiﬁcantly higher than that of the Control group (39.00 ±7.50 µm and 19.54 ± 2.73 µm), hMERTK group (48.45 ± 6.98 µm and 24.51 ± 2.52 µm,) and hMERTK/Gas6 group (49.35 ± 5.78 µm and 25.77 ± 4.95 µm,). The rats were ﬁnally sacriﬁced and the eyeballs were taken out. The H&E staining images of retina were shown in Figure 9A. The results evidenced that the ONL nuclear layers in control group (3 ± 0.71) was signiﬁcantly less than that of the hMERTK group (6.11 ± 1.23), hMERTK/Gas6 group (6.56 ± 1.24), and hMERTK/Gas6 NPs group (10.89 ± 1.36) (Figure 9B). A statistically signiﬁcant diﬀerence was observed between the hMERTK/Gas6 NP group and the other 2 treatment groups. However, there was no signiﬁcant diﬀerence between the hMERTK and hMERTK/Gas6 group. And the result of the ONL thickness is consistent with the result of ONL nuclear layers (Supplementary Figure 2). Altogether, these results suggest that the combined treatment of AAV2-BEST1-hMERTK and Gas6 NPs had a greater protective eﬀect on photoreceptors. DISCUSSION In this study, we prepared Gas6 protein-loaded mPEG-PLGA nanoparticles (Gas6 NPs), which allowed for localized and sustained Gas6 protein release to overcome the short half-life of Frontiers in Medicine \| www.frontiersin.org December 2021 \| Volume 8 \| Article 794299 7 Wu et al. Gas6 NPs Enhances Retinal Phagocytosis RE 7 \| B-wave of dark-adapted (scotopic) ERG data of the representative RCS rats at 12 dpi and 19 dpi. (A) Scotopic ERG response (12 dpi) at a ser y from 0.01 to 30 cds/m2. Control: n = 7; MerTK: n = 9; MerTK/Gas6: n = 7; MerTK/Gas6 NPs: n = 6. (B) Scotopic ERG response (19 dpi) at a se y from 0.01 to 30 cds/m2. Control: n = 6; MerTK: n = 5; MerTK/Gas6: n = 3; MerTK/Gas6 NPs: n = 4. P < 0.05, *P < 0.005. B-wave of dark-adapted (scotopic) ERG data of the representative RCS rats at 12 dpi and 19 dpi. (A) Scotopic ERG response (12 dpi) at a series o m 0.01 to 30 cds/m2. Control: n = 7; MerTK: n = 9; MerTK/Gas6: n = 7; MerTK/Gas6 NPs: n = 6. (B) Scotopic ERG response (19 dpi) at a series m 0.01 to 30 cds/m2. Control: n = 6; MerTK: n = 5; MerTK/Gas6: n = 3; MerTK/Gas6 NPs: n = 4. P < 0.05, *P < 0.005. FIGURE 7 \| B-wave of dark-adapted (scotopic) ERG data of the representative RCS rats at 12 dpi and 19 dpi. (A) Scotopic ERG response (12 dpi) at a series of intensity from 0.01 to 30 cds/m2. Control: n = 7; MerTK: n = 9; MerTK/Gas6: n = 7; MerTK/Gas6 NPs: n = 6. (B) Scotopic ERG response (19 dpi) at a series of intensity from 0.01 to 30 cds/m2. Control: n = 6; MerTK: n = 5; MerTK/Gas6: n = 3; MerTK/Gas6 NPs: n = 4. P < 0.05, *P < 0.005. December 2021 \| Volume 8 \| Article 794299 8 Frontiers in Medicine \| www.frontiersin.org Wu et al. Gas6 NPs Enhances Retinal Phagocytosis GURE 8 \| Architecture in the retina of the representative RCS rats at 12 dpi and 19 dpi. DISCUSSION (A) Representative retinal cross section B-scan OCT image (the locatio as indicated with a black line in ocular fundus) at 12 dpi and 19 dpi; (B) Quantitative measurements of retinal thickness in different locations at 12 dpi and 19 dp erage thickness of temporal retina at 200, 400, and 600 microns at 12 dpi and 19 dpi. P < 0.05 vs. control group; #P < 0.05 vs. hMERTK/Gas6 NPs group. (Contin FIGURE 8 \| Architecture in the retina of the representative RCS rats at 12 dpi and 19 dpi. (A) Representative retinal cross section B-scan OCT image (the location was indicated with a black line in ocular fundus) at 12 dpi and 19 dpi; (B) Quantitative measurements of retinal thickness in different locations at 12 dpi and 19 dpi; (C) Average thickness of temporal retina at 200, 400, and 600 microns at 12 dpi and 19 dpi. P < 0.05 vs. control group; #P < 0.05 vs. hMERTK/Gas6 NPs group. (Continued) December 2021 \| Volume 8 \| Article 794299 Frontiers in Medicine \| www.frontiersin.org 9 Gas6 NPs Enhances Retinal Phagocytosis Wu et al. FIGURE 9 \| Histologic structure of RCS rats at 19 dpi. (A) Hematoxylin–eosin (H&E) staining of retina. (B) The ONL nuclear layers of the retina. The data are expressed as mean ± SD. P < 0.05 vs. control group; #P < 0.05 vs. hMERTK/Gas6 NPs group. GCL, Ganglion cell layer; IPL, Inner plexiform layer; INL, Inner nuclear layer; OPL, Outer plexiform layer; ONL, Outer nuclear layer; RPE, Retinal pigment epithelium. P < 0.05, bar, 50 µm. FIGURE 9 \| Histologic structure of RCS rats at 19 dpi. (A) Hematoxylin–eosin (H&E) staining of retina. (B) The ONL nuclear layers of the retina. The data are expressed as mean ± SD. P < 0.05 vs. control group; #P < 0.05 vs. hMERTK/Gas6 NPs group. GCL, Ganglion cell layer; IPL, Inner plexiform layer; INL, Inner nuclear layer; OPL, Outer plexiform layer; ONL, Outer nuclear layer; RPE, Retinal pigment epithelium. *P < 0.05, bar, 50 µm. nanoparticles (Gas6 NPs) and conducted a series of evaluations to determine the safety and function of Gas6 NPs in vitro. Our results demonstrated that both Gas6 protein and Gas6 NPs induced phagocytosis in hfRPE cells, which is consistent with previous reports (13). Frontiers in Medicine \| www.frontiersin.org Characterization of Gas6 NPs Fourier transform infrared (FTIR) spectra were recorded using an FTIR spectrometer (Spectrum One, PerkinElmer). The average size of Gas6 NPs was determined by dynamic light scattering (DLS) using a ZetaSizer Nano ZS (Malvern Instruments Ltd., Malvern, Worcestershire, United Kingdom). Samples were appropriately diluted with distilled water, and subsequently measured at 633 nm at 25◦C and a constant angle of 90◦. Stability experiments were regularly performed for the duration of 1 week by measuring the size of Gas6 NPs in phosphate-buﬀered saline (PBS) solution at room temperature. The concentration of Gas6 NPs used in the stability experiments was the same as that in the in vivo study. Furthermore, the morphology of Gas6 NPs was conﬁrmed using a transmission electron microscope (TEM; JEM-200CX, JEOL Ltd., Tokyo, Japan). Our ﬁndings strongly suggest that Gas6 NPs could enhance RPE phagocytosis in vitro and be used for enhancing the eﬀects of gene therapy for MERTK-associated RP in vivo. We propose that the promotion of RPE phagocytosis via the sustained released of Gas6 may enhance the therapeutic eﬃciency of gene therapy in RCS rats. The potential mechanisms underlying the manner in which the Gas6 protein enhances the therapeutic eﬃciency of MERTK gene therapy is unknown and should be addressed by further studies. However, it has been reported that Gas6 is a multi-functional circulating protein with multiple roles related to inﬂammation the and immune system (25), and retinitis pigmentosa is usually accompanied by inﬂammation. Of note, once it is identiﬁed that Gas6 enhances the therapeutic eﬀect of MERTK gene therapy by up-regulating phagocytosis and improving the retinal microenvironment, early administration of retinal microenvironment via Gas6 sustained release system could be applicable to other subtypes of retinal degenerative diseases. In vitro Gas6 Release Gas6 protein release kinetics were determined by the Human Gas6 ELISA Kit (Sigma-Aldrich Corp., St. Louis, Missouri, USA). Brieﬂy, Gas6 NPs were immersed in PBS at 4◦C with gentle shaking. At determined intervals, the suspension was centrifuged and the supernatant was replaced and collected. The Gas6 concentration in the supernatant was determined by the Human Gas6 Elisa Kit. The in vitro release reactions were carried out in triplicate for each sample. DISCUSSION Moreover, our results conﬁrm the bioactivity of the released Gas6 protein, which provides a solid foundation for the application of Gas6 NPs in enhancing the therapeutic eﬀect of gene therapy. the Gas6 protein in vivo. We identiﬁed that Gas6 NPs preserved Gas6 protein bioactivity and promoted RPE phagocytosis in vitro. Additionally, we assessed whether Gas6 NPs enhanced the therapeutic eﬀects of gene therapy in RCS rat model of MERTK-associated retinal dystrophy. Our results demonstrate that the co-administration of AAV2-BEST1-hMERTK and Gas6 NPs could protect photoreceptors from degeneration in RCS rats. Consequently, the ERG response was remarkably ameliorated and more structure of retina was preserved in the hMERTK/Gas6 NPs group. These ﬁndings strongly suggest that Gas6 NPs are a promising method for the sustained release of Gas6 protein and could enhance the therapeutic eﬀect of gene therapy for MERTK-associated RP. To conﬁrm the eﬀect of Gas6 NPs in vivo, we designed the co-administration strategy of Gas6 NPs with MERTK gene and subsequently evaluated its therapeutic eﬀects in RCS rats. Our results evidenced that the hMERTK/Gas6 group had a similar ERG response to that of the hMERTK group, while the hMERTK/Gas6 NPs group attained the highest amplitude based on ERG response. This suggested the sustained release of Gas6 protein within the therapeutic window for an extended period was essential for enhancing visual function. The mPEG- PLGA formulation would oﬀer numerous advantages, including the protection of Gas6 from degradation or elimination and the ability to deliver Gas6 locally to the retina, it is likely to enhance the therapeutic eﬀect of gene therapy. Moreover, the dark-adapted ERG response of the control, hMERTK, and hMERTK/Gas6 groups was signiﬁcantly weakened at certain light intensities at 19 dpi compared to 12 dpi, whereas no MERTK gene replacement therapy is considered as a promising treatment for MERTK-associated RP and demonstrated therapeutic eﬃcacy (20–23). Recent studies have indicated that the use of Gas6 enhances retinal phagocytosis via the MERTK receptor either alone or in combination with other speciﬁc ligands for receptor tyrosine kinases, which may enhance the therapeutics eﬀects of gene therapy (13). Whereas, Gas6 as a recombinant protein has many disadvantages such as short in vivo half-lives and chemical instability, may necessitate frequent administration over time to maintain long term eﬀect. Determination of the Encapsulation Efﬁciency The encapsulation eﬃciency percentage (EE%) of Gas6 NPs was calculated as follows: EE% = Gas6total −Gas6free Gas6total ∗100% In summary, we designed the hMERTK/Gas6 NPs co- administration system and evaluated its therapeutic eﬀects on RP treatment. We developed injectable Gas6 NPs with mPEG-PLGA via the double emulsion technique. The Gas6 NPs facilitated the development of a localized Gas6 delivery system with improved retention time. In vitro studies demonstrated that the Gas6 NPs remarkably increased the phagocytic function of hfRPE cells. Meanwhile, in vivo studies evidenced that the co-administration of Gas6 NPs with MERTK gene replacement therapy remarkably ameliorated the functional recovery of the ERG response and preserved more retinal structure. where Gas6total is the total amount of Gas6 in the nanoparticle and suspension, while Gas6free is the amount of free Gas6 in the suspension. Enzyme-linked immune absorbent assay (ELISA) was applied to detect Gas6free in the suspension. DISCUSSION Therefore, we encapsulated Gas6 into mPEG-PLGA December 2021 \| Volume 8 \| Article 794299 Frontiers in Medicine \| www.frontiersin.org 10 Gas6 NPs Enhances Retinal Phagocytosis Wu et al. Ltd. (Shanghai, China). Moreover, polyvinyl alcohol (PVA), pluronic F68, esteramide (EA), and dichloromethane (DCM), all having a purity of at least 99%, were purchased from Sigma Chemical Corp. (St. Louis, Missouri, United States). mPEG- PLGA nanoparticles loaded with Gas6 were prepared using the double emulsion (w/o/w) technique. Brieﬂy, 10 mg mPEG- PLGA was dissolved in 0.5 mL EA, after which 0.1 mL Gas6 solution (3 g/dL) was added. This mixture was transferred to a centrifuge tube and emulsiﬁed by sonication for 3 min at 80 W. Thereafter, the resultant emulsion was slowly added to 0.7 mL 2% (m/v) PVA and 0.3 mL 2% (m/v) pluronic F68, which was then stirred vigorously for 10 min. The mixture was subsequently emulsiﬁed via sonication for 5 min at 250 W. Both emulsiﬁcation steps were performed in an ice bath. After the solvent was evaporated by applying vacuum, the Gas6 NPs were collected by centrifugation at 10,000 rcf. for 10 min and then washed twice using distilled water. such diﬀerence was observed in the hMERTK/Gas6 NPs group. It is possible that the sustained release of the Gas6 protein induced a prolonged eﬀect within the retinal environment, and consequently enhanced RPE phagocytosis via the MERTK receptor for an extended period. To our knowledge, it is the ﬁrst time we proposed the concept of sustained release of Gas6 protein within the therapeutic window for enhancing the eﬀects of gene therapy for MERTK- associated RP. Here, the mPEG-PLGA was used to encapsulate Gas6 protein, which has good biocompatible and its degradation depends on molecular weight, conformation, and copolymer composition (24). The Gas6 protein release kinetics obtained in this study demonstrate that mPEG-PLGA nanoparticles are a suitable protein delivery system and, more speciﬁcally, an eﬀective sustained-release system. While, nearly 70% of Gas6 had been released from Gas6 NPs by day 12, after which its release slowed down. In the clinical setting, we need the release of Gas6 to be sustained for a longer duration in order to reduce the frequency of administration and increase patient compliance and comfort. This requires us to optimize this sustained-release system in future research to ﬁnally achieve an extended-release system. Cell Culture and Animals Cell Culture and Animals Human fetal RPE (hfRPE) cells were kindly provided by Professor Guoping Fan(University of California, Los Angeles, California, US) (26). These cells were cultured in DMEM/F12(1:1) medium supplemented with 15% fetal bovine serum (FBS) (SH3007003HI; Thermo Fisher Scientiﬁc) at 37◦C in a humidiﬁed incubator with a 5% CO2/95% air atmosphere. The medium was replaced every 2 days. RCS rats were also kindly provided by Professor Guoping Fan and kept in Capital Medical University,Beijing,China. The animals were kept in temperature-controlled rooms with a 12- h light/dark cycle and were provided with standard food and water ad libitum. The study was approved and monitored by the Institutional Animal Care and Use Committee of the Capital Medical University (IACUC; AEEI-2018-198), and conformed to the National Institute of Health Guide for the Care and Use of Laboratory Animals as well as the Association for Research in Vision and Ophthalmology (ARVO) Statement for the Use of Animals in Ophthalmic and Vision Research. Animal Experiments p The pAAV2-BEST1-hMERTK vector was prepared by Vigene Biosciences Inc. Brieﬂy, the human MERTK gene and BEST1 promoter were ampliﬁed from a Human Retina Marathon- Ready cDNA Library (Clontech) and cloned into the pUC19 vector. The sequence of the construct was veriﬁed by Sanger sequencing. Thereafter, the expression cassette, containing the BEST1 promoter, hMERTK cDNA, and 3xFLAG tag, was excised by restriction enzymes and cloned into the pAAV plasmid. The pAAV2-BEST1-MERTK plasmid was then transfected into HEK293 cells and the virus was harvested. The ﬁnal concentration of AAV2-BEST1-hMERTK was 3.37 × 1013 vg/mL. The map of pAAV2-BEST1-hMERTK and pAAV2-BEST1-EGFP were showed in Supplementary Figure 3. The sequence of BEST1 promoter was list in Supplementary Table 1. Phagocytosis Assay g y y The hfRPE cells were plated at a density of 5 × 104 cells/well into poly-l-lysine coated 24-well plates and allowed to adhere overnight. Cells were then pretreated with Gas6 NPs dispersion (ﬁnal concentration of Gas6 was 500 ng/mL) or recombinant Gas6 protein (ﬁnal concentration 500 ng/mL) for 1 h. Thereafter, 5 µL of 1 µm ﬂuorescein isothiocyanate (FITC)-labeled carboxylate-modiﬁed microspheres (1933365; 1:10 dilution; Invitrogen, USA) was added, and initially incubated for 1.5 h and 3 h at 37◦C with a 5% CO2 atmosphere. The medium was then removed and cells were washed 6 times to remove excess beads. Cells were subsequently labeled with 4′,6- diamidino-2-phenylindole (DAPI) and tetramethylrhodamine (TRITC) phalloidin (CA1610; Solarbio, Beijing, China). Four images were taken of each well using a Leica SP5 microscope. Additionally, the amounts of beads per cell was counted in 40 cells per condition. p y Subretinal injections of the AAV2-BEST1-hMERTK virus were administered to RCS rats at P19 (0 dpi). RCS rats were divided into 4 diﬀerent groups based on the administered treatment, namely Control group (AAV2-BEST1-EGFP), hMERTK group (AAV2-BEST1-hMERTK), hMERTK/Gas6 group (AAV2-BEST1-hMERTK & Gas6 protein), and hMERTK/Gas6 NPs group (AAV2-BEST1-hMERTK & Gas6 NPs). Rats were anesthetized using an intraperitoneal injection containing 30 mg/kg pentobarbital sodium and 5 mg xylazine hydrochloride. Pupils were dilated using 0.2 mg/mL tropicamide phenylephrine (Santen Pharmaceutical Co., Ltd., Shiga Plant, Shiga, Japan) and topically anesthetized with 0.5% proparacaine (Santen Pharmaceutical Co., Ltd.). A 31-gauge insulin syringe Cell Viability Assay The hfRPE cells (0.5 × 104 cells/well) were allowed to adhere overnight after plated in 96-well plates before being incubated for 7 days. Cells were then treated with Gas6 NPs dispersion (ﬁnal concentration of Gas6 was 500 ng/mL) and incubated for an additional 7 days. Cell viability was tested daily. Brieﬂy, the culture medium was replaced by 200 µL 10% CCK8 reagent (CA1210; Solarbio, Beijing, China) and incubated for 2 h at 37◦C. The optical density (OD) value was measured at 450 nm using an ELISA reader (BioTek, Winooski, Vermont, USA) according to the manufacturer’s instructions. The relative proliferation rate was calculated as follows: RPR = (ODTEST/ODControl) ×100% MATERIALS AND METHODS Preparation of Gas6 NPs mPEG-PLGA (mPEG5000-PLGA 75/25 [70000]) was purchased from Shanghai Zhong-Liang Oil and Fat Chemical Co., December 2021 \| Volume 8 \| Article 794299 Frontiers in Medicine \| www.frontiersin.org 11 Gas6 NPs Enhances Retinal Phagocytosis Wu et al. for 1 h. Thereafter, 15 µL of 1 µm ﬂuorescein isothiocyanate (FITC)-labeled carboxylate-modiﬁed microspheres (1933365; 1:10 dilution; Invitrogen, USA) was added, and incubated for 1.5 h and 3 h at 37◦C with a 5% CO2 atmosphere. Thereafter, proteins were extracted from cells and their total concentration was measured using a BCA Protein Assay Kit (CWBIO) according to the manufacturer’s instructions. Equal quantities (40 µg) of proteins per gel lane were separated on 10% polyacrylamide gels by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and then transferred to polyvinylidene ﬂuoride membranes using an electroblotting apparatus (Bio-Rad). Membranes were blocked using a solution containing 5% non-fat milk and TBS- Tween20 and then incubated separately at 4◦C overnight with the following primary antibodies: GAPDH (sc-25778; 1:1000; SantaCruz Biotechnology), active GTP-Rac1 (26903; 1:200; Neweast Biosciences), Rac1 (ab155938; 1:600; Abcam), p- FAK861 (44-626G; 1:800; Invitrogen), FAK (AHO0502; 1:200; Invitrogen), p-AKT473 (4060s; 1:500; CST), and AKT (9272; 1:500; CST). Membranes were then incubated with horseradish- peroxidase (HRP)-conjugated secondary antibody (goat anti- mouse/rabbit IgG antibody; G21240/G21234; 1:1000; Invitrogen) for 1 h at room temperature. Membranes were then washed 3 times (10 min per wash) with 0.1% TBS-Tween20 after each antibody application. Thereafter, immuno-labeled proteins were detected using the ECL Plus Detection System (Invitrogen) according to the manufacturer’s instructions. The band was analyzed using Image-Pro Plus (IPP) software. Cell Culture and Animals Human fetal RPE (hfRPE) cells were kindly provided by Professor Guoping Fan(University of California, Los Angeles, California, US) (26). These cells were cultured in DMEM/F12(1:1) medium supplemented with 15% fetal bovine serum (FBS) (SH3007003HI; Thermo Fisher Scientiﬁc) at 37◦C in a humidiﬁed incubator with a 5% CO2/95% air atmosphere. The medium was replaced every 2 days. Optical Coherence Tomography p g p y OCT was performed using a Micron III (Phoenix Research Labs, Pleasanton, CA). Animals were anesthetized and their pupils were dilated using 0.2 mg/mL tropicamide phenylephrine. The corneal surface was protected using a 1.5% hydroxyethylcellulose solution. The rat ocular fundus was monitored using the fundus camera of the Micron. Representative OCT images of the retina were taken horizontally across the optic nerve head (ONH), and the imaging location was marked on the image with a black line. Thirty images were averaged to eliminate projection artifacts. The acquired OCT images were quantitatively analyzed using the InSight software (Phoenix Research Labs). The thickness of the outer nuclear layer (ONL) was measured at 100, 200, 400-µm of temporal retina and 100, 200, 400, 600-µm of nasal retina from the ONH. The average value of the ONL in nasal retina at 200, 400, and 600 microns was used represent the average thickness of ONL. ETHICS STATEMENT The animal study was reviewed and approved by Institutional Animal Care and Use Committee of the Capital Medical University (IACUC; AEEI-2018-198). Western Blotting Analysis g The hfRPE cells were plated at a density of 5 × 105 cells/well into poly-l-lysine coated 6-well plates and allowed to adhere overnight. Cells were then pretreated with Gas6 NPs dispersion (ﬁnal concentration of Gas6 was 500 ng/mL) or recombinant Gas6 protein (ﬁnal concentration 500 ng/mL) December 2021 \| Volume 8 \| Article 794299 Frontiers in Medicine \| www.frontiersin.org 12 Gas6 NPs Enhances Retinal Phagocytosis Wu et al. was used to carefully puncture the temporal corneoscleral limbus. Thereafter, a 33-gauge blunt needle (Hamilton) was used to administer the subretinal injection. The blunt needle tip was inserted through the sclerotic puncture and aimed at the contralateral subretinal space. Subsequently, 1 µL of AAV2- BEST1-EGFP (control group) or AAV2-BEST1-hMERTK virus (hMERTK, hMERTK/Gas6, and hMERTK/Gas6 NPs groups) was injected into the subretinal space of the RCS rats’ right eye. Tobramycin and dexamethasone eye ointment (ALCON) were applied once daily for 3 days after injection. moist chamber containing rabbit recombinant monoclonal anti- hMERTK antibody (ab52968; 1:200; Abcam) at 4◦C overnight. Thereafter, sections were rinsed 3 times with PBS and incubated with Alexa Fluor 594 anti-rabbit secondary antibody (R37119; 1:1000; Invitrogen) in the dark for 1 h at room temperature. Cell nuclei were then stained with DAPI and ﬂuorescent signals were visualized using a Zeiss ﬂuorescence microscope (Observer Z1). DATA AVAILABILITY STATEMENT The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding authors. Histological Analysis To quantify the layers and the thickness of outer nuclear layer (ONL) after combined treatment, hematoxylin and eosin (H&E) staining was performed. Rats were anesthetized using 400 mg/kg chloral hydrate, after which their eyes were enucleated, ﬁxed in 4% paraformaldehyde for 4 h, washed 3 times with PBS, and embedded in paraﬃn. Eight-micron-thick paraﬃn sections were used for H&E staining. Stained slices were visualized by a Leica microscope. The ONL layers was counted in a double-blind manner. The thickness of the ONL was measured using Image- Pro Plus (IPP) software. There are 3 animals in each group, and the layers and thickness of ONL of each animal were measured in 3 diﬀerent slices. At 7 dpi, 2 µL PBS (control and hMERTK groups), 2 µL Gas6 protein (hMERTK/Gas6 group; 0.5 µg/uL), or 2 µL Gas6 NPs (hMERTK/Gas6 NPs group; 0.5 µg Gas6/µL) was administered via intravitreal injection. The ERG and OCT were recorded at 12 dpi and 19 dpi. Animals exhibiting retinal bleeding and cataracts were excluded. Electroretinographic Analysis g p y The Espion Visual Electrophysiology System (Diagnosys, USA) was used to record the electroretinogram. After at least 12 h of dark adaptation, animals were anesthetized and their pupils were dilated using 0.2 mg/mL tropicamide phenylephrine. Animals were placed on a regulated heating pad throughout the experiment. Electroretinograms (ERGs) were recorded by means of a golden ring that made contact with the corneal surface through a layer of 0.2% carbomer. Additionally, needle electrodes were inserted into the cheeks and tails of animals and served as the reference and ground leads, respectively. Scotopic testing at eight-intensity stimulus-response series (0.01 cd.s/m2 to 30 cd.s/m2) were presented; the resulting b-wave amplitudes were measured from the trough of the a-wave to the crest of the b-wave. Statistical Analysis All data were presented as the mean ± standard deviation (SD). Independent Samples T-test analysis was used to compare the CCK8 absorbance between the control group and Gas6 NPs group. One-way analysis of variance (ANOVA) and Fisher’s least signiﬁcant diﬀerence (LSD) test were used to determine whether signiﬁcant diﬀerences exist between the ERG amplitudes of the diﬀerent treatment groups. Two-way analysis of variance and Fisher’s least signiﬁcant diﬀerence (LSD) test were used to determine whether signiﬁcant diﬀerences exist among the OCT results. The independent Samples T-test analysis was used to compare the diﬀerences between the cell viability of each group. Statistical analyses were performed by SPSS 20.0 (SPSS Inc., Chicago, Illinois, USA). P < 0.05 indicated statistical signiﬁcance. REFERENCES factor release for induced angiogenesis. Acta Biomater. (2015) 13:88–100. doi: 10.1016/j.actbio.2014.11.002 factor release for induced angiogenesis. Acta Biomater. (2015) 13:88–100. doi: 10.1016/j.actbio.2014.11.002 1. Dias MF, Joo K, Kemp JA, Fialho SL, Jessica A, Woo SJ, et al. Molecular genetics and emerging therapies for retinitis pigmentosa: basic research and clinical perspectives. Prog Rtein Eye Res. (2018) 63:107– 31. doi: 10.1016/j.preteyeres.2017.10.004 1. Dias MF, Joo K, Kemp JA, Fialho SL, Jessica A, Woo SJ, et al. Molecular genetics and emerging therapies for retinitis pigmentosa: basic research and clinical perspectives. Prog Rtein Eye Res. (2018) 63:107– 31. doi: 10.1016/j.preteyeres.2017.10.004 17. Zhang Z, Tongchusak S, Mizukami Y, Kang YJ, Ioji T, Touma M, et al. Induction of anti-tumor cytotoxic T cell responses through PLGA- nanoparticle mediated antigen delivery. Biomaterials. (2011) 32:3666– 78. doi: 10.1016/j.biomaterials.2011.01.067 j y 2. Sahel J, Bonnel S, Mrejen S, Paques M. Retinitis pigmentosa and other dystrophies. Dev Ophthalmol. (2010) 47:160–7. doi: 10.1159/000320079 18. Wu C, Baldursdottir S, Yang M, Mu H. Lipid and PLGA hybrid microparticles as carriers for protein delivery. J Drug Deliv Sci Technol. (2018) 43:65– 72. doi: 10.1016/j.jddst.2017.09.006 2. Sahel J, Bonnel S, Mrejen S, Paques M. Retinitis pigmentosa and other dystrophies. Dev Ophthalmol. (2010) 47:160–7. doi: 10.1159/000320079 3. Charbel Issa P, Bolz HJ, Ebermann I, Domeier E, Holz FG, Scholl HPN. Characterisation of severe rod-cone dystrophy in a consanguineous family with a splice site mutation in the MERTK gene. Br J Ophthalmol. (2009) 93:920–5. doi: 10.1136/bjo.2008.147397 3. Charbel Issa P, Bolz HJ, Ebermann I, Domeier E, Holz FG, Scholl HPN. Characterisation of severe rod-cone dystrophy in a consanguineous family with a splice site mutation in the MERTK gene. Br J Ophthalmol. (2009) 93:920–5. doi: 10.1136/bjo.2008.147397 19. Kurowska-Stolarska M, Alivernini S, Garcia Melchor E, Elmesmari A, Tolusso B, Tange C, et al. MicroRNA-34a dependent regulation of AXL controls the activation of dendritic cells in inﬂammatory arthritis. Nat Commun. (2017) 8:1–13. doi: 10.1038/ncomms15877 4. Abu-Saﬁeh L, Alrashed M, Anazi S, Alkuraya H, Khan AO, Al-Owain M, et al. Autozygome-guided exome sequencing in retinal dystrophy patients reveals pathogenetic mutations and novel candidate disease genes. Genome Res. (2013) 23:236–47. doi: 10.1101/gr.144105.112 4. Abu-Saﬁeh L, Alrashed M, Anazi S, Alkuraya H, Khan AO, Al-Owain M, et al. Autozygome-guided exome sequencing in retinal dystrophy patients reveals pathogenetic mutations and novel candidate disease genes. Genome Res. (2013) 23:236–47. doi: 10.1101/gr.144105.112 20. REFERENCES Koch S, Sothilingam V, Garrido MG, Tanimoto N, Becirovic E, Koch F, et al. Gene therapy restores vision and delays degeneration in the CNGB11/1 mouse model of retinitis pigmentosa. Hum Mol Genet. (2012) 21:4486– 96. doi: 10.1093/hmg/dds290 5. Patel N, Aldahmesh MA, Alkuraya H, Anazi S, Alsharif H, Khan AO, et al. Expanding the clinical, allelic, and locus heterogeneity of retinal dystrophies. Genet. Med. (2016) 18:554–62. doi: 10.1038/gim.2015.127 21. LaVail MM, Yasumura D, Matthes MT, Yang H, Hauswirth WW, Deng WT, et al. Gene therapy for MERTK-associated retinal degenerations. Adv Exp Med Biol. (2016) 854:487–93. doi: 10.1007/978-3-319-17121-0_65 g 6. Bok D, Hall MO. The role of the pigment epithelium in the etiology of inherited retinal dystrophy in the rat. J Cell Biol. (1971) 49:664– 82. doi: 10.1083/jcb.49.3.664 22. Wert KJ, Davis RJ, Sancho-Pelluz J, Nishina PM, Tsang SH. Gene therapy provides long-term visual function in a pre-clinical model of retinitis pigmentosa. Hum Mol Genet. (2013) 22:558–67. doi: 10.1093/hmg/dds466 7. Zhang JX, Wang NL, Lu QJ. Development of gene and stem cell therapy for ocular neurodegeneration. Int J Ophthalmol. (2015) 3:622–30. doi: 10.3980/j.issn.2222-3959.2015.03.33 23. Deng WT, Dinculescu A, Li Q, Boye SL, Li J, Gorbatyuk MS, et al. Tyrosine-mutant AAV8 delivery of human MERTK provides long-term retinal preservation in RCS rats. Invest Ophthalmol Vis Sci. (2012) 53:1895– 904. doi: 10.1167/iovs.11-8831 8. Boye SE, Boye SL, Lewin AS, Hauswirth WW. A comprehensive review of retinal gene therapy. Mol Ther. (2013) 21:509–19. doi: 10.1038/mt.2012.280 24. Csaba N, Sánchez A, Alonso MJ. PLGA: Poloxamer and PLGA: Poloxamine blend nanostructures as carriers for nasal gene delivery. J Control Release. (2006) 113:164–72. doi: 10.1016/j.jconrel.2006.03.017 9. Conlon TJ, Deng WT, Erger K, Cossette T, Pang JJ, Ryals R, et al. Preclinical potency and safety studies of an AAV2-mediated gene therapy vector for the treatment of MERTK associated retinitis pigmentosa. Hum Gene Ther Clin Dev. (2013) 24:23–8. doi: 10.1089/humc.2013.037 25. Lemke G. Biology of the TAM receptors. Cold Spring Harb Perspect Biol. (2013) 5:a009076. doi: 10.1101/cshperspect.a009076 10. Ghazi NG, Abboud EB, Nowilaty SR, Alkuraya H, Alhommadi A, Cai H, et al. Treatment of retinitis pigmentosa due to MERTK mutations by ocular subretinal injection of adeno-associated virus gene vector: results of a phase I trial. Hum Genet. (2016) 135:327–43. doi: 10.1007/s00439-016-1637-y 26. Liao JL, Yu J, Huang K, Hu J, Diemer T, Ma Z, et al. Molecular signature of primary retinal pigment epithelium and stem-cell-derived RPE cells. FUNDING We greatly thank Prof. Qingjun Lu for advice on study design. This work was supported by National Key R&D Program of China (2017YFA0104100 and 2016YFC0905201), Beijing Municipal Institute of Public Medical Research Development and Reform Pilot Project (2018-2), Beijing Hospitals Authority Youth Program (QML20180208), and Beijing Institute of Ophthalmology Key Program (2019003). These funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmed. 2021.794299/full#supplementary-material The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmed. 2021.794299/full#supplementary-material The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmed. 2021.794299/full#supplementary-material Tissue Immunoﬂuorescence Staining Tissue Immunoﬂuorescence Staining Samples were cut into 10 µm sections using a cryostat (Leica Microsystems, Wetzlar, Germany). Thereafter, sections were rinsed 3 times with PBS for 5 min at room temperature, and then blocked with 5% bovine serum albumin (BSA) for 30 min at room temperature. Sections were subsequently transferred to a SW and YM conducted the experiments, analysis the data, and wrote the paper. JZ and NW designed the experiments and revised the paper. QL and XY provided the materials. All authors contributed to manuscript revision, read, and approved the submitted version. December 2021 \| Volume 8 \| Article 794299 Frontiers in Medicine \| www.frontiersin.org 13 Wu et al. Gas6 NPs Enhances Retinal Phagocytosis REFERENCES Hum Mol Genet. (2010) 19:4229–38. doi: 10.1093/hmg/ddq341 11. Law AL, Parinot C, Chatagnon J, Gravez B, Sahel J-A, Bhattacharya SS, et al. Cleavage of Mer Tyrosine Kinase (MERTK) from the cell surface contributes to the regulation of retinal phagocytosis. J Biol Chem. (2015) 290:4941–52. doi: 10.1074/jbc.M114.628297 Conﬂict of Interest: The authors declare that the research was conducted in the absence of any commercial or ﬁnancial relationships that could be construed as a potential conﬂict of interest. 12. Pelaez D. Stem Cells for microenvironmental modulation and retinal regeneration. Curr Tissue Eng. (2016) 5:52– 9. doi: 10.2174/2211542004666150713191117 Publisher’s Note: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their aﬃliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. 13. Albert R, Kristóf E, Zahuczky G, Tóth M, Veréb Z, Oláh B, et al. Triamcinolone regulated apopto-phagocytic gene expression patterns in the clearance of dying retinal pigment epithelial cells. A key role of MERTK in the enhanced phagocytosis. BBA Gen Subjects. (2015) 1850:435– 46. doi: 10.1016/j.bbagen.2014.10.026 14. Putney SD, Burke PA. Improving protein therapeutics with sustained-release formulations. Nat. Biotechnol. (1998) 16:153–7. doi: 10.1038/nbt0298-153 Copyright © 2021 Wu, Mao, Liu, Yan, Zhang and Wang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. 15. Rauck BM, Novosat TL, Oudega M, Wang Y. Biocompatibility of a coacervate- based controlled release system for protein delivery to the injured spinal cord. Acta Biomater. (2015) 11:204–11. doi: 10.1016/j.actbio.2014.09.037 16. Li Z, Qu T, Ding C, Ma C, Sun H, Li S, et al. Injectable gelatin derivative hydrogels with sustained vascular endothelial growth December 2021 \| Volume 8 \| Article 794299 Frontiers in Medicine \| www.frontiersin.org 14
https://openalex.org/W3112217385	http://sedici.unlp.edu.ar/bitstream/handle/10915/119728/Documento_completo.pdf?sequence=1	English	null	Kinetic Studies of Sodium and Metforminium Decavanadates Decomposition and In Vitro Cytotoxicity and Insulin- Like Activity	Inorganics	2,020	cc-by	17,265	Received: 22 October 2020; Accepted: 2 December 2020; Published: 8 December 2020 Abstract: The kinetics of the decomposition of 0.5 and 1.0 mM sodium decavanadate (NaDeca) and metforminium decavanadate (MetfDeca) solutions were studied by 51V NMR in Dulbecco’s modiﬁed Eagle’s medium (DMEM) medium (pH 7.4) at 25 ◦C. The results showed that decomposition products are orthovanadate [H2VO4]−(V1) and metavanadate species like [H2V2O7]2−(V2), [V4O12]4−(V4) and [V5O15]5−(V5) for both compounds. The calculated half-life times of the decomposition reaction were 9 and 11 h for NaDeca and MetfDeca, respectively, at 1 mM concentration. The hydrolysis products that presented the highest rate constants were V1 and V4 for both compounds. Cytotoxic activity studies using non-tumorigenic HEK293 cell line and human liver cancer HEPG2 cells showed that decavanadates compounds exhibit selectivity action toward HEPG2 cells after 24 h. The eﬀect of vanadium compounds (8–30 µM concentration) on the protein expression of AKT and AMPK were investigated in HEPG2 cell lines, showing that NaDeca and MetfDeca compounds exhibit a dose-dependence increase in phosphorylated AKT. Additionally, NaDeca at 30 µM concentration stimulated the glucose cell uptake moderately (62%) in 3T3-L1 adipocytes. Finally, an insulin release assay in βTC-6 cells (30 µM concentration) showed that sodium orthovanadate (MetV) and MetfDeca enhanced insulin release by 0.7 and 1-fold, respectively. Keywords: polyoxometalates; decavanadate; cytotoxicity; insulin-like activity; diabetes therapy; vanadium biochemistry; vanadium speciation Keywords: polyoxometalates; decavanadate; cytotoxicity; insulin-like activity; diabetes therapy; vanadium biochemistry; vanadium speciation Received: 22 October 2020; Accepted: 2 December 2020; Published: 8 December 2020 Kinetic Studies of Sodium and Metforminium Decavanadates Decomposition and In Vitro Cytotoxicity and Insulin- Like Activity Aniela M. Silva-Nolasco 1,2, Luz Camacho 2 , Rafael Omar Saavedra-Díaz 1, Oswaldo Hernández-Abreu 1 , Ignacio E. León 3 and Irma Sánchez-Lombardo 1, Aniela M. Silva-Nolasco 1,2, Luz Camacho 2 , Rafael Omar Saavedra-Díaz 1, Oswaldo Hernández-Abreu 1 , Ignacio E. León 3 and Irma Sánchez-Lombardo 1,* 1 Centro de Investigación de Ciencia y Tecnología Aplicada de Tabasco, División Académica de Ciencias Básicas (CICTAT), Universidad Juárez Autónoma de Tabasco, Carretera Cunduacán-Jalpa km. 1 Col. La Esmeralda, Cunduacán 86690, Tabasco, Mexico; animonsino@gmail.com (A.M.S.-N.); rafael.saavedra@ujat.mx (R.O.S.-D.); oswaldo.hernandez@ujat.mx (O.H.-A.) 2 Laboratorio de Nutrición Experimental, Instituto Nacional de Pediatría, Ciudad de Mexico 04530, Mexico; camacho.luz@gmail.com 2 Laboratorio de Nutrición Experimental, Instituto Nacional de Pediatría, Ciudad de Mexico 04530, Mex camacho.luz@gmail.com g 3 Centro de Química Inorgánica CEQUINOR (CONICET, UNLP), Bv 120 1465, La Plata 1900, Argentina; iel86@yahoo.com.ar 3 Centro de Química Inorgánica CEQUINOR (CONICET, UNLP), Bv 120 1465, La Plata 1900, Argentina; iel86@yahoo.com.ar y * Correspondence: irma.sanchez@ujat.mx 1. Introduction Polyoxometalates (POMs) have several applications in biology and medicine. Interactions between the highly charged POM molecules and biological molecules frequently occur through hydrogen-bonding and electrostatic interactions [1]. Moreover, POMs have shown pharmacological activities in vitro and in vivo, such as antitumor, antimicrobial, and antidiabetic [2,3]. Their roles in biological systems are non-functional or functional kind of interactions with biomolecules [4], like the tungstate cluster that helps to solve the X-ray structure of ribosome [5] or the insulin-like properties of the decavanadates [6]. www.mdpi.com/journal/inorganics Inorganics 2020, 8, 67; doi:10.3390/inorganics8120067 www.mdpi.com/journal/inorganics 2 of 20 20 Inorganics 2020, 8, 67 Inorganics 2020, 8, In recent years, several organic and inorganic decavanadate compounds have been synthesized, exhibiting a wide structural supramolecular diversity in one, two or three dimensions [7–9]. However, the interaction of decavanadates with biological targets under physiological conditions are scarce reported since the decavanadate anion can be formed at vanadium concentrations up to 0.1 mM and in the pH range of 2–6 [10], and some organic decavanadates compounds are water-insoluble [11]. g p y properties of the decavanadates [6]. In recent years, several organic and inorganic decavanadate compounds have been synthesized, exhibiting a wide structural supramolecular diversity in one, two or three dimensions [7–9]. However, the interaction of decavanadates with biological targets under physiological conditions are scarce reported since the decavanadate anion can be formed at vanadium concentrations up to 0.1 mM and in the pH range of 2–6 [10] and some organic decavanadates compounds are water-insoluble In biological studies, buﬀer solutions are extensively used, although just a few studies have addressed the speciation of the decomposition products of the decavanadate compounds in such reaction media. The decomposition of the decameric species at neutral pH can be followed by 51V NMR showing a decrease in the peaks associated with the three magnetic independent vanadium nuclei of the decavanadate VA, VB and VC (Figure 1), albeit an increase of the signals for the metavanadate peak [H2VO4]−(V1) and the appearance of the orthovanadates species signals like [H2V2O7]2−(V2), [V4O12]4−(V4) and [V5O15]5−(V5) [12–14]. Moreover, monomeric vanadate is always present in decavanadate solutions at neutral pH [15]. The decavanadate decomposition rate is faster in acid than in basic solutions [16,17]. In the latter, the reaction proceeds via base-dependent or base-independent paths, and it depends on the counterions present in the solution [17]. 1. Introduction Schematic structure of (a) decavanadate anion [V10O28]6−, (b) orthovanadate [H2VO4]- (V1), metavanadate species like (c) [H2V2O7]2−(V2), (d) [V4O12]4−(V4), (e) [V5O15]5−(V5) and (f) diprotonated metformin (Metf). Vanadium speciation is complicated under physiological conditions, many known forms of vanadium V4+ and V5+ species have been shown to readily interconvert through redox and hydrolytic reactions, and it is, therefore, difficult to determine which are the active species [18]. Additionally, in biological studies, the active vanadium species will depend on the sample preparation and handling, that is, whether the compounds were dissolved in media or buffer before addition to the cell culture and for how long the complexes have been in solution before adding aliquots to the medium [19]. Metabolic diseases like diabetes mellitus type 2 (DM2) and cancer are non-communicable Vanadium speciation is complicated under physiological conditions, many known forms of vanadium V4+ and V5+ species have been shown to readily interconvert through redox and hydrolytic reactions, and it is, therefore, diﬃcult to determine which are the active species [18]. Additionally, in biological studies, the active vanadium species will depend on the sample preparation and handling, that is, whether the compounds were dissolved in media or buﬀer before addition to the cell culture and for how long the complexes have been in solution before adding aliquots to the medium [19]. yp diseases (NCD) that have become one of the major health hazards of the modern world [20]. Carcinogenesis occurs when normal cells receive genetic “hits”, after which a full neoplastic phenotype of growth, invasion, and metastasis develops. Diabetes may influence this process through chronic inflammation, endogenous or exogenous hyperinsulinemia, or hyperglycemia, but potential biologic links between the two diseases are incompletely understood [21]. The development Metabolic diseases like diabetes mellitus type 2 (DM2) and cancer are non-communicable diseases (NCD) that have become one of the major health hazards of the modern world [20]. Carcinogenesis occurs when normal cells receive genetic “hits”, after which a full neoplastic phenotype of growth, invasion, and metastasis develops. Diabetes may inﬂuence this process through chronic inﬂammation, endogenous or exogenous hyperinsulinemia, or hyperglycemia, but potential biologic links between the two diseases are incompletely understood [21]. The development of innovative therapeutic modalities [22] that increase the eﬀectiveness of clinical drugs like cis-platin or metformin hydrochloride and arrest their chemoresistance or side eﬀects is a topic trend for scientists. 1. Introduction mM and in the pH range of 2 6 [10], and some organic decavanadates compounds are water-insoluble [11]. In biological studies, buffer solutions are extensively used, although just a few studies have addressed the speciation of the decomposition products of the decavanadate compounds in such reaction media. The decomposition of the decameric species at neutral pH can be followed by 51V NMR showing a decrease in the peaks associated with the three magnetic independent vanadium nuclei of the decavanadate VA, VB and VC (Figure 1), albeit an increase of the signals for the metavanadate peak [H2VO4]− (V1) and the appearance of the orthovanadates species signals like [H2V2O7]2− (V2), [V4O12]4− (V4) and [V5O15]5− (V5) [12–14]. Moreover, monomeric vanadate is always present in decavanadate solutions at neutral pH [15]. The decavanadate decomposition rate is faster in acid than in basic solutions [16,17]. In the latter, the reaction proceeds via base-dependent or base- independent paths, and it depends on the counterions present in the solution [17]. Figure 1. Schematic structure of (a) decavanadate anion [V10O28]6−, (b) orthovanadate [H2VO4]- (V1), metavanadate species like (c) [H2V2O7]2− (V2), (d) [V4O12]4− (V4), (e) [V5O15]5− (V5) and (f) diprotonated metformin (Metf). a) VA O O O O O O VB VB VC VC VC VC O O O O O O O O O O O O O VA VB O O O O O O VB O O O b) V O O V V V V O O- O OH -O O OH O- V O O O O O- O O O- -O V V O V V O O- O V O O O O- O O- -O O O -O O V OH -O OH O 6- c) d) e) f) N NH+ NH2+ NH2 NH2 Figure 1. Schematic structure of (a) decavanadate anion [V10O28]6−, (b) orthovanadate [H2VO4]- (V1), metavanadate species like (c) [H2V2O7]2−(V2), (d) [V4O12]4−(V4), (e) [V5O15]5−(V5) and (f) diprotonated metformin (Metf). b) b) V V O V V O O- O V O O O O- O O- -O O O -O O e) f) e) d) Figure 1. Schematic structure of (a) decavanadate anion [V10O28]6−, (b) orthovanadate [H2VO4]- (V1), metavanadate species like (c) [H2V2O7]2− (V2), (d) [V4O12]4− (V4), (e) [V5O15]5− (V5) and (f) diprotonated metformin (Metf). Figure 1. 1. Introduction In this context, AMP-activated kinase (AMPK) signaling has become a promising therapeutic target in hepatocellular Inorganics 2020, 8, 67 3 of 20 carcinoma [23]. Another interesting target is the identiﬁcation of exploitable vulnerabilities for the treatment of hyperactive phosphatidylinositol 3-kinase (PI3K/AKT) tumors [24], and combining inhibitors of the pentose phosphate pathway (PPP) may represent a promising approach for selectively causing oxidative stress-induced cell killing in ovarian and lung cancer cells [25]. The medicinal potentiality of vanadium compounds is a challenging task that demands investigation [26] and in general few groups have pursued it. The insulin-like eﬀects of vanadium have been tested in vitro and in vivo [27,28]; however, the applied necessary dose of vanadium still was close to the levels at which side eﬀects are observed [29]. In fact, there is only one vanadium compound that has been tested in humans, the bis(ethylmaltolato)oxovanadium(IV) (BEOV). In general, 20 mg of vanadium compound was well tolerated [30], but at the end of Phase IIa clinical trial, the trial was abandoned due to renal problems of some patients [31]. However, several questions about the transport and mode of action of the vanadium compounds need to be addressed [28] due to the distinct action mechanism that regulates glucose metabolism by vanadium [32]. g g y In this work, we have studied the kinetics of the decomposition of 0.5 and 1 mM sodium decavanadate (NaDeca) and metforminium decavanadate (MetfDeca) in Dulbecco’s modiﬁed Eagle’s medium (DMEM) solution at pH 7.4 by 51V NMR, with the aim to understand the medium and the vanadium concentration eﬀects in both, the decomposition rate and the inﬂuence in the ratio of the ﬁnal products, namely V1, V2, V4 and V5. To our knowledge, the ammonium decavanadate compound decomposition in MES; MES = 2-(N-morpholino)ethanesulfonic acid by 51V NMR is the only report that describes the decomposition reaction [14]. Thus, NaDeca stability has not been extensively studied in buﬀer solutions. NaDeca and MetfDeca compounds are composed of the highly negative charged decavanadate and the positive counter ions. The counter ions bonding with the decameric moiety are ionic [33]. In that regard, the same biological activity of both compounds was expected if metformin hydrochloride (Metf) was pharmacologically an inactive molecule. Nevertheless, due to Metf antidiabetic properties, diﬀerent results were expected in the biological activity of the NaDeca and MetfDeca compounds. 1. Introduction MetfDeca compound in vivo exhibited hypoglycemic and lipid-lowering properties in type 1 diabetes mellitus (T1DM) [34] and type 2 diabetes mellitus (T2DM) models [35]. However, some questions were not addressed in those studies, like if MetV and MetfDeca regulated hyperglycemia and oxidative stress with the same action mechanisms, MetfDeca stability and toxicological eﬀects [35]. With the aim to address some of the former questions and to estimate if two diﬀerent counter ions could play a role as activators or inhibitors in the biological activity of decavanadates, we investigated how the decomposition products in DMEM medium at pH 7.4 can promote damage on the cell viability of HEK293 human embryonic kidney cells and HEPG2 human liver cancer cells. A comparison of these results with the cytotoxic eﬀect of sodium orthovanadate and metformin hydrochloride was also performed. In addition, the activation of AKT and AMPK pathways for the HEPG2 cell line by the vanadium compounds were studied in order to establish if the hydrolysis products promote the same activation mechanism in the metabolic pathways. Finally, glucose uptake in 3T3L-1 diﬀerentiated adipocytes study is presented along with an insulin release assay in βTC-6 cells at 30 µM concentration of the vanadium compounds, with the purpose of identifying if the same active species are promoting the desirable eﬀects in each case. 2.1. Characterization of the Sodium and Metforminium Decavanadate Solutions 2.1. Characterization of the Sodium and Metforminium Decavanadate Solutions The metforminium decavanadate (MetfDeca) (C4H13N5)3V10O28·8H2O and the sodium decavanadate (NaDeca) Na6V10O28·18H2O were prepared according to previously reported procedures [33,36]. The 51V NMR spectra for 1 mM concentration of NaDeca and MetfDeca compounds were recorded The metforminium decavanadate (MetfDeca) (C4H13N5)3V10O28·8H2O and the sodium decavanadate Deca) Na6V10O28·18H2O were prepared according to previously reported procedures [33,36]. The metforminium decavanadate (MetfDeca) (C4H13N5)3V10O28·8H2O and the sodium decavanadate (NaDeca) Na6V10O28·18H2O were prepared according to previously reported procedures [33,36]. The 51V NMR spectra for 1 mM concentration of NaDeca and MetfDeca compounds were recorded at pH 4 in 10% DMSO-d6 and 90% H2O (v/v), showing three signals at −420, −494, −510 ppm that were assigned to decameric species [V10O28]6−(V10), attributed to the three diﬀerent vanadium atoms of the (NaDeca) Na6V10O28·18H2O were prepared according to previously reported procedures [33,36]. The 51V NMR spectra for 1 mM concentration of NaDeca and MetfDeca compounds were recorded at pH 4 in 10% DMSO-d6 and 90% H2O (v/v), showing three signals at −420, −494, −510 ppm that were assigned to decameric species [V10O28]6−(V10), attributed to the three diﬀerent vanadium atoms of the p p g p y p p The 51V NMR spectra for 1 mM concentration of NaDeca and MetfDeca compounds were recorded at pH 4 in 10% DMSO-d6 and 90% H2O (v/v), showing three signals at −420, −494, −510 ppm that were assigned to decameric species [V10O28]6−(V10), attributed to the three diﬀerent vanadium atoms of the 4 of 20 e Inorganics 2020, 8, 67 The 51V decavanadate structure V10A, V10B and V10C respectively and one signal at −556 ppm assigned to the diprotonated monomeric species [H2VO4]−(V1) (Figure 2a) [37]. The 51V NMR spectra for both 0.5 and 1.0 mM concentration samples for NaDeca and MetfDeca complexes show the same species present in the solution; additionally, the complexes are stable through time. These results are in agreement with the reported V10A, V10B and V10C peaks that were observed for 10 mM NaDeca solution in D2O at pH 3.1 and in Middlebrook 7H9 broth medium supplemented with 10% ADC enrichment (5% BSA, 2% dextrose, 5% catalase), glycerol (0.2%, v/v) and Tween-80 (0.05%, v/v) at pH 6.5. [13] In contrast, NaDeca and MetfDeca are not stable in the DMEM medium at pH 7. 2.1. Characterization of the Sodium and Metforminium Decavanadate Solutions Their hydrolysis products are orthovanadate [H2VO4]- (V1) and metavanadate species like [H2V2O7]2−(V2), [V4O12]4−(V4) and [V5O15]5−(V5) that are observed at −556, −570, −578 and −586 ppm, respectively (Figure 2b) like previously reported for 1 mM solution of (NH4)6 [V10O28]·6H2O in MES buﬀer (0.1 M), NaCl (0.5 M) at pH 8 [14]. that were assigned to decameric species [V10O28]6− (V10), attributed to the three different vanadium atoms of the decavanadate structure V10A, V10B and V10C respectively and one signal at −556 ppm assigned to the diprotonated monomeric species [H2VO4]− (V1) (Figure 2a) [37]. The 51V NMR spectra for both 0.5 and 1.0 mM concentration samples for NaDeca and MetfDeca complexes show the same species present in the solution; additionally, the complexes are stable through time. These results are in agreement with the reported V10A, V10B and V10C peaks that were observed for 10 mM NaDeca solution in D2O at pH 3.1 and in Middlebrook 7H9 broth medium supplemented with 10% ADC enrichment (5% BSA, 2% dextrose, 5% catalase), glycerol (0.2%, v/v) and Tween-80 (0.05%, v/v) at pH 6.5. [13] In contrast, NaDeca and MetfDeca are not stable in the DMEM medium at pH 7. Their hydrolysis products are orthovanadate [H2VO4]- (V1) and metavanadate species like [H2V2O7]2− (V2), [V4O12]4− (V4) and [V5O15]5− (V5) that are observed at −556, −570, −578 and −586 ppm, respectively (Figure 2b) like previously reported for 1 mM solution of (NH4)6 [V10O28]·6H2O in MES buffer (0.1 M), NaCl (0.5 M) at pH 8 [14]. (a) (b) 51 (a) (b) Figure 2. 51V NMR spectra of (a) 1 mM NaDeca (top) and MetfDeca (bottom) in 10% DMSO-d6 and 90% H2O (v/v) DMSO at pH 4, (b) 1 mM NaDeca (top) and MetfDeca (bottom) in DMEM medium at pH 7.4. . Kinetic Studies by 51V NMR (a) (b) (b) Figure 2. 51V NMR spectra of (a) 1 mM NaDeca (top) and MetfDeca (bottom) in 10% DMSO-d6 and 90% H2O (v/v) DMSO at pH 4, (b) 1 mM NaDeca (top) and MetfDeca (bottom) in DMEM medium at pH 7.4. . Kinetic Studies by 51V NMR 2.2. Kinetic Studies by 51V NMR 2.2. Kinetic Studies by 51V NMR In vanadium(V) solutions, diﬀerent oligomeric vanadate species can occur simultaneously, depends on several factors such as vanadate concentration, pH and ionic strength [12], so at 0.5 and 5 of 20 a io vanad Inorganics 2020, 8, 67 p in the concentr 1 mM of NaDeca and MetfDeca, the V10 and V1 species were present at pH 4, but the hydrolysis of both compounds in DMEM medium allowed us to follow by 51V NMR the formation and the increment in the concentration over time at 25 ◦C of the orthovanadate, V1 and metavanadate species V2, V4 and V5 at −556, −570, −578 and −586 ppm, respectively. The kinetics of the decomposition of 1.0 mM NaDeca and MetfDeca (10 mM total vanadium) are plotted in Figure 3a, where the vanadium concentration for V10 species was calculated by integration of the V10A (2 vanadium atoms), V10B (4 vanadium atoms) and V10C (4 vanadium atoms) resonances at −420, −494 and −510 ppm, respectively, and the rate constants for the three decavanadate signals VA, VB and VC are shown with a negative sign by convention in Table 1. For comparison, the increase in concentration of the V1 and V4 vanadate species as a function of time are plotted in Figure 3b. Interestingly, the reaction is faster at 0.5 mM concentration of decavanadate than at 1 mM for NaDeca and MetfDeca compounds (Table 1). The rate constants of 0.5 mM NaDeca (2.28 ± 0.08) × 10−3 and (1.72 ± 0.07) × 10−4 for the appearance of V4 and V5 species, respectively, are three and four times higher than the ones calculated for 0.5 mM MetfDeca compound (7.63 ± 0.8) × 10−4 and (4.09 ± 0.3) × 10−5 for V4 and V5 species, respectively. Surprisingly, the rate constants for the appearance of the V4 and V5 species (Table 1) do not diﬀer signiﬁcantly for 1 mM NaDeca compared with 1 mM MetfDeca. V5 at 556, 570, 578 and 586 ppm, respectively. The kinetics of the decompos ca and MetfDeca (10 mM total vanadium) are plotted in Figure 3a, where the on for V10 species was calculated by integration of the V10A (2 vanadium atom atoms) and V10C (4 vanadium atoms) resonances at −420, −494 and −510 ppm, re te constants for the three decavanadate signals VA, VB and VC are shown with nvention in Table 1. 2.2. Kinetic Studies by 51V NMR For comparison, the increase in concentration of the V1 and V a function of time are plotted in Figure 3b. Interestingly, the reaction is faster on of decavanadate than at 1 mM for NaDeca and MetfDeca compounds (Tab nts of 0.5 mM NaDeca (2.28 ± 0.08) × 10−3 and (1.72 ± 0.07) × 10−4 for the appear ecies, respectively, are three and four times higher than the ones calculated fo compound (7.63 ± 0.8) × 10−4 and (4.09 ± 0.3) × 10−5 for V4 and V5 species, re ly, the rate constants for the appearance of the V4 and V5 species (Table 1) do ly for 1 mM NaDeca compared with 1 mM MetfDeca. 1 mM of NaDeca and MetfDeca, the V10 and V1 species were present at pH 4, but the hydrolysis of both compounds in DMEM medium allowed us to follow by 51V NMR the formation and the increment in the concentration over time at 25 ◦C of the orthovanadate, V1 and metavanadate species V2, V4 and V5 at −556, −570, −578 and −586 ppm, respectively. The kinetics of the decomposition of 1.0 mM NaDeca and MetfDeca (10 mM total vanadium) are plotted in Figure 3a, where the vanadium concentration for V10 species was calculated by integration of the V10A (2 vanadium atoms), V10B (4 vanadium atoms) and V10C (4 vanadium atoms) resonances at −420, −494 and −510 ppm, respectively, and the rate constants for the three decavanadate signals VA, VB and VC are shown with a negative sign by convention in Table 1. For comparison, the increase in concentration of the V1 and V4 vanadate species as a function of time are plotted in Figure 3b. Interestingly, the reaction is faster at 0.5 mM concentration of decavanadate than at 1 mM for NaDeca and MetfDeca compounds (Table 1). The rate constants of 0.5 mM NaDeca (2.28 ± 0.08) × 10−3 and (1.72 ± 0.07) × 10−4 for the appearance of V4 and V5 species, respectively, are three and four times higher than the ones calculated for 0.5 mM MetfDeca compound (7.63 ± 0.8) × 10−4 and (4.09 ± 0.3) × 10−5 for V4 and V5 species, respectively. Surprisingly, the rate constants for the appearance of the V4 and V5 species (Table 1) do not diﬀer signiﬁcantly for 1 mM NaDeca compared with 1 mM MetfDeca. V5 at 556, 570, 578 and 586 ppm, respectively. 2.2. Kinetic Studies by 51V NMR (a) 51V NMR decomposition of 1 mM NaDeca and MetfDeca in Dulbecco’s modiﬁed Eagle’s medium (DMEM) medium, plotted as vanadium concentration (10 mM total vanadium concentration) associated with the decameric species V10A (circle), V10B (square), V10C (triangle) over time at 25 ◦C. (b) Hydrolysis of (left) 1 mM and (right) 0.5 mM NaDeca and MetfDeca in DMEM medium followed by the formation of the orthovanadate V1 at −556 ppm (red circle for NaDeca and blue circle for MetfDeca) and metavanadate species V4 at −578 ppm (red triangle for NaDeca and blue triangle for MetfDeca) over time at 25 ◦C. 0 20 40 60 80 100 1.0×10-4 3.2×10-4 1.0×10-3 t, min [Vanadium], M b) 1 mM -556Nadeca -556Metfdeca -578Nadeca -578Metfdeca 0 20 40 60 80 100 5.0×10-4 1.0×10-3 1.5×10-3 t, min [Vanadium], M 0.5 mM -556Nadeca -556Metfdeca -578Nadeca -578Metfdeca 0 20 40 60 80 100 1.0×10-4 3.2×10-4 1.0×10-3 t, min [Vanadium], M b) 1 mM -556Nadeca -556Metfdeca -578Nadeca -578Metfdeca [Vanadium], M 0 20 40 60 80 100 5.0×10-4 1.0×10-3 1.5×10-3 t, min [Vanadium], M 0.5 mM -556Nadeca -556Metfdeca -578Nadeca -578Metfdeca a) 51V NMR decomposition of 1 mM NaDeca and MetfDeca in Dulbecco’s modifie (DMEM) medium, plotted as vanadium concentration (10 mM total v ion) associated with the decameric species V10A (circle), V10B (square), V10C (trian °C. (b) Hydrolysis of (left) 1 mM and (right) 0.5 mM NaDeca and MetfDeca i ollowed by the formation of the orthovanadate V1 at −556 ppm (red circle for Na Figure 3. (a) 51V NMR decomposition of 1 mM NaDeca and MetfDeca in Dulbecco’s modiﬁed Eagle’s medium (DMEM) medium, plotted as vanadium concentration (10 mM total vanadium concentration) associated with the decameric species V10A (circle), V10B (square), V10C (triangle) over time at 25 ◦C. (b) Hydrolysis of (left) 1 mM and (right) 0.5 mM NaDeca and MetfDeca in DMEM medium followed by the formation of the orthovanadate V1 at −556 ppm (red circle for NaDeca and blue circle for MetfDeca) and metavanadate species V4 at −578 ppm (red triangle for NaDeca and blue triangle for MetfDeca) over time at 25 ◦C. Figure 3. (a) 51V NMR decomposition of 1 mM NaDeca and MetfDeca in Dulbecco’s modiﬁed Eagle’s rcle for MetfDeca) and metavanadate species V4 at −578 ppm (red triangle for NaDeca an e for MetfDeca) over time at 25 °C. 2.2. Kinetic Studies by 51V NMR The kinetics of the decompos ca and MetfDeca (10 mM total vanadium) are plotted in Figure 3a, where the on for V10 species was calculated by integration of the V10A (2 vanadium atom atoms) and V10C (4 vanadium atoms) resonances at −420, −494 and −510 ppm, re e constants for the three decavanadate signals VA, VB and VC are shown with nvention in Table 1. For comparison, the increase in concentration of the V1 and V a function of time are plotted in Figure 3b. Interestingly, the reaction is faster on of decavanadate than at 1 mM for NaDeca and MetfDeca compounds (Tab nts of 0.5 mM NaDeca (2.28 ± 0.08) × 10−3 and (1.72 ± 0.07) × 10−4 for the appear cies, respectively, are three and four times higher than the ones calculated fo compound (7.63 ± 0.8) × 10−4 and (4.09 ± 0.3) × 10−5 for V4 and V5 species, re ly, the rate constants for the appearance of the V4 and V5 species (Table 1) do y for 1 mM NaDeca compared with 1 mM MetfDeca. 0 20 40 60 80 100 2×10-4 4×10-4 6×10-4 8×10-4 1×10-3 t, min [Vanadium], M 1mM Nadeca -420 -494 -510 a) 0 20 40 60 80 100 2×10-4 4×10-4 6×10-4 8×10-4 1×10-3 t, min [Vanadium], M 1mM Metfdeca -420 -494 -510 Figure 3. Cont. 0 20 40 60 80 100 2×10-4 4×10-4 6×10-4 8×10-4 1×10-3 t min [Vanadium], M 1mM Nadeca -420 -494 -510 a) a) 0 20 40 60 80 100 2×10-4 4×10-4 6×10-4 8×10-4 1×10-3 t, min [Vanadium], M 1mM Metfdeca -420 -494 -510 Figure 3. Cont. [Vanadium], M 6 of 20 Inorganics 2020, 8, 67 , , (a) 51V NMR decomposition of 1 mM NaDeca and MetfDeca in Dulbecco’s modifie (DMEM) medium, plotted as vanadium concentration (10 mM total v tion) associated with the decameric species V10A (circle), V10B (square), V10C (trian 5 °C. (b) Hydrolysis of (left) 1 mM and (right) 0.5 mM NaDeca and MetfDeca in followed by the formation of the orthovanadate V1 at −556 ppm (red circle for Na 0 20 40 60 80 100 1.0×10-4 3.2×10-4 1.0×10-3 t, min [Vanadium], M b) 1 mM -556Nadeca -556Metfdeca -578Nadeca -578Metfdeca 0 20 40 60 80 100 5.0×10-4 1.0×10-3 1.5×10-3 t, min [Vanadium], M 0.5 mM -556Nadeca -556Metfdeca -578Nadeca -578Metfdeca Figure 3. 2.2. Kinetic Studies by 51V NMR ecomposition of NaDeca and MetfDeca show first-order dependence versus ti Deca at 0.5 and 1 mM concentration, the calculated half-life time of the decomp dium at 25 °C is 9 h. In contrast, the calculated lifetime for MetfDeca is 9 h and 1 mM concentration, respectively (Table 1). These results are in line with the ha omposition of decameric species found by Ramos et al., where for 10 μM dec on in different buffers pH 7–7.5, the half-life time is between 5 to 10 h. In that The decomposition of NaDeca and MetfDeca show ﬁrst-order dependence versus time. In the case of NaDeca at 0.5 and 1 mM concentration, the calculated half-life time of the decomposition in DMEM medium at 25 ◦C is 9 h. In contrast, the calculated lifetime for MetfDeca is 9 h and 11 h for 0.5 mM and 1 mM concentration, respectively (Table 1). These results are in line with the half-life time for the decomposition of decameric species found by Ramos et al., where for 10 µM decavanadate concentration in diﬀerent buﬀers pH 7–7.5, the half-life time is between 5 to 10 h. In that study, the authors performed a stabilization study of the decavanadate species with the G-actin protein, and due to the coordination of the protein with the decameric species, its half-life time was increased ﬁve times from 5 to 27 h at 10 µM of decavanadate concentration, however, in the same study the addition of 200 µM of ATP to the medium prevented the actin polymerization by V10 and the half-life time decreased from 27 to 10 h [12]. p rformed a stabilization study of the decavanadate species with the G-actin pr coordination of the protein with the decameric species, its half-life time was incr 5 to 27 h at 10 μM of decavanadate concentration, however, in the same study th of ATP to the medium prevented the actin polymerization by V10 and the hal from 27 to 10 h [12]. The decomposition rate of the decavanadate moiety is sensitive to the cations present in solution [16], the fast reaction in acid media can be accelerated by alkali metal cations and slowed down by large cations such as tetra-alkylammonium ions due to the formation of ionic-pairs with the protonated decavanadate to form [VO2]+ in seconds [16]. 2.2. Kinetic Studies by 51V NMR In basic media, the reaction is slower than in acid media, but the decomposition reaction proceeds via base independent (k10) and base dependent (k2) paths (Equation (1)). In the absence of sodium ions, the rate of reaction is independent of [OH−] [17]. In this work, it seems that the base-dependent decomposition path is active as well, because for 7 of 20 × 10−3 Inorganics 2020, 8, 67 NaDeca and MetfDeca, the observed rate of decomposition is not increasing with decavanadate concentration in both cases (Table 1), and the presence of a high sodium concentration in the DMEM media (µ = 0.1 M NaCl), produces an increase in the decomposition rate via a reactive alkali-metal decavanadate species (k2) Equation (1) [17,38]. 0.001 M 494 (V10B) ( 1.13 ± 0.1) × 10 ( 8.39 ± 0.2) × 10 −510 (V10c) (−1.16 ± 0.2) × 10−3 (−9.49 ± 0.5) × 10−4 −556 (V1) (2.62 ± 0.6) × 10−3 (2.00 ± 0.2) × 10−3 −570 (V2) (6.81 ± 0.4) × 10−4 (5.34 ± 0.4) × 10−4 −578 (V4) (1.96 ± 0.2) × 10−3 (1.85 ± 0.09) × 10−3 Table 1. Summary of rate constants for the decomposition of 0.5 and 1.0 mM NaDeca and MetfDeca compound hydrolysis in DMEM medium at 25 ◦C and pH 7.4. 2.2. Kinetic Studies by 51V NMR Decavanadate Compound Concentration 51V NMR Signal a kobs, min−1 NaDeca kobs, min−1 MetfDeca 0.0005 M −420 (V10A) (−1.25 ± 0.03) × 10−3 (−1.40 ± 0.03) × 10−3 −494 (V10B) (−1.26 ± 0.2) × 10−3 (−1.48 ± 0.7) × 10−3 −510 (V10c) (−1.82 ± 0.4) × 10−3 (−1.44 ± 0.4) × 10−3 −556 (V1) (4.59 ± 0.7) × 10−3 (2.97 ± 0.4) × 10−3 −570 (V2) (7.07 ± 0.7) × 10−4 (6.17 ± 0.8) × 10−4 −578 (V4) (2.28 ± 0.08) × 10−3 (7.63 ± 0.8) × 10−4 −586 (V5) (1.72 ± 0.07) × 10−4 (4.09 ± 0.3) × 10−5 0.001 M −420 (V10A) (−1.46 ± 0.7) × 10−3 (−1.36 ± 0.2) × 10−3 −494 (V10B) (−1.13 ± 0.1) × 10−3 (−8.39 ± 0.2) × 10−4 −510 (V10c) (−1.16 ± 0.2) × 10−3 (−9.49 ± 0.5) × 10−4 −556 (V1) (2.62 ± 0.6) × 10−3 (2.00 ± 0.2) × 10−3 −570 (V2) (6.81 ± 0.4) × 10−4 (5.34 ± 0.4) × 10−4 −578 (V4) (1.96 ± 0.2) × 10−3 (1.85 ± 0.09) × 10−3 −586 (V5) (1.89 ± 0.1) × 10−4 (1.38 ± 0.8) × 10−4 a For calculating the rate of consumption of the decavanadate complexes, three diﬀerent resonances were used, −420, −494 and −510, whereas for calculating the rate of appearance for the V1, V2, V4 and V5, only one resonance was used. −586 (V5) (1.89 ± 0.1) × 10−4 (1.38 ± 0.8) × 10−4 a For calculating the rate of consumption of the decavanadate complexes, three different resonances were used, −420, −494 and −510, whereas for calculating the rate of appearance for the V1, V2, V4 and V5, only one resonance was used. The decomposition rate of the decavanadate moiety is sensitive to the cations present in solution 16], the fast reaction in acid media can be accelerated by alkali metal cations and slowed down by arge cations such as tetra-alkylammonium ions due to the formation of ionic-pairs with the protonated decavanadate to form [VO2]+ in seconds [16]. In basic media, the reaction is slower than n acid media, but the decomposition reaction proceeds via base independent (k1’) and base dependent (k2) paths (Equation (1)). In the absence of sodium ions, the rate of reaction is independent of [OH−] [17]. 2.2. Kinetic Studies by 51V NMR In this work, it seems that the base-dependent decomposition path is active as well, because for NaDeca and MetfDeca, the observed rate of decomposition is not increasing with decavanadate concentration in both cases (Table 1), and the presence of a high sodium concentration n the DMEM media (μ = 0.1 M NaCl), produces an increase in the decomposition rate via a reactive alkali-metal decavanadate species (k2) Equation (1) [17,38]. Table 1. Summary of rate constants for the decomposition of 0.5 and 1.0 mM NaDeca and MetfDeca compound hydrolysis in DMEM medium at 25 ◦C and pH 7.4. 586 (V5) (1.89 ± 0.1) × 10 4 (1.38 ± 0.8) × 10 4 a For calculating the rate of consumption of the decavanadate complexes, three different resonances were used −420 −494 and −510 whereas for calculating the rate of appearance for the V1 V2 V4 and Table 1. Summary of rate constants for the decomposition of 0.5 and 1.0 mM NaDeca and MetfDeca compound hydrolysis in DMEM medium at 25 ◦C and pH 7.4. 586 (V5) (1.89 ± 0.1) × 10 (1.38 ± 0.8) × 10 a For calculating the rate of consumption of the decavanadate complexes, three different resonances were used −420 −494 and −510 whereas for calculating the rate of appearance for the V1 V2 V4 and a For calculating the rate of consumption of the decavanadate complexes, three diﬀerent resonances were used, −420, −494 and −510, whereas for calculating the rate of appearance for the V1, V2, V4 and V5, only one resonance was used. the DMEM media (μ = 0.1 M NaCl), produces an increase in the decomposition rate via a reactiv kali-metal decavanadate species (k2) Equation (1) [17,38]. −d[V10O286−]tot/dt = [k10 + k2[OH−]][V10O286−]tot (1) d[V4O124−]/dt = [k10 + k2[OH−]][V10O286−]tot (2) −d[V10O286−] tot/dt = [k1’ + k2[OH−] ] [V10O286−] tot (1) d[V4O124−]/dt = [k1’ + k2[OH−]][V10O286−]tot (2) (1) (2) (1) (2) (1) (1) (2) (2) Goddard and Druskovich’s [17,38] decomposition experiments were followed by UV-Vis techniques, although metavanadate species formation was not reported. Decavanadate 51V NMR signals are wide, and the spectrum acquisition takes longer than the UV-Vis one. However, metavanadate species formation can be followed by 51V NMR. 2.2. Kinetic Studies by 51V NMR In Table 1, NaDeca hydrolysis products formation rates are moderately faster than the ones calculated for MetfDeca, and the reaction rate is not increasing with the decavanadate concentration, so Equation (1) for the decomposition reaction was rewritten as Equation (2), where the reaction rate was expressed in terms of the metavanadate species formation. Based on the literature and our results, we proposed that in high alkali metal concentration, like in DMEM medium, the sodium ions form an ionic aggregate with the V10 species (Scheme 1), which then reacts with the hydroxide ion [17]. In this work, M+ is the sodium ion, and the M’ is the metformin cation (C4H12N5+), which at pH 7 is monoprotonated [33]. Goddard and Druskovich’s [17,38] decomposition experiments were followed by UV-Vis techniques, although metavanadate species formation was not reported. Decavanadate 51V NMR signals are wide, and the spectrum acquisition takes longer than the UV-Vis one. However, metavanadate species formation can be followed by 51V NMR. In Table 1, NaDeca hydrolysis products formation rates are moderately faster than the ones calculated for MetfDeca, and the reaction rate is not increasing with the decavanadate concentration, so Equation (1) for the decomposition reaction was rewritten as Equation (2), where the reaction rate was expressed in terms of the metavanadate species formation. Based on the literature and our results, we proposed that in high alkali metal concentration, like in DMEM medium, the sodium ions form an ionic aggregate with the V10 species (Scheme 1), which then reacts with the hydroxide ion [17]. In this work, M+ is the sodium ion, and the M’ is the metformin cation (C4H12N5+), which at pH 7 is monoprotonated [33]. Scheme 1. Putative reaction mechanism for decavanadate decomposition reaction in DMEM medium at pH 7.4. Scheme 1. Putative reaction mechanism for decavanadate decomposition reaction in DMEM medium at pH 7.4. Scheme 1. Putative reaction mechanism for decavanadate decomposition reaction in DMEM medium at pH 7.4. Scheme 1. Putative reaction mechanism for decavanadate decomposition reaction in DMEM medium at pH 7.4. 2.3. Cell Viability solution at pH 7 the insulin relea To investigate the cytotoxicity of vanadium compounds against non-tumoral and tumoral human cells and potential anticancer activity, the compounds NaDeca, MetfDeca, MetV and Metf were tested against HEK293 human embryonic kidney cells and HEPG2 human liver cancer cells. In Figure 4a, the percentage of cell viability vs. compound concentration for the four compounds against HEK293 is shown. The IC50 value found for NaDeca was 40 ± 4 µM, for MetfDeca was 85 ± 5 µM, for sodium MetV was 181 ± 7 µM and for Metf was 420 ± 11 µM. In the case of the HEPG2, the cytotoxicity dose dependence is shown in Figure 4b. The highest cytotoxic activity was observed for NaDeca, with an IC50 value of 9.0 ± 0.7 µM, follow by the MetfDeca with an IC50 of 29 ± 0.7 µM, and IC50 values of 93 ± 5 and 540 ± 4 µM for MetV and Metf, respectively. 2.3. Cell Viability To investigate the cytotoxicity of vanadium compounds against non-tumoral and tumoral human cells and potential anticancer activity, the compounds NaDeca, MetfDeca, MetV and Metf were tested against HEK293 human embryonic kidney cells and HEPG2 human liver cancer cells. In Figure 4a, the percentage of cell viability vs. compound concentration for the four compounds against HEK293 is shown. The IC50 value found for NaDeca was 40 ± 4 μM, for MetfDeca was 85 ± 5 μM, for sodium MetV was 181 ± 7 μM and for Metf was 420 ± 11 μM. In the case of the HEPG2, the cytotoxicity dose dependence is shown in Figure 4b. The highest cytotoxic activity was observed for NaDeca, with an IC50 value of 9.0 ± 0.7 μM, follow by the MetfDeca with an IC50 of 29 ± 0.7 μM, and IC50 values of 93 ± 5 and 540 ± 4 μM for MetV and Metf respectively μ , p y Figure 4. Cell viability assay at different vanadium compounds concentrations after treatment for 24 h (a) HEK293 and (b) HEPG2 cells. The cell viability of each treatment group was compared with the corresponding untreated control, which was normalized to 100% of cell viability. Error bars represent the standard deviation for triplicate runs (n = 3). 2.2. Kinetic Studies by 51V NMR The base dependent equation can be rewritten as: The base dependent equation can be rewritten as: The base dependent equation can be rewritten as: The base dependent equation can be rewritten as: The base dependent equation can be rewritten as: The base dependent equation can be rewritten as: Rate = k2[OH−][V10O286−]tot = k2M[OH−][V10O286−]tot (3) Rate = k2[OH−][V10O286−]tot = k2M[OH−][V10O286−]tot (3) (3) (3) where where [V10O286−]tot = [(C4H12N5)V10O285−] + [NaV10O285−] + [Na2V10O284−] (4) [V10O286−]tot = [(C4H12N5)V10O285−] + [NaV10O285−] + [Na2V10O284−] (4) (4) (4) 8 of 20 Inorganics 2020, 8, 67 In DMEM medium, it seems likely that the ion-pair association is stronger with the metformin cation (C4H12N5)V10O285−at 1 mM MetfDeca concentration, which does not form at the same rate as the Na2V10O284−active species to react with the free OH−anion (Scheme 1) like 1 mM NaDeca, where the total V10 anion concentration (Equation (4)) is almost in the higher ionic aggregate Na2V10O284− species, and on that way can follow the base-dependent path (Equation (3)). However, the same calculated values for the decomposition rate of 0.5 mM and 1 mM NaDeca (Table 1) suggest that the concentration of Na2V10O284−species remains the same under the buﬀer conditions; several ion-pairs can be proposed by the combination of monovalent cation and hexavalent anion. Nevertheless, Schwarzenbach and Geier [39] showed that the alkali metal cations formed the ion-pair complexes MHV10O284−, MV10O285−, and M2V10O284−base on their formation constants 91% of the decavanadate is in the ion-pair form M2V10O284−and 9% in the form MV10O285−for M = Li or Na [39]. Inorganics 2020, 8, x FOR PEER REVIEW 8 of 20 In DMEM medium, it seems likely that the ion-pair association is stronger with the metformin cation (C4H12N5)V10O285− at 1 mM MetfDeca concentration, which does not form at the same rate as the Na2V10O284− active species to react with the free OH− anion (Scheme 1) like 1 mM NaDeca, where the total V10 anion concentration (Equation (4)) is almost in the higher ionic aggregate Na2V10O284− species, and on that way can follow the base-dependent path (Equation (3)). 2.2. Kinetic Studies by 51V NMR However, the same calculated values for the decomposition rate of 0.5 mM and 1 mM NaDeca (Table 1) suggest that the concentration of Na2V10O284−species remains the same under the buffer conditions; several ion-pairs In vanadium speciation diagrams, at total vanadium concentration lower than 5 µM, the decavanadate anion is not formed [10], but some meta and orthovanadate species are present in solution at neutral pH. In that regard, this kinetic study was performed to have an approximate of the constant rate values at which the oligomer vanadium species were formed and, with some cautions in the interpretations of the data, would allow us to compare the biological activity of MetV (V1) and Metf versus NaDeca and MetfDeca to show if the hydrolysis products produce a different biological response than the orthovanadate (V1) and to quantify if MetfDeca compound promotes a synergistic effect between its components that increase the decavanadate antidiabetic properties. In that regard, the biological experiments that are shown in the next sections were performed in DMEM solution at pH 7.4, and the cells were incubated with the compounds for 24 h, with the exception of the insulin release assay, where the cells were incubated with the compounds for one hour. concentration of Na2V10O28 species remains the same under the buffer conditions; several ion-pairs can be proposed by the combination of monovalent cation and hexavalent anion. Nevertheless, Schwarzenbach and Geier [39] showed that the alkali metal cations formed the ion-pair complexes MHV10O284−, MV10O285−, and M2V10O284− base on their formation constants 91% of the decavanadate is in the ion-pair form M2V10O284− and 9% in the form MV10O285− for M = Li or Na [39]. In vanadium speciation diagrams, at total vanadium concentration lower than 5 μM, the decavanadate anion is not formed [10], but some meta and orthovanadate species are present in solution at neutral pH. In that regard, this kinetic study was performed to have an approximate of the constant rate values at which the oligomer vanadium species were formed and, with some cautions in the interpretations of the data, would allow us to compare the biological activity of MetV (V1) and Metf versus NaDeca and MetfDeca to show if the hydrolysis products produce a different biological response than the orthovanadate (V1) and to quantify if MetfDeca compound promotes a synergistic effect between its components that increase the decavanadate antidiabetic properties. 2.2. Kinetic Studies by 51V NMR In that regard, the biological experiments that are shown in the next sections were performed in DMEM 2.3. Cell Viability solution at pH 7 the insulin relea 0 50 100 150 0 20 40 60 80 100 120 Concentration (µM) Cell viability (% of control) HEK293 Nadeca Metfdeca MetV a) Metf 0 50 100 150 0 20 40 60 80 100 120 Concentration (µM) Cell viability (% of control) HEPG2 Nadeca Metfdeca MetV b) Metf Figure 4. Cell viability assay at diﬀerent vanadium compounds concentrations after treatment for 24 h (a) HEK293 and (b) HEPG2 cells. The cell viability of each treatment group was compared with the corresponding untreated control, which was normalized to 100% of cell viability. Error bars represent the standard deviation for triplicate runs (n = 3). 0 50 100 150 0 20 40 60 80 100 120 Concentration (µM) Cell viability (% of control) HEPG2 Nadeca Metfdeca MetV b) Metf μ 0 50 100 150 0 20 40 60 80 100 120 Concentration (µM) Cell viability (% of control) HEK293 a) a) Figure 4. Cell viability assay at different vanadium compounds concentrations after treatment for 24 h (a) HEK293 and (b) HEPG2 cells. The cell viability of each treatment group was compared with the corresponding untreated control, which was normalized to 100% of cell viability. Error bars represent the standard deviation for triplicate runs (n = 3). Figure 4. Cell viability assay at diﬀerent vanadium compounds concentrations after treatment for 24 h (a) HEK293 and (b) HEPG2 cells. The cell viability of each treatment group was compared with the corresponding untreated control, which was normalized to 100% of cell viability. Error bars represent the standard deviation for triplicate runs (n = 3). 9 of 20 Inorganics 2020, 8, 67 As can be seen in Figure 4a,b, the cell viability decreases in a dose manner response. The IC50 of the vanadium compounds against HEPG2 cells is around the same value as other compounds previously reported, like cis-platin (15.9 µM) [40] and monomeric V4+ compounds [41–43]. As can be seen in Figure 4a,b, the cell viability decreases in a dose manner response. The IC50 of the vanadium compounds against HEPG2 cells is around the same value as other compounds previously reported, like cis-platin (15.9 µM) [40] and monomeric V4+ compounds [41–43]. The metformin hydrochloride does not reduce the cellular viability in the range of concentrations that the decavanadate compounds do; the NaDeca compound exhibits more activity than MetfDeca and MetV regardless of the cell line after 24 h. 2.3. Cell Viability solution at pH 7 the insulin relea The cytotoxicity of the tested compounds against HEPG2 is diﬀerent for the non-tumorigenic HEK293 cells indicating that the toxicity of the compounds exhibits a good correlation on selectivity toward HEPG2 cancer cells in 24 h (see Table 2). The three vanadium and Metf compounds do not aﬀect the viability of the HEK293 cells; this is an important result from this work, which may have an impact due to the new strategies intended to reduce the renal toxicity induced by cisplatin [44,45]. Table 2. Cytotoxic activity (IC50) and selectivity index (SI) of compounds against HEK293 and HEPG2 cells after 24 h. Compound HEK293 Lower Cell Viability (%) HEK293 IC50 (µM) HEPG2 Lower Cell Viability (%) HEPG2 IC50 (µM) SI (IC50 HEK293/ (IC50 HEPG2) NaDeca 61 40 ± 4 22 9 ± 0.7 4.4 MetfDeca 67 85 ± 5 28 29 ± 0.7 2.9 MetV 88 181 ± 7 51 93 ± 5 1.9 Metf 100 420 ± 11 98 540 ± 4 0.77 Table 2. Cytotoxic activity (IC50) and selectivity index (SI) of compounds against HEK293 and HEPG2 cells after 24 h. In the case of decavanadate compounds, the IC50 seems strongly dependent on the type of cell line and the counter ion; for example, the IC50 of the Na4[(HOCH2CH2)3NH]2[V10O28]·6H2O towards HEPG2 cell line is 16.4 ± 3 µg/mL while for human cervical cancer cell line (Hela cells) is 53.1 ± 12.1 µg/mL [46], the compounds [(H2tmen)3V10O28]·6H2O and [(H2en)3V10O28]·2H2O were tested in human normal hepatocytes L02, and their IC50 values are 6.5 ± 0.6 and 7.2 ± 0.7 µM, respectively indicating that are cytotoxic for the L02 human cell line [47]. In 2018 Nunes and coworkers studied the cytotoxicity effect of three decavanadates compounds in African green monkey kidney (Vero) cells, and the three compounds exhibit low effect; 200 µM of the compounds reduced 50% of the Vero cells viability in 96 h. The compounds tested were the decavanadate complexes of sodium, nicotinamidium [(3-Hpca)4H2V10O28]·2H2O·2(3-pca) and isonicotinamidium [(4-Hpca)4H2V10O28]·2(4-pca) [48]. However, in the three studies presented before [46–48] for decavanadate compounds, the effect of the counter ion in the cytotoxic studies was not studied. It seems that the decavanadate compounds—or their decomposition products V1, V2 and V4— decreased the viability of hepatocarcinoma HEPG2 cells faster than the normal HEK293 cells (Table 2) after 24 h. 2.3. Cell Viability solution at pH 7 the insulin relea The cytotoxicity of cancer and normal cells can be attributed to a diﬀerent mechanism like Wang and coworkers reported in 2010 [49] that for 100 µM of MetV in MEM (minimum essential medium) in normal hepatocytes L02, the cell arrest mechanism is ROS-dependent and for HEPG2 is ROS-independent to mediated ERK (extracellular signal-regulated protein kinase activation) after 72 h. In the present study, the Metf cation association with the decavanadate moiety promotes some kind of protection against the normal HEK293 cells. However, the dissociation of the ion-pairs NaV10O285−, (C4H12N5)V10O285−and the further hydrolysis to V1 and other products will not protect the vanadium atoms for the reduction into V4+ that could signiﬁcantly increase the ROS levels and the apoptosis for the normal cells. 2.4. Proteins Expression Protein kinase B (AKT) is a crucial mediator of insulin-resistant glucose and lipid digestion [50]. To evaluate the eﬀect that decavanadate compounds have in phosphatidylinositol 3-kinase (PI3K/AKT) and AMPK pathways in HEPG2 cells, a Western blot examination was performed. 10 of 20 Inorganics 2020, 8, 67 The cells were cultured with various concentrations of the compounds for 24 h without insulin. In Figure 5a we can see that NaDeca highly phosphorylates AKTα while the MetfDeca, MetV and Metformin show moderate activity. Thus, NaDeca and MetfDeca compounds exhibit a dose-dependence increase in phosphorylated AKT (p-AKT) as shown in Figure 5c, where 8, 16, and 30 µM of NaDeca induced a 2, 4 and 6-fold-increase in the phosphorylation, respectively. In contrast, the expression of the AMPK, a cellular metabolism energy sensor, by its phosphorylation p-AMPKα is not signiﬁcantly elevated by the compounds (Figure 5d). However, NaDeca in 8 µM concentration exhibits around 33% of the increase in the AMPK phosphorylation. The low percentage of phosphorylation in AMPK by the vanadium compounds and metformin (Figure 5d) can be explained as follows: in hepatocellular carcinoma (HCC), the pathway function is downregulated [51], it seems like a low level of AMPK is required to maintain viability during the metabolic stress of tumor cells by diﬀerent mechanisms [52]. Inorganics 2020, 8, x FOR PEER REVIEW 10 of 20 The cells were cultured with various concentrations of the compounds for 24 h without insulin. In Figure 5a we can see that NaDeca highly phosphorylates AKTα while the MetfDeca, MetV and Metformin show moderate activity. Thus, NaDeca and MetfDeca compounds exhibit a dose- dependence increase in phosphorylated AKT (p-AKT) as shown in Figure 5c, where 8, 16, and 30 μM of NaDeca induced a 2, 4 and 6-fold-increase in the phosphorylation, respectively. In contrast, the expression of the AMPK, a cellular metabolism energy sensor, by its phosphorylation p-AMPKα is not significantly elevated by the compounds (Figure 5d). However, NaDeca in 8 μM concentration exhibits around 33% of the increase in the AMPK phosphorylation. The low percentage of phosphorylation in AMPK by the vanadium compounds and metformin (Figure 5d) can be explained as follows: in hepatocellular carcinoma (HCC), the pathway function is downregulated [51], it seems like a low level of AMPK is required to maintain viability during the metabolic stress of tumor cells b diff t h i [52] y [ ] (a) (b) (c) (d) Figure 5. 2.4. Proteins Expression Effect of vanadium compounds on the protein expression of (a) protein kinase B (AKT) representative Western blot and (b) AMP-activated kinase (AMPK) representative Western blot in HEPG2 human hepatocarcinoma cells. Metf 2 mM and vanadium compounds 8, 16 and 30 μM concentrations, respectively, from left to right. Quantitative data of (c) p-AKT/t-AKT and t- AKT/actine (d) p-AMPK/t-AMPK and t-AMPK/actine. All the values are the mean ± SD. * p < 0.05, p < 0.01 and * p < 0.001 vs. untreated control cells. Control Metf MetV Metfdeca Nadeca 0 2 4 6 8 Control 2 mM 30 µM 8 µM 16 µM * * *** p-AKT/t-AKT (fold of control) Control Metf MetV Metfdeca Nadeca 0.0 0.5 1.0 1.5 2.0 Control 2 mM 30 µM 8 µM 16 µM * * * * * * t-AKT/actine (fold of control) ControlMetf MetV Metfdeca Nadeca 0.0 0.5 1.0 1.5 2.0 Control 2 mM 30 µM 8 µM 16 µM p-AMPK/t-AMPK (fold of control) Control Metf MetV Metfdeca Nadeca 0.0 0.5 1.0 1.5 2.0 Control 2 mM 30 µM 8 µM 16 µM t-AMPK/actine (fold of control) * * Figure 5. Eﬀect of vanadium compounds on the protein expression of (a) protein kinase B (AKT) representative Western blot and (b) AMP-activated kinase (AMPK) representative Western blot in HEPG2 human hepatocarcinoma cells. Metf 2 mM and vanadium compounds 8, 16 and 30 µM concentrations, respectively, from left to right. Quantitative data of (c) p-AKT/t-AKT and t-AKT/actine (d) p-AMPK/t-AMPK and t-AMPK/actine. All the values are the mean ± SD. * p < 0.05, p < 0.01 and * p < 0.001 vs. untreated control cells. (a) (b) (b) (a) (a) Control Metf MetV Metfdeca Nadeca 0 2 4 6 8 * * *** p-AKT/t-AKT (fold of control) ControlMetf MetV Metfdeca Nadeca 0.0 0.5 1.0 1.5 2.0 Control 2 mM 30 µM 8 µM 16 µM p-AMPK/t-AMPK (fold of control) (c) Control Metf MetV Metfdeca Nadeca 0.0 0.5 1.0 1.5 2.0 * * * * * * t-AKT/actine (fold of control) (d) Control Metf MetV Metfdeca Nadeca 0.0 0.5 1.0 1.5 2.0 Control 2 mM 30 µM 8 µM 16 µM t-AMPK/actine (fold of control) * * (d) (c) Figure 5. 2.4. Proteins Expression Effect of vanadium compounds on the protein expression of (a) protein kinase B (AKT) representative Western blot and (b) AMP-activated kinase (AMPK) representative Western blot in HEPG2 human hepatocarcinoma cells. Metf 2 mM and vanadium compounds 8, 16 and 30 μM concentrations, respectively, from left to right. Quantitative data of (c) p-AKT/t-AKT and t- AKT/actine (d) p-AMPK/t-AMPK and t-AMPK/actine. All the values are the mean ± SD. * p < 0.05, p < 0 01 and * p < 0 001 vs untreated control cells Figure 5. Eﬀect of vanadium compounds on the protein expression of (a) protein kinase B (AKT) representative Western blot and (b) AMP-activated kinase (AMPK) representative Western blot in HEPG2 human hepatocarcinoma cells. Metf 2 mM and vanadium compounds 8, 16 and 30 µM concentrations, respectively, from left to right. Quantitative data of (c) p-AKT/t-AKT and t-AKT/actine (d) p-AMPK/t-AMPK and t-AMPK/actine. All the values are the mean ± SD. * p < 0.05, p < 0.01 and * p < 0.001 vs. untreated control cells. 2.4. Proteins Expression In the present work, the activation of AKT by 2 mM of Metf and by 8–30 μM of MetfDeca is moderate; in the case of Metf, p-AKT is increased by 21% while for 30 μM of MetfDeca, the increase is 80%; however, we observed that the NaDeca formation rate of metavanadate species is moderately faster than MetfDeca under the same experimental conditions (Table 1), due to the weaker ionic pairs for NaDeca than for MetfDeca (scheme 1), so if all the 8 μM NaDeca decomposition product is V1 the p-AKT fold should be ten times 0.21, the fold value that we found experimentally for the MetV is 2.1 In the present work, the activation of AKT by 2 mM of Metf and by 8–30 µM of MetfDeca is moderate; in the case of Metf, p-AKT is increased by 21% while for 30 µM of MetfDeca, the increase is 80%; however, we observed that the NaDeca formation rate of metavanadate species is moderately faster than MetfDeca under the same experimental conditions (Table 1), due to the weaker ionic pairs for NaDeca than for MetfDeca (Scheme 1), so if all the 8 µM NaDeca decomposition product is V1 the p-AKT fold should be ten times 0.21, the fold value that we found experimentally for the MetV is 2.1 (Figure 5c), this clearly indicates that 8 µM of NaDeca is decomposed to 80 µM of V1. Nevertheless, the decomposition of NaDeca at higher concentration solutions shows lower amounts of V1 produced, based on the p-AKT fold activity. If we double NaDeca concentration to 16 µM, the experimental fold 11 of 20 Inorganics 2020, 8, 67 value is 3.3 for the decameric compound, and for MetV is 0.21, while for 30 µM of MetV, the p-AKT fold value is 0.45 and for NaDeca is 6. It seems that higher vanadium concentrations induce the formation of larger oligomers like V2, V4 and V5. Thus the active species could be a combination of the orthovanadate and the metavanadate units, with diﬀering AKT signaling activation mechanisms promoted by diﬀerent vanadium species. p y p Activation of the IR kinases by vanadium compounds exhibit diﬀerent mechanisms, which depend on the type of cell and also the oxidation state of the metal [53,54]. 2.4. Proteins Expression Recently, several lines of evidence suggest that cancer cells upregulate the oxidative pentose phosphate pathway (PPP) to support cell growth and survival, by consequence exhibited increased PPP ﬂux, NADPH/NADPC ratio, and ROS [25], in the liver 30% of the glucose oxidation occurs via PPP, so, it is not surprising that the AKT activation could be in a phosphatidylinositol 3-kinase (PI3K)-dependent manner by ROS [24]. However, in our control experiments the AKT signal in HEPG2 cells has not been activated (Figure 5c), suggesting that AKT phosphorylation by MetV, MetfDeca and NaDeca can be attributed to the activation of PTB-1B by orthovanadate (V1) [32]. On the other hand, for NaDeca compound, the tetramer species is formed at the same speed than V1 (Table 1), V4 could be the one that is reduced [55] and the vanadium (IV) species VO2+ is activating the AKT pathway in a PI3K-dependent manner by ROS, like in the case of VOSO4 that exhibited a 17-fold increase in the phosphorylation of AKT in HEPG2 cells at 25 µM concentration [56]. In 2015 Levina and coworkers performed a speciation study by XANES spectroscopy, where for 1 mM of orthovanadate in HEPG2 cells with DMEM medium after 24 h, 50% of the initial vanadium was found as tetrahedral species of V5+ (V1, V2, V4 and V5 are tetrahedral), 30% as V4+ moieties with a coordination number of six and 20% as V4+ with a coordination number of ﬁve [19], this study supports our observation that after 24 h not more decavanadate species are present in solution. It also supports our hypothesis that not all the vanadium in solution is present in the highest oxidation state (V5+) and some has been reduced to V4+ promoting diﬀerent mechanism of AKT activation, particularly for the NaDeca compound. 2.6. Insulin Release Assa 2.6. Insulin Release Assay 2.6. Insulin Release Assay The effect on insulin release of NaDeca, MetfDeca, MetV at 30 μM concentration was studied in βTC-6 cells. Glucose 10 mM and repaglinide 30 μM concentration were used as control. The latter was used due to the pharmacological activity, such as blocking ATP-dependent K+ channels and stimulate the release of insulin from the pancreas in a dose-dependent manner [59]. Figure 7 shows that MetV and MetfDeca enhanced insulin release by 0.7 and 1-fold relative to glucose control. In addition, both vanadium compounds showed more activity than the repaglinide at the same concentration, while NaDeca shows lower activity at 30 μM concentration than the glucose and repaglinide controls. βTC-6 cells secrete insulin in response to glucose; however, this cell line derived from transgenic mice develop a high hexokinase activity [60]; in normal pancreatic β-cells isolated from mouse islets, the effects of NaVO3 were studied at 0.1–1 mM concentration [61], the authors found that vanadate did not affect basal insulin release, although, vanadate potentiated the glucose effect by a different mechanism than blocking the sodium pump or affecting the AmpC levels [61]. 60 The eﬀect on insulin release of NaDeca, MetfDeca, MetV at 30 µM concentration was studied in βTC-6 cells. Glucose 10 mM and repaglinide 30 µM concentration were used as control. The latter was used due to the pharmacological activity, such as blocking ATP-dependent K+ channels and stimulate the release of insulin from the pancreas in a dose-dependent manner [59]. Figure 7 shows that MetV and MetfDeca enhanced insulin release by 0.7 and 1-fold relative to glucose control. In addition, both vanadium compounds showed more activity than the repaglinide at the same concentration, while NaDeca shows lower activity at 30 µM concentration than the glucose and repaglinide controls. βTC-6 cells secrete insulin in response to glucose; however, this cell line derived from transgenic mice develop a high hexokinase activity [60]; in normal pancreatic β-cells isolated from mouse islets, the eﬀects of NaVO3 were studied at 0.1–1 mM concentration [61], the authors found that vanadate did not aﬀect basal insulin release, although, vanadate potentiated the glucose eﬀect by a diﬀerent mechanism than blocking the sodium pump or aﬀecting the AmpC levels [61]. y The effect on insulin release of NaDeca, MetfDeca, MetV at 30 μM concentration was studied in βTC-6 cells. Glucose 10 mM and repaglinide 30 μM concentration were used as control. 2.5. Glucose Uptake Assay The decomposition products promoted the 12 of 20 20 ition Inorganics 2020, 8, 67 Inorganics 2020, 8, in the presence inhibition of PTP and the activation of SSAO that regulates the translocation of the GLUT4 transport and stimulates glucose transport [58], like in the case of the vanadium compounds tested in this work, where the GLUT4 transport is translocated and the glucose is transported by the cell. were V1, V4 and [V(OH)2(OO)2(OH)2]2−. The decomposition products promoted the inhibition of PTP and the activation of SSAO that regulates the translocation of the GLUT4 transport and stimulates glucose transport [58], like in the case of the vanadium compounds tested in this work, where the GLUT4 transport is translocated and the glucose is transported by the cell. and the activation of SSAO that regulates the translocation of the GLUT4 transport and stimulates glucose transport [58], like in the case of the vanadium compounds tested in this work, where the GLUT4 transport is translocated and the glucose is transported by the cell. Figure 6. Effect of vanadium compounds on the glucose uptake in 3T3-L1 differentiated adipocytes at 16, 30 μM for vanadium compounds and 2 mM for Metf in the absence of insulin. All the values are the mean ± SD. * p < 0.05 and ** p < 0.01 vs. untreated control cells. Control Nadeca Metfdeca MetV Metf Insulin 0 20 40 60 80 100 Intensity (%) 16 µM 30 µM 100 nM * * ** 2 mM Figure 6. Eﬀect of vanadium compounds on the glucose uptake in 3T3-L1 diﬀerentiated adipocytes at 16, 30 µM for vanadium compounds and 2 mM for Metf in the absence of insulin. All the values are the mean ± SD. * p < 0.05 and ** p < 0.01 vs. untreated control cells. Figure 6. Effect of vanadium compounds on the glucose uptake in 3T3-L1 differentiated adipocytes at 16, 30 μM for vanadium compounds and 2 mM for Metf in the absence of insulin. All the values are the mean ± SD. * p < 0.05 and ** p < 0.01 vs. untreated control cells. Control Nadeca Metfdeca MetV Metf Insulin 0 20 40 60 80 100 Intensity (%) 16 µM 30 µM 100 nM * * ** 2 mM Figure 6. 2.5. Glucose Uptake Assay To establish whether MetV, MetfDeca, NaDeca and Metf compounds stimulate glucose uptake on adipocytes, the eﬀect on the 2-NDBG cell uptake in 3T3-L1 diﬀerentiated adipocytes was evaluated. The experiments were performed at 16, 30 µM concentration for vanadium compounds and 2 mM for Metf in the absence of insulin. Insulin (100 nM) was used as a positive control. As it can be seen in Figure 6, NaDeca (30 µM) stimulates the glucose cell uptake on 62%, MetfDeca on 52%, MetV on 37% and Metf (2 mM) on 33%, while control conditions stimulate around to 20%. At 16 µM, NaDeca stimulates 29% and is the only compound that shows a notable diﬀerence between both concentrations. Our results suggest that the uptake is moderate due to the low concentration of the compounds; it has been shown that elevated concentrations of decavanadate 100 µM [6] and vanadate 325 µM were required for stimulation of glucose uptake in rat adipocytes, the later associated with IR Tyr auto-phosphorylation [53]. The activation of the insulin receptor substrates (IRS) has been demonstrated to occur in a dose-dependent manner in cardiomyocytes for MetV [54] and in 3T3-L1 cells for VOSO4 [57] due to diﬀerent mechanisms of actions. Our results indicate that the PI3K pathway was activated due to the activation of IRS-1 by PTPB1 phosphorylation for MetfDeca, MetV and Metf by a combination of diﬀerent mechanisms that includes PTPB1 phosphorylation, and for NaDeca by a ROS production, where V1 and the metavanadate species are involved, the ROS production by a decavanadate compound and the activation of the semicarbazide-sensitive amine oxidase (SSAO)/vascular adhesion protein-1 (VAP-1) was reported by Ybarola [58] the compound hexakis(benzylammonium) decavanadate showed that can stimulate glucose uptake in rat adipocytes in a dose dependent manner EC50 150 µM, an in vitro assay they conﬁrmed that hexakis(benzylammonium) decavanadate is oxidized in the same extension by SSAO enzyme as benzylamine and vanadate, using 51V NMR the authors also found that for 10 mM of the compound in the presence of 2.5 mM of H2O2 at pH 7.4, the major products of the decavanadate decomposition were V1, V4 and [V(OH)2(OO)2(OH)2]2−. 2.5. Glucose Uptake Assay Effect of vanadium compounds on the glucose uptake in 3T3-L1 differentiated adipocytes at 16, 30 μM for vanadium compounds and 2 mM for Metf in the absence of insulin. All the values are the mean ± SD. * p < 0.05 and ** p < 0.01 vs. untreated control cells. Figure 6. Eﬀect of vanadium compounds on the glucose uptake in 3T3-L1 diﬀerentiated adipocytes at 16, 30 µM for vanadium compounds and 2 mM for Metf in the absence of insulin. All the values are the mean ± SD. * p < 0.05 and ** p < 0.01 vs. untreated control cells. Figure 6. Effect of vanadium compounds on the glucose uptake in 3T3-L1 differentiated adipocytes at 16, 30 μM for vanadium compounds and 2 mM for Metf in the absence of insulin. All the values are the mean ± SD. * p < 0.05 and ** p < 0.01 vs. untreated control cells. 2.6. Insulin Release Assa 2.6. Insulin Release Assay 2.6. Insulin Release Assay The latter was used due to the pharmacological activity, such as blocking ATP-dependent K+ channels and stimulate the release of insulin from the pancreas in a dose-dependent manner [59]. Figure 7 shows that MetV and MetfDeca enhanced insulin release by 0.7 and 1-fold relative to glucose control. In addition, both vanadium compounds showed more activity than the repaglinide at the same concentration, while NaDeca shows lower activity at 30 μM concentration than the glucose and repaglinide controls. βTC-6 cells secrete insulin in response to glucose; however, this cell line derived from transgenic mice develop a high hexokinase activity [60]; in normal pancreatic β-cells isolated from mouse islets, the effects of NaVO3 were studied at 0.1–1 mM concentration [61], the authors found that vanadate did not affect basal insulin release, although, vanadate potentiated the glucose effect by a different mechanism than blocking the sodium pump or affecting the AmpC levels [61]. Control Glucose Repaglinide Nadeca Metfdeca MetV 0 20 40 Insulin ng/mL 10 mM 30 µM * * * * Control Glucose Repaglinide Nadeca Metfdeca MetV 0 20 40 60 Insulin ng/mL 10 mM 30 µM * * * * Figure 7. Eﬀect of the vanadium compounds (30 µM concentration) on the insulin release in βTC-6 cells with 10 mM of glucose and 30 µM of repaglinide. All the values are the mean ± SD. * p < 0.05 vs. untreated control cells. Figure 7. Eﬀect of the vanadium compounds (30 µM concentration) on the insulin release in βTC-6 cells with 10 mM of glucose and 30 µM of repaglinide. All the values are the mean ± SD. * p < 0.05 vs. untreated control cells. Figure 7. Eﬀect of the vanadium compounds (30 µM concentration) on the insulin release in βTC-6 cells with 10 mM of glucose and 30 µM of repaglinide. All the values are the mean ± SD. * p < 0.05 vs. untreated control cells. 13 of 20 Inorganics 2020, 8, 67 Figure 7. Eff In 1999, Proks and coworkers performed an experiment in diﬀerent types of cloned KATP channels expressed in Xenopus oocytes [62]. 2.6. Insulin Release Assa 2.6. Insulin Release Assay 2.6. Insulin Release Assay Their results showed that sodium decavanadate in 2 mM concentration made by a solution of Na3VO4 at pH 7.2 modulated KATP channel activity via the SUR subunit, the Hill coeﬃcients for both activation and inhibition of KATP currents suggested that the cooperativity action of more than one vanadate molecule was involved in these eﬀects. They also found that the eﬀects were abolished by boiling the solution where the vanadate polymers where virtually absent [62]. Our results indicate that vanadium species promote more than one insulin release mechanism in βTC-6 cells (Figure 8), MetV, NaDeca, and MetfDeca decomposition in V1 augment insulin secretion by tyrosine phosphorylation of IRS-1 and IRS-2 [63,64], while in the second mechanism, vanadium oligomers can be active blocking ATP-dependent K+ channels, however, we propose that the active species in the decavanadate solutions are the vanadium dimers V2, although, V2 formation rate is slower (Table 1), it can be present in considerable amounts blocking ATP-dependent K+ channels [62,65]. In the case of V4, the higher oligomer formation is promoted by the decomposition reaction of the NaDeca compound (Table 1), and the tetramer V4 has not followed any of the two mechanisms (Figure 8). It has been shown that vanadium compounds like VOSO4 and NaVO3 (1.6–100 µM) stimulated ROS production in isolated rat liver mitochondria [66]. In 2013 Hosseini and coworkers showed that V5+ (25–200 µM) interaction with respiratory complex III is the major source of V5+ induced ROS in rat liver mitochondria [67]. Interestingly, the concentration of ROS formation highly increases with 200 µM of sodium metavanadate in 60 min while with just 50 µM, it is not the case [67]. We hypothesized that NaDeca at 30 µM concentration product V4 has some interaction with the cell mitochondria like its membrane depolarization [68] through ROS production that inhibits the insulin release by NaDeca compound. untreated control cells. In 1999, Proks and coworkers performed an experiment in different types of cloned KATP channels expressed in Xenopus oocytes [62]. Their results showed that sodium decavanadate in 2 mM concentration made by a solution of Na3VO4 at pH 7.2 modulated KATP channel activity via the SUR subunit, the Hill coefficients for both activation and inhibition of KATP currents suggested that the cooperativity action of more than one vanadate molecule was involved in these effects. 2.6. Insulin Release Assa 2.6. Insulin Release Assay 2.6. Insulin Release Assay They also found that the effects were abolished by boiling the solution where the vanadate polymers where virtually absent [62]. Our results indicate that vanadium species promote more than one insulin release mechanism in βTC-6 cells (Figure 8), MetV, NaDeca, and MetfDeca decomposition in V1 augment insulin secretion by tyrosine phosphorylation of IRS-1 and IRS-2 [63,64], while in the second mechanism, vanadium oligomers can be active blocking ATP-dependent K+ channels, however, we propose that the active species in the decavanadate solutions are the vanadium dimers V2, although, V2 formation rate is slower (Table 1), it can be present in considerable amounts blocking ATP- dependent K+ channels [62,65]. In the case of V4, the higher oligomer formation is promoted by the decomposition reaction of the NaDeca compound (Table 1), and the tetramer V4 has not followed any of the two mechanisms (Figure 8). It has been shown that vanadium compounds like VOSO4 and NaVO3 (1.6–100 μM) stimulated ROS production in isolated rat liver mitochondria [66]. In 2013 Hosseini and coworkers showed that V5+ (25–200 μM) interaction with respiratory complex III is the major source of V5+ induced ROS in rat liver mitochondria [67]. Interestingly, the concentration of ROS formation highly increases with 200 μM of sodium metavanadate in 60 min while with just 50 μM, it is not the case [67]. We hypothesized that NaDeca at 30 μM concentration product V4 has some interaction with the cell mitochondria like its membrane depolarization [68] through ROS production that inhibits the insulin release by NaDeca compound. V O O V V O O- O V O O O O- O -O -O V OH -O OH O Glucose KATP-channel K+ Glycolysis Pyruvate Membrane depolarization Ca2+ Ca2+ TCA cycle Respiratory chain ATP/ADP Other Metabolites Insuline granule exocytosis V10 V4 V1 Promotes insulin release DMEM medium, pH 7.4 Decomposition reaction Gluts PI3-K IR IRS-2 V V OH -O O OH O- O O V2 ROS Figure 8. Illustration of the vanadium species mechanisms of action on the insulin release in βTC-6 cells. The total vanadium concentration is 300 μM, NaDeca decomposition reaction is moderately faster than MetfDeca reaction (Table 1), so NaDeca and MetfDeca majority decomposition products are V1 and V4; however, some V2 is present, the putative mechanism of action for V1 is that enhanced tyrosine phosphorylation, and on that way, V1 species is able to further augment insulin secretion. 2.6. Insulin Release Assa 2.6. Insulin Release Assay 2.6. Insulin Release Assay A second putative mechanism involves inhibition of the KATP channel by V2 species. V O O V V O O- O V O O O O- O -O -O V OH -O OH O Glucose KATP-channel K+ Glycolysis Pyruvate Membrane depolarization Ca2+ Ca2+ TCA cycle Respiratory chain ATP/ADP Other Metabolites Insuline granule exocytosis V10 V4 V1 Promotes insulin release DMEM medium, pH 7.4 Decomposition reaction Gluts PI3-K IR IRS-2 V V OH -O O OH O- O O V2 ROS Figure 8. Illustration of the vanadium species mechanisms of action on the insulin release in βTC-6 cells. The total vanadium concentration is 300 µM, NaDeca decomposition reaction is moderately faster than MetfDeca reaction (Table 1), so NaDeca and MetfDeca majority decomposition products are V1 and V4; however, some V2 is present, the putative mechanism of action for V1 is that enhanced tyrosine phosphorylation, and on that way, V1 species is able to further augment insulin secretion. A second putative mechanism involves inhibition of the KATP channel by V2 species. Membrane depolarization Glucose Other Metabolites Figure 8. Illustration of the vanadium species mechanisms of action on the insulin release in βTC-6 cells. The total vanadium concentration is 300 μM, NaDeca decomposition reaction is moderately faster than MetfDeca reaction (Table 1), so NaDeca and MetfDeca majority decomposition products are V1 and V4; however, some V2 is present, the putative mechanism of action for V1 is that enhanced tyrosine phosphorylation, and on that way, V1 species is able to further augment insulin secretion. A second putative mechanism involves inhibition of the KATP channel by V2 species. Figure 8. Illustration of the vanadium species mechanisms of action on the insulin release in βTC-6 cells. The total vanadium concentration is 300 µM, NaDeca decomposition reaction is moderately faster than MetfDeca reaction (Table 1), so NaDeca and MetfDeca majority decomposition products are V1 and V4; however, some V2 is present, the putative mechanism of action for V1 is that enhanced tyrosine phosphorylation, and on that way, V1 species is able to further augment insulin secretion. A second putative mechanism involves inhibition of the KATP channel by V2 species. 3.1. Chemicals and Reagents Ammonium metavanadate (NH4VO3), sodium metavanadate (NaVO3), hydrochloric acid (HCl 37% w/v in H2O), dimethyl sulfoxide (DMSO), d6-DMSO, deuterium oxide (D2O), 3-(4,5-dimethylthiazol-2-yl) 14 of 20 Inorganics 2020, 8, 67 -2,5-diphenyltetrazolium Bromide (MTT) 98%, 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), potassium chloride (KCl), sodium chloride (NaCl), ethylenediaminetetraacetic acid disodium salt (EDTA), ethylene glycol-bis(2-aminoethylether)-N,N,N0,N0-tetraacetic acid (EGTA), β-glycerol-phosphate, triton X-100, NaF, sodium pyrophosphate dibasic, sodium orthovanadate (Na3VO4) and 1,4-dithiothreitol (DDT) were purchased from Sigma-Aldrich (St Louis, MO, USA). Phenylmethylsulfonyl fluoride (PMSF) from Calbiochem (San Diego, CA, USA). COMPLETE (protease inhibitor cocktail) from ROCHE (Mannheim, Germany). Dulbecco’s modified Eagle’s medium (DMEM) high glucose, fetal bovine serum (FBS) and penicillin/streptomycin from Gibco (Gaithersburg, MD, USA). All the cell lines used were purchased from ATCC (HEP-G2 HB-8065, 3T3-L1 CL-173, Beta-TC-6 RL 11506) (Manassas, VA, USA). Metformin hydrochloride (C4H11N5·HCl) was isolated directly from commercial brand tablets. The metforminium decavanadate (MetfDeca) (C4H13N5)3V10O28·8H2O was prepared according to the literature [33]. The sodium decavanadate (NaDeca) Na6V10O28·18H2O was prepared by suspending NaVO3 (0.12 g, 1 mmol) in distilled water (30 mL). After the suspension was stirred at room temperature for 1 h, the pH was adjusted to 4 by the addition of HCl (1 M). The resulting orange solution was ﬁltered, and the ﬁltrate was allowed to evaporate at 4 ◦C. Orange crystals were obtained after one week, according to a previously reported procedure [36]. The concentrations of the stock solutions for the biological studies in water for metformin hydrochloride (Metf), sodium metavanadate (MetV), metforminium decavanadate (MetfDeca) were 30 mM, whereas for sodium decavanadate (NaDeca) was 15 mM. The metforminium decavanadate crystals are water-insoluble, so it was solubilized in 10% DMSO before the addition of water. For the 51V NMR studies, 10% DMSO-d6 was used. 3.2. Kinetic Studies The kinetics of the decomposition reaction of sodium and metforminium decavanadates in DMEM medium at 25 ◦C was determined by 51V NMR at 0.5 and 1.0 mM of decavanadate concentration. The spectra were acquired using 0.5 mL as a ﬁnal volume with 10% DMSO-d6 in a Bruker Ascend 600 MHz spectrometer. 51V spectra were recorded using parameters reported previously [12,69] at 157.85 MHz. The chemical shifts were obtained using an external reference using 100 mM Na3VO4 solution in 1.0 M NaOH ([VO4]3−signal at −541 ppm) [70]. The concentrations of each vanadate species Vx were calculated from the fractions of the total integrated areas using the following equation: [Vx] = (Ax/At) × ([Vt]/n), where Ax corresponds to the area measured for the x vanadate species with n as the oligomer number (number of vanadium atoms), At is the sum of the measured areas and [Vt] is the total vanadate concentration [71]. In the case of the decameric species, three signals at −420, −494 and −510 ppm were integrated for 2, 4 and 4 vanadium atoms, respectively [72]. The rate constants were calculated by the initial rates method, where the species concentration Vx was plotted over time (100 min), the 51V NMR spectra were acquired every 20 min, and the reaction was started when the decavanadate compound aliquot was added to the DMEM medium. 3.3. Cell Viability Assay Cell viability of the three vanadium compounds NaDeca, MetfDeca, MetV and metformin hydrochloride against HEPG2 and HEK293 was tested using MTT assay (Sigma-Aldrich, St Louis, MO, USA). The cells were placed in a 96-well micro-assay culture plate (ULTRACRUZ, Santa Cruz Dallas, TX, USA) at a density of 1 × 105 cells per well in 0.2 mL of DMEM-high glucose culture medium supplemented with fetal bovine serum FBS (10%) and penicillin/streptomycin (1%), and grown at 37 ◦C in a humidiﬁed 5% CO2 incubator for 24 h. After this, the cells were treated with 0.002 mL of each compound per well by triplicate; sequential dilutions 1:2 were made for each compound, DMSO was used as a blank. The cells were incubated for 24 h. The surviving cells were determined. We added 0.01 mL of MTT (5 mg/mL in phosphate-buﬀered saline) to each well, and the cells were incubated for 3 h at 37 ◦C in a humidiﬁed 5% CO2 incubator. After this time, the medium was removed from the cells, and 0.1 mL of DMSO was added to each well, and the cells were incubated for 1 h. The cells 15 of 20 Inorganics 2020, 8, 67 viability was determined by measure their ability to reduce MTT (yellow) to formazan product (violet). The absorbance was quantiﬁed at 600 nm by a Modulus microplate Luminometer spectrophotometer (Turner BioSystems, Sunnyvale, CA, USA). viability was determined by measure their ability to reduce MTT (yellow) to formazan product (violet). The absorbance was quantiﬁed at 600 nm by a Modulus microplate Luminometer spectrophotometer (Turner BioSystems, Sunnyvale, CA, USA). 3.6. Glucose Uptake Assay 3T3-L1 diﬀerentiated adipocytes cells were seeded in a 96-well plate (ULTRACRUZ, Santa Cruz) 1 × 105 cells per well. The next day media was changed to starving media (DMEM without supplementation, no glucose), compounds were added at 16 and 30 µM ﬁnal concentration and incubated 20 h at standard conditions. Cells were incubated with or without 100 nM insulin for 1 h. After this, 300 µM of 2-NBDG (Invitrogen by Thermo Fisher Scientiﬁc) were added to each well and incubated 20 min at 37 ◦C and 5% CO2, cells were washed once with PBS, and 100 µL/well of fresh PBS were added. Fluorescence was read at 485/535 nm (Modulus Microplate Luminometer). 3.5. Adipocyte Diﬀerentiation Preadipocytes 3T3-L1 were obtained from ATCC and diﬀerentiated, as previously described [73]. Brieﬂy, cells were grown to conﬂuency in a 75 cm ﬂask (CORNING) with DMEM medium supplemented with 10% calf serum (Biowest, Riverside, MO, USA) and standard temperature and CO2 conditions (37 ◦C and 5% CO2). Two days after reaching conﬂuency, media was replaced to induce diﬀerentiation (DMEM supplemented with 10% fetal bovine serum (FBS), 1.0 µg/mL human insulin, 0.5 mM 3-isobutyl-1-methylxanthine and 1 µM dexamethasone). After 48 h, media was changed with DMEM supplemented with 10% fetal bovine serum and 1.0 µg/mL human insulin and cells were incubated for 48 h. Finally, the media was replaced with DMEM supplemented with 10% FBS for 4 days, media was refreshed every 2 days. 3.4. Western Blot Analysis The cells were placed in 6-well micro-assay culture plates at a density of 5 × 105 cells per well in 3 mL of DMEM-high glucose culture medium supplemented with fetal bovine serum FBS (10%) and penicillin/streptomycin (1%); the cells were treated with 8, 16 and 30 µM of each compound and the cells were grown at 37 ◦C in a humidiﬁed 5% CO2 incubator for 24 h. Cultured cells were washed with 1 mL of cold phosphate buﬀer solution (PBS). For AMPK assays cells were lysed by 0.25 mL of ice-cold HEPES lysis buﬀer: HEPES (50 mM, pH 7.4), EDTA (1 mM), EGTA (1 mM), KCl (50 mM), glycerol (5 mM), Triton X100 (0.1% w/v), NaF (50 mM), NaPPi (5 mM), Na3VO4 (1 mM), DDT (1 mM), PMSF (0.2 mM) and COMPLETE 1X as protease inhibitor. Homogenates were centrifuged at 16,128× g for 20 min at 4 ◦C in an Eppendorf centrifuge 5804R. Supernatants were collected for their protein quantitation by Lowry method; 50 µg of protein were separated by 10% SDS-page and transferred to PVDF for blotting using the following antibodies (cell signaling 1:1000) anti-pAKT (Ser473), anti-p-AMPKα (Thr172), anti-AMPKα, anti-AKT (PKBα) and anti-β-actin at 4 ◦C overnight. Blots were visualized with HRP-conjugated goat anti-rabbit IgG or HRP-conjugated goat anti-mouse IgG at room temperature for one hour. Actin was used as loading controls for the total protein content. Proteins were visualized and quantiﬁed in a Bio-Rad ChemiDoc XRS (Bio-Rad, Hercules, CA, USA). with the Quantity One software (Version 4.5, Bio-Rad, Hercules, CA, USA). 4. Conclusions performed the biological studies; I.E.L. and O.H.-A. reviewed and edited the manuscript; I.S.-L. wrote most of the manuscript and carried out the kinetics analysis. All authors have read and agreed to the published version of the manuscript. Funding: This research received no external funding. A.M.S.-N. and I.S.-L. would like to thank the UJAT for funding A.M.S.-N. fellowship. Acknowledgments: We specially thanks to Boris Rodenak Kladniew to participate in the biological da Acknowledgments: We specially thanks to Boris Rodenak Kladniew to participate in the biological data analysis. Acknowledgments: We specially thanks to Boris Rodenak Kladniew to participate in the biological data analysis. Conﬂicts of Interest: The authors declare no conﬂict of interest Conﬂicts of Interest: The authors declare no conﬂict of interest. Conﬂicts of Interest: The authors declare no conﬂict of interest. Conﬂicts of Interest: The authors declare no conﬂict of interest. s of Interest: The authors declare no conﬂict of interest. 3.7. Insulin Release Assay Studies were performed with βTC-6 cells. The cells were placed in 24-well micro-assay culture plates at a density of 2.5 × 105 cells per well in DMEM culture medium; the cells were incubated overnight at 37 ◦C in a humidiﬁed 5% CO2 incubator. The next day medium was changed, and 10 mM glucose and 30 µM repaglinide were added as controls. Compounds were added at 30 µM ﬁnal concentration and incubated for one hour at standard conditions. The insulin quantiﬁcation was made with a mouse insulin ELISA kit (ALPCO, INSMS-E01, ALPCO, Salem NH, USA). 16 of 20 Inorganics 2020, 8, 67 16 of 20 3.8. Statistical Analysis Data were presented as mean ± SEM of three independent experiments. Statistical signiﬁcance of data was analyzed by Student’s t-test and one-way analysis of variance (ANOVA). A probability of the value of p < 0.05 was considered as statistically signiﬁcant. Calculations and ﬁgures were made using Grad Pad Prism version 8 (GraphPad Software, San Diego, CA, USA). 4. Conclusions Vanadium solution chemistry represents a challenge due to its complexity. However, new therapeutic approaches can be explored with decavanadate compounds in biological reaction media, vanadium therapeutic potential in different diseases like DM2, cancer, metabolic syndrome and cardiovascular diseases should be addressed. Decavanadate decomposition products like V2 and V4 action mechanisms in cytotoxic activity, AMPK and AKT expression still have open questions; however, V1 is well known as a glucose uptake promoter and insulin release agent. Nevertheless, the combination of orthovanadate and methavanadate species can increase the desirable therapeutic effects of vanadium, as shown in this work. Our results show that at least two mechanisms are promoted AKT activation by NaDeca, and MetfDeca hydrolysis products in HEPG2 cells, the ﬁrst one with the orthovanadate (V1) species involved in PTP-1B mediated AKT activation, while the second mechanism involves the activation of the AKT pathway in a PI3K-dependent manner by ROS, in this regard, we hypothesized that V4 could be involved in a vanadium reduction process that promotes the ROS exacerbation in HEPG2 cells in DMEM medium and that ROS production results in a decrement of the cell viability in normal (HEK293) and carcinogenic cells (HEPG2). In this sense, our results indicate that a combination of at least two mechanisms is associated with the glucose uptake in 3T3L-1 diﬀerentiated adipocytes that includes PTP-1B phosphorylation and ROS production in the case of NaDeca. MetfDeca and MetV at 30 µM concentration enhanced insulin release in βTC-6 cells; surprisingly, the NaDeca compound is almost inactive in the assay. Our results suggest that MetfDeca decomposition products (V1 and V2) promote more than one insulin release mechanism in the DMEM medium. The ﬁrst proposed mechanism is that V1 augment insulin secretion by tyrosine phosphorylation of the IRS, and in a second putative mechanism, vanadium oligomers like V2 can be active, blocking ATP-dependent K+ channels. However, V4 species that are produced by the decomposition reaction of NaDeca and MetfDeca are not following either mechanism. The data presented in this paper demonstrate that decavanadate decomposition products are able to promote diﬀerent biological mechanisms of action, than the ones promoted by orthovandate (MetV) and metformin hydrochloride (Metf). Thus, more chemical and biological experiments are necessary to establish the active species and their composition with the aim to explore new therapies in the treatment of some metabolic diseases. Author Contributions: A.M.S.-N. and R.O.S.-D. performed the kinetics experiments; A.M.S.-N. and L.C. p y g p y g y g design to artificial enzymes. Curr. Opin. Biotechnol. 2019, 58, 92–99. [CrossRef] 2. ˇColovi´c, M.B.; Lackovi´c, M.; Lalatovi´c, J.; Mougharbel, A.S.; Kortz, U.; Krsti´c, D.Z. Polyoxometalates in Biomedicine: Update and Overview. Curr. Med. Chem. 2020, 27, 362–379. [CrossRef] References 1. Van Rompuy, L.S.; Parac-Vogt, T.N. Interactions between polyoxometalates and biological systems: From drug design to artificial enzymes. Curr. Opin. Biotechnol. 2019, 58, 92–99. [CrossRef] 2. ˇColovi´c, M.B.; Lackovi´c, M.; Lalatovi´c, J.; Mougharbel, A.S.; Kortz, U.; Krsti´c, D.Z. Polyoxometalates in Biomedicine: Update and Overview. Curr. Med. Chem. 2020, 27, 362–379. [CrossRef] 17 of 20 Inorganics 2020, 8, 67 17 of 20 3. Aureliano, M. Decavanadate contribution to vanadium biochemistry: In vitro and in vivo studies. Inorg. Chim. Acta 2014, 420, 4–7. [CrossRef] 4. Bijelic, A.; Rompel, A. The use of polyoxometalates in protein crystallography—An attempt to widen a well-known bottleneck. Coord. Chem. Rev. 2015, 299, 22–38. [CrossRef] [PubMed] 5. Crans, D.C.; Sánchez-Lombardo, I.; McLauchlan, C.C. Exploring Wells-Dawson Clusters Associated with the Small Ribosomal Subunit. Front. Chem. 2019, 7, 462–476. [CrossRef] [PubMed] 6. Aureliano, M.; Crans, D.C. Decavanadate (V10O286−) and oxovanadates: Oxometalates with many biological activities. J. Inorg. Biochem. 2009, 103, 536–546. [CrossRef] [PubMed] 7. Nakamura, S.; Ozeki, T. Hydrogen-bonded aggregates of protonated decavanadate anions in their tetraalkylammonium salts. J. Chem. Soc. Dalton Trans. 2001, 472–480. [CrossRef] 8. Ferreira da Silva, J.L.; Fátima Minas da Piedade, M.; Teresa Duarte, M. Decavanadates: A building-block for supramolecular assemblies. Inorg. Chim. Acta 2003, 356, 222–242. [CrossRef] 9. Bošnjakovi´c-Pavlovi´c, N.; Prévost, J.; Spasojevi´c-de Biré, A. Crystallographic Statistical Study of Decavanadate Anion Based-Structures: Toward a Prediction of Noncovalent Interactions. Cryst. Growth Des. 2011, 11, 3778–3789. [CrossRef] 10. Povar, I.; Spinu, O.; Zinicovscaia, I.; Pintilie, B.; Ubaldini, S. Revised Poubaix diagrams for the vanadium–water system. J. Electrochem. Sci. Eng. 2019, 9, 620. [CrossRef] 11. Kojima, T.; Antonio, M.; Ozeki, T. Solvent-Driven Association and Dissociation of the Hydrogen-Bonded Protonated Decavanadates. J. Am. Chem. Soc. 2011, 133, 7248–7251. [CrossRef] [PubMed] 12. Ramos, S.; Manuel, M.; Tiago, T.; Duarte, R.; Martins, J.; Gutiérrez-Merino, C.; Moura, J.J.G.; Aureliano, M. Decavanadate interactions with actin: Inhibition of G-actin polymerization and stabilization of decameric vanadate. J. Inorg. Biochem. 2006, 100, 1734–1743. [CrossRef] [PubMed] 13. Samart, N.; Arhouma, Z.; Kumar, S.; Murakami, H.A.; Crick, D.C.; Crans, D.C. Decavanadate Inhibits Mycobacterial Growth More Potently Than Other Oxovanadates. Front. Chem. 2018, 6. [CrossRef] [PubMed] 14. Krivosudský, L.; Roller, A.; Rompel, A. Tuning the interactions of decavanadate with thaumatin, lysozyme, proteinase K and human serum proteins by its coordination to a pentaaquacobalt(II) complex cation. New J. Chem. 2019, 43, 17863–17871. [CrossRef] 15. Soares, S.S.; Martins, H.; Duarte, R.O.; Moura, J.J.G.; Coucelo, J.; Gutiérrez-Merino, C.; Aureliano, M. References Jakusch, T.; Kiss, T. In vitro study of the antidiabetic behavior of vanadium compounds. Coord. Chem. Rev. 2017, 351, 118–126. [CrossRef] 29. Thompson, K.H.; Orvig, C. Vanadium in diabetes: 100 years from Phase 0 to Phase I. J. Inorg. Biochem. 2006, 100, 1925–1935. [CrossRef] 30. Thompson, K.H.; Lichter, J.; LeBel, C.; Scaife, M.C.; McNeill, J.H.; Orvig, C. Vanadium treatment of type 2 diabetes: A view to the future. J. Inorg. Biochem. 2009, 103, 554–558. [CrossRef] 31. Rehder, D. Transport, Accumulation, and Physiological Eﬀects of Vanadium. In Detoxiﬁcation of Heavy Metals; Sherameti, I., Varma, A., Eds.; Springer: Berlin/Heidelberg, Germany, 2011; pp. 205–220. 32. Gruzewska, K.; Michno, A.; Pawelczyk, T.; Bielarczyk, H. Essentiality and toxicity of vanadium supplements in health and pathology. J. Physiol. Pharmacol. 2014, 65, 603–611. 33. Sánchez Lombardo, I.; Sánchez Lara, E.; Pérez Benítez, A.; Mendoza, Á.; Bernès, S.; González Vergara, E. Synthesis of Metforminium(2+) Decavanadates-Crystal Structures and Solid-State Characterization. Eur. J. Inorg. Chem. 2014, 2014, 4581–4588. [CrossRef] 34. Treviño, S.; González-Vergara, E. Metformin-decavanadate treatment ameliorates hyperglycemia and redox balance of the liver and muscle in a rat model of alloxan-induced diabetes. New J. Chem. 2019, 43, 17850–17862. [CrossRef] 35. Treviño, S.; Sánchez-Lara, E.; Sarmiento-Ortega, V.E.; Sánchez-Lombardo, I.; Flores-Hernández, J.Á.; Pérez-Benítez, A.; Brambila-Colombres, E.; González-Vergara, E. Hypoglycemic, lipid-lowering and metabolic regulation activities of metforminium decavanadate (H2Metf)3 [V10O28]·8H2O using hypercaloric-induced carbohydrate and lipid deregulation in Wistar rats as biological model. J. Inorg. Biochem. 2015, 147, 85–92. [CrossRef] [PubMed] 36. Durif, A.; Averbuch-Pouchot, M.T.; Guitel, J.C. Structure d’un Decavanadate d’Hexasodium Hydrate. Acta Crystallogr. Sect. B Struct. Sci. 1980, 36, 680–682. [CrossRef] 37. Howarth, O.W.; Jarrold, M. Protonation of the decavanadate(6-) ion: A vanadium-51 nuclear magnetic resonance study. J. Chem. Soc. Dalton Trans. 1978, 5, 503–506. [CrossRef] 38. Goddard, J.B.; Gonas, A.M. Kinetics of the dissociation of decavanadate ion in basic solutions. Inorg. Chem. 1973, 12, 574–579. [CrossRef] 39. Schwarzenbach, G.; Geier, G. 97. Die Raschacidiﬁerung und -alkalisierung von Vanadaten. Helv. Chim. Acta 1963, 46, 906–926. [CrossRef] 40. Pascale, F.; Bedouet, L.; Baylatry, M.; Namur, J.; Laurent, A. Comparative Chemosensitivity of VX2 and HCC Cell Lines to Drugs Used in TACE. Anticancer Res. 2015, 35, 6497–6503. 41. Mohamadi, M.; Yousef Ebrahimipour, S.; Torkzadeh-Mahani, M.; Foro, S.; Akbari, A. A mononuclear diketone-based oxido-vanadium(IV) complex: Structure, DNA and BSA binding, molecular docking and anticancer activities against MCF-7, HPG-2, and HT-29 cell lines. RSC Adv. 2015, 5, 101063–101075. [CrossRef] 42. References Vanadium distribution, lipid peroxidation and oxidative stress markers upon decavanadate in vivo administration. J. Inorg. Biochem. 2007, 101, 80–88. [CrossRef] [PubMed] 16. Clare, B.W.; Kepert, D.L.; Watts, D.W. Acid decomposition of decavanadate: Speciﬁc salt eﬀects. J. Chem. Soc. Dalton Trans. 1973, 2481–2487. [CrossRef] 17. Druskovich, D.M.; Kepert, D.L. Base decomposition of decavanadate. J. Chem. Soc. Dalton Trans. 1975, 947–951. [CrossRef] 18. Crans, D.; Rithner, C.; Theisen, L. Application of time-resolved vanadium-51 2D NMR for quantitation of kinetic exchange pathways between vanadate monomer, dimer, tetramer, and pentamer. J. Am. Chem. Soc. 1990, 112, 2901–2908. [CrossRef] 19. Levina, A.; McLeod, A.I.; Pulte, A.; Aitken, J.B.; Lay, P.A. Biotransformations of Antidiabetic Vanadium Prodrugs in Mammalian Cells and Cell Culture Media: A XANES Spectroscopic Study. Inorg. Chem. 2015, 54, 6707–6718. [CrossRef] 0. Saklayen, M.G. The Global Epidemic of the Metabolic Syndrome. Curr. Hypertens. Rep. 2018, 20, 12. [Cross 20. Saklayen, M.G. The Global Epidemic of the Metabolic Syndrome. Curr. Hypertens. Rep. 2018, 20, 12. [CrossRef] 21. Giovannucci, E.; Harlan, D.M.; Archer, M.C.; Bergenstal, R.M.; Gapstur, S.M.; Habel, L.A.; Pollak, M.; Regensteiner, J.G.; Yee, D. Diabetes and Cancer: A consensus report. Diabetes Care 2010, 33, 1674–1685. [CrossRef] [PubMed] 22. Tschöp, M.H. Metabolic Disease Therapies. Cell Metab. 2017, 26, 579–583. 22. Tschöp, M.H. Metabolic Disease Therapies. Cell Metab. 2017, 26, 579–583. 23. Lee, C.-W.; Wong, L.L.-Y.; Tse, E.Y.-T.; Liu, H.-F.; Leong, V.Y.-L.; Lee, J.M.-F.; Hardie, D.G.; Ng, I.O.-L.; Ching, Y.-P. AMPK Promotes p53 Acetylation via Phosphorylation and Inactivation of SIRT1 in Liver Cancer Cells. Cancer Res. 2012, 72, 4394. [CrossRef] [PubMed] 24. Koundouros, N.; Poulogiannis, G. Phosphoinositide 3-Kinase/Akt Signaling and Redox Metabolism in Cancer. Front. Oncol. 2018, 8, 160–169. [CrossRef] 25. Zheng, W.; Feng, Q.; Liu, J.; Guo, Y.; Gao, L.; Li, R.; Xu, M.; Yan, G.; Yin, Z.; Zhang, S.; et al. Inhibition of 6-phosphogluconate Dehydrogenase Reverses Cisplatin Resistance in Ovarian and Lung Cancer. Front. Pharmacol. 2017, 8, 421–432. [CrossRef] 18 of 20 18 of 20 Inorganics 2020, 8, 67 26. Rehder, D. Implications of vanadium in technical applications and pharmaceutical issues. Inorg. Chim. Acta 2017, 455, 378–389. [CrossRef] 27. Crans, D.C.; Henry, L.; Cardiﬀ, G.; Posner, B.I. Developing Vanadium as an Antidiabetic or Anticancer Drug: A Clinical and Historical Perspective. In Essential Metals in Medicine: Therapeutic Use and Toxicity of Metal Ions in the Clinic, 1st ed.; Carver, P.L., Ed.; De Gruyter: Berlin, Germany; Boston, MA, USA, 2019; pp. 203–230. 28. 45. Fujita, H.; Hirose, K.; Sato, M.; Fujioka, I.; Fujita, T.; Aoki, M.; Takai, Y. Metformin attenuates hypoxia-induced resistance to cisplatin in the HepG2 cell line. Oncol. Lett. 2019, 17, 2431–2440. [CrossRef] [PubMed] References Molecular Pathways: Is AMPK a Friend or a Foe in Cancer? Clin. Cancer Res. 2015, 21, 3836–3840. [CrossRef] 53. Lu, B.; Ennis, D.; Lai, R.; Bogdanovic, E.; Nikolov, R.; Salamon, L.; Fantus, C.; Le-Tien, H.; Fantus, I.G. Enhanced sensitivity of insulin-resistant adipocytes to vanadate is associated with oxidative stress and decreased reduction of vanadate (+5) to vanadyl (+4). J. Biol. Chem. 2001, 276, 35589–35598. [CrossRef] 54. Tardif, A.; Julien, N.; Chiasson, J.-L.; Coderre, L. Stimulation of glucose uptake by chronic vanadate pretreatment in cardiomyocytes requires PI 3-kinase and p38 MAPK activation. Am. J. Physiol. Endocrinol. Metab. 2003, 284, E1055–E1064. [CrossRef] [PubMed] 55. Žiži´c, M.; Miladinovi´c, Z.; Stani´c, M.; Hadžibrahimovi´c, M.; Živi´c, M.; Zakrzewska, J. 51V NMR investigation of cell-associated vanadate species in Phycomyces blakesleeanus mycelium. Res. Microbiol. 2016, 167, 521–528. [CrossRef] [PubMed] 56. Zhao, Q.; Chen, D.; Liu, P.; Wei, T.; Zhang, F.; Ding, W. Oxidovanadium(IV) sulfate-induced glucose uptake in HepG2 cells through IR/Akt pathway and hydroxyl radicals. J. Inorg. Biochem. 2015, 149, 39–44. [CrossRef] 56. Zhao, Q.; Chen, D.; Liu, P.; Wei, T.; Zhang, F.; Ding, W. Oxidovanadium(IV) sulfate-induced glucose uptake in HepG2 cells through IR/Akt pathway and hydroxyl radicals. J. Inorg. Biochem. 2015, 149, 39–44. [CrossRef] 57. Seale, A.P.; de Jesus, L.A.; Park, M.C.; Kim, Y.S. Vanadium and insulin increase adiponectin production in 3T3 L1 adipocytes Pharmacol Res 2006 54 30 38 [CrossRef] [PubMed] 57. Seale, A.P.; de Jesus, L.A.; Park, M.C.; Kim, Y.S. Vanadium and insulin increase adiponectin production in 3T3-L1 adipocytes. Pharmacol. Res. 2006, 54, 30–38. [CrossRef] [PubMed] 58. Yraola, F.; García-Vicente, S.; Marti, L.; Albericio, F.; Zorzano, A.; Royo, M. Understanding the mechanism of action of the novel SSAO substrate (C7NH10)6(V10O28)·2H2O, a prodrug of peroxovanadate insulin mimetics. Chem. Biol. Drug. Des. 2007, 69, 423–428. [CrossRef] 59. Marín-Peñalver, J.J.; Martín-Timón, I.; Sevillano-Collantes, C.; Del Cañizo-Gómez, F.J. Update on the treatment of type 2 diabetes mellitus. World J. Diabetes 2016, 7, 354–395. [CrossRef] 60. Skelin, M.; Rupnik, M.; Cencic, A. Pancreatic beta cell lines and their applications in diabetes mellitus research. Altex 2010, 27, 105–113. [CrossRef] 61. Zhang, A.Q.; Gao, Z.Y.; Gilon, P.; Nenquin, M.; Drews, G.; Henquin, J.C. Vanadate stimulation of insulin release in normal mouse islets. J. Biol. Chem. 1991, 266, 21649–21656. 62. Proks, P.; Ashﬁeld, R.; Ashcroft, F.M. Interaction of vanadate with the cloned beta cell K(ATP) channel. J. Biol. Chem. 1999, 274, 25393–25397. [CrossRef] 63. References Dhahagani, K.; Mathan Kumar, S.; Chakkaravarthi, G.; Anitha, K.; Rajesh, J.; Ramu, A.; Rajagopal, G. Synthesis and spectral characterization of Schiﬀbase complexes of Cu(II), Co(II), Zn(II) and VO(IV) containing 4-(4-aminophenyl)morpholine derivatives: Antimicrobial evaluation and anticancer studies. Spectrochim. Acta A Mol. Biomol. Spectrosc. 2014, 117, 87–94. [CrossRef] 43. Correia, I.; Adão, P.; Roy, S.; Wahba, M.; Matos, C.; Maurya, M.R.; Marques, F.; Pavan, F.R.; Leite, C.Q.F.; Avecilla, F.; et al. Hydroxyquinoline derived vanadium(IV and V) and copper(II) complexes as potential anti-tuberculosis and anti-tumor agents. J. Inorg. Biochem. 2014, 141, 83–93. [CrossRef] [PubMed] 44. Amuthan, V.D.A.; Chandrashekara, S.S.; Venkata, R.; Richard, L. Cytoprotective Activity of Neichitti (Vernonia cinerea) in Human Embryonic Kidney (HEK293) Normal Cells and Human Cervix Epitheloid Carcinoma (HeLa) Cells against Cisplatin Induced Toxicity: A Comparative Study. J. Clin. Diagn. Res. 2019, 13, KC01–KC06. [CrossRef] 45. Fujita, H.; Hirose, K.; Sato, M.; Fujioka, I.; Fujita, T.; Aoki, M.; Takai, Y. Metformin attenuates hypoxia-induced resistance to cisplatin in the HepG2 cell line. Oncol. Lett. 2019, 17, 2431–2440. [CrossRef] [PubMed] 19 of 20 19 of 20 Inorganics 2020, 8, 67 46. Cheng, M.; Li, N.; Wang, N.; Hu, K.; Xiao, Z.; Wu, P.; Wei, Y. Synthesis, structure and antitumor studies of a novel decavanadate complex with a wavelike two-dimensional network. Polyhedron 2018, 155, 313–319. [CrossRef] 7. Li, Y.-T.; Zhu, C.-Y.; Wu, Z.-Y.; Jiang, M.; Yan, C.-W. Synthesis, crystal structures and anticancer activitie two decavanadate compounds. Transit. Met. Chem. 2010, 35, 597–603. [CrossRef] 48. Missina, J.M.; Gavinho, B.; Postal, K.; Santana, F.S.; Valdameri, G.; de Souza, E.M.; Hughes, D.L.; Ramirez, M.I.; Soares, J.F.; Nunes, G.G. Eﬀects of Decavanadate Salts with Organic and Inorganic Cations on Escherichia coli, Giardia intestinalis, and Vero Cells. Inorg. Chem. 2018, 57, 11930–11941. [CrossRef] 49. Wang, Q.; Liu, T.-T.; Fu, Y.; Wang, K.; Yang, X.-G. Vanadium compounds discriminate hepatoma and normal hepatic cells by diﬀerential regulation of reactive oxygen species. J. Biol. Inorg. Chem. 2010, 15, 1087–1097. [CrossRef] 50. Hao, J.; Huang, K.; Chen, C.; Liang, Y.; Wang, Y.; Zhang, X.; Huang, H. Polydatin Improves Glucose and Lipid Metabolisms in Insulin-Resistant HepG2 Cells through the AMPK Pathway. Biol. Pharm. Bull. 2018, 41, 891–898. [CrossRef] 51. Ferretti, A.C.; Hidalgo, F.; Tonucci, F.M.; Almada, E.; Pariani, A.; Larocca, M.C.; Favre, C. Metformin and glucose starvation decrease the migratory ability of hepatocellular carcinoma cells: Targeting AMPK activation to control migration. Sci. Rep. 2019, 9, 2815. [CrossRef] 52. Hardie, D.G. References Gogg, S.; Chen, J.; Efendic, S.; Smith, U.; Ostenson, C. Eﬀects of phosphotyrosine phosphatase inhibition on insulin secretion and intracellular signaling events in rat pancreatic islets. Biochem. Biophys. Res. Commun. 2001, 280, 1161–1168. [CrossRef] [PubMed] 64. Chen, J.; Ostenson, C.G. Inhibition of protein-tyrosine phosphatases stimulates insulin secretion in pancreatic islets of diabetic Goto-Kakizaki rats. Pancreas 2005, 30, 314–317. [CrossRef] [PubMed] 65. Liu, H.-K.; Green, B.D.; McClenaghan, N.H.; McCluskey, J.T.; Flatt, P.R. Long-Term Beneﬁcial Eﬀects of Vanadate, Tungstate, and Molybdate on Insulin Secretion and Function of Cultured Beta Cells. Pancreas 2004, 28, 264–268. [CrossRef] [PubMed] 20 of 20 20 of 20 Inorganics 2020, 8, 67 66. Zhao, Y.; Ye, L.; Liu, H.; Xia, Q.; Zhang, Y.; Yang, X.; Wang, K. Vanadium compounds induced mitochondria permeability transition pore (PTP) opening related to oxidative stress. J. Inorg. Biochem. 2010, 104, 371–378. [CrossRef] 67. Hosseini, M.J.; Shaki, F.; Ghazi-Khansari, M.; Pourahmad, J. Toxicity of vanadium on isolated r mitochondria: A new mechanistic approach. Metallomics 2013, 5, 152–166. [CrossRef] 68. Soares, S.S.; Gutiérrez-Merino, C.; Aureliano, M. Mitochondria as a target for decavanadate toxicity in Sparus aurata heart. Aquat. Toxicol. 2007, 83, 1–9. [CrossRef] 69. Crans, D.C.; Bunch, R.L.; Theisen, L.A. Interaction of trace levels of vanadium(IV) and vanadium(V) in biological systems. J. Am. Chem. Soc. 1989, 111, 7597–7607. [CrossRef] 70. Griﬃn, E.; Levina, A.; Lay, P.A. Vanadium(V) tris-3,5-di-tert-butylcatecholato complex: Links between speciation and anti-proliferative activity in human pancreatic cancer cells. J. Inorg. Biochem. 2019, 201, 110815. [CrossRef] 71. Aureliano, M.; Tiago, T.; Gândara, R.M.C.; Sousa, A.; Moderno, A.; Kaliva, M.; Salifoglou, A.; Duarte, R.O.; Moura, J.J.G. Interactions of vanadium(V)–citrate complexes with the sarcoplasmic reticulum calcium pump. J. Inorg. Biochem. 2005, 99, 2355–2361. [CrossRef] 72. Sánchez-Lombardo, I.; Baruah, B.; Alvarez, S.; Werst, K.R.; Segaline, N.A.; Levinger, N.E.; Crans, D.C. Size and shape trump charge in interactions of oxovanadates with self-assembled interfaces: Application of continuous shape measure analysis to the decavanadate anion. New J. Chem. 2016, 40, 962–975. [CrossRef] and shape trump charge in interactions of oxovanadates with self-assembled interfaces: Application of continuous shape measure analysis to the decavanadate anion. New J. Chem. 2016, 40, 962–975. [CrossRef] 73. Student, A.K.; Hsu, R.Y.; Lane, M.D. Induction of fatty acid synthetase synthesis in diﬀerentiating 3T3-L1 preadipocytes. J. Biol. Chem. 1980, 255, 4745–4750. [PubMed] p y J , , [ ] 73. Student, A.K.; Hsu, R.Y.; Lane, M.D. Induction of fatty acid synthetase synthesis in diﬀerentiating 3T3-L1 preadipocytes. J. Biol. References Chem. 1980, 255, 4745–4750. [PubMed] Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional aﬃliations. © 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/). © 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
https://openalex.org/W2018170899	https://www.int-arch-photogramm-remote-sens-spatial-inf-sci.net/XL-5/151/2014/isprsarchives-XL-5-151-2014.pdf	English	null	Water Surface and Velocity Measurement-River and Flume	The international archives of the photogrammetry, remote sensing and spatial information sciences/International archives of the photogrammetry, remote sensing and spatial information sciences	2,014	cc-by	5,619	1. INTRODUCTION The modelling of water flow in natural watercourses by computational fluid dynamics (CFD) incorporates several uncertainties due to the great complexity of the phenomena involved at assorted scales. An overlooked factor reducing water conveyance and therefore increasing flooding incidence is natural waterborne vegetation, which varies both temporally and spatially. A funded project at Loughborough University is therefore examining the influence of such riparian vegetation upon water flow, requiring detailed measurement of both the 3D water surface and surface flow field velocities. A laboratory flume provides the ideal environment to capture detailed empirical data in a controlled environment. However, there is a concern that real processes and interactions can be overlooked unless limited experimental work is conducted on a real river to avoid issues relating to scale. Thomson et al. (1999) generated elevation maps representing the water surface of a reach of the North Saint Vrain Creek, Colorado at a range of discharges. Their data appear to have been collected by level and staff using a series of cross-sectional surveys, typically adopted at the time. The availability of total stations and terrain modelling software subsequently helped to improve the ease of data collection, by removing the need to measure along a linear line. However, such approaches still required the manual movement of the prism and consequently a surface which strictly was not captured at an instant in time. The ability of the camera to freeze motion has long been recognised. Fraser and McGee (1995) used synchronised large format CRC-1 cameras to capture 71 floating targets during the filling of a lock in the USA. Quoted accuracy was 2 cm, achieved at a frequency of 8 Hz during the eight minutes required for a 26m to lift cycle. Chandler et al. (1996) also used two synchronized analogue cameras in an oblique perspective. Two Hasselblad ELX cameras, modified to include a reseau plate, were used to measure the confluence of a pro-glacial meltwater channel, immediately downstream of the Upper Arolla glacier in Switzerland. They used 60 polystyrene fishing floats as marker points, constrained by six fishing lines. Images were scanned and off–the-shelf image processing software was used to measure the centroids of each target. A self-calibrating bundle adjustment was then used to derive XYZ coordinates and consequent digital elevation models (DEMs). 1. INTRODUCTION This paper reports on the use of close range digital photogrammetry, to measure both the topographic surface created by a dynamic and flowing water and surface velocity vectors representing the surface flow field. A key aspect was to develop a technique that could be used in both a laboratory flume environment but could also be applied on a real river. Access to just inexpensive hardware and software is required, through the use of a pair of synchronised digital SLR cameras and proprietary close range photogrammetric software. ABSTRACT: Understanding the flow of water in natural watercourses has become increasingly important as climate change increases the incidence of extreme rainfall events which cause flooding. Vegetation in rivers and streams reduce water conveyance and natural vegetation plays a critical role in flood events which needs to be understood more fully. A funded project at Loughborough University is therefore examining the influence of vegetation upon water flow, requiring measurement of both the 3-D water surface and flow velocities. Experimental work therefore requires the measurement of water surface morphology and velocity (i.e. speed and direction) in a controlled laboratory environment using a flume but also needs to be adaptable to work in a real river. Measuring the 3D topographic characteristics and velocity field of a flowing water surface is difficult and the purpose of this paper is to describe recent experimental work to achieve this. After reviewing past work in this area, the use of close range digital photogrammetry for capturing both the 3D water surface and surface velocity is described. The selected approach uses either two or three synchronised digital SLR cameras in combination with PhotoModeler for data processing, a commercial close range photogrammetric package. One critical aspect is the selection and distribution of appropriate floating marker points, which are critical if automated and appropriate measurement methods are to be used. Two distinct targeting approaches are available: either large and distinct specific floating markers or some fine material capable of providing appropriate texture. Initial work described in this paper uses specific marker points, which also provide the potential measuring surface velocity. The paper demonstrates that a high degree of measurement and marking automation is possible in a flume environment, where lighting influences can be highly controlled. When applied to a real river it is apparent that only lower degrees of automation are practicable. The study has demonstrated that although some automation is possible for point measurement, point matching needs to be manually guided in a natural environment where lighting cannot be controlled. single point through time, normally for providing flood warnings. Measuring the true water surface therefore requires sampling elevations at multiple locations in space, at an appropriate density and, ideally captured at an instant in time. The scientific literature reveals only a few instances where this has been achieved. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-5, 2014 ISPRS Technical Commission V Symposium, 23 – 25 June 2014, Riva del Garda, Italy The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-5, 20 ISPRS Technical Commission V Symposium, 23 – 25 June 2014, Riva del Garda, Italy WATER SURFACE AND VELOCITY MEASUREMENT- RIVER AND FLUME J.H. Chandlera * , E. Ferreiraa, R. Wackrowa and K Shionoa a Dept. Civil and Building Engineering, Loughborough University, LE11 3TU, UK– J.H.Chandler@lboro.ac.uk Commission V, WG V5 KEY WORDS: Hydrology, Change Detection, Automation, Sequences, Multisensor The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-5, 2014 ISPRS Technical Commission V Symposium, 23 – 25 June 2014, Riva del Garda, Italy The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-5, 2014 ISPRS Technical Commission V Symposium, 23 – 25 June 2014, Riva del Garda, Italy Chandler et al., 1996) then it is possible to determine both surface morphology and velocity, if a sequence of images is required. However, the number and distribution of points is critical if a dense surface representation is to be created. Automation in the measurement chain can be challenging, particularly if the frequency of an image sequence is incompatible with the predicted flow velocity. Alternatively, if a measurable surface texture can be distributed evenly that generates image patches which are consistent from different camera positions and not affected by random spectral reflections, then there is the potential of creating a dense morphological representation fully automatically, using either stereo (e.g. Han and Endreny, 2014) or even perhaps MVS DEM generation methods. Surface water velocities could also be potentially possible using a PIV-based approach, although discrete targets could not be tracked individually. Stereo image sequences have been used to capture the dynamics of waves in a range of studies. Piepmeir and Waters (2004) review the use of stereopairs methods for measuring water waves created in laboratory flumes. In particular, they describe the problem of creating texture on the surface, which is sufficiently distinct and unique to facilitate measurement. They note how various authors have “polluted” the water, utilizing specialized lighting to provide suitable texture that can be measured. This is a key problem for measuring a material that is generally reflective and transparent. Their own solution involved generating a fine mist to provide an appropriate texture and minimize reflections, but do not provide the details necessary to achieve this. Both Santel et al. (2004) and de Vries et al. (2010) apply stereo methods to capture breaking surf waves in the more challenging outdoor coastal environment. Both studies used comparatively simple point matching methods to generate digital surface models. Santel (2004) achieved this in an area of 200×200 m2, represented by 20,000 conjugate points generated from 200 manually measured seed locations. Accuracies were difficult to estimate because of the lack of control but comparisons with a tide gauge suggested differences of 5 cm were typical. de Vries et al. (2010) applied stereo photogrammetry also, but in both the flume and field environment; taking advantage of a pier to support the two synchronised cameras. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-5, 2014 ISPRS Technical Commission V Symposium, 23 – 25 June 2014, Riva del Garda, Italy They fully describe the use of a standard normalised correlation method to achieve image correspondences. Accuracies were also assessed by comparing elevations with the tidal gauging station, but by averaging elevation across time and space. A more useful comparison was possible in the flume study, through the use of a pressure sensor mounted on the flume bed. Water depth comparisons through time revealed good correspondences and an overall RMSE error of 34mm. The authors stressed the importance of illumination and visual texture also. In particular, the benefits of an overcast sky creating naturally diffuse illumination of the sea surface. Examining the images acquired by Santel et al. (2004), it can be also seen that overcast conditions were prevalent. The work conducted so far has used the former targeting solution, involving the sourcing of appropriate floating marker points. Selection of such markers has always been challenging and there are a range of requirements, which are often mutually exclusive. To allow full automation in the measurement process targets need to appear sufficiently distinct and spherical in an image sequence. If working in a river, then ideally they need to be biodegradable or at least collectable to avoid polluting the natural environment. This is not such an issue when working on a laboratory flume, but any material must be retrievable and must not damage expensive infrastructure such as filters or pumps. After trying a range of materials, including biodegradable packing chips (Chandler et al, 2008), translucent plastic spheres originally manufactured for roller ball-based deodorants were selected. These are mass manufactured and therefore reasonably cost-effective. They are also available in a range of sizes and approximately 200 spheres of both 20 mm and 10 mm diameter were purchased from Weener Plastics, UK Ltd for less than £100. The spheres also had an unexpected but very desirable quality when imaged using a standard SLR with Flash. The translucent and spherical shape appears to generate a small bright spot from the reflected flashlight (Figure 1). This is located at the centre of the sphere and at a fixed distance above the actual water surface and critically is independent of viewing direction. This allows the image point measured on multiple synchronised frames to represent a true point in 3-D space. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-5, 2014 ISPRS Technical Commission V Symposium, 23 – 25 June 2014, Riva del Garda, Italy Of perhaps increased relevance, Han and Endreny (2014) describe recent work in which they were able to use automated DEM extraction using a synchronised pair of digital images to measure the water surface. This was achieved using fine wax powder particles (0.3-0.8 mm diameter), which appears to create an appropriate texture under normal lighting conditions. The technique has been applied to a flume environment and they report excellent accuracies of 0.3 mm using cameras mounted just 1.3 m above the water surface. However, distributing non-biodegradable wax particles does not appear appropriate when working in the field environment. synchronised frames to represent a true point in 3 D space. Figure 1 Sharp target complete with bright central spot 3. THE MEASURING SYSTEM Examination of the literature and indeed past work conducted by the first author suggested that a viable system based on digital photogrammetry could be developed. A key issue controlling the photogrammetry design is the type of targeting or more correctly “seeding particles” to be used. The overall adopted methodology is described in the subsequent section. 2. WATER SURFACE AND VELOCITY MEASUREMENT- PAST WORK Water elevation or “stage” is routinely measured at many river locations but is always restricted to recording elevation at a This contribution has been peer-reviewed. doi:10.5194/isprsarchives-XL-5-151-2014 151 The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-5, 2014 ISPRS Technical Commission V Symposium, 23 – 25 June 2014, Riva del Garda, Italy 3.2 Imaging of the channel mounted temporarily on stakes hammered into the bank. Two Nikon D80 digital cameras were available for basic image acquisition, with a third Nikon D7000 potentially available. These are moderately priced digital SLR type cameras (10 and 15 MP) and were equipped with either a 24 or 28 mm fixed focus lens. Cameras can be mounted either vertically or obliquely on standard camera tripods. These provide the flexibility to achieve a convergent configuration for either flume or field application, with a desired base to distance ratio of approximately 1:7. The flowing water clearly creates a dynamic measurement problem, so obtaining synchronisation between the two exposures is critical. This has been achieved by constructing an electronically controlled triggering mechanism that creates a coincident electronic pulse along two standard Nikon MC-DC1 cables. Tests revealed that coincident exposures can be achieved to up to 1/1000th of a second, although all automated features (focus/exposure etc.) must be disabled. In addition, all other camera settings must be identical otherwise synchronisation accuracies drop to 1/300th of a second. For the flume work, two control bars each with 10 RAD targets were placed upon the flume sides and again coordinated using the total station (Figure 4). The control survey then provides a real world coordinate system, where the z-axis is oriented to local gravity at a known scale. Figure 3 Control survey and general layout for field application Two exposure and imaging strategies have been adopted so far. Primary data acquisition has involved using the inbuilt flash in “standard mode”, which achieves a distinct sharp image of the sphere frozen in space (Figure 1 and Figure 4). Typical exposure settings for the darker and more demanding laboratory application include: shutter priority, f-16, 1/60th second, ISO: 1000. The alternative has created a deliberate blur by using “SLOW flash synchronisation”. Typical exposure settings include: aperture priority, f-11, 1/10th second, ISO: 1600. This creates a sharp image of the sphere at the initial instant of exposure, but then a noticeable and measurable trail which is a function of the exposure time and sphere velocity (Figure 2). It was found necessary to vary the ISO setting to achieve a trail of an appropriate length, which can then allow determination of surface water speed and direction of flow at a point. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-5, 2014 ISPRS Technical Commission V Symposium, 23 – 25 June 2014, Riva del Garda, Italy of the channel mounted temporarily on stakes hammered into the bank. 3.4 Seeding distribution/collection One of the challenges that had to be overcome was to distribute the floating marker points evenly across the water surface during photo acquisition. Initial attempts that involved simply throwing the spheres manually proved unsatisfactory. Distribution was uneven and once wet, spheres tended to stick together or conglomerate. The difficulty was eased by constructing four narrow wooden “strip containers” which could be preloaded with spheres just prior to the test. These straddled the flume and could be tipped to provide a consistently spaced row of spheres. By repeating this sequentially using all preloaded strip containers, an even distribution down the channel could be achieved with practice (Figure 4). Figure 2 Deliberateley blurred targets for velocity determination 3.2 Imaging Figure 3 Control survey and general layout for field application Figure 4 Control layout for laboratory application with evenly distributed “sharp” targets p p Figure 2 Deliberateley blurred targets for velocity determination Figure 4 Control layout for laboratory application with evenly distributed “sharp” targets 3.1 Targeting/seeding particles Figure 1 Sharp target complete with bright central spot Two distinct targeting approaches appear to be viable, each having their own strengths and weaknesses. If large and visually distinct targets are adopted (e.g. Fraser and McGee, 1995; This contribution has been peer-reviewed. doi:10.5194/isprsarchives-XL-5-151-2014 This contribution has been peer-reviewed. doi:10.5194/isprsarchives-XL-5-151-2014 152 The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-5, 2014 ISPRS Technical Commission V Symposium, 23 – 25 June 2014, Riva del Garda, Italy The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-5, 2014 ISPRS Technical Commission V Symposium, 23 – 25 June 2014, Riva del Garda, Italy The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-5, 2014 ISPRS Technical Commission V Symposium, 23 – 25 June 2014, Riva del Garda, Italy 3.6 Surface velocity data-processing The measurement necessary for detecting surface velocity was broadly similar, but required greater manual intervention. Initial stages necessary to detect and determine the three-dimensional positions of the bright spots was identical. These represented the positions of the spheres at the instant of the flash (Figure 2). The end of the trail then had to be estimated and measured manually. The length of this 3-D vector divided by the exposure time (recorded in the jpeg EXIF header) then provides the velocity of the sphere. The assumption then has to be made that the sphere is moving at the same velocity of the water, which appears appropriate. The processing necessary for water surface determination, required the following key stages: 1. identifying appropriate synchronised pair, key criteria including: accurate synchronisation, even distribution of seed points across the flume/river and targets located within areas of interest (i.e. in the vicinity of vegetative elements); 2. An “Automated Smart Points” project was used to load in the required image pair and establish an initial orientation. 3. Automatically measure “RAD” targets and process to confirm that these have been measured and detected correctly. It was decided to convey the length and orientation of these velocity vectors by simply generating an orthophoto in the XY plane from the oblique imagery (Figure 7). 4. Introduce the desired datum using “External Geometry Explorer” by loading and an appropriate “Coded Target Definition File” containing the control coordinates. 5. Automatically marks positions of the seed points using “Automatic Target Marking”. Critical parameters included: diameter range in pixels and circularity or “Fit Error”. (Typical parameters: White Dots, Fit Error: 0.4, Diameter: 5-20 pixels). 4. DISCUSSION Although the previous section describes the targeting solution and key processing stages adopted, there are a range of critical controls which need to be discussed. In addition, an understanding of how the derived morphological and superficial velocity data will be used in modelling of the flow, before planned future work is identified. 6. Three water surface points were then manually “Referenced” across the pair to obtain three 3-D coordinates to define an approximate plane representing the water surface. For convenience, these points were placed on a new layer named: “wsurf”. p y 7. The remaining surface points were then “Automatically Referenced”, guided using the approximate plane defined by the three “wsurf” points. 3.5 Water surface data-processing Although some manual interaction was required, it was found that user assisted automatic processing methods could typically generate 30 to 40 water surface points in approximately 10 to 20 minutes. However, it was found that lighting played a critical control (Section 4.1). If automated methods were not viable or became too time consuming, then manual marking and referencing provided a workable alternative. The PhotoModeler scanner software was available and chosen for all photogrammetric data processing and attempts were made to automate as many of the required data processing stages as possible. The two cameras were calibrated using the PhotoModeler camera calibration option, simply by acquiring convergent imagery of a RAD-based control field at the required fixed focusing distance. This is a fully automated process and recovers all of the elements of inner orientation necessary to model the internal camera geometry. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-5, 2014 ISPRS Technical Commission V Symposium, 23 – 25 June 2014, Riva del Garda, Italy The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-5, 2014 ISPRS Technical Commission V Symposium, 23 – 25 June 2014, Riva del Garda, Italy The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-5, 20 ISPRS Technical Commission V Symposium, 23 – 25 June 2014, Riva del Garda, Italy 9. Data was then “exported” and a fixed vertical offset was subtracted to account for the difference between the bright measured spot and the actual water surface (4mm for the 20 mm diameter spheres). 9. Data was then “exported” and a fixed vertical offset was subtracted to account for the difference between the bright measured spot and the actual water surface (4mm for the 20 mm diameter spheres). The spars guided the spheres into two fishing nets and proved both portable and effective (Figure 5). both portable and effective (Figure 5). Figure 5 Two hinged wooden spas used for collecting spheres Figure 6. Automated measurement using PhotoModeler Figure 5 Two hinged wooden spas used for collecting spheres Figure 6. Automated measurement using PhotoModeler 3.3 Control Although control is not essential for the photogrammetry, obtaining appropriate scaled and oriented data is critical for relating measurements to gravity induced flows. Control could be achieved using PhotoModeler “Ringed Automatically Detected” or RAD targets of appropriate size, each easily coordinated using a standard reflectorless Total Station (Figure 3). For the river application, targets were located on both sides The plastic spheres are non-biodegradable and it is therefore unethical to allow them to escape into the natural environment. Spheres could be easily collected in both the flume and field by using two wooden and hinged spars which straddled the river. This contribution has been peer-reviewed. doi:10.5194/isprsarchives-XL-5-151-2014 153 The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-5, 2014 ISPRS Technical Commission V Symposium, 23 – 25 June 2014, Riva del Garda, Italy 4.1 Critical controls turbulent flow field in natural watercour corridors and at the same time to derive impro a variety of simulations tools and conceptual m Spatial measurement is challenging and the Figure 8. Exaggerated water surface profile in downstream d marker points. However, if automation is sought, then spurious bright spots created by the reflected light and areas of reduced contrast due to “flair”, need to be avoided. In the flume environment this was achieved by setting up a plastic canopy (e.g. a garden gazebo proved effective!) was positioned and placed over the top of the test area so that it prevented the multiple ceiling lights being reflected into the lens. This approach was not practicable in the outdoor field environment, where the scale of application was much larger. Although initial fieldwork tests proved successful, two other approaches are recommended and will be tested. Measurements could be restricted to days where the conditions are both overcast but bright. The alternative and more versatile option will be to use plane polarising filters to hopefully reduce unwanted reflections. turbulent flow field in natural watercourses and riparian corridors and at the same time to derive improved parameters in a variety of simulations tools and conceptual models. Spatial measurement is challenging and the major details and Figure 8. Exaggerated water surface profile in downstream direction (left)- real riv re both d more polarising filters to corridors and at the same time to derive improved parameters in a variety of simulations tools and conceptual models. Figure 8. Exaggerated water surface profile in downstream direction (left)- real riv Spatial measurement is challenging and the major details and problems faced during data acquisition and processing has been given in previous sections. It is apparent that independently of the degree of automation achieved, such techniques are feasible to apply in both flume and field conditions. It is desirable to somehow strengthen the linkage between what is in its essence a remote sensing technology and hydraulic real world applications by enhancing and/or combining methods. In fact, there is a wide range of potential applications where the developed techniques can be applied, including: stage and discharge measurements for river basin management plans, CFD model validation or even more fundamental aquatic studies, among others. 4.1 Critical controls One of the main reasons why measuring a water surface presents challenges is the naturally reflective properties of water. The water surface acts as a dynamic and multifaceted mirror, which creates an infinite number of reflections which vary continually, both spatially and temporally. 8. The project was then “Processed” and consequent 3-D model and residuals examined (Figure 6) to ensure no gross errors had been introduced. This contribution has been peer-reviewed. doi:10.5194/isprsarchives-XL-5-151-2014 This contribution has been peer-reviewed. doi:10.5194/isprsarchives-XL-5-151-2014 154 The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-5, 2014 ISPRS Technical Commission V Symposium, 23 – 25 June 2014, Riva del Garda, Italy The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-5, 2014 ISPRS Technical Commission V Symposium, 23 – 25 June 2014, Riva del Garda, Italy Figure 7 Scaled orthophoto showing velocity vectors in the XY plane, flume application ISPRS Technical Commission V Symposium, 23 – 25 June 2014, Riva del Garda, Italy Figure 7 Scaled orthophoto showing velocity vectors in the XY plane, flume application using diverse living vegetation specimens with different biomechanical properties and under both submerged and emergent conditions. It is thus expected to enhance, through empirical evidence, the current knowledge regarding the turbulent flow field in natural watercourses and riparian It also has to be recognised that a different pattern of reflected points is captured by each image in a synchronised pair. If manual measurement approaches are adopted then the human brain can interpret and identify the required target floating marker points. However, if automation is sought, then spurious marker points. However, if automation is sought, then spurious bright spots created by the reflected light and areas of reduced contrast due to “flair”, need to be avoided. In the flume environment this was achieved by setting up a plastic canopy (e.g. a garden gazebo proved effective!) was positioned and placed over the top of the test area so that it prevented the multiple ceiling lights being reflected into the lens. This approach was not practicable in the outdoor field environment, where the scale of application was much larger. Although initial fieldwork tests proved successful, two other approaches are recommended and will be tested. Measurements could be restricted to days where the conditions are both overcast but bright. The alternative and more versatile option will be to use plane polarising filters to hopefully reduce unwanted reflections. 4.3 Proposed future work The measurement system described in this paper is evolving and further experimental work will hopefully refine procedures. Currently, use is just being made of two synchronised cameras but it has always been recognised that a third camera would provide valuable additional redundancy. A new triggering mechanism has been constructed to allow a third camera, a Nikon D7000. It is hoped that this should improve accuracy and allow processes to be further automated. However, achieving perfect synchronisation appears challenging, probably because slightly differing electronic systems associated with the two camera types, create differential time delays. de Vries, S., Hill, D.F., de Schipper, M.A. and Stive, M.J.F., 2011. Remote sensing of surf zone waves using stereo imaging. Coastal Engineering, 58(3), pp.239–250. http://linkinghub.elsevier.com/retrieve/pii/S0378383910001444 accessed 22nd November 2013. Fraser, C. and McGee, R.G., 1995. Dynamic measurement of undulating water surfaces in a lock fill. Photogrammetric Engineering and Remote Sensing, 61(1): 83-87. Lane, S. N., Richards, K. S. and Chandler, J. H., 1993. Developments in photogrammetry- the geomorphological potential. Progress in Physical Geography, 17(3): 306-328. As discussed in Section 3.1, large and visually distinct seeds have been used in tests conducted to date. Utilising automated DEM generation methods have not proved practicable because of the lack of texture on the surface. Santel (2004) and de Vries et al., (2011) have reported success with automated DEM generation but it is probable that lighting conditions were uniquely suited. Han and Endreny (2014) describe the use of fine wax powder to create surface texture, which could be of use in the laboratory. This is an approach which could yield excellent results, particularly if Multi-View Stereo (MVS) DEM generation methods are utilised. This exact seeding material appears difficult to obtain currently, and so the search remains for a cheap, lightweight and easily distributable seeding material, which is also biodegradable. As this paper is published we are currently trying cork particles. Any other suggestions? PhotoModeler Scanner, 2013. Eos Systems Inc. http://www.photomodeler.com/, accessed 18th October 2013. Piepmeir, J.A. and Waters, J., 2004. Analysis of stereo vision based measurements of laboratory water waves. Geoscience and Remote Sensing Symposium, IGARSS '04. Proceedings. IEEE International 5: 3588-91. Santel, F., Linder, W., Heipke, C., 2004. Stereoscopic 3-D image sequence analysis of sea surfaces. International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 35(5): 643-649. http://www.isprs.org/ proceedings/XXXV/congress/comm5/papers/643.pdf accessed, 6th December 2013. 7. REFERENCES Figure 7 represents the speed and direction of flow at the water surface, for one of the flume based image sets. The orthophoto suggest that the magnitude of these flow velocities do not vary spatially across and around the vegetation, at least for this particular test. This might indicate the presence of some structures on the water surface. Furthermore, independent velocity measurements derived using a 2D/3D side looking Acoustic Doppler Velocimetry (ADV) system (Vectrino), demonstrate that when compared to the approaching flow, longitudinal velocities increase outside the wake zone, and are compatible with the photogrammetric estimate. Chandler, J.H., Lane, S.N. and Richards, K.S., 1996. The determination of water surface morphology at river channel conferences using automated digital photogrammetry and their consequent use in numerical flow modelling, International Archives of Photogrammetry and Remote Sensing, Vienna, Austria, Vol. XXXI Part B7, pp. 99-104. Chandler, J.H., Wackrow, R., Sun, X., Shiono, K.and Rameshwaran, P., 2008. Measuring a dynamic and flooding river surface by close range digital photogrammetry. International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 37(B8): 211–216 (on CD- ROM). http://www.isprs.org/proceedings/XXXVII/ congress/tc8.aspx accessed, 18th November 2013. 6. ACKNOWLEDGEMENTS None of the work described could have been possible without the financial support provided by EPSRC Grant: EP/K004891/1. Further details of the project can be found: http://floods.lboro.ac.uk/ The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-5, 2014 ISPRS Technical Commission V Symposium, 23 – 25 June 2014, Riva del Garda, Italy The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-5, 2014 ISPRS Technical Commission V Symposium, 23 – 25 June 2014, Riva del Garda, Italy The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-5, 2014 ISPRS Technical Commission V Symposium, 23 – 25 June 2014, Riva del Garda, Italy Figures 7 and 8 show some initial outcomes. By surfacing the three-dimensional morphological data using conventional Delaunay triangulation, it is possible to generate water surface profiles through any desirable plane. For river flows, the principal direction of the plane needs obviously to be oriented towards the downstream direction. Figure 8 represents a downstream cross-sectional view which dissects a bush introduced into the flow in the river channel. (Note: Pixel value = elevation). It is apparent that surface elevation increases just before the bush at a distance of 2.5 m. This is a direct consequence of flow blockage caused by the vegetation which produces a local increase of the pressure field in the vicinity of the bush. There is also a depression as the flow dissipates around and behind the vegetation at a distance of approximately 0.7 m, which is consistent with the expected flow field created by such an obstacle. both seeding and lighting, which consequently impact upon image quality, appears to play a critical control on the level of automation that can be achieved. Future proposed work will attempt to further automate the measurement procedures. Options include examining the potential of Multi-View Stereo by acquiring synchronised image triplets, but this will require identification and sourcing of an appropriate surface seeding material. 4.2 Flow modelling As stated in the introduction, the key aim of the funded research project is to consider the role of vegetation on river flooding. Specifically, to quantify the impact of riparian vegetation on fluid flow in terms of turbulence, bed friction and water surface variation. One objective is to further develop existing three- dimensional numerical flow models to simulate flow, which currently rarely take into account the frictional effects of vegetative elements. In this context, a comprehensive data set comprising point velocities, boundary shear stresses and the free surface deformation and velocity is presently being acquired This contribution has been peer-reviewed. doi:10.5194/isprsarchives-XL-5-151-2014 155 The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-5, 2014 ISPRS Technical Commission V Symposium, 23 – 25 June 2014, Riva del Garda, Italy 5. CONCLUSION Han, B. and Endreny, T.A., 2014. River surface water topography mapping at the sub Melanie to resolution and precision with close range photogrammetry: laboratory scale application. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 7(2): 602–608. Han, B. and Endreny, T.A., 2014. River surface water topography mapping at the sub Melanie to resolution and precision with close range photogrammetry: laboratory scale application. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 7(2): 602–608. This paper has demonstrated the practicalities of deriving the three-dimensional surface topography and water surface velocity of a dynamic and flowing water surface. The technique is flexible and can be applied in both the flume and field environment. Commercial software appears able to automate many aspects of the photogrammetry-based solution, although This contribution has been peer-reviewed. doi:10.5194/isprsarchives-XL-5-151-2014 This contribution has been peer-reviewed. doi:10.5194/isprsarchives-XL-5-151-2014 156
https://openalex.org/W2162356213	http://publikationen.ub.uni-frankfurt.de/files/20258/1471_2148_8_162.pdf	English	null	Resolution among major placental mammal interordinal relationships with genome data imply that speciation influenced their earliest radiations	BMC evolutionary biology	2,008	cc-by	11,447	BioMed Central BioMed Central Open A Research article Resolution among major placental mammal interordinal relationships with genome data imply that speciation influenced their earliest radiations Björn M Hallström* and Axel Janke Open Access Address: Department of Cell and Organism Biology, Division of Evolutionary Molecular Systematics, University of Lund, Sölvegatan 29, S-223 62 Lund Sweden Address: Department of Cell and Organism Biology, Division of Evolutionary Molecular Systematics, University of Lund, Sölvegatan 29, S-223 62 Lund, Sweden Email: Björn M Hallström* - bjorn.hallstrom@cob.lu.se; Axel Janke - axel.janke@cob.lu.se * Corresponding author Received: 22 January 2008 Accepted: 27 May 2008 is article is available from: http://www.biomedcentral.com/147 © 2008 Hallström and Janke; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. J This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Background years (Myr) later [2]. The resolution of these narrow tem- poral occurrences required more data than available to the study of Nikolaev et al. [3], however the placement of Xenarthra and Afrotheria remains challenged [12]. In comparison, more distant branching events such as that of monotreme, marsupial and placental mammals can be significantly resolved with as little as 8000 aa sites of nuclear coding genes [14]. Molecular dating on phyloge- nomic data estimated that 30–40 Myr separated the splits among the three mammalian infra-classes [14]. This leaves enough time to accumulate phylogenetically rele- vant sequence differences and to reduce the effects of lin- eage sorting and prevents introgression. g Recently the resolution of the mammalian tree made a quantum leap forward with the analysis of protein coding sequences of whole genome data [1-4]. The genome data allowed collecting the protein coding sequences from some 3000 genes, which is equivalent to 2.2 million nucleotides (Mnt), representing ≈10% of all coding sequences [2]. During the last 15 years since the influen- tial review of Novacek [5] on the mammalian evolution, many previously uncertain relationships are now becom- ing consistently resolved by different data sets and analyt- ical approaches. However, lately released mammal genome data have not been investigated by phylogenomic analyses. While the monophyly of the clade Laurasiatheria is well established, phylogenomic analysis did not yet resolve the relationships within this clade. The internal branches of dog, cow, bat, and shrew received only 86% and 93% bootstrap support, respectively, in a recent phylogenomic analyses [3]. This non-significant support may be due to the "insufficient" amount of 200 knt of coding sequence that was available at the time of the analysis. However, the splits may also have occurred at a very narrow time win- dow of a few millions years or less, and more data than expected may be required to resolve such tight diver- gences. Presently the monophyly of each of the four major clades, Euarchontoglires, Laurasiatheria, Xenarthra, and Afrothe- ria is supported by most analyses based on sequence data [2,6] and genome level data such as LINE and LTR ele- ments [7]. The smallest clade, Xenarthra, consists of only two orders, Cingulata (armadillos) and Pilosa (sloths and anteaters) [8]. The Afrotheria [9] comprises the traditional orders Sirenia, Proboscidea, Hyracoidea, and Tubuliden- tata plus the members from the paraphyletic Lipotyphla (Insectivora): the Tenrecidae, Macroscelidea and Chryso- chloridea. Background The Afrotheria and Xenarthra are grouped in the clade named Xenafrotheria [10], a clade that is sup- ported by phylogenomic analyses [2] and retroposon and indel analysis [11]. Other studies, however, find support for a basal Xenarthra clade from retroposon data [7], or a basal Afrotheria clade from some sequence data analyses [3,12]. The two remaining clades, Euarchontoglires (Pri- mates, Rodentia, Lagomorpha, Scandentia, Dermoptera) and Laurasiatheria (all remaining orders) together com- prise the species rich taxon Boreoplacentalia, which is supported by all current analyses. The new name Boreo- placentalia has been suggested for this clade for being more consistent and specific than the previous name Boreoeutheria [13]. It is noteworthy that the few million years between mam- mal ordinal divergences are in the order of the average duration of mammalian species and their typical specia- tion times [15,16]. The fact that the splits between the extant orders occurred several tens of million years ago does not diminish the problem that these splits represent divergences among species or even lineages that were at the brink of becoming genetically separated species. Therefore, speciation adds an additional level of compli- cation for resolving such narrow divergences. Not only is the time for such divergences relatively short to accumu- late sufficient sequence differences that are needed to resolve these splits, but lineage sorting [17], species hybridization [18] and hybrid speciation [19] can make the resolution of significant parts of the mammalian tree all but impossible. Many branches of the mammalian tree remained difficult to resolve, even with the analysis of several 100 thousand nt (knt) of sequence data. The most basal placental mam- mal divergences were inconclusive even with the analysis of some 200 knt of coding sequence data [3]. The analysis of a ten-fold larger dataset, 2.2 Mnt of protein coding sequence data supported the grouping of Xenarthra with the Afrotheria in the higher order clade Xenafrotheria [2]. Jackknife analysis showed that at least 600 knt sites of pro- tein coding data were needed to significantly resolve this relationship [2]. The notoriously difficult to resolve split between Xenafrotheria and Boreoplacentalia took place at ≈100 million years ago (Mya). Abstract Background: A number of the deeper divergences in the placental mammal tree are still inconclusively resolved despite extensive phylogenomic analyses. A recent analysis of 200 kbp of protein coding sequences yielded only limited support for the relationships among Laurasiatheria (cow, dog, bat and shrew), probably because the divergences occurred only within a few million years from each other. It is generally expected that increasing the amount of data and improving the taxon sampling enhance the resolution of narrow divergences. Therefore these and other difficult splits were examined by phylogenomic analysis of the hitherto largest sequence alignment. The increasingly complete genome data of placental mammals also allowed developing a novel and stringent data search method. Results: The rigorous data handling, recursive BLAST, successfully removed the sequences from gene families, including those from well-known families hemoglobin, olfactory, myosin and HOX genes, thus avoiding alignment of possibly paralogous sequences. The current phylogenomic analysis of 3,012 genes (2,844,615 nucleotides) from a total of 22 species yielded statistically significant support for most relationships. While some major clades were confirmed using genomic sequence data, the placement of the treeshrew, bat and the relationship between Boreoeutheria, Xenarthra and Afrotheria remained problematic to resolve despite the size of the alignment. Phylogenomic analysis of divergence times dated the basal placental mammal splits at 95–100 million years ago. Many of the following divergences occurred only a few (2–4) million years later. Relationships with narrow divergence time intervals received unexpectedly limited support even from the phylogenomic analyses. Conclusion: The narrow temporal window within which some placental divergences took place suggests that inconsistencies and limited resolution of the mammalian tree may have their natural explanation in speciation processes such as lineage sorting, introgression from species hybridization or hybrid speciation. These processes obscure phylogenetic analysis, making some parts of the tree difficult to resolve even with genome data. Page 1 of 13 (page number not for citation purposes) http://www.biomedcentral.com/1471-2148/8/162 BMC Evolutionary Biology 2008, 8:162 Page 2 of 13 (page number not for citation purposes) Results The first step in the data collection, searching for human protein-coding genes that do not have obviously closely related members of a gene family resulted in a set of 14,302 human sequences, from the original 24,108 unique genes. These sequences were used for identifying related sequences from the remaining 21 species. After selecting against genes from multi-gene families in any of the species (step two) and using only alignments in which more than 16 species are represented, 8,813 multiple sequence alignments remained. From this data set sequence alignments that were shorter than 300 bp or observed aa distances (p) larger than 30% for any species pair were removed. This left a data set of 3,012 multiple sequence alignments. The total size of this alignment was 2,844,615 nucleotides for 22 species. The average length of the individual sequences is 944 ± 748 nt and the aver- age p distance between human and platypus is 18.6 ± 4.32%, with a maximum of 30% (Figure 1), which is expected from the filtering constraints. When inspected by eye the base and amino acid frequen- cies appear to be very similar among the species. However, due to the large size of the dataset the chi-square test may become overly strict. Compositional homogeneity was rejected for many species both for nucleotide and amino acid data (see Table 1). The compositional heterogeneity remained even when the data were R/Y-coded. The evolutionary model and the number of rate heteroge- neity classes were estimated from previous phylogenomic analyses [2]. Exact tests such as implemented in MODEL- TEST [20] could not be performed due to computational constraints and the inability of most programs to analyze a data set of this magnitude. A manual analysis would have been prohibitively time consuming to perform and we suspect would have resulted in the same model selec- tion. The phylogenetic analyses of the data produced an unam- biguous picture of most placental mammals relation- ships. The ML tree in Figure 2 was constructed from first and second codon positions (NT12) and analyzed under a GTR 8Γ+I model of sequence evolution for the 2,844,615 nt long sequences in TF. Although in this and most other analyses the bat groups with the cow, the branch is shown as unresolved in Figure 2, because the placement of the bat on the mammalian tree is indefinite in some analyses. Background However, the two clades diverged into Xenarthra and Afrotheria, and Laurasiathe- ria and Euarchontoglires respectively only 3–4 million The continuously increasing amount of sequence data from protein coding genes now makes it possible to pro- vide a phylogenomic analysis of placental mammals for 19 species on the basis of several million characters long alignments. The quantity of new genome data also allows using a new and more stringent assembly of the data, which aims at avoiding data from gene families and thus paralogous sequences in the alignment. We investigate if the resolution of the basal split among placental mam- mals [2] prevails with the increased taxon sampling and if other difficult to resolve divergences such as those within Page 2 of 13 (page number not for citation purposes) http://www.biomedcentral.com/1471-2148/8/162 http://www.biomedcentral.com/1471-2148/8/162 BMC Evolutionary Biology 2008, 8:162 http://www.biomedcentral.com/1471-2148/8/162 the Laurasiatheria and Euarchontoglires can be resolved with statistical confidence with an increased amount of phylogenomic data and new analytical approaches [12]. families that had more than 75% sequence similarity and eliminated these. It can be argued that more distant sequences will not be found when searching for such sequences in other species with the same cutoff value of 10-12. Page 3 of 13 (page number not for citation purposes) Results Bayesian analysis of the same data set, on first plus second codon positions or on aa sequence data reconstructed the same topology. Also R/Y coded sequences that were analyzed using a two-state [21] ML 4Γ+I model reconstructed the topology shown in Figure 2. ML analysis with a non-stationary model in PAML [22] confirmed the topology in Figure 2. Finally, the same tree was also reconstructed after removing sites with more than three different amino acids across all species. The rigorous filtering by recursive BLAST search removed sequences that are part of gene families notably reduced the amount of data. The filtering was rather effective, even though there is no formal proof for its efficiency yet. Already after the first filtering step (human sequence against human genome) sequences of the hemoglobin, myosin, olfactory receptor, HOX gene and other sequences that are part of known gene families were elim- inated from the data set. It is conceivable that by this approach data from other, less known gene families were also excluded from the analysis. The recursive BLAST search with a cutoff value of 10-12 identified genes in gene Distance distribution of the human-platypus aa sequence alignment Figure 1 Distance distribution of the human-platypus aa sequence alignment. 0% 5% 10% 15% 20% 25% 30% 9% 8% 6% 5% 4% 7% 3% 2% 1% 0% human-platypus distance fraction of alignment Placing the treeshrew on the mammalian tree by sequence analysis was somewhat problematic. ML reconstructed an alternative position of the treeshrew, outside the primates plus Glires clade, when the nt data were not partitioned by codon position. An extended ML analysis on aa and first and second codon position sequences (NT12) could, however, differentiate between alternative placements of the treeshrew on the tree (Table 2a, Figure 3, tree 1–3) for the different datasets. A sister group position to the pri- mates is the best ML option for most analyses. The tree- shrew as sister group to primates plus Glires can be rejected by other datasets with statistical confidence. Also Distance distribution of the human-platypus aa sequenc alignment Figure 1 Distance distribution of the human-platypus aa sequence alignment. Page 3 of 13 (page number not for citation purposes) Page 3 of 13 (page number not for citation purposes) http://www.biomedcentral.com/1471-2148/8/162 BMC Evolutionary Biology 2008, 8:162 Table 1: Nt composition (f) and chi-2 probabilities (P) for homogeneous character composition for nt, R/Y-coded and aa sequence data. Results Neither a codon analysis of the concatenated dataset or a separate analysis of 402 individual genes, a total of 501,012 nt of sequence data, for which full taxon sampling was available, clarified the position of the treeshrew, because none of the topol- ogies in Table 2a received significant support. the grouping of the treeshrew with the Glires can be rejected by ML analysis of aa and NT12. Neither a codon analysis of the concatenated dataset or a separate analysis of 402 individual genes, a total of 501,012 nt of sequence data, for which full taxon sampling was available, clarified the position of the treeshrew, because none of the topol- ogies in Table 2a received significant support. ML tree on NT12 analysis and ML bootstrap support values (nt/aa) for branches that do not receive maximum support Figure 2 ML tree on NT12 analysis and ML bootstrap support values (nt/aa) for branches that do not receive maxi- mum support. Platypus Opossum Rabbit Squirrel Rat Mouse Guinea pig Treeshrew Galago Macaque Human Chimpanzee Shrew Hedgehog Bat Cow Dog Cat Tenrec Elephant Armadillo Chicken 43/98 Glires Laurasiatheria Boreoplacentalia Xenafro- theria Xen- arthra Afro- theria Euarchonta Alternative phylogenetic positions of the guinea pig among the rodents can clearly be rejected by an SH test on all analyzed data types (Table 2b, Figure 3, tree 4–6). Glires Euarchonta The highly unconventional phylogenetic position of the bat as sister group to the cow and thus probably to all Arti- odactyla and Cetacea is recovered by most analyses and datasets (Table 2c, Figure 3, tree 7–11). The only alterna- tive position is as sister group to the lipotyphla (hedgehog plus shrew) which cannot be rejected by the SH test on aa or codon sequence data. It is the favored topology for sep- arate aa analysis of individual genes. Albeit, the support is very limited. Varied results were obtained from ML analy- ses on data that were partitioned according to the average aa distance. Without following logic the bat grouped either with the cow or the shrew plus hedgehog. Analyses of partitions with over 10% aa distance grouped the bat indistinguishably by ML on any of the neighboring branches. Although most analyses favor a grouping of bat plus cow, the branch leading to the bat, cow and carni- vores is shown as unresolved in Figure 2. Results Species f(A) f(T) f(G) f(C) f(R) f(Y) P(nt) P(R/Y) P(aa) Human 27.6 24.0 24.7 23.8 52.2 47.8 0.00 1.33 49.1a Chimpanzee 27.6 24.0 24.7 23.8 52.2 47.8 0.00 1.23 27.5a Macaque 27.5 24.0 24.7 23.8 52.2 47.8 0.00 90.20a 90.3a Galago 27.5 24.0 24.6 23.9 52.2 47.8 0.00 86.57a 4.10 Treeshrew 27.1 23.4 25.0 24.5 52.2 47.8 0.00 41.72a 91.6a Rat 26.7 22.8 25.3 24.8 52.0 48.0 0.00 0.00 0.00 Mouse 26.8 23.2 25.3 24.7 52.1 47.9 0.00 0.37 0.02 Guinea pig 27.1 23.6 25.0 24.3 52.1 47.9 5.03a 33.09a 63.7a Squirrel 27.5 24.1 24.6 23.9 52.1 48.0 0.00 1.19 30.4a Rabbit 26.6 22.8 25.6 25.0 52.2 47.8 0.00 88.4a 0.01 Cat 26.5 22.9 25.6 25.1 52.0 48.0 0.00 0.00 0.00 Dog 27.2 23.6 24.9 24.2 52.1 47.9 0.00 28.22a 93.8a Cow 26.7 23.0 25.4 24.9 52.1 47.9 0.00 1.42 0.04 Microbat 26.5 22.8 25.6 25.2 52.1 47.9 0.00 0.07 0.00 Shrew 26.7 23.0 25.4 25.0 52.0 48.0 0.00 0.05 40.7a Hedgehog 27.2 23.6 24.9 24.4 52.1 47.9 0.00 0.10 48.9a Elephant 27.4 23.8 24.8 23.0 52.2 47.8 0.00 68.44a 7.28a Tenrec 26.3 22.5 25.7 25.5 52.0 48.0 0.00 0.00 0.00 Armadillo 27.5 23.8 24.7 24.0 52.2 47.8 0.00 71.50a 77.3a Opossum 28.4 24.9 23.8 22.9 52.2 47.8 0.00 35.09a 0.00 Platypus 26.5 22.3 25.7 25.5 52.2 47.8 0.00 17.3a 0.00 Chicken 28.8 24.5 24.3 22.5 53.0 47.0 0.00 0.00 0.00 a Species for which compositional homogeneity could not be rejected with 5% significance, and are thus assumed to be homogeneous. ) and chi-2 probabilities (P) for homogeneous character composition for nt, R/Y-coded and aa sequence Table 1: Nt composition (f) and chi-2 probabilities (P) for homogeneous character composition for nt, R/Y-coded and aa sequence data. composition (f) and chi-2 probabilities (P) for homogeneous character composition for nt, R/Y-coded and a Table 1: Nt composition (f) and chi-2 probabilities (P) for homogeneous character composition for nt data. a Species for which compositional homogeneity could not be rejected with 5% significance, and are thus assumed to be homogeneous. the grouping of the treeshrew with the Glires can be rejected by ML analysis of aa and NT12. Results ML tree on NT12 analysis and ML bootstrap support values (nt/aa) for branches that do not receive maximum support Figure 2 ML tree on NT12 analysis and ML bootstrap support values (nt/aa) for branches that do not receive maxi- mum support. ML tree on NT12 analysis and ML bootstrap support values (nt/aa) for branches that do not receive maximum support Figure 2 ML tree on NT12 analysis and ML bootstrap support values (nt/aa) for branches that do not receive maxi- mum support. ML analyses of all nt sequences (NT123), NT12, codon model analysis and separate analysis of nt sequences favor Page 4 of 13 (page number not for citation purposes) Page 4 of 13 (page number not for citation purposes) http://www.biomedcentral.com/1471-2148/8/162 BMC Evolutionary Biology 2008, 8:162 a split between Xenafrotheria and the remaining placen- tals (Table 2d, tree 12). However, analysis of aa sequences ever so slightly favor a split between Afrotheria and the remaining placentals (Table 2d, tree 13), but cannot reject alternative hypothesis. Partitioned ML analysis of codon positions and separate analysis of aa sequences support a split between Xenarthra and the remaining placentals (Table 2d, tree 14). ML bootstrap support values for the branches that do not receive maximum support are shown in Figure 2. Thus, bootstrap analysis unambiguously supports the relation- ships among Xenafrotheria, Glires and primates. As expected, nt sequences find only limited support for the treeshrew as sister group to the primates, but even the analysis of aa sequences does not provide maximal boot- strap support for this relationship. While nt sequences provide the strongest bootstrap support (91%) for placing the cow and bat on a common branch, ML bootstrap anal- ing only by 31% and instead favors to group the bat with the shrew plus hedgehog on a common branch. This rela- tionship receives 64% bootstrap support. Divergence times were calculated for the tree topology as shown in Figure 1. Estimated ages for all divergences are depicted in a chronogram in Figure 4. The exact values for different methods and their standard deviations for the r8s dates based on aa and nt sequences are shown in Table 3. Virtually the same divergence times among Boreopla- centalia were calculated for the two alternatively rooted trees (Table 4). For some dates the r8s program tend to provide 4–5 Myr younger time estimates than TF. Results How- ever, the dates provided by TF generally conform within a few 100,000 years to those previously published on phyl- ogenomic analyses [2], and most importantly, only mar- ginal differences are observed for the relative dates. The standard deviations appear to be unrealistically small, but reflect the amount of data that was used for the analysis. Table 2: Support for alternative positions (see fig. 3) of controversial relationships of treeshrew (a), guinea pig (b), bat (c), and Boreoplacentalia, Afrotheria and Xenarthra (d). SH probabilities are shown for analyses of concatenated sequences and bootstrap probabilities are shown for the separate analyses of individual sequences (marked: sep). a) AA NT123 NT12 NT123 cdp codon NT123 sep AA sep Tree 1 0.0286 0.4355 0.013 0.1077 0.613 0.1853 0.1320 Tree 2 1.0 0.6312 1.0 1.0 1.0 0.8146 0.3325 Tree 3 0 1.0 0 0 0.048 0.0001 0.5355 b) AA NT123 NT12 NT123 cdp codon NT123 sep AA sep Tree 4 1.0 1.0 1.0 1.0 1.0 1.0 1.0 Tree 5 0 0.0002 0 0 0 0 0 Tree 6 0 0 0 0 0 0 0 c) AA NT123 NT12 NT123 cdp codon NT123 sep AA sep Tree 7 1.0 1.0 1.0 1.0 1.0 0.9512 0.2728 Tree 8 0.003 0 0 0 0 0 0 Tree 9 0 0 0 0 0 0 0 Tree 10 0.625 0 0 0 0.182 0.0488 0.7273 Tree 11 0 0 0 0 0 0 0 d) AA NT123 NT12 NT123 cdp codon NT123 sep AA sep Tree 12 0.5854 1.0 1.0 0.0486 1.0 0.7549 0.0346 Tree 13 1.0 0.0040 0 0.0442 0.094 0.0170 0.2342 Tree 14 0.6176 0.0458 0 1.0 0.079 0.2281 0.7312 Table 2: Support for alternative positions (see fig. 3) of controversial relationships of treeshrew (a), guinea pig (b), bat (c), and Boreoplacentalia, Afrotheria and Xenarthra (d). SH probabilities are shown for analyses of concatenated sequences and bootstrap probabilities are shown for the separate analyses of individual sequences (marked: sep). ing only by 31% and instead favors to group the bat with the shrew plus hedgehog on a common branch. This rela- tionship receives 64% bootstrap support. a split between Xenafrotheria and the remaining placen- tals (Table 2d, tree 12). However, analysis of aa sequences ever so slightly favor a split between Afrotheria and the remaining placentals (Table 2d, tree 13), but cannot reject alternative hypothesis. Page 5 of 13 (page number not for citation purposes) Results Partitioned ML analysis of codon positions and separate analysis of aa sequences support a split between Xenarthra and the remaining placentals (Table 2d, tree 14). a split between Xenafrotheria and the remaining placen- tals (Table 2d, tree 12). However, analysis of aa sequences ever so slightly favor a split between Afrotheria and the remaining placentals (Table 2d, tree 13), but cannot reject alternative hypothesis. Partitioned ML analysis of codon positions and separate analysis of aa sequences support a split between Xenarthra and the remaining placentals (Table 2d, tree 14). Divergence times were calculated for the tree topology as shown in Figure 1. Estimated ages for all divergences are depicted in a chronogram in Figure 4. The exact values for different methods and their standard deviations for the r8s dates based on aa and nt sequences are shown in Table 3. Virtually the same divergence times among Boreopla- centalia were calculated for the two alternatively rooted trees (Table 4). For some dates the r8s program tend to provide 4–5 Myr younger time estimates than TF. How- ever, the dates provided by TF generally conform within a few 100,000 years to those previously published on phyl- ogenomic analyses [2], and most importantly, only mar- ginal differences are observed for the relative dates. The standard deviations appear to be unrealistically small, but reflect the amount of data that was used for the analysis. Most divergence time estimates are only negligibly differ- ML bootstrap support values for the branches that do not receive maximum support are shown in Figure 2. Thus, bootstrap analysis unambiguously supports the relation- ships among Xenafrotheria, Glires and primates. As expected, nt sequences find only limited support for the treeshrew as sister group to the primates, but even the analysis of aa sequences does not provide maximal boot- strap support for this relationship. Results While nt sequences provide the strongest bootstrap support (91%) for placing the cow and bat on a common branch, ML bootstrap anal- ysis of aa sequences data support the bat plus cow group- Page 5 of 13 (page number not for citation purposes) Page 5 of 13 (page number not for citation purposes) http://www.biomedcentral.com/1471-2148/8/162 BMC Evolutionary Biology 2008, 8:162 Partial trees illustrating the alternative topologies that were statistically evaluated to determine the phylogenetic position of the tree shrew (1–3), the relationships within Rodentia (4–6), the position of the bat (7–11), and the Xenarthra, Afrotheria and Boreoplacentalia relationships (12–14) Figure 3 Partial trees illustrating the alternative topologies that were statistically evaluated to determine the phyloge- netic position of the tree shrew (1–3), the relationships within Rodentia (4–6), the position of the bat (7–11), and the Xenarthra, Afrotheria and Boreoplacentalia relationships (12–14). Results cies [8,24-26], immunological data [27], sequence data of single genes [28], sequence data from whole mitochon- drial genomes [29], concatenation of several nuclear encoded genes [6], hundreds of nuclear genes from expressed sequence tag data [14,30], and now phyloge- nomic analyses, which cover 10% of the coding region of the genome. The use of rare genomic characters such as retroposed elements [7] or indels [11] as cladistic markers has become an important addition to the analysis of sequence data. ent from analysis of much smaller nuclear or mitoge- nomic alignments [23]. A notable exception is the age of the lagomorph and rodent divergence, which has previ- ously been estimated 5–20 Myr younger by phylogenomic analysis [2]. It should be observed that many of the diffi- cult to resolve divergences are splits that are separated by only a few Myr. These involve e.g. the treeshrew, bat, pri- mate, Glires, and the Boreoplacentalia and Xenafrotheria divergence. Results Cat Dog Cow Bat Hedgehog Shrew Cat Dog Cow Bat Hedgehog Shrew Cat Dog Cow Bat Hedgehog Shrew Cat Dog Cow Bat Hedgehog Shrew Cat Dog Cow Bat Hedgehog Shrew Rabbit Treeshrew PRIMATES RODENTS Rabbit Treeshrew PRIMATES RODENTS Rabbit Treeshrew PRIMATES RODENTS Guinea pig Rat Mouse Squirrel Rabbit Guinea pig Rat Mouse Squirrel Rabbit Guinea pig Rat Mouse Squirrel Rabbit 7 8 9 10 11 4 5 6 2 1 3 Xenarthra Afrotheria Outgroup Boreoplacentalia 13 Afrotheria Xenarthra Outgroup Boreoplacentalia 14 Afrotheria Boreoplacentalia Outgroup Xenarthra 12 Rabbit Treeshrew PRIMATES RODENTS 2 Rabbit Treeshrew PRIMATES RODENTS Rabbit Treeshrew PRIMATES RODENTS Rabbit Treeshrew PRIMATES RODENTS Guinea pig Rat Mouse Squirrel Rabbit Guinea pig Rat Mouse Squirrel Rabbit Guinea pig Rat Mouse Squirrel Rabbit 4 5 6 2 1 3 Rabbit Treeshrew PRIMATES RODENTS 1 3 3 g Guinea pig Rat Mouse Squirrel Rabbit 5 Guinea pig Rat Mouse Squirrel Rabbit 4 Guinea pig Rat Mouse Squirrel Rabbit 6 Cat Dog Cow Bat Hedgehog Shrew Cat Dog Cow Bat Hedgehog Shrew Cat Dog Cow Bat Hedgehog Shrew 8 9 10 Cat Dog Cow Bat Hedgehog Shrew 7 8 7 9 10 8 Afrotheria Xenarthra Outgroup Boreoplacentalia 14 Xenarthra Afrotheria Outgroup Boreoplacentalia 13 Afrotheria Boreoplacentalia Outgroup Xenarthra 12 12 13 14 Partial trees illustrating the alternative topologies that were statistically evaluated to determine the phylogenetic position of the tree shrew (1 3), the relationships within Rodentia (4 6), the position of the bat (7 11), and the Xenarthra, Afrotheria and Boreoplacentalia relationships (12 14) Figure 3 Partial trees illustrating the alternative topologies that were statistically evaluated to determine the phyloge- netic position of the tree shrew (1–3), the relationships within Rodentia (4–6), the position of the bat (7–11), and the Xenarthra, Afrotheria and Boreoplacentalia relationships (12–14). cies [8,24-26], immunological data [27], sequence data of single genes [28], sequence data from whole mitochon- drial genomes [29], concatenation of several nuclear encoded genes [6], hundreds of nuclear genes from expressed sequence tag data [14,30], and now phyloge- nomic analyses, which cover 10% of the coding region of the genome. The use of rare genomic characters such as retroposed elements [7] or indels [11] as cladistic markers has become an important addition to the analysis of sequence data. Discussion Table 3: Divergence time estimates for branches a-t in figure 4, were based on aa and nt sequences us a refers to the cow-bat branching which is shown unresolved in figure 4. bEstimated date is on the edge of the calibration interval and thus no standard deviation could be calculated. a refers to the cow-bat branching which is shown unresolved in figure 4. bEstimated date is on the edge of the calibration interval and thus no stand most phylogenetic analyses orthology is determined by a sequence similarity criterion [6,32] or choosing from a pre-determined database of orthologous sequences, such as COG [33] and the orthologs matrix project, OMA [34]. These databases usually use similarity criteria too and are sometimes combined with manual curation for increasing the confidence in the orthology of the sequences. chromosome maps for most genomes make it currently impractical to utilize synteny information and orthology can only be established when the species phylogeny is known [31]. However, the mammalian phylogeny is often unclear and the very issue of such studies. Therefore, in Table 4: Divergence time estimates from nt sequences using TF, for trees with alternative placements of Xenarthra and Afrotheria (trees 12–14 in figure 3). Divergence Tree 12 Tree 13 Tree 14 a 162.3 161.6 161.8 b 138.4 138.4 138.4 c 99.93 ND ND d 95.31 95.31 95.30 e 97.85 ND ND f 88.4 87.82 87.74 g 88.86 88.44 88.64 h 79.74 88.15 83.1 i 83.47 82.76 82.65 j 86.53 85.81 85.98 k 76.25 75.64 75.91 l 85.87 85.41 85.63 m 74.04 73.38 73.29 n 78.53 77.87 77.69 oa 81.1 80.66 80.82 p 59.98 59.51 59.68 q 69.62 68.97 68.89 r 28.88 27.4 27.38 s 12.3 12.3 12.3 t 9 9 9 The branches c and e do not exist and were not determined (ND). Table 4: Divergence time estimates from nt sequences using TF, for trees with alternative placements of Xenarthra and Afrotheria (trees 12–14 in figure 3). In the mammalian genome, the ratio of gene to gene fam- ily is about two and the size of about half of the gene fam- ilies have changed during mammalian evolution [35]. The globin genes are a textbook example of such gene families [36]. The human beta hemoglobin family has six mem- bers that have been duplicated at different times during evolution of mammals and they fulfill different functions. The branches c and e do not exist and were not determined (ND). The branches c and e do not exist and were not determined (ND). a refers to the cow-bat branching Page 7 of 13 (page number not for citation purposes) Discussion The steadily increasing amount of genomic sequence data from placental mammals does not only allow an ever- deeper insight into their evolution, but it also allows development of rigorous data handling procedures. A cen- tral aim in all phylogenomic analyses is selecting ortholo- gous sequences for data analysis. The lack of complete The study of placental mammal relationships has drawn the attention of generations of scientists. However, the resolution of the mammalian tree has been problematic and many relationships remained unresolved or only poorly supported. Initially phylogenetic studies were made on morphological data from recent and fossil spe- Page 6 of 13 (page number not for citation purposes) Page 6 of 13 (page number not for citation purposes) http://www.biomedcentral.com/1471-2148/8/162 BMC Evolutionary Biology 2008, 8:162 Table 3: Divergence time estimates for branches a-t in figure 4, were based on aa and nt sequences using TF and r8s. Divergence TF aa (Mya) TF nt (Mya) r8s aa (Mya) r8s nt (Mya) a 160.3 162.3 160.22 ± 0.34 160.34 ± 0.27 b 138.4 138.4 138.4 ± 0.00b 138.4 ± 0.00b c 100.5 99.93 95.0 ± 0.00b 95.0 ± 0.00b d 95.3 95.31 89.74 ± 0.16 89.57 ± 0.07 e 98.03 97.85 93.20 ± 0.30 93.20 ± 0.07 f 87.44 88.4 81.67 ± 0.18 80.91 ± 0.14 g 88.75 88.86 83.61 ± 0.18 83.21 ± 0.09 h 83.21 79.74 77.69 ± 0.24 77.56 ± 0.17 i 81.24 83.47 75.88 ± 0.19 74.86 ± 0.19 j 84.69 86.53 79.96 ± 0.22 79.00 ± 0.18 k 75.41 76.25 71.55 ± 0.21 71.29 ± 0.11 l 85.18 85.87 80.94 ± 0.30 80.44 ± 0.12 m 72.62 74.04 64.45 ± 0.17 63.46 ± 0.25 n 76.27 78.53 72.69 ± 0.21 71.47 ± 0.18 oa 80.11 81.1 76.36 ± 0.29 76.00 ± 0.15 p 60.56 59.98 57.17 ± 0.28 56.50 ± 0.11 q 67.3 69.62 60.31 ± 0.18 59.47 ± 0.26 r 26.73 28.88 26.76 ± 0.20 26.21 ± 0.12 s 12.3 12.3 12.3 ± 0.00b 12.3 ± 0.00b t 9 9 7.85 ± 0.15 7.67 ± 0.10 a refers to the cow-bat branching which is shown unresolved in figure 4. bEstimated date is on the edge of the calibration interval and thus no standard deviation could be calculated. ence time estimates for branches a-t in figure 4, were based on aa and nt sequences using TF and r8s. http://www.biomedcentral.com/1471-2148/8/162 Divergences are labeled with letters (a-t) and the exact dates and the values estimated by other methods and sequence data are shown in Table 3. Chronogra Figure 4 that other, less thoroughly studied, gene families were also eliminated from the analysis during this process. Thus, the risk for including paralogous sequences in the current analysis was greatly reduced by the recursive BLAST approach. The phylogenetic position of the treeshrew (Scandentia) within the Euarchontoglires was problematic even when analyzing nearly 3 Mnt of protein coding data. The uncer- tainty of placing the treeshrews among placental mam- mals is evident from the numerous positions it had been placed on the tree in previous studies. The treeshrews have been placed as the first divergence among Euarchontag- lires [40,41], as sister group to Lagomorpha or rodents [42], sister group to Dermoptera [6], sister group to pri- mates [5] or part of the traditional Archonta clade outside a ((Primates, Dermoptera), Chiroptera) clade [43]. Two recent studies on rare genomic changes, found support for a grouping the treeshrew with primates, from four indels [40] and five retroposed elements [41] respectively. This conforms with our sequence analysis based findings. The data collection and filtering produced a 2.8 Mbp long dataset composed of 3012 alignments of inferred tran- scripts from genes for 21 mammalian and one avian spe- cies with a moderate evolutionary rate. This alignment is well suited for studying deep phylogenetic divergences in the placental tree. The compositional bias of the data remained to be a major obstacle in the analysis, which could not be overcome for most sequences by recoding the sequences to R/Y. However, congruency of the recon- structed trees with analyses using a non-stationary model, indicate that compositional bias does not influence the tree topology to a greater extend. Further and less well- defined systematic errors are a general threat to phyloge- netic analysis [38] and are the source for strong support for erroneous relationships. Removing sites with more than three different aa, i.e relatively variable sites, or ana- lyzing different data partitions according to evolutionary The placement of the bat as sister group to the cow in most ML analyses is highly unexpected and has not been pro- posed in any other analyses. Currently the result can only be regarded as peculiar. http://www.biomedcentral.com/1471-2148/8/162 http://www.biomedcentral.com/1471-2148/8/162 BMC Evolutionary Biology 2008, 8:162 http://www.biomedcentral.com/1471-2148/8/162 Chronogram showing the estimated divergence times Figure 4 Chronogram showing the estimated divergence times. The figure is based on divergence times estimated from nt sequences using TF. Divergences are labeled with letters (a-t) and the exact dates and the values estimated by other methods and sequence data are shown in Table 3. Platypus Opossum Rabbit Squirrel Rat Mouse Guinea pig Treeshrew Galago Macaque Human Chimpanzee Shrew Hedgehog Bat Cow Dog Cat Tenrec Elephant Armadillo 160 140 120 100 80 60 40 20 0 Mya a b c d e f g h i j k l q n p s r t m rate are a recommend cure to the problem. However, these analyses did not change the topology. The taxon sampling currently available for genome analysis does not yet allow for investigating if one or several rouge taxa influence the ML reconstruction. The reconstructed phylogeny is largely congruent with that of previous studies. Most analyses favor a basal diver- gence of Xenafrotheria and Boreoplacentalia. This split is supported by phylogenomic sequence analyses, two retro- poson insertions and indels [2,4,11]. The revised naming of the Boreoeutheria clade is preferable to the previously used term, because of the improved consistency and logic. Recent naming of new clades were largely based on molec- ular phylogenetics, which by definition exclude the stem group members of a clade. Thus, the term "eutheria" in the formerly used Boreoeutheria encompasses also stem group eutherians only known from the fossil record. In comparison "placentalia" describes the respective clade more accurately, referring only to crown group members, which are accessible for molecular studies [13]. The elephant and tenrec group, as expected, in the Afroth- eria clade [9]. The Boreoplacentalia are split into Laurasia- theria and Eurchontoglires, the latter with a well- supported Glires clade. Within the rodents the guinea pig fell between Myomorpha (rat and mouse) and Sciuromor- pha (squirrel), which is consistent with some previous findings of partially constraint tree analyses [6], but inconsistent with some mitogenomic studies, which find the squirrel (Sciuridae) closer to the muroid mouse and rat to the exclusion of the guinea pig [39]. Chronogram showing the estimated divergence times Figure 4 Chronogram showing the estimated divergence times. The figure is based on divergence times estimated from nt sequences using TF. Discussion Other mammals, such as goat and sheep, which have dif- ferent history of gene duplications, have as many as 13 members of the beta hemoglobin family [37]. Thus, there is a considerable risk that the orthology of sequences from gene families cannot be securely established, unless by laborious case-to-case studies for each gene and each spe- cies involved. Such an approach is prohibitive for large- scale phylogenetic analyses. The rigorous approach of the data assembly, recursive BLAST, in our analysis excludes the vast majority of sequences from multi-gene families, which are the source of paralogous sequences, from the analysis. Already the first filtering step successfully removed sequences from typical gene families. While this is no formal proof that all genes from gene families were removed, it is conceivable Page 7 of 13 (page number not for citation purposes) Page 7 of 13 (page number not for citation purposes) http://www.biomedcentral.com/1471-2148/8/162 http://www.biomedcentral.com/1471-2148/8/162 http://www.biomedcentral.com/1471-2148/8/162 http://www.biomedcentral.com/1471-2148/8/162 http://www.biomedcentral.com/1471-2148/8/162 BMC Evolutionary Biology 2008, 8:162 hog+shrew). This grouping receives inconclusive 64% bootstrap support. Mitogenomic and nuclear gene data generally find the bat (Chiroptera) as the sister group to Cetferungulata [3,6,44], which is the currently preferred hypothesis. The notable exception to the general consen- sus came from a recent retroposon analysis that found support for a sister group relationship of Chiroptera to Perissodactyla plus Carnivora [45]. If this particular rela- tionship was supported by our sequence data, the bat would have been placed on the carnivore branch, because there is no species from Perissodactyla present in our data set. However, ML analyses can exclude a relationship of the bat and carnivores and most alternative positions with high statistical confidence. pool of another. It has been estimated that such hybridi- zation can occur up to 7–17 Myr after separation in birds [47] and thus probably also in mammals. Cattle-zebu (Bos taurus – Bos indicus) hybrids are common and fully fertile despite that the respective lineages have been separated for 1–2 Myr [48]. There are surprisingly many examples where viable, fertile and even well adapted offspring form by species hybridization in mammals [18,49,50]. It has been estimated that 10% of all animals (6% of European mammals), actually hybridize with at least one other spe- cies [47]. Many more may have gone through a phase of hybridization that remains undetected, because the spe- cies that now are observed to hybridize are usually the evolutionarily youngest. Another process that can obscure phylogenetic analysis is lineage sorting. Lineage sorting is the result of an ancestral polymorphism that survives a speciation event, succeeded by fixation of different alleles in the descendant lineages. This gives rise to a gene history that is incongruent with the species' history [17]. The extent to which lineage sort- ing jumbles the genome of closely related species has recently been investigated for the anthropoid genomes, i.e. the genomes of human, chimpanzee, gorilla, and orangutan. For about one quarter of the human genome gorilla and not chimpanzee is the closest genetic relative [51]. In 0.04% of the studied genes the human sequence share closest ancestry even with the orangutan. Lineage sorting is currently the favored hypothesis to explain the mosaic of the primate genomes [51,52]. http://www.biomedcentral.com/1471-2148/8/162 However, other natural processes such as hybridization have been used to explain the fact that some parts of the genome are more similar between human and gorilla than between human and chimpanzee [53,54]. Systematic errors can never be formally excluded as a cause for reconstructing inaccurate relationships. There- fore we have investigated the effect of two major sources of systematic error, namely highly variable sites and highly variable sequences [38]. The exclusion or separate analyses of these had no clear consequence on the recon- structed topology. Until genome data from further species become available for analysis, we regard the phylogenetic position of the bat as unsettled by genomic sequence data. The divergences among the placental mammal orders in this study occurred between ≈100 and ≈80 Mya, thus well inside the Cretaceous (140–65 Mya) and agrees with pre- vious molecular based estimates on mitogenomic and phylogenomic data [2,11,23] but is in steep contrast with some recent interpretations of the fossil record [46]. Many ordinal and higher level divergences occur within a few Myr from each other. TF estimates the Xenafrotheria and Boreoplacentalia divergence ≈100 Mya and the first split within these clades is estimated to ≈98 Mya and ≈95 Mya. The r8s estimates are somewhat younger, but present the same picture. Similarly, the Euarchontoglires clade diverges at ≈87 Mya into Glires and Euarchonta, and at ≈85 Mya into treeshrew (Scandentia) and the primate lin- eage. These splits are remarkably close to each other and would prove problematic to resolve even from recent spe- cies. One major problem of resolving short divergence time intervals is that an insufficient amount of phyloge- netic information accumulates in these short intervals. This obstacle can usually be overcome simply by increas- ing the amount of data, which is the current strategy in phylogenomic analyses. Finally, a process with a similar effect on the genome as species hybridization is hybrid speciation [19], which may be a new and radical explanation for the concoction of the mammalian genome. In this process two previously isolated sister-species hybridize and give offspring to a new and possibly better-adapted species [19]. While such a process is common in plants, there is growing evidence that this also occurs in animals, including vertebrates [50,55]. Which of these processes, lineage sorting or hybridization, has the dominating effect on the mammalian genome cannot be determined yet and needs to be studied in more detail. http://www.biomedcentral.com/1471-2148/8/162 The only alternative place of the bat that cannot be rejected by ML analysis of aa sequences is on a common branch to the Lipotyphla (hedge- Page 8 of 13 (page number not for citation purposes) Page 8 of 13 (page number not for citation purposes) http://www.biomedcentral.com/1471-2148/8/162 Probably the only solution is the collec- tion of large amounts of information as in phylogenomic analyses in the hope that the massive amount of data evens out events that lead to a genome history that is inconsistent with the species history. Large phylogenomic analyses produce a tree that, at least on average, reflects the species history. nettii (galago), Tupaia belangeri (treeshrew), Rattus norvegi- cus (Norway rat), Mus musculus (house mouse), Cavia porcellus (guinea pig), Spermophilus tridecemlineatus (ground squirrel), Oryctolagus cuniculus (rabbit), Felis catus (cat), Canis familiaris (dog), Bos taurus (cow), Myotis luci- fugus (little brown bat), Sorex araneus (shrew), Erinaceus europaeus (western european hedgehog), Loxodonta afri- cana (African elephant), Echinops telfairi (tenrec), Dasypus novemcinctus (armadillo), Monodelphis domestica (opos- sum), Ornithorhynchus anatinus (platypus) and Gallus gal- lus (chicken). These species represent 12 of the 21 extant eutherian orders. One species from each of Metatheria (marsupials) and Prototheria (monotremes), and finally Aves (bird) served as outgroups to the placental mam- mals. In order to avoid including paralogous sequences in the alignment we developed a new search strategy that aims at excluding sequences that stem from recent gene duplica- tions. This approach can identify and eliminate dupli- cated sequences that exist in several rather similar copies within each genome. This decreases the risk of including paralogous sequences in the analysis. Completely or nearly completely sequenced genome data are becoming available for phylogenomic analysis of mammalian relationships at an increasing rate. This allows ever more detailed reconstruction of their history and their major branches are becoming with time more consistently recovered. However, some mammalian rela- tionships will require more detailed studies of their his- tory, taking into account that the genome is not a fixed entity but malleable by speciation events. The first step of the data collection process is a "recursive BLAST search". It was conducted by searching each human sequence against all human sequences using BLAST [58] with a cutoff E-value of 10-12. Sequences that had identi- fied transcripts originating from more than one gene were excluded from further investigation. In cases where several transcripts of different lengths from the same gene were found, supposedly representing different splicing vari- ants, only the longest sequence was retained. http://www.biomedcentral.com/1471-2148/8/162 In any case, these processes seem to have a sizeable impact on the resolution of short internal branches of the mammalian tree and phylogenetic reconstruction in gen- eral. The studies of the jumbled anthropoid genomes [51,53,54] are the first to put some exact values on the amount of gene trees that are incongruent with the species tree. The other, more severe, problem that cannot as easily be solved is connected to genomic processes around specia- tion. Before closely related species become completely genetically isolated they can hybridize with each other. This poses a problem for phylogenetic analysis, because the species history can be obscured by introgression, the incorporation of genes from one species into the gene Page 9 of 13 (page number not for citation purposes) http://www.biomedcentral.com/1471-2148/8/162 BMC Evolutionary Biology 2008, 8:162 It must be recognized that there is a non-negligible chance that a fair amount of single loci have a history that is not compatible with the species history. Therefore introgres- sion or linage sorting are problematic especially to super- tree analyses [56] and analyses from rare genomic events [57]. In these approaches of phylogenetic reconstruction many branches are only supported by single or at most a few genes or loci. Incompatibilities in the support from the analysis of retroposon insertion for basal and ordinal mammalian divergences [7,11,45] probably document introgression or linage sorting events that confound the genome history. Probably the only solution is the collec- tion of large amounts of information as in phylogenomic analyses in the hope that the massive amount of data evens out events that lead to a genome history that is inconsistent with the species history. Large phylogenomic analyses produce a tree that, at least on average, reflects the species history. It must be recognized that there is a non-negligible chance that a fair amount of single loci have a history that is not compatible with the species history. Therefore introgres- sion or linage sorting are problematic especially to super- tree analyses [56] and analyses from rare genomic events [57]. In these approaches of phylogenetic reconstruction many branches are only supported by single or at most a few genes or loci. Incompatibilities in the support from the analysis of retroposon insertion for basal and ordinal mammalian divergences [7,11,45] probably document introgression or linage sorting events that confound the genome history. Conclusion h l Phylogenomic analysis of 3,012 genes (2,844,615 nucle- otides) from 19 placental mammal species could signifi- cantly resolve most relationships and date their basal divergences in the Cretaceous at 100 – 90 Mya. However, the mammalian divergences that occurred in very narrow time windows of 2–4 Myr remain surprisingly difficult to resolve, even by the huge amount of genomic data. Diver- gences of lineages that are now considered to be orders, started as speciation events even if they occurred millions of years ago. Recent species that have been separated for about the same amount of time can still hybridize and have fully fertile offspring, leading to introgression, or may be affected by lineage sorting that results in a gene history that is different from the species history. Thus, some deep mammalian divergences that are separated by only a few Myr appear to be affected by speciation proc- esses that obscure the phylogenetic reconstruction. In the second step, human sequences that passed the cri- teria of the first step were used for searching for similar genes in other species. The human sequences were used in a BLAST search against each of the species' databases with a cutoff E-value of 10-12. The sequence with the highest E- value score was kept for further examination. In the third step, the sequence from step two was used for a recursive BLAST search with a cutoff E-value of 10-12 against the same species. Similar to step one, only sequences with a single hit were kept for further analysis. Finally, genes for which sequences could be found for fewer than 16 of the 22 species were removed. Sequences shorter than 300 bp were also removed. The remaining sequences were translated to amino acids and multiple sequence alignments were created using MUSCLE [59]. Sequences with more than 30% observed (p) distance at the aa level for any species pair were removed. The aa alignments were then used as a template for creating nucleotide sequence alignments using the original, untranslated sequences. Methods di d Predicted cDNA sequences from chicken and all mamma- lian genomes with assemblies and gene builds in release 46 of ENSEMBL were downloaded from ftp:// ftp.ensembl.org/pub/current_fasta/. In total 22 species were included, Homo sapiens (human), Pan troglodytes (chimpanzee), Macaca mulatta (macaque), Otolemur gar- Page 10 of 13 (page number not for citation purposes) http://www.biomedcentral.com/1471-2148/8/162 http://www.biomedcentral.com/1471-2148/8/162 http://www.biomedcentral.com/1471-2148/8/162 BMC Evolutionary Biology 2008, 8:162 Prior to the phylogenetic reconstruction the data set was analyzed for identifying compositional biases and other anomalies that could possibly influence the tree recon- struction. A chi-square test on compositional homogene- ity was performed on the base frequencies, R/Y recoded nucleotide sequences, (A and G sites were coded R (purine) and T and C were coded to Y (pyrimidine)) and aa sequences. In addition the sequence data were ana- lyzed by maximum likelihood (ML) using the PAML v.3.15 program baseml [22] with a non-stationary model (nhomo = 3), that allows for different base composition on different branches [60]. The R/Y recoding and ML anal- ysis with a non-stationary model was done for detecting reconstruction artifacts that may be caused by composi- tional biases, but did not replace the standard ML analy- sis. using the WAG2000 model [66] applying rate heterogene- ity (8Γ + I). Uncertain or controversial relationships were further analyzed by an extended ML analysis where differ- ent possible topologies were statistically evaluated in TF using Shimodaira-Hasegawa probabilities, pSH [67]. For these comparative analyses an additional analysis with a codon substitution model was performed in PAML with rate heterogeneity (8Γ). ML Bootstrap support values were calculated from 100 replicates of aa and nt sequences. The analysis was done on the Bioportal cluster on University of Oslo running TF for 28 days. The data were randomly divided into several partitions of approximately 300,000 nt (100 kaa) for Bayesian infer- ence using MrBayes 3.1.2 [68], running for 1,000,000 Markov chain Monte Carlo generations with one cold and three heated chains, discarding the first 100,000 genera- tions as burn-in and then sampling each 100th tree. R/Y coded sequences that were analyzed using a two-state ML model [21] with (4Γ+I). Partitioned analysis was per- formed in TF and MrBayes with data partition according to codon positions. The data were also partitioned into alignments with an overall aa distance of 2%, 4%, 6%, 8%, 10% and >12% and analyzed separately by ML. Methods di d Finally the data were also analyzed by ML after removing sites with more than three different aa sites. The alignment was also inspected for any grossly deviating distances between human and platypus that might have escaped detection during the data collection process. The p distances between human and platypus were calculated in a sliding window of 800 amino acids, that was moved 400 steps for each calculation until the end of the concate- nated alignment was reached. These two species were arbi- trarily chosen to represent one of the most distant mammal species pairs in this analysis and for one (human) of which the most complete genomic data are available. In addition to the analysis of the concatenated data we analyzed alignments from individual genes with full taxon sampling (402 alignments, 501,012 bp) separately with PAML, both for nt and aa sequence data. The likeli- hood values were then combined with totalml [63] and the bootstrap probabilities were recorded. Due to the size of the data set most programs were appar- ently not able to analyze the fully concatenated data prop- erly. ML analyses could be performed on the full alignment with Treefinder version June 2007 [61] only. Other programs such as TREE-PUZZLE [62], PHYML [63] or PAML either crashed or required a seemingly infinite amount time to finish a single analysis. For the ML analy- sis with PAML the data were randomly partitioned into manageable data packages and the results combined with the "totalml" program from PHYML. The nt tree was reconstructed by TF with the GTR model [64] with rate heterogeneity assuming eight classes of gamma distrib- uted rate categories [65] and one class of invariable sites (8Γ + I). The analyses of aa sequences were performed Divergence times were estimated from both aa and nt sequence data using the nonparametric rate smoothing method on a logarithmic scale (NPRS-LOG) imple- mented in TF and using the r8s program [69] applying the NPRS method and the POWELL algorithm. The eight fos- sil-based age constraints that were used to calibrate the tree were taken from Benton & Donoghue [70] and are listed in Table 5. Mean values and their standard devia- Table 5: Calibration points used for dating placental mammal divergences. Authors' contributions 23. Arnason U, Janke A: Mitogenomic analyses of eutherian rela- tionships. Cytogenet Genome Res 2002, 96:20-32. BMH performed data collection and preparation, and conducted the phylogenetic analyses. AJ had the initial idea. AJ and BMH wrote the manuscript. p y g 24. Owen R: On the anatomy of vertebrates Volume I and II. Longmans, Green, London; 1866. 25. Gregory WK: The orders of mammals. Bull Am Mus Nat Hist 1910, 27:1-524. 26. Simpson GG: The principles of classification and a classifica- tion of mammals. Bull Am Mus Nat Hist 1945, 85:1-272. References 1. Cannarozzi G, Schneider A, Gonnet G: A phylogenomic study of human, dog, and mouse. PLoS Comput Biol 2007, 3:1e2. 1. Cannarozzi G, Schneider A, Gonnet G: A phylogenomic study of human, dog, and mouse. PLoS Comput Biol 2007, 3:1e2. p 30. Kullberg M, Hallström B, Arnason U, Janke A: Expressed sequence tags as a tool for phylogenetic analysis of placental mammal evolution. PLoS ONE 2007, 2:e775. g p 2. Hallström BM, Kullberg M, Nilsson MA, Janke A: Phylogenomic data analyses provide evidence that Xenarthra and Afrothe- ria are sister groups. Mol Biol Evol 2007, 24:2059-2068. 31. Zmasek CM, Eddy SR: RIO: Analyzing proteomes by auto- mated phylogenomics using resampled inference of orthologs. BMC Bioinformatics 2002, 3:14. g p 3. Nikolaev S, Montoya-Burgos JI, Margulies EH, Program NCS, Rougemont J, Nyffeler B, Antonarakis S: Early History of Mam- mals Is Elucidated with the ENCODE Multiple Species Sequencing Data. PloS Genet 2007, 3:1. g 32. Remm M, Storm CE, Sonnhammer EL: Automatic clustering of orthologs and inparalogs from pairwise species comparisons. J Mol Biol 2001, 314:1041-1052. q g 4. Wildman DE, Uddin M, Opazo JC, Liu G, Lefort V, Guindon S, Gascuel O, Grossman LI, Romero R, Goodman M: Genomics, biogeogra- phy, and the diversification of placental mammals. Proc Natl Acad Sci USA 2007, 104:14395-14400. 33. Tatusov RL, Natale DA, Garkavtsev IV, Tatusova TA, Shankavaram UT, Rao BS, Kiryutin B, Galperin MY, Fedorova ND, Koonin EV: The COG database: new developments in phylogenetic classifica- tion of proteins from complete genomes. Nucleic Acids Res 2001, 29:22-28. 5. Novacek MJ: Mammalian phylogeny: shaking the tree. Nature 1992, 356:121-125. 6. Murphy WJ, Eizirik E, O'Brien SJ, Madsen O, Scally M, Douady CJ, Teeling E, Ryder OA, Stanhope MJ, de Jong WW, Springer MS: Res- olution of the early placental mammal radiation using Baye- sian phylogenetics. Science 2001, 294:2348-2351. 34. Dessimoz C, Cannarozzi GM, Gil M, Margadant D, Roth A, Schneider A, Gonnet GH: OMA, A Comprehensive, Automated Project for the Iden- tification of Orthologs from Complete Genome Data: Introduction and First Achievements LNCS 3678: Comparative Genomics: RECOMB 2005 International Workshop, RCG Dublin, Ireland; 2005. 7. Kriegs JO, Churakov G, Kiefmann M, Jordan U, Brosius J, Schmitz J: Retroposed elements as archives for the evolutionary his- tory of placental mammals. PLoS Biol 2006, 4:4e91. p, , ; 35. References Demuth JP, De Bie T, Stajich JE, Cristianini N, Hahn MW: The evo- lution of mammalian gene families. PLoS ONE 2006, 1:e85. y p 8. McKenna MC, Bell SK: Classification of mammals – above the species level New York: Columbia University Press; 1997. 36. Burmester T, Ebner B, Weich B, Hankeln T: Cytoglobin: A Novel Globin Type Ubiquitously Expressed in Vertebrate Tissues. Mol Biol Evol 2002, 19:416-421. 9. Stanhope MJ, Waddell VG, Madsen O, de Jong W, Hedges SB, Cleven GC, Kao D, Springer MS: Molecular evidence for multiple ori- gins of Insectivora and for a new order of endemic African insectivore mammals. Proc Natl Acad Sci USA 1998, 95:9967-9972. 37. Aguileta G, Bielawski JP, Yang Z: Proposed standard nomencla- ture for the alpha- and beta-globin gene families. Genes Genet Syst 2006, 81:367-371. 10. Asher RJ: Insectivoran-grade placental mammals: character evolution and fossil history. In The Rise of Placental Mammals Edited by: Rose KD, Archibald JD. Baltimore and London: The Johns Hopkins University Press; 2005:50-70. y 38. Rodriguez-Ezpeleta N, Brinkmann H, Burger G, Roger AJ, Gray MW, Philippe H, Lang BF: Toward resolving the eukaryotic tree: the phylogenetic positions of jakobids and cercozoans. Curr Biol 2007, 17:1420-1425. p y 11. Murphy WJ, Pringle TH, Crider TA, Springer MS, Miller W: Using genomic data to unravel the root of the placental mammal phylogeny. Genome Res 2007, 174:413-421. 39. Horner DS, Lefkimmiatis K, Reyes A, Gissi C, Saccone C, Pesole G: Phylogenetic analyses of complete mitochondrial genome sequences suggest a basal divergence of the enigmatic rodent Anomalurus. BMC Evol Biol 2007, 7:16. p y g y 12. Nishihara H, Okada N, Hasgawa M: Rooting the eutherian tree: the power and pitfalls of phylogenomics. Genome Biol 2007, 8:R199. 40. Janecka JE, Miller W, Pringle TH, Wiens F, Zitzmann A, Helgen KM, Springer MS, Murphy WJ: Molecular and genomic data identify the closest living relative of primates. Science 2007, 318:792-794. 13. Arnason U, Adegoke J, Gullberg A, Harley E, Janke A, Kullberg M: Mitogenomic Relationships of Placental Mammals and Molecular Estimates of their Divergences. Gene in press. 41. Kriegs JO, Churakov G, Jurka J, Brosius J, Schmitz J: Evolutionary history of 7SL RNA-derived SINEs in Supraprimates. Trends Genet 2007, 23:158-161. 14. Kullberg M, Hallström B, Arnason U, Janke A: the platypus reject the Marsupionta and acknowledge the Theria hypothesis. Zoologica Scripta 2008, 37:115-127. 42. Corneli PS: Complete Mitochondrial Genomes and Eutherian Evolution. Acknowledgements 27. Nutall GHF: Blood immunity and blood relationship Cambridge: Cam- bridge University Press; 1904. The University of Oslo allowing us to use their Bioportal cluster. We are very grateful to Ulfur Arnason and Maria A. Nilsson for comments on the manuscript. The work was supported by the Lindström foundation. g y 28. Goodman M, Czelusniak J, Beeber JE: Phylogeny of primates and other eutherian orders: a cladistic analysis using amino acid and nucleotide sequence data. Cladistics 1985, 1:171-185. q 29. Janke A, Feldmaier-Fuchs G, Thomas WK, von Haeseler A, Pääbo S: q 29. Janke A, Feldmaier-Fuchs G, Thomas WK, von Haeseler A, Pääbo S: The marsupial mitochondrial genome and the evolution of placental mammals. Genetics 1994, 137:243-256. Methods di d Split Minimum age (Mya) Maximum age (Mya) Eutheria Metatheria 125.2 138.4 Boreoplacentalia Xenafrotheria 95.3 113 Euarchontoglires Laurasiatheria 95.3 113 Lagomorpha Rodentia 61.5 100.5 Caniformia Feliformia 43 63 Apes Old World monkeys 23 34 Rat Mouse 11.0 12.3 Human Chimpanzee 6.5 10 Table 5: Calibration points used for dating placental mammal divergences. http://www.biomedcentral.com/1471-2148/8/162 http://www.biomedcentral.com/1471-2148/8/162 BMC Evolutionary Biology 2008, 8:162 tions were calculated in r8s from the branch lengths of 100 bootstrapped ML analyses of aa and nt sequences. tions were calculated in r8s from the branch lengths of 100 bootstrapped ML analyses of aa and nt sequences. 21. Felsenstein J: Evolutionary trees from DNA sequences: a max- imum likelihood approach. J Mol Evol 1981, 17:368-376. 22. Yang Z: PAML: A program package for phylogenetic analysis by maximum likelihood. Comput Appl Biosci 1997, 13:555-556. 45. Nishihara H, Hasegawa M, Okada N: Pegasoferae, an unexpected mammalian clade revealed by tracking ancient retroposon insertions. Proc Natl Acad Sci USA 2006, 103:9929-9934. BMC Evolutionary Biology 2008, 8:162 http://www.biomedcentral.com/1471-2148/8/162 http://www.biomedcentral.com/1471-2148/8/162 http://www.biomedcentral.com/1471-2148/8/162 46. Wible JR, Rougier GW, Novacek MJ, Asher RJ: Cretaceous euthe- rians and Laurasian origin for placental mammals near the K/T boundary. Nature 2007, 447:1003-1006. y 47. Mallet J: Hybridization as an invasion of the genome. Trends Ecol Evol 2005, 20:229-237. 48. Hiendleder S, Lewalski H, Wolf E, Janke A: Complete mitochon- drial genomes of Bos taurus and Bos indicus cattle provide novel aspects for taxonomy and domestication and reveal extent of intraspecific variation. Cytogenet Cell Genet 2008, 120:150-156. 49. Arnason U, Spilliaert R, Palsdottir A, Arnason A: Molecular identi- fication of hybrids between the two largest whale species, the blue whale (Balaenoptera musculus) and the fin whale (B. physalus). Hereditas 1991, 115:183-189. ( p y ) 50. Roy MS, Girman DG, Taylor AC, Wayne RK: The use of museum specimens to reconstruct the genetic variability and rela- tionships of extinct populations. Experientia 1994, 50:551-557. p p p p 51. Ebersberger I, Galgoczy P, Taudien S, Taenzer S, Platzer M, von Haeseler A: Mapping human genetic ancestry. Mol Biol Evol 2007, 24:2266-2276. 52. Barton NH: Evolutionary biology: how did the human species form? Curr Biol 2006, 1616:647-650. 53. Patterson N, Richter DJ, Gnerre S, Lander ES, Reich D: Genetic evi- dence for complex speciation of humans and chimpanzees. Nature 2006, 441:1103-1108. 54. Osada N, Wu CI: Inferring the mode of speciation from genomic data: a study of the great apes. Genetics 2005, 169:259-564. 55. Nolte AW, Freyhof J, Stemshorn KC, Tautz D: An invasive lineage of sculpins, Cottus sp. (Pisces, Teleostei) in the Rhine with new habitat adaptations has originated from hybridization between old phylogeographic groups. Proc Biol Sci 2005, 272:2379-2387. 56. Bininda-Emonds OR, Cardillo M, Jones KE, MacPhee RD, Beck RM, Grenyer R, Price SA, Vos RA, Gittleman JL, Purvis A: The delayed rise of present-day mammals. Nature 2007, 446:507-512. p y 57. Boore JL: The use of genome-level characters for phyloge- netic reconstruction. Trends Ecol Evol 2006, 21:439-446. 58. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ: Basic local alignment search tool. J Mol Biol 1990, 215:403-410. 59 Ed RC MUSCLE l i l li i h hi h 59. Edgar RC: MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Res 2004, 32:1792-1797. 60. Yang Z, Roberts D: On the use of nucleic acid sequences to infer early branchings in the tree of life. Mol Biol Evol 1995, 12:451-458. BMC Evolutionary Biology 2008, 8:162 61. Jobb G, von Haeseler A, Strimmer K: TREEFINDER4: a powerful graphical analysis environment for molecular phylogenetics. BMC Evol Biol 2004, 4:18. 62. Strimmer K, von Haeseler A: Quartet puzzling: a quartet maxi- mum likelihood method for reconstructing tree topologies. Mol Biol Evol 1996, 13:964-969. 63. Guindon S, Gascuel S: A simple, fast and accurate algorithm to estimate large phylogenies by maximum likelihood. Syst Biol 2003, 52:696-704. 64. Lanave C, Preparata G, Saccone C, Serio G: A new method for cal- culating evolutionary substitution rates. J Mol Evol 1984, 20:86-93. 65. Yang Z: Maximum likelihood phylogenetic estimation from DNA sequences with variable rates over sites: approximate methods. J Mol Evol 1994, 39:306-314. Publish with BioMed Central and every scientist can read your work free of charge "BioMed Central will be the most significant development for disseminating the results of biomedical research in our lifetime." Sir Paul Nurse, Cancer Research UK Your research papers will be: available free of charge to the entire biomedical community peer reviewed and published immediately upon acceptance cited in PubMed and archived on PubMed Central yours — you keep the copyright Submit your manuscript here: http://www.biomedcentral.com/info/publishing_adv.asp BioMedcentral Page 13 of 13 (page number not for citation purposes) Publish with BioMed Central and every scientist can read your work free of charge "BioMed Central will be the most significant development for disseminating the results of biomedical research in our lifetime." Sir Paul Nurse, Cancer Research UK Your research papers will be: available free of charge to the entire biomedical community peer reviewed and published immediately upon acceptance cited in PubMed and archived on PubMed Central yours — you keep the copyright Submit your manuscript here: http://www.biomedcentral.com/info/publishing_adv.asp BioMedcentral Publish with BioMed Central and every scientist can read your work free of charge J 66. Whelan S, Goldman N: A general empirical model of protein evolution derived from multiple protein families using a maximum likelihood approach. Mol Biol Evol 2001, 18:691-699. pp 67. Shimodaira H, Hasegawa M: Multiple comparisons of log-likeli- hoods with applications to phylogenetic inference. Mol Biol Evol 1999, 16:1114-1116. 68. Huelsenbeck JP, Ronquist F: MrBayes: Bayesian inference of phy- logeny. Bioinformatics 2001, 17:754-755. g y 69. Sanderson MJ: R8s: inferring absolute rates of molecular evo- lution and divergence times in the absence of a molecular clock. Bioinformatics 2002, 19:301-302. 70. Benton MJ, Donoghue PC: Paleontological evidence to date the tree of life. References J Mamm Evol 2002, 9:281-305. g p 15. Curnoe D, Thorne A, Coate JA: Timing and tempo of primate speciation. J Evol Biol 2006, 19:59-65. speciation. J Evol Biol 2006, 19:59-65. 43. Shoshani J, McKenna MC: Higher taxonomic relationships among extant mammals based on morphology, with selected comparisons of results from molecular data. Mol Phylogenet Evol 1998, 9:572-584. 16. van Dam JA, Abdul Aziz H, Alvarez Sierra MA, Hilgen FJ, Hoek Ostende LW van den, Lourens LJ, Mein P, Meulen AJ van der, Pelaez- Campomanes P: Long-period astronomical forcing of mammal turnover. Nature 2006, 443:687-691. Campomanes P: Long-period astronomical forcing of mammal turnover. Nature 2006, 443:687-691. y g 44. Pumo DE, Finamore PS, Franek WR, Phillips CJ, Tarzami S, Balzarano D: Complete mitochondrial genome of a neotropical fruit bat, Artibeus jamaicensis, and a new hypothesis of the rela- tionships of bats to other eutherian mammals. J Mol Evol 1998, 47:709-717. 17. Nei M: Molecular evolutionary genetics New York: Columbia University Press; 1987. 18. Gray AP: Mammalian hybrids: a check-list with bibliography 2nd edition. Commonwealth Agricultural Bureaux; 1972. Commonwealth Agricultural Bureaux; 1972. 19. Mallet J: Hybrid speciation. Nature 2007, 446:279-283. 45. Nishihara H, Hasegawa M, Okada N: Pegasoferae, an unexpected mammalian clade revealed by tracking ancient retroposon insertions. Proc Natl Acad Sci USA 2006, 103:9929-9934. Nishihara H, Hasegawa M, Okada N: Pegasoferae, an unexpected mammalian clade revealed by tracking ancient retroposon insertions. Proc Natl Acad Sci USA 2006, 103:9929-9934. 20. Posada D, Crandall KA: Modeltest: testing the model of DNA substitution. Bioinformatics 1998, 14:817-818. Page 12 of 13 (page number not for citation purposes) Page 12 of 13 (page number not for citation purposes) BMC Evolutionary Biology 2008, 8:162 BMC Evolutionary Biology 2008, 8:162 Mol Biol Evol 2007, 24:26-53. Page 13 of 13 (page number not for citation purposes)
https://openalex.org/W4309469401	https://www.nature.com/articles/s41598-022-23392-6.pdf	English	null	The effect of intravitreal brolucizumab on choroidal thickness in patients with neovascular age-related macular degeneration	Scientific reports	2,022	cc-by	5,928	The effect of intravitreal brolucizumab on choroidal thickness in patients with neovascular age‑related macular degeneration OPEN Ki Woong Bae 1, Dong Ik Kim 1 & Daniel Duck‑Jin Hwang 1,2* In this study we evaluated the effect of intravitreal brolucizumab injections on choroidal thickness in patients with neovascular age-related macular degeneration (nAMD) who previously showed an incomplete response to anti-vascular endothelial growth factor treatment. A total of thirty-four eyes from 34 patients were included in this study. The patients received an average of 2.4 ± 1.1 brolucizumab injections with the mean follow-up period of 4.9 ± 2.0 months. After their first brolucizumab treatment, the central foveal thickness (CFT) and subfoveal choroidal thickness (SFCT) were significantly decreased from 431.6 ± 190.0 μm and 193.9 ± 75.1 μm to 274.6 ± 109.4 μm (P < 0.001) and 169.4 ± 71.1 μm (P < 0.001), respectively. However, there were no improvements in visual acuity. Patients were divided into three subgroups according to the number of brolucizumab treatments: one, two, and three or more injections. In all three subgroups, the CFT and SFCT were significantly reduced compared to baseline at all time points of brolucizumab injections. In conclusion, choroidal thickness was significantly reduced after intravitreal brolucizumab injections as a switching treatment in patients with nAMD. Age-related macular degeneration (AMD) is one of the leading cause of blindness worldwide1. Neovascular AMD (nAMD) is characterized by retinal vascular leakage and fluid accumulation associated with choroidal neovascularization (CNV)2. To date, various treatment modalities have been attempted to inhibit CNV-induced exudation. Currently, intravitreal injection of anti-vascular endothelial growth factor (VEGF) agents is the first- line treatment for nAMD. Brolucizumab is the most recently developed anti-VEGF agent used for nAMD3. Brolucizumab is the smallest molecules among the available anti-VEGF drugs and can be administered at higher concentrations than other anti-VEGF agents such as ranibizumab or aflibercept4. According to two pivotal clinical trials, including HAWK and HARRIER, brolucizumab was not inferior to aflibercept in terms of visual outcomes and demonstrated more favorable anatomical effects in post hoc analyses3,5,6.fl f y Intravitreal injections of anti-VEGF agents may affect choroidal thickness. It has been reported that aflibercept reduced choroidal thickness to a greater extent than ranibizumab7–9. Koizumi et al. showed that the decrease in choroidal thickness from aflibercept treatment was associated with better visual and anatomic outcomes at one year10. However, a thinner choroid may be related to macular atrophy in long-term follow-up, which could result in severe vision loss11,12. www.nature.com/scientificreports www.nature.com/scientificreports Scientific Reports \| (2022) 12:19855 Results B li Baseline characteristics. The baseline characteristics of the 34 patients with nAMD are summarized in Table 1. Among them, 27 were male (79.4%) and the mean age was 70.6 ± 6.9 years. Polypoidal choroidal vascu- lopathy (PCV) was the most frequent AMD subtype (26 eyes, 76.5%), followed by typical AMD (6 eyes, 17.6%) and retinal angiomatous proliferation (RAP) (2 eyes, 5.9%). The mean number of brolucizumab injections was 2.4 ± 1.1, ranged 1–4 times. All eyes were non-treatment-naïve, and the mean number of previous anti-VEGF (non-brolucizumab) injections was 17.8 ± 10.1 (range, 3–40). The mean follow-up period was 4.9 ± 2.0 months (range, 1.8–8.0). Visual outcomes and central foveal thickness (CFT) after intravitreal brolucizumab injec‑ tion. At the baseline, the average best corrected visual acuity (BCVA), converted to the logarithm of the minimal angle of resolution (logMAR), was 0.42 ± 0.27 (range 0.05–0.82) and 0.42 ± 0.32 (range 0.05–1.30) (P = 0.921) one month after the first injection. There was no significant change in vision, even after additional brolucizumab treatment. Among the 34 study eyes, the initial CFT was 431.6 ± 190.0 μm, which significantly decreased to 274.6 ± 109.4 μm (P < 0.001) after the first brolucizumab treatment. The contralateral 17 eyes that had no retinal pathology including epiretinal membrane, age-related macular degeneration, macular hole, or history of vitrec- tomy, had an average initial CFT of 286.2 ± 52.6 μm and SFCT of 246.5 ± 65.3 μm during the same period with no significant change at one month follow up.hhi i The temporal changes in retinal thickness are presented in Figs. 1, 2, and 3. The CFT was significantly decreased after additional brolucizumab injections compared to baseline CFT; after the second and third brolu- cizumab injections, the CFTs were 256.8 ± 106.4 μm (n = 19, P < 0.001) and 338.8 ± 115.4 μm (n = 9, P = 0.015), respectively. Temporal changes in subfoveal choroidal thickness after intravitreal brolucizumab injec‑ tion. For 34 eyes, the mean SFCT before brolucizumab treatment was 193.9 ± 75.1 μm which decreased 12.7% from the baseline to 169.4 ± 71.1 μm (P < 0.001) after the first injection. The SFCT was significantly decreased after additional brolucizumab injections compared to baseline SFCT; after the second and third brolucizumab treatments to 175.8 ± 74.9 μm (n = 19, P = 0.002) and 170.1 ± 90.5 μm (n = 9, P = 0.012), respectively. Results B li Figure 4 shows an example of brolucizumab treatment. Adverse events. Among the 34 eyes, adverse events were reported in five eyes (14.7%) of five patients (6.0% from the total 83 injections) during the observation period. Of the five eyes with intraocular inflammation (IOI), three eyes showed only anterior uveitis and two eyes showed additional signs of vitreous opacity and occlusive vasculitis. Treatment for IOI was adjusted according to the severity of the adverse events. In cases with only mild IOI without signs of posterior uveitis or vasculitis, topical corticosteroids were initiated and tapered. Two patients with retinal vasculitis were treated with local and systemic corticosteroids. All five patients with IOI recovered, and none of them had persistent visual disturbances. www.nature.com/scientificreports/ www.nature.com/scientificreports/ Table 1. Demographic and baseline characteristics (n = 34). Values are presented as N (percentage) or mean ± standard deviation. AMD age-related macular degeneration; PCV polypoidal choroidal vasculopathy; RAP retinal angiomatous proliferation; logMAR logarithm of the minimal angle of resolution; IOP intraocular pressure. Age (years) 70.6 ± 6.9 (range, 56–85) Sex, male: female 27:7 Type of AMD (N, %) PCV: typical AMD: RAP 26 (76.5), 6 (17.6), 2 (5.9) Previous number of intravitreal injections 17.8 ± 10.1 (range, 3–40) Injection number of brolucizumab 2.4 ± 1.1 (range, 1–4) Adverse events (N, %) 5 (14.7) Mean follow up period (months) 4.9 ± 2.0 (range, 1.8–8.0) Hypertension (N, %) 14 (41.2) Diabetes mellitus (N, %) 7 (20.6) Spherical equivalent (diopters) + 0.44 ± 1.35 (range,− 3.00 − + 2.75) Corrected visual acuity (logMAR) 0.42 ± 0.27 (range, 0.05–0.82) IOP (mmHg) 14.2 ± 2.5 Age (years) 70.6 ± 6.9 (range, 56–85) Sex, male: female 27:7 Type of AMD (N, %) PCV: typical AMD: RAP 26 (76.5), 6 (17.6), 2 (5.9) Previous number of intravitreal injections 17.8 ± 10.1 (range, 3–40) Injection number of brolucizumab 2.4 ± 1.1 (range, 1–4) Adverse events (N, %) 5 (14.7) Mean follow up period (months) 4.9 ± 2.0 (range, 1.8–8.0) Hypertension (N, %) 14 (41.2) Diabetes mellitus (N, %) 7 (20.6) Spherical equivalent (diopters) + 0.44 ± 1.35 (range,− 3.00 − + 2.75) Corrected visual acuity (logMAR) 0.42 ± 0.27 (range, 0.05–0.82) IOP (mmHg) 14.2 ± 2.5 Table 1. Demographic and baseline characteristics (n = 34). Values are presented as N (percentage) or mean ± standard deviation. AMD age-related macular degeneration; PCV polypoidal choroidal vasculopathy; RAP retinal angiomatous proliferation; logMAR logarithm of the minimal angle of resolution; IOP intraocular pressure. Table 1. Demographic and baseline characteristics (n = 34). Values are presented as N (percentage) or mean ± standard deviation. AMD age-related macular degeneration; PCV polypoidal choroidal vasculopathy; RAP retinal angiomatous proliferation; logMAR logarithm of the minimal angle of resolution; IOP intraocular pressure. The effect of intravitreal brolucizumab on choroidal thickness in patients with neovascular age‑related macular degeneration OPEN Therefore, it is important to monitor the change of choroidal thickness during anti-VEGF treatment. Some case series studies found that intravitreal brolucizumab injection reduced choroidal thickness12–15. However, little is known about the change in choroidal thickness of patients with nAMD who have already been treated with other anti-VEGF agents and switched to brolucizumab. Therefore, we evaluated temporal changes of subfoveal choroidal thickness (SFCT) in patients with nAMD who were treated with brolucizumab as a switching therapy due to an incomplete response to previous treatment. 1Department of Ophthalmology, Hangil Eye Hospital, #35 Bupyeong‑Daero, Bupyeong‑Gu, Incheon 21388, Korea. 2Department of Ophthalmology, Catholic Kwandong University College of Medicine, Incheon, Korea. *email: daniel.dj.hwang@gmail.com \| https://doi.org/10.1038/s41598-022-23392-6 Scientific Reports \| (2022) 12:19855 ificreports/ Table 1. Demographic and baseline characteristics (n = 34). Values are presented as N (percentage) or mean ± standard deviation. AMD age-related macular degeneration; PCV polypoidal choroidal vasculopathy; RAP retinal angiomatous proliferation; logMAR logarithm of the minimal angle of resolution; IOP intraocular pressure. Age (years) 70.6 ± 6.9 (range, 56–85) Sex, male: female 27:7 Type of AMD (N, %) PCV: typical AMD: RAP 26 (76.5), 6 (17.6), 2 (5.9) Previous number of intravitreal injections 17.8 ± 10.1 (range, 3–40) Injection number of brolucizumab 2.4 ± 1.1 (range, 1–4) Adverse events (N, %) 5 (14.7) Mean follow up period (months) 4.9 ± 2.0 (range, 1.8–8.0) Hypertension (N, %) 14 (41.2) Diabetes mellitus (N, %) 7 (20.6) Spherical equivalent (diopters) + 0.44 ± 1.35 (range,− 3.00 − + 2.75) Corrected visual acuity (logMAR) 0.42 ± 0.27 (range, 0.05–0.82) IOP (mmHg) 14.2 ± 2.5 Discussion We evaluated the effect of brolucizumab on choroidal thickness in patients with nAMD who previously had an incomplete response to other anti-VEGF agents. The CFT and SFCT significantly decreased after the first brolu- cizumab injection, while BCVA did not show improvement compared with baseline. Patients were divided into three subgroups according to the number of brolucizumab treatments: one, two, and three or more injections. Scientific Reports \| (2022) 12:19855 \| https://doi.org/10.1038/s41598-022-23392-6 www.nature.com/scientificreports/ Figure 1. Change of central foveal thickness (CFT) and subfoveal choroidal thickness (SFCT) after intravitreal brolucizumab treatment stratified by number of injections. Data were presented with mean and standard deviation. Eyes were divided into three groups according to the number of brolucizumab injections. CFT and SFCT were significantly decreased after brolucizumab injection compared to the baseline in all three groups (P < 0.05). P value was obtained from paired T test or Wilcoxon signed rank test. Figure 1. Change of central foveal thickness (CFT) and subfoveal choroidal thickness (SFCT) after intravitreal brolucizumab treatment stratified by number of injections. Data were presented with mean and standard deviation. Eyes were divided into three groups according to the number of brolucizumab injections. CFT and SFCT were significantly decreased after brolucizumab injection compared to the baseline in all three groups (P < 0.05). P value was obtained from paired T test or Wilcoxon signed rank test. n all three subgroups, the CFT and SFCT were significantly reduced compared to baseline at all time points of rolucizumab injections.l Yun et al. reported a study in which three monthly injections of aflibercept or ranibizumab were administered to patients with nAMD9. They showed a significant reduction in choroidal thickness in both groups and the change was greater in the aflibercept group. Recently, some reports have shown changes of choroidal thickness in patients with nAMD after brolucizumab treatment. Fukuda et al. performed a study comparing the effects of three monthly aflibercept or brolucizumab injections in 52 treatment-naïve PCV patients13. In both groups, the choroidal thickness was significantly reduced by 35.6 μm and 38.9 μm, respectively. However, this difference was not significant between two groups. Matsumoto et al. evaluated the effect of brolucizumab injections in 42 eyes with treatment-naïve type 1 CNV14. The patients received three monthly injections of brolucizumab. The choroidal thickness decreased by 10.6% after the first injection and by 15.5% after three injections. Ota et al. Discussion investigated the effect of brolucizumab in patients with nAMD who showed resistance to aflibercept treatment15. After the first brolucizumab injection, choroidal thickness decreased by 12.9 μm (7.1%). Tamashiro et al. reported the results of three monthly injections of brolucizumab in patients with nAMD12. The subjects were divided into two subgroups: treatment-naïve and switched. Choroidal thickness decreased significantly in both groups, and the degree of change was greater in the treatment-naïve group (36.1 μm, 15.6% vs. 12.1 μm, 5.3%). h d l d d h A h d h b l b h g g g g μ μ In this study, we included patients with nAMD who were treated with brolucizumab as a switching treatment. Unlike other studies where monthly injections were performed, the treatment interval was two or three months in this study population. For a total of 34 eyes, brolucizumab injection was administered at least once, 19 eyes at least twice, and 9 eyes at 3 or 4 times. Since the injection interval was longer than that in previous studies, we could evaluate the effect of brolucizumab on choroidal thickness for a relatively longer period, with an average of 4.9 months and up to 8 months. The reduction in choroidal thickness in this study was 12.7%, which was generally similar to that reported in previous studies. This is lower than the results of studies from treatment-naïve patients, which may be because we included non-naïve patients who had been treated with other anti-VEGF agents. Brolucizumab is the smallest molecules among the available anti-VEGF drugs and can be administered at higher concentrations4. Therefore, it could achieve a more favorable anatomical outcome than other anti-VEGF agents5,6. Previous studies on the effect of brolucizumab treatment in patients with nAMD showed a reduction Scientific Reports \| (2022) 12:19855 \| https://doi.org/10.1038/s41598-022-23392-6 www.nature.com/scientificreports/ Figure 2. Temporal change in central subfoveal thickness (CFT) compared to baseline stratified by number of intravitreal brolucizumab injections. (A) Data presented as the average. The mean CFT of all patients was significantly decreased compared to baseline at all three-timepoints; after one, two, and three injections (P < 0.05). (B) Eyes were divided into three groups according to the number of brolucizumab injections. The CFT was significantly decreased after brolucizumab injection compared to the baseline in all three groups (P < 0.05). Discussion However, there was no significant change in the CFT between the time of #1 injection and #2 injections in patients with two injections of brolucizumab (n = 10, P = 0.169). Also, there was no significant change in the CFT in subjects with three injections of brolucizumab among three time points (n = 9); at the time of #1 injection and #2 injections (P = 0.953), #1 injection and #3 injections (P = 0.066), and #2 injections and #3 injections (P = 0.110), respectively. P value was obtained from paired T test or Wilcoxon signed rank test. Figure 2. Temporal change in central subfoveal thickness (CFT) compared to baseline stratified by number of intravitreal brolucizumab injections. (A) Data presented as the average. The mean CFT of all patients was significantly decreased compared to baseline at all three-timepoints; after one, two, and three injections (P < 0.05). (B) Eyes were divided into three groups according to the number of brolucizumab injections. The CFT was significantly decreased after brolucizumab injection compared to the baseline in all three groups (P < 0.05). However, there was no significant change in the CFT between the time of #1 injection and #2 injections in patients with two injections of brolucizumab (n = 10, P = 0.169). Also, there was no significant change in the CFT in subjects with three injections of brolucizumab among three time points (n = 9); at the time of #1 injection and #2 injections (P = 0.953), #1 injection and #3 injections (P = 0.066), and #2 injections and #3 injections (P = 0.110), respectively. P value was obtained from paired T test or Wilcoxon signed rank test. in choroidal thickness. This study also confirmed that brolucizumab injections reduced choroidal thickness even in patients who were refractory to other anti-VEGF treatments. However, it is not clear whether this choroidal thickness reduction in nAMD is positive or negative for prognosis. Koizumi et al. reported that a decrease in choroidal thickness was related to improved visual outcomes after aflibercept injections10. In contrast, Sadda et al. showed that a thinner choroid was a risk factor for the development of macular atrophy11. Future studies are needed to determine the long-term effects of the changes in choroidal thickness after brolucizumab treatment.hh This study has some limitations. The study design is retrospective and included a small number of patients. Discussion In addition, the AMD types were heterogeneous and all the subjects were Korean; therefore, the results may not be generalizable to other racial or ethnic groups. However, the observation period was up to 8 months, which is meaningful as it is the longest follow-up among studies that investigate the changes in choroidal thickness after brolucizumab treatment. In addition, we consecutively recruited the subjects who were treated with broluci- zumab as a switching therapy. The enrolled patients had various AMD types and different treatment periods with other anti-VEGF agents. Therefore, this study showed real-world data from actual nAMD practice. In summary, choroidal thickness was significantly reduced after intravitreal brolucizumab injections as a switching treatment in patients with nAMD. Methodsh However, there was no significant change in the SFCT between the time of #1 injection and #2 injections in patients with two injections of brolucizumab (n = 10, P = 0.799). Also, there was no significant change in the SFCT in patients with three injections of brolucizumab among three time points (n = 9); at the time of #1 injection and #2 injections (P = 0.441), #1 injection and #3 injections (P = 0.953), and #2 injections and #3 injections (P = 0.110), respectively. P value was obtained from paired T test or Wilcoxon signed rank test. Figure 3. Temporal change in subfoveal choroidal thickness (SFCT) compared to baseline stratified by number of intravitreal brolucizumab injections. (A) Data were presented as the average. The mean SFCT of all patients significantly decreased compared to baseline at all three-timepoints; after one, two, and three injections (P < 0.05). (B) Eyes were divided into three groups according to the number of brolucizumab injections. The SFCT was significantly decreased after brolucizumab injection compared to the baseline in all three groups (P < 0.05). However, there was no significant change in the SFCT between the time of #1 injection and #2 injections in patients with two injections of brolucizumab (n = 10, P = 0.799). Also, there was no significant change in the SFCT in patients with three injections of brolucizumab among three time points (n = 9); at the time of #1 injection and #2 injections (P = 0.441), #1 injection and #3 injections (P = 0.953), and #2 injections and #3 injections (P = 0.110), respectively. P value was obtained from paired T test or Wilcoxon signed rank test. glaucoma, uveitis, and vitreoretinal disease, were also excluded. Previous ocular trauma or surgery, except for cataract extraction, were also exclusion criteria in this study. All patients had been previously treated with other anti-VEGF agents, such as bevacizumab, ranibizumab, and/or aflibercept. However, despite those previous fre- quent intravitreal injections, fluid accumulation persisted on spectral domain optical coherence tomography (SD-OCT), and another treatment option was needed. Therefore, anti-VEGF drug was switched to broluci- zumab. The baseline visit was regarded as the day of the initial intravitreal brolucizumab injection. Medical records and SD-OCT results were reviewed at baseline, at the time of each brolucizumab injection, one month after treatment, and at the final visit. glaucoma, uveitis, and vitreoretinal disease, were also excluded. Methodsh This retrospective study was approved by the Institutional Review Board of Hangil Eye Hospital and adhered to the tenets of the Declaration of Helsinki. The requirement to obtain informed consent from study participants was waived by the institutional review board given the retrospective nature of the study. Patients. This retrospective, observational, consecutive case series study enrolled patients between April 2021 and December 2021. A total of thirty-four eyes from 34 patients with nAMD were included. Patients with geographic atrophy or disciform scars in the macula were excluded. Eyes with other ocular diseases, such as Scientific Reports \| (2022) 12:19855 \| https://doi.org/10.1038/s41598-022-23392-6 www.nature.com/scientificreports/ Figure 3. Temporal change in subfoveal choroidal thickness (SFCT) compared to baseline stratified by number of intravitreal brolucizumab injections. (A) Data were presented as the average. The mean SFCT of all patients significantly decreased compared to baseline at all three-timepoints; after one, two, and three injections (P < 0.05). (B) Eyes were divided into three groups according to the number of brolucizumab injections. The SFCT was significantly decreased after brolucizumab injection compared to the baseline in all three groups (P < 0.05). However, there was no significant change in the SFCT between the time of #1 injection and #2 injections in patients with two injections of brolucizumab (n = 10, P = 0.799). Also, there was no significant change in the SFCT in patients with three injections of brolucizumab among three time points (n = 9); at the time of #1 injection and #2 injections (P = 0.441), #1 injection and #3 injections (P = 0.953), and #2 injections and #3 injections (P = 0.110), respectively. P value was obtained from paired T test or Wilcoxon signed rank test. Figure 3 Temporal change in subfoveal choroidal thickness (SFCT) compared to baseline stratified Figure 3. Temporal change in subfoveal choroidal thickness (SFCT) compared to baseline stratified by number of intravitreal brolucizumab injections. (A) Data were presented as the average. The mean SFCT of all patients significantly decreased compared to baseline at all three-timepoints; after one, two, and three injections (P < 0.05). (B) Eyes were divided into three groups according to the number of brolucizumab injections. The SFCT was significantly decreased after brolucizumab injection compared to the baseline in all three groups (P < 0.05). Methodsh Previous ocular trauma or surgery, except for cataract extraction, were also exclusion criteria in this study. All patients had been previously treated with other anti-VEGF agents, such as bevacizumab, ranibizumab, and/or aflibercept. However, despite those previous fre- quent intravitreal injections, fluid accumulation persisted on spectral domain optical coherence tomography (SD-OCT), and another treatment option was needed. Therefore, anti-VEGF drug was switched to broluci- zumab. The baseline visit was regarded as the day of the initial intravitreal brolucizumab injection. Medical records and SD-OCT results were reviewed at baseline, at the time of each brolucizumab injection, one month after treatment, and at the final visit. ti PCV was diagnosed based on the presence of polypoidal lesions, with or without branching vascular networks. Cases that exhibited retinal-retinal or retinal-choroidal anastomoses were classified as type 3 neovascularization (RAP)16,17. The remaining patients who were not diagnosed with either PCV or RAP were classified as having typical nAMD with type 1 or type 2 CNV18. Protocols for optical coherence tomography (OCT) scans. SD-OCT examination with macular thickness mapping was performed using a Spectralis OCT (Heidelberg Engineering, Heidelberg, Germany). A central volume scan with a 25-scan pattern and macular thickness map protocol was performed. CFT was obtained from the central 1 mm subfield in the macular thickness map presented by the OCT machine. SFCT, defined as the perpendicular distance from Bruch’s membrane to the sclera-choroidal junction at the subfovea, was measured manually using a caliper provided by the OCT machine. The average thickness of the horizontal and vertical scans was defined as the SFCT19. Intravitreal injection methods. All injections were performed on the day of the OCT examination. Under topical anesthesia with proparacaine (0.5%) eye drops, the bulbar conjunctiva and fornices were rinsed with 5% povidone-iodine, followed by the application of a sterile drape and lid speculum. After a drop of 5% povidone-iodine was applied, 6 mg (0.05 mL) of brolucizumab was injected into the pars plana. The needle was https://doi.org/10.1038/s41598-022-23392-6 Scientific Reports \| (2022) 12:19855 \| www.nature.com/scientificreports/ www.nature.com/scientificreports/ Figure 4. A representative case of a 69 year old male with polypoidal choroidal vasculopathy. The patient had previously received more than 10 treatments with ranibizumab and aflibercept, however macular exudation was persistent and treatment agent was switched to brolucizumab. (A) Fluorescein angiography and (B) indocyanine green angiography showing a subfoveal polyp with a branching vascular network. www.nature.com/scientificreports/ removed carefully, and the injection site was compressed using a sterile cotton applicator to prevent reflux. Anti- biotic eye drops were applied four times per day for one week after injection. removed carefully, and the injection site was compressed using a sterile cotton applicator to prevent reflux. Anti- biotic eye drops were applied four times per day for one week after injection. Data analysis. Statistical analyses were performed using a commercially available software package (IBM SPSS Statistics 25.0; SPSS Inc.,Chicago, IL, USA). Data were analyzed by dividing them into continuous and descriptive variables. Quantitative variables were presented as mean ± standard deviation, and qualitative vari- ables were reported as absolute frequencies and percentages. Temporal changes in BCVA and retinal thickness were also evaluated. Statistical significance was defined as P < 0.05. Data availability Th d l The data are not available for public access because of patient privacy concerns, but are available from the cor- responding author upon reasonable request. Received: 22 July 2022; Accepted: 31 October 2022 Received: 22 July 2022; Accepted: 31 October 2022 Acknowledgementshi g The authors declare no financial support for this study. The authors declare no financial support for this study. References https://doi.org/10.1136/bjophthalmol-2014-306432 (2015).t g g p p g 7. Koizumi, H. et al. Aflibercept therapy for polypoidal choroidal vasculopathy: Short-term results of a multicentre study. Br. J Ophthalmol. 99, 1284–1288. https://doi.org/10.1136/bjophthalmol-2014-306432 (2015).t p , p g j p ( ) 8. Yamazaki, T., Koizumi, H., Yamagishi, T. & Kinoshita, S. Subfoveal choroidal thickness after ranibizumab therapy for neovascular age-related macular degeneration: 12-month results. Ophthalmology 119, 1621–1627. https://doi.org/10.1016/j.ophtha.2012.02. 022 (2012).l p p g j p ( ) 8. Yamazaki, T., Koizumi, H., Yamagishi, T. & Kinoshita, S. Subfoveal choroidal thickness after ranibizumab therapy for neovascular age-related macular degeneration: 12-month results. Ophthalmology 119, 1621–1627. https://doi.org/10.1016/j.ophtha.2012.02. 022 (2012).l 9. Yun, C. et al. Comparison of intravitreal aflibercept and ranibizumab injections on subfoveal and peripapillary choroidal thickness in eyes with neovascular age-related macular degeneration. Graefes Arch. Clin. Exp. Ophthalmol. 254, 1693–1702. https://doi.org/ 10.1007/s00417-015-3260-3 (2016).l 9. Yun, C. et al. Comparison of intravitreal aflibercept and ranibizumab injections on subfoveal and peripapillary choroidal thickness in eyes with neovascular age-related macular degeneration. Graefes Arch. Clin. Exp. Ophthalmol. 254, 1693–1702. https://doi.org/ 10.1007/s00417-015-3260-3 (2016). ( ) 10. Koizumi, H. et al. Subfoveal choroidal thickness during Aflibercept therapy for neovascular age-related macular degeneration: Twelve-month results. Ophthalmology 123, 617–624. https://doi.org/10.1016/j.ophtha.2015.10.039 (2016). 10. Koizumi, H. et al. Subfoveal choroidal thickness during Aflibercept therapy for neovascular age-related macular degeneration: Twelve-month results. Ophthalmology 123, 617–624. https://doi.org/10.1016/j.ophtha.2015.10.039 (2016). 1. Sadda, S. R. et al. Spectral-domain OCT analysis of risk factors for macular atrophy development in the HARBOR study for neo- vascular age-related macular degeneration. Ophthalmology 127, 1360–1370. https://doi.org/10.1016/j.ophtha.2020.03.031 (2020) g g p gy p g j p 12. Tamashiro, T. et al. Subfoveal choroidal thickness after brolucizumab therapy for neovascular age-related macular degeneration: A short-term multicenter study. Graefes Arch. Clin. Exp. Ophthalmol. https://doi.org/10.1007/s00417-021-05517-1 (2022). 2. Tamashiro, T. et al. Subfoveal choroidal thickness after brolucizumab therapy for neovascular age-related macular degeneration A short-term multicenter study. Graefes Arch. Clin. Exp. Ophthalmol. https://doi.org/10.1007/s00417-021-05517-1 (2022).l y f g 13. Fukuda, Y. et al. Comparison of outcomes between 3 monthly Brolucizumab and Aflibercept injections for polypoidal chor vasculopathy. Biomedicines https://doi.org/10.3390/biomedicines9091164 (2021). 14. Matsumoto, H., Hoshino, J., Mukai, R., Nakamura, K. & Akiyama, H. Short-term outcomes of intravitreal brolucizumab for treatment-naive neovascular age-related macular degeneration with type 1 choroidal neovascularization including polypoidal choroidal vasculopathy. Sci. Rep. 11, 6759. https://doi.org/10.1038/s41598-021-86014-7 (2021).l p y p p g ( ) 15. Ota, H. et al. References Switching from aflibercept to brolucizumab for the treatment of refractory neovascular age-related macular de eration. Jpn. J. Ophthalmol. https://doi.org/10.1007/s10384-022-00908-1 (2022).t p p p g 16. Lee, K., Park, Y. G. & Park, Y. H. Visual prognosis after pneumatic displacement of submacular hemorrhage according to age-related macular degeneration subtypes. Retina 40, 2304–2311. https://doi.org/10.1097/IAE.0000000000002762 (2020).f 17. Lee, J. H., Kim, J. H., Kim, J. W., Kim, C. G. & Lee, D. W. Difference in treatment burden of neovascular age-related macular degeneration among different types of neovascularization. Graefes Arch. Clin. Exp. Ophthalmol. 259, 1821–1830. https://doi.org/ 10.1007/s00417-020-05028-5 (2021). ( ) 18. Hwang, D. D. et al. Distinguishing retinal angiomatous proliferation from polypoidal choroidal vasculopathy with a deep neural network based on optical coherence tomography. Sci. Rep. 11, 9275. https://doi.org/10.1038/s41598-021-88543-7 (2021).f p g p y p p g ( ) 19. Yamashita, T. et al. Repeatability and reproducibility of subfoveal choroidal thickness in normal eyes of Japanese using dif SD-OCT devices. Invest. Ophthalmol. Vis. Sci. 53, 1102–1107. https://doi.org/10.1167/iovs.11-8836 (2012). Acknowledgementshi References 1. Lim, L. S., Mitchell, P., Seddon, J. M., Holz, F. G. & Wong, T. Y. Age-related macular degeneration. Lancet 379, 1728–1738. https:// doi.org/10.1016/S0140-6736(12)60282-7 (2012). g ( ) ( ) 2. Schmidt-Erfurth, U. et al. Guidelines for the management of neovascular age-related macular degeneration by the European society of retina specialists (EURETINA). Br. J. Ophthalmol. 98, 1144–1167. https://doi.org/10.1136/bjophthalmol-2014-305702 (2014). g 2. Schmidt-Erfurth, U. et al. Guidelines for the management of neovascular age-related macular degeneration by the European society f l (EURETINA) B J O h h l l 6 h //d / 6/b h h l l ( ) 2. Schmidt-Erfurth, U. et al. Guidelines for the management of neovascular age-related macular degeneration by the European society of retina specialists (EURETINA). Br. J. Ophthalmol. 98, 1144–1167. https://doi.org/10.1136/bjophthalmol-2014-305702 (2014). 2. Schmidt-Erfurth, U. et al. Guidelines for the management of neovascular age-related macular degeneration by the European society of retina specialists (EURETINA). Br. J. Ophthalmol. 98, 1144–1167. https://doi.org/10.1136/bjophthalmol-2014-305702 (2014). 3 Dugel P U et al HAWK and HARRIER: Phase 3 multicenter randomized double-masked trials of brolucizumab for neovascular of retina specialists (EURETINA). Br. J. Ophthalmol. 98, 1144–1167. https://doi.org/10.1136/bjophthalmol-2014-305702 (2014) 3. Dugel, P. U. et al. HAWK and HARRIER: Phase 3, multicenter, randomized, double-masked trials of brolucizumab for neovascula age-related macular degeneration. Ophthalmology 127, 72–84. https://doi.org/10.1016/j.ophtha.2019.04.017 (2020).l 3. Dugel, P. U. et al. HAWK and HARRIER: Phase 3, multicenter, randomized, double-masked trials of brolucizumab for neovascul age-related macular degeneration. Ophthalmology 127, 72–84. https://doi.org/10.1016/j.ophtha.2019.04.017 (2020).l g g p gy p g j p 4. Dugel, P. U. et al. Brolucizumab versus Aflibercept in participants with neovascular age-related macular degeneration: A rand omized trial. Ophthalmology 124, 1296–1304. https://doi.org/10.1016/j.ophtha.2017.03.057 (2017). p gy p g j p ( ) 5. Sharma, A. et al. Brolucizumab-early real-world experience: BREW study. Eye (Lond) 35, 1045–1047. https://doi.org/10.1038/ s41433-020-1111-x (2021). p gy g j 5. Sharma, A. et al. Brolucizumab-early real-world experience: BREW study. Eye (Lond) 35, 1045–1047. https://doi.org/10.1038/ s41433-020-1111-x (2021). l ll l l l b h l 6. Avaylon, J., Lee, S. & Gallemore, R. P. Case series on initial responses to intravitreal Brolucizumab in patients with recalcitrant chronic wet age-related macular degeneration. Int. Med. Case Rep. J. 13, 145–152. https://doi.org/10.2147/IMCRJ.S252260 (2020) g g p J , p g J ( ) 7. Koizumi, H. et al. Aflibercept therapy for polypoidal choroidal vasculopathy: Short-term results of a multicentre study. Br. J. Ophthalmol. 99, 1284–1288. Methodsh (C) Optical coherence tomography (OCT) at baseline showing fibrovascular retinal pigment epithelium detachment accompanied with intra and subretinal fluid. The white arrows indicate the inner scleral border and yellow arrows demonstrate the subfoveal choroidal thickness (SFCT). Central foveal thickness (CFT) was 333 μm and SFCT was 244 μm. (D) One month after the first brolucizumab treatment, CFT and SFCT were decreased to 253 and 187 μm, respectively. Complete fluid resolution was presented in the OCT image. Best corrected visual acuity (BCVA) also improved from 0.5 to 0.2 logarithm of the minimal angle of resolution (logMAR). No adverse reactions related with brolucizumab injection were reported. (E) Three months after first injection, intra and subretinal fluid was found in OCT scans and an additional injection was performed. CFT and SFCT were 317 and 196 μm and both were still reduced compared to baseline. (F) At last visit, six months after baseline, mild intraretinal Figure 4. A representative case of a 69 year old male with polypoidal choroidal vasculopathy. The patient had previously received more than 10 treatments with ranibizumab and aflibercept, however macular exudation was persistent and treatment agent was switched to brolucizumab. (A) Fluorescein angiography and (B) indocyanine green angiography showing a subfoveal polyp with a branching vascular network. (C) Optical coherence tomography (OCT) at baseline showing fibrovascular retinal pigment epithelium detachment accompanied with intra and subretinal fluid. The white arrows indicate the inner scleral border and yellow arrows demonstrate the subfoveal choroidal thickness (SFCT). Central foveal thickness (CFT) was 333 μm and SFCT was 244 μm. (D) One month after the first brolucizumab treatment, CFT and SFCT were decreased to 253 and 187 μm, respectively. Complete fluid resolution was presented in the OCT image. Best corrected visual acuity (BCVA) also improved from 0.5 to 0.2 logarithm of the minimal angle of resolution (logMAR). No adverse reactions related with brolucizumab injection were reported. (E) Three months after first injection, intra and subretinal fluid was found in OCT scans and an additional injection was performed. CFT and SFCT were 317 and 196 μm and both were still reduced compared to baseline. (F) At last visit, six months after baseline, mild intraretinal fluid was remained and a third brolucizumab injection was administrated. BCVA was 0.15 logMAR. CFT and SFCT were 300 and 204 μm, respectively. https://doi.org/10.1038/s41598-022-23392-6 Scientific Reports \| (2022) 12:19855 \| www.nature.com/scientificreports/ Author contributions D.D.H. conceived and designed the study. D.D.H. and K.W.B. prepared the manuscript. D. D. H. collected the data, and K.W.B. and D.I.K. verified the data. D.D.H. and K.W.B. performed statistical analyses. All authors reviewed the manuscript prior to submission. Competing interests The authors declare no competing interests. Additional information Correspondence and requests for materials should be addressed to D.D.-J.H. Reprints and permissions information is available at www.nature.com/reprints. © The Author(s) 2022 Competing interests h p g The authors declare no competing interests. Reprints and permissions information is available at www.nature.com/reprints. Scientific Reports \| (2022) 12:19855 \| https://doi.org/10.1038/s41598-022-23392-6 www.nature.com/scientificreports/ Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. © The Author(s) 2022 https://doi.org/10.1038/s41598-022-23392-6 Scientific Reports \| (2022) 12:19855 \|
https://openalex.org/W2982794273	http://pdf.blucher.com.br/designproceedings/9cidi/4.0236.pdf	Portuguese	null	A cultura do design aplicada ao desenvolvimento de negócios: diagnóstico da Casa da Alfândega	Blucher Design Proceedings	2,019	cc-by	7,019	A cultura do design aplicada ao desenvolvimento de negócios: diagnóstico da Casa da Alfândega The culture of design applied to business development: diagnosis of Casa da Alfândega Diego Borges da Silva & Ricardo Triska design, gestão de design, design de serviço, modelo de negócios, Casa da Alfândega Este estudo fornece indícios para a aplicação do design de serviço e de suas diretrizes metodológicas a partir de diagnóstico da Casa da Alfândega de Santa Catarina, organização que promove a cultural e a manutenção das tradições, lendas e histórias desse estado. Para atingir seu objetivo, primeiro foram apresentadas as contribuições teóricas e metodológicas e, num segundo momento, a aplicação do estudo de caso. O uso da metodologia GODP serviu para organizar e oferecer uma sequência de ações que permitiram com que o design fosse concebido de forma consciente, respondendo de forma assertiva e consistente aos objetivos fixados para a prática projetual. Já o uso da ferramenta Business Model Canvas facilitou a construção de uma nova tela de modelo de negócios. Os resultados obtidos permitiram concluir que a cultura do design agrega valor ao serviço em parceria com o modelo de negócios. Assim, foi proposto um plano de intervenção para a Casa da Alfândega, com recomendações de otimização. design, design management, service design, business model, Casa da Alfândega This study provides indications for the application of the service design and its methodological guidelines from the diagnosis of Casa da Alfândega of Santa Catarina, an organization that promotes cultural and maintenance of the traditions, legends and histories of this state. To reach its objective, first the theoretical and methodological contributions were presented and, in a second moment, the application of the case study. The use of the GODP methodology served to organize and offer a sequence of actions that allowed the design to be consciously designed, responding assertively and consistently to the objectives set for the design practice. Already the use of the Business Model Canvas tool facilitated the construction of a new business model screen. The results obtained allowed us to conclude that the design culture adds value to the service in partnership with the business model. Thus, an intervention plan for the Casa da Alfândega was proposed, with optimization recommendations. 9o CONGIC 9th Information Design Student Conference 9o CIDI 9th Information Design International Conference 9o CIDI 9th Information Design International Conference 2 Design de Serviço para negócios Existe uma relação de proximidade entre design de serviço e modelo de negócios. Partindo desse pressuposto, Mager (2015) defende que design de serviço é essencial para todas as organizações que querem inovar e melhorar suas estratégias de serviço e a experiência do usuário. Já para a Osterwalter (2010), sempre que uma organização projeta um novo produto, serviço ou experiência, torna-se essencial desenhar seu modelo de negócio. O surgimento de modelo de negócios, baseados em serviços, transformou as relações de trabalho no mundo. Essa transformação foi possibilitada pelas novas tecnologias de informação e comunicação, pela especialização das organizações e dos profissionais, pelas regulamentações globais e pelo aumento do uso do serviço externo pelas entidades em escalas múltiplas. (Girotra & Netessine, 2013, p.8). Ao longo dos últimos duzentos anos, principalmente nas últimas décadas, presencia- se um aumento e queda das interações dedicadas à produção local de bens, com mais confiança em interações cognitivas e sociais cada vez mais complexas com outros (Maglio & Spohrer, 2013, p.667). Isso representa a ascensão do chamado “setor de serviço” da economia. Assim, o pensamento voltado ao design entrou no centro como inovação estratégica, em que apresenta uma estrutura para o pensamento de design que é relevante para a gestão de negócios e estratégias de design. Por isso, Brown (2010, p.3), indaga sobre a possibilidade de pensar em uma nova abordagem, que realmente contribua para uma mudança massiva na forma como organizações e usuários pensam e agem para solucionar os problemas. Na visão de Brown (2010, p.30), o usuário compra, usa e depois descarta. Entrega valor uma ou duas vezes, porém apenas quando tem contato com aquele serviço. E, nessa lógica de serviço, essa entrega de valor é pelo tempo. Ou seja, quanto tempo o usuário está interagindo com aquele serviço. O contato com o serviço muda e por isso se faz necessária uma reflexão sobre a temporalidade na lógica de serviço e o que acontece. Quando existe a lógica de serviço, é preciso entregar um produto para alguém com background que está acompanhando o mesmo serviço. Poderia ter apenas uma pessoa e o serviço estar ruim. O serviço permite atender mais usuários, com o mesmo produto, ao mesmo tempo e, por meio dele, podem surgir novas oportunidades de se criar um novo modelo de negócio. Silva, D. B. & Triska, R. \| A cultura do design aplicada ao desenvolvimento de negócios O objeto deste estudo foi a Casa da Alfândega de Santa Catarina, que desde 1988 abriga o projeto Galeria do Artesanato, que visa dar continuidade ao artesanato de Santa Catarina por meio da comercialização e da preservação da cultura e valorização do saber fazer do povo catarinense. Por fim, foram apresentados os resultados do diagnóstico da Casa da Alfândega, etapa realizada em 18 meses, que culminou com a apresentação do Plano de Intervenção, com orientações e diretrizes para a realização de ações internas e externas visando a melhoria do serviço e da gestão da Casa da Alfândega. Anais do 9º Congresso Internacional de Design da Informação \| CIDI 2019 Proceedings of the 9th Information Design International Conference Anais do 9o Congresso Nacional de Iniciação Científica em Design da Informação \| CONGIC 2019 Proceedings of the 9th Information Design Student Conference 1 Introdução É comum que o serviço seja visto ou elaborado como elemento essencial em uma estratégia. Os negócios de menor porte, que por definição estão mais vulneráveis nesse ambiente, podem utilizar o serviço ao cliente como uma ferramenta indispensável de diferenciação. Frequentemente, nessas organizações o serviço é uma alternativa que apresenta os menores custos de implementação, aliados a resultados de desempenho. É comum que o serviço seja visto ou elaborado como elemento essencial em uma estratégia. Os negócios de menor porte, que por definição estão mais vulneráveis nesse ambiente, podem utilizar o serviço ao cliente como uma ferramenta indispensável de diferenciação. Frequentemente, nessas organizações o serviço é uma alternativa que apresenta os menores custos de implementação, aliados a resultados de desempenho. Nessa dinâmica de relacionamento entre organização e usuário, a aplicação do design de serviço e de suas diretrizes metodológicas orienta o processo de desenvolvimento de um novo serviço alinhado à gestão, porém, com entregas que se caracterizam por inovar e agregar mais valor às organizações. Um projeto de design de serviço coloca o usuário no centro de qualquer intervenção, e entender o usuário e projetar uma melhor experiência passa a ser aspecto central em qualquer projeto de serviço. Nesse sentido, este estudo pretendeu discutir o papel do design de serviço em uma organização artesanal, visto que o design surge como uma abordagem centrada no usuário, a partir de uma visão multidisciplinar, de colaboração e do pensamento tangível, que gera caminhos para soluções inovadoras. Anais do 9º CIDI e 9º CONGIC Luciane Maria Fadel, Carla Spinillo, Anderson Horta, Cristina Portugal (orgs.) Sociedade Brasileira de Design da Informação – SBDI Belo Horizonte \| Brasil \| 2019 ISBN 978-85-212-1728-2 Proceedings of the 9th CIDI and 9th CONGIC Luciane Maria Fadel, Carla Spinillo, Anderson Horta, Cristina Portugal (orgs.) Sociedade Brasileira de Design da Informação – SBDI Belo Horizonte \| Brazil \| 2019 ISBN 978-85-212-1728-2 Silva, D. B. & Triska, R. \| A cultura do design aplicada ao desenvolvimento de negócios Silva, D. B. & Triska, R. \| A cultura do design aplicada ao desenvolvimento de negócios Silva, D. B. & Triska, R. \| A cultura do design aplicada ao desenvolvimento de negócios A difusão de modelo de negócios recentemente desenvolvidos e altamente relacionados com serviço pode ser especialmente analisada em ambientes onde muitas organizações estão prestes a mudar para novos modelos de negócios, como a contratação de serviço completo. E, para Kesting e Günzel-Jensen (2015, p.288), isso afeta toda a organização, visto que é mais provável que uma empresa não possa oferecer uma proposta de valor voltada ao design de serviço enquanto seus processos internos permanecem padronizados e não ajustados. Neste caso, organizações que colocam os usuários em primeiro plano, estão construindo novos negócios alinhados com o mundo contemporâneo, sem implicar em negócios menos lucrativos. Por outro lado, organizações que estão adotando essa visão a partir do design, estão atingindo níveis de performance e rentabilidade muito superiores às organizações tradicionais. Além disso, organizações que aplicam o design para definir e conduzir seus negócios estão conseguindo inovar de modo mais eficiente e efetivo, tornando-se organizações mais inovadoras. Em termos práticos, usar o design para compreender as necessidades dos usuários, bem como para repensar problemas complexos, leva a insights que representam vantagem competitiva. Essa visão é fundamental para qualquer organização que queira ser inovadora, por isso, tem-se como objeto deste estudo a Casa da Alfândega de Santa Catarina. Anais do 9º Congresso Internacional de Design da Informação \| CIDI 2019 Proceedings of the 9th Information Design International Conference Anais do 9o Congresso Nacional de Iniciação Científica em Design da Informação \| CONGIC 2019 Proceedings of the 9th Information Design Student Conference 2 Design de Serviço para negócios Por exemplo, o Netflix mudou a forma como se consome entretenimento; o Uber, como se usa transporte; e o Airbnb, como os usuários se hospedam quando viajam. Ou seja, novas possibilidades de se consumir, interagir e se conectar com o serviço ou produto. Em menos de uma década não existiam espaços de coworking. Hoje, saímos da cultura o que é meu é meu, para um hub de inovação, sendo um espaço para novos negócios. A partir disso, se veem emergir novos modelos de negócios, que podem ser considerados como novos arranjos sociais, em que as pessoas se unem para suprir uma necessidade. Ainda, dá para ir mais além, como passar de modelos de eficiência, como o Fleety – primeira rede de carros compartilhados do Brasil – para modelos mais disruptivos. Então, passa de produtos e tecnologia para soluções e essa reorganização social faz emergir o design de serviço. 1801 3 Metodologia GODP Nesta segunda etapa do estudo é apresentada uma breve fundamentação da metodologia GODP – Guia de Orientação para Desenvolvimento de Projetos (Figura 1), proposta por Merino (2016) e o NGD – Núcleo de Gestão de Design, da UFSC. Anais do 9º Congresso Internacional de Design da Informação \| CIDI 2019 Proceedings of the 9th Information Design International Conference Anais do 9o Congresso Nacional de Iniciação Científica em Design da Informação \| CONGIC 2019 Proceedings of the 9th Information Design Student Conference 1802 Figura 1: Etapas do GODP (Merino, 2016). A metodologia GODP possibilitou o desenvolvimento do projeto organizado em três momentos: Inspiração (-1/0/1), Ideação (2/3) e Implementação (4/5/6). Estes momentos ocorreram de maneira sequencial e foram detalhados conforme as etapas ocorriam. Entretanto, como primeiro passo da prática projetual, Merino (2016, p.8) orienta identificar os Blocos de Referência (Figura 2) que auxiliam a organizar e a escolher a ferramenta para cada bloco: produto, usuário e contexto. A autora justifica esse passo, sendo “[...] um desafio que é o grande volume de informações, considerando que projetamos algo (produto), para alguém (usuário) em algum lugar (contexto)”. Figura 1: Etapas do GODP (Merino, 2016). Figura 1: Etapas do GODP (Merino, 2016). A metodologia GODP possibilitou o desenvolvimento do projeto organizado em três momentos: Inspiração (-1/0/1), Ideação (2/3) e Implementação (4/5/6). Estes momentos ocorreram de maneira sequencial e foram detalhados conforme as etapas ocorriam. Entretanto, como primeiro passo da prática projetual, Merino (2016, p.8) orienta identificar os Blocos de Referência (Figura 2) que auxiliam a organizar e a escolher a ferramenta para cada bloco: produto, usuário e contexto. A autora justifica esse passo, sendo “[...] um desafio que é o grande volume de informações, considerando que projetamos algo (produto), para alguém (usuário) em algum lugar (contexto)”. 1802 Silva, D. B. & Triska, R. \| A cultura do design aplicada ao desenvolvimento de negócios Silva, D. B. & Triska, R. \| A cultura do design aplicada ao desenvolvimento de negócios Silva, D. B. & Triska, R. \| A cultura do design aplicada ao desenvolvimento de negócios Figura 2: Bloco de Referência (Merino, 2016, p.20). A seguir, serão apresentados os resultados obtidos por meio do estudo de caso. Figura 2: Bloco de Referência (Merino, 2016, p.20). Figura 2: Bloco de Referência (Merino, 2016, p.20). A seguir, serão apresentados os resultados obtidos por meio do estudo de caso. 4 Apresentação dos resultados Esta seção se refere à execução do estudo aplicado, direcionado por um diagnóstico da Casa da Alfândega de Santa Catarina. As primeiras intervenções aconteceram em setembro de 2015 e duraram 18 meses, até a concepção do Plano de Intervenção. A seguir, são apresentados os resultados, por meio da metodologia GODP, que se apresenta em três momentos: (1) inspiração; (2) ideação; (3) implementação. Anais do 9º Congresso Internacional de Design da Informação \| CIDI 2019 g g Anais do 9o Congresso Nacional de Iniciação Científica em Design da Informação \| CONGIC 2019 h g ç Proceedings of the 9th Information Design Student Conference Anais do 9o Congresso Nacional de Iniciação Científica em Design da Informação \| CONGIC 2019 Proceedings of the 9th Information Design Student Conference Anais do 9º Congresso Internacional de Design da Informação \| CIDI 2019 Proceedings of the 9th Information Design International Conference Anais do 9o Congresso Nacional de Iniciação Científica em Design da Informação \| CONGIC 201 g g ç eedings of the 9th Information Design International Conferen ais do 9o Congresso Nacional de Iniciação Científica em Des Momento inspiração Neste momento foi realizado um percurso configurado em três etapas (tabela 1): oportunidades, prospecção e levantamento de dados. Essa etapa se caracterizou por entender a cultura e a organização sob a perspectiva do usuário, identificando o real problema de design que precisava ser resolvido. Tabela 1: Momento inspiração. ETAPA AÇÃO REALIZADA (-1) Oportunidades Um primeiro contato com a FCC, responsável pela administração da Casa da Alfândega, em que foram verificadas oportunidades para se oferecer ao usuário uma experiência cultural da diversidade criativa do artesanato catarinense. (0) Prospecção Contou-se com o apoio da administração da Casa da Alfândega, a fim de levantar dados preliminares e, assim, definir a problemática que iria nortear este estudo. (1) Levantamento de dados Foram coletadas informações a partir da realização de visitas a campo, entre setembro de 2015 e fevereiro de 2017, com observações assistemáticas, entrevistas com colaboradores e registros fotográficos. Nesta etapa, foram aplicados os Blocos de Referência, para se chegar ao diagnóstico, por meio do produto, do usuário e do contexto. A tabela demonstrou as fases definidas para o momento inspiração (-1, 0, 1), em que se propuseram ações de melhoria na Casa da Alfândega, a partir da aproximação das ferramentas de design e de gestão. Segundo Mozota, Klöpsch e Costa (2009, p.39), o design e a gestão são voltados à resolução de problemas, à criação, a atividades coordenadoras e sistêmicas. O design consiste em pensar e pesquisar a coerência do sistema de objetos. E cada problema percebido implica em descobrir um equilíbrio entre toda uma gama de serviços. 1803 Para um melhor entendimento da metodologia utilizada, foram definidos os Blocos de Referência, sendo que o bloco produto analisou o artesanato; o bloco usuário, o artesão; Silva, D. B. & Triska, R. \| A cultura do design aplicada ao desenvolvimento de negócios Silva, D. B. & Triska, R. \| A cultura do design aplicada ao desenvolvimento de negócios e o bloco contexto, a Casa da Alfândega. e o bloco contexto, a Casa da Alfândega. Bloco produto: artesanato Por ser um espaço que se destina à comercialização de produtos artesanais, a maioria dos produtos são classificados como artesanato de referência cultural, por fazer o “resgate ou releitura de elementos culturais tradicionais da região onde é produzido”. (PAB, 2012, p.29). Bloco usuário: artesão Tendo em vista que este diagnóstico se propôs a verificar as fragilidades da Casa da Alfândega, optou-se aqui por mensurar o usuário sob o ponto de vista do artesão e não do cliente. O PAB (2012, p.11) entende por artesão “o trabalhador que de forma individual exerce um ofício manual, transformando matéria-prima bruta ou manufaturada em produto acabado”. Até fevereiro de 2017 a Casa da Alfândega contava com 108 artesãos no quadro de comercialização de produtos artesanais. Quanto ao gênero, 75 eram mulheres e 32 homens. Quanto ao fator geográfico (Figura 3), Santa Catarina tem 292 municípios e apenas 17 cidades estão representadas no espaço, somando apenas 6,2% do total de municípios. Figura 3: Distribuição geográfica. Figura 3: Distribuição geográfica. Embora pareça um número relativamente inexpressivo, essas 17 cidades são responsáveis por grande parte do Produto Interno Bruto (PIB) no setor terciário, segundo dados do SEBRAE/SC (2013). A maioria é considerada polo, com grande movimentação de renda e turismo. Artesãos também colaboram com a organização da Casa, onde se revezam para demonstrar suas mais variadas técnicas, permitindo que o usuário conheça de perto a elaboração e confecção dos produtos, bem como ouvir histórias do folclore catarinense. Anais do 9º Congresso Internacional de Design da Informação \| CIDI 2019 Proceedings of the 9th Information Design International Conference Anais do 9o Congresso Nacional de Iniciação Científica em Design da Informação \| CONGI Proceedings of the 9th Information Design Student Conference g ç Proceedings of the 9th Information Design Student Conference Bloco contexto: Casa da Alfândega Desde 1988, está instalado no prédio da Alfândega o projeto Galeria do Artesanato, que tem como objetivo efetuar o resgate da cultura de Santa Catarina por meio da Silva, D. B. & Triska, R. \| A cultura do design aplicada ao desenvolvimento de negócios comercialização do artesanato, além da preservação, do resgate e da continuidade do artesanato local, sendo um ponto de disseminação cultural. Assim, uma das preocupações manifestadas pela Fundação Catarinense de Cultura é: Melhor qualificar o bem cultural produzido, sem esvaziá-lo de seu sentido original (simbólico), que é a razão fundamental do mesmo existir. Melhor qualificar no sentido de oferecer ao visitante, e possível comprador, a experiência cultural da diversidade criativa, ainda que manifestada sob um mesmo significado. Falamos então de sintaxes diferentes para uma mesma semântica. (Santa Catarina, 2015). Com uma área de 1.300 m2, construído em dois andares, o prédio da Casa da Alfândega é dividido em quatro áreas (Figura 4): área 1, área 2, área 3 e área 4. Figura 4: Planta baixa do prédio da Alfândega. A área 1 (pavimento térreo) está cedida para o projeto Galeria do Artesanato, local onde são comercializados os produtos artesanais. Esse espaço também conta com o recurso de um mezanino, que é utilizado pela área administrativa da Casa. A área 2 é destinada ao recebimento das mercadorias e, esse local, serve de espaço para a triagem das peças que são entregues, a cada 14 dias, conforme normas internas, além de espaço para dois banheiros (masculino e feminino) e uma área para refeitório. A área 3 no momento está desativada. E na área 4 (pavimento superior), fica instalada a 11ª Coordenadoria Regional do Instituto do Patrimônio Histórico e Artístico Nacional – IPHAN/SC. Figura 4: Planta baixa do prédio da Alfândega. A área 1 (pavimento térreo) está cedida para o projeto Galeria do Artesanato, local onde são comercializados os produtos artesanais. Esse espaço também conta com o recurso de um mezanino, que é utilizado pela área administrativa da Casa. A área 2 é destinada ao recebimento das mercadorias e, esse local, serve de espaço para a triagem das peças que são entregues, a cada 14 dias, conforme normas internas, além de espaço para dois banheiros (masculino e feminino) e uma área para refeitório. A área 3 no momento está desativada. Anais do 9º Congresso Internacional de Design da Informação \| CIDI 2019 Proceedings of the 9th Information Design International Conference Anais do 9o Congresso Nacional de Iniciação Científica em Design da Informação \| CONGIC 2019 Proceedings of the 9th Information Design Student Conference Anais do 9º Congresso Internacional de Design da Informação \| CIDI 2019 Proceedings of the 9th Information Design International Conference Bloco contexto: Casa da Alfândega E na área 4 (pavimento superior), fica instalada a 11ª Coordenadoria Regional do Instituto do Patrimônio Histórico e Artístico Nacional – IPHAN/SC. Por se tratar de uma construção tombada pelo IPHAN, desde 1975, os administradores da Casa da Alfândega devem cumprir o que estabelece a Portaria nº 420/2010, expedida em 24 de dezembro de 2010, que dispõe sobre os procedimentos a serem observados para a concessão de autorização para realização de intervenções em bens edificados tombados e nas respectivas áreas de entorno. Esse processo torna mais burocrática qualquer intervenção na parte interna ou externa da construção, limitando, por exemplo, o uso de placas com identidade visual, sinalização e acessibilidade. A organização do espaço interno da Casa da Alfândega também foi considerada neste diagnóstico. Se, por um lado, o volume de informações visuais nesse espaço causa um frenesi sensorial, por outro, provoca dificuldade em apreciar os produtos que estão sendo ofertados. Desse modo, torna-se cansativo aos sentidos, desinteressante para a cognição e pouco atrativo ao desejo. Assim, chega-se ao final deste diagnóstico com a apresentação de uma síntese (Figura 5) dos itens abordados nos blocos de referência e que serviram de base para a execução desta etapa de diagnóstico. 1805 Silva, D. B. & Triska, R. \| A cultura do design aplicada ao desenvolvimento de negócios Figura 5: Diagnóstico preliminar. Figura 5: Diagnóstico preliminar. Foi possível perceber que produtos e serviços possuem deficiência. Assim, como parte deste momento de inspiração, verificou-se a necessidade de desenvolver um Plano de Intervenção que esteja alinhado com o planejamento estratégico da organização, a fim de priorizar os objetivos e delimitar os fatores passíveis de ação do design de serviço. Figura 5: Diagnóstico preliminar. Foi possível perceber que produtos e serviços possuem deficiência. Assim, como parte deste momento de inspiração, verificou-se a necessidade de desenvolver um Plano de Intervenção que esteja alinhado com o planejamento estratégico da organização, a fim de priorizar os objetivos e delimitar os fatores passíveis de ação do design de serviço. Anais do 9º Congresso Internacional de Design da Informação \| CIDI 2019 Proceedings of the 9th Information Design International Conference Anais do 9o Congresso Nacional de Iniciação Científica em Design da Informação \| CONGIC 2019 Proceedings of the 9th Information Design Student Conference Momento ideação Definido o diagnóstico, vem a parte de gerar ideias e conceitos para o serviço. É nesta fase que são usadas ferramentas simples e rápidas para visualizar processos. O design de serviço é iterativo não apenas durante o processo das etapas apresentadas, mas também dentro de cada etapa. (Stickdorn & Schneider, 2014, p.132). Nesta etapa “podem ser utilizadas técnicas analíticas que permitirão definir as estratégias do projeto”. De posse dessas estratégias, “são definidos os conceitos globais do projeto, sendo geradas alternativas preliminares”. (Merino, 2016, p.14). As ações realizadas neste momento foram apresentadas a seguir (Tabela 2). 1806 Silva, D. B. & Triska, R. \| A cultura do design aplicada ao desenvolvimento de negócios Tabela 2: Momento Ideação ETAPA AÇÃO REALIZADA (2) Organização Os dados coletados foram organizados e analisados, identificando oportunidades e, principalmente, discrepância do que realmente é executado na Casa da Alfândega (realizado) para estruturação do documento com o que se objetiva para a Casa (idealizado). Nesta etapa também foi aplicada a ferramenta Canvas. (3) Criação De posse desse diagnóstico (momento inspiração), começou a etapa de geração de conceitos e alternativas de projeto, bem como busca por ideias e seleção de propostas, a fim de se estabelecer o Plano de Intervenção. Tabela 2: Momento Ideação Durante o momento de Ideação (2, 3) as informações levantadas foram organizadas e analisadas com auxílio da ferramenta Canvas. Para isso, foi organizado um workshop com alguns colaboradores da Casa da Alfândega. A motivação e engajamento dos colaboradores é de extrema importância para a implementação sustentável do serviço. Por isso, eles devem ser envolvidos desde o início do processo. Assim, “o ideal é que os funcionários contribuam para a prototipagem de alguns momentos do serviço a fim de terem uma visão clara do conceito”. (Stickdorn & Schneider, 2014, p.136). É importante uma avaliação posterior da mudança do serviço para avaliar seu progresso. Com isso, demonstra a iteratividade do processo de design de serviço, até mesmo após a “finalização” do projeto. Anais do 9º Congresso Internacional de Design da Informação \| CIDI 2019 Proceedings of the 9th Information Design International Conference Anais do 9o Congresso Nacional de Iniciação Científica em Design da Informação \| CONGIC 2019 Proceedings of the 9th Information Design Student Conference Business Model Canvas Por meio do desenvolvimento deste Canvas, em momento que antecede a aplicação da abordagem do design de serviço, ficou claro que os colaboradores e a artesã que participaram deste workshop conseguiram visualizar a organização e perceber novos caminhos para alcançar uma melhor gestão na Casa da Alfândega. Além do mais, com essa aplicação foi possível verificar problemas que foram percebidos durante os 18 meses em que este estudo foi realizado, em conjunto com a supervisora da Casa da Alfândega, mas que sequer são percebidos no dia a dia pelos colaboradores e artesãos. Por meio do desenvolvimento deste Canvas, em momento que antecede a aplicação da abordagem do design de serviço, ficou claro que os colaboradores e a artesã que participaram deste workshop conseguiram visualizar a organização e perceber novos caminhos para alcançar uma melhor gestão na Casa da Alfândega. Além do mais, com essa aplicação foi possível verificar problemas que foram percebidos durante os 18 meses em que este estudo foi realizado, em conjunto com a supervisora da Casa da Alfândega, mas que sequer são percebidos no dia a dia pelos colaboradores e artesãos. Com base nesse Canvas foi possível idealizar o Plano de Intervenção que atenda às deficiências encontradas e que compreendido no design de serviço, atue a curto, médio ou longo prazo. Neste sentido, foram estruturadas e organizadas as ações que no terceiro momento, Implementação, deram origem ao Plano de Intervenção. Business Model Canvas O debate sobre modelo de negócios inicia-se por suas definições, refletindo diferentes pontos de vista sobre as aplicações das diversas atividades compreendidas por modelo de negócios. Um dos modelos mais popularizados é o Business Model Canvas, de Osterwalder (2004), que postula que modelo de negócios pode ser melhor descrito por meio de nove blocos básicos, que mostram a lógica de como a organização pretende gerar melhores lucros. Nesta etapa foi realizado um workshop, com o objetivo de demonstrar o uso da ferramenta Canvas e os benefícios que podem ser alcançados na organização com a utilização desta. O workshop foi realizado em 20 de fevereiro de 2017, na Casa da Alfândega, durou aproximadamente 2 horas e contou com a participação três colaboradores da Casa da Alfândega: a supervisora; a assistente administrativa que também é artesã; e a responsável pelo controle de estoque, além de uma artesã convidada para esta atividade, que faz parte do projeto desde 1988. Em conjunto com os colaboradores da Casa da Alfândega, chegou-se ao resultado da tela de modelo de negócios da organização (Figura 6), pautado na ferramenta Canvas, onde os blocos em cinza se referem à etapa PARA QUEM?; os blocos em vermelho se referem à etapa O QUÊ?; os blocos em verde à etapa QUANTO?; e, os blocos em azul à etapa COMO? 1807 Silva, D. B. & Triska, R. \| A cultura do design aplicada ao desenvolvimento de negócios Figura 6: Canvas gerado no workshop. Figura 6: Canvas gerado no workshop. Por meio do desenvolvimento deste Canvas, em momento que antecede a aplicação da abordagem do design de serviço, ficou claro que os colaboradores e a artesã que participaram deste workshop conseguiram visualizar a organização e perceber novos caminhos para alcançar uma melhor gestão na Casa da Alfândega. Além do mais, com essa aplicação foi possível verificar problemas que foram percebidos durante os 18 meses em que este estudo foi realizado, em conjunto com a supervisora da Casa da Alfândega, mas que sequer são percebidos no dia a dia pelos colaboradores e artesãos. Com base nesse Canvas foi possível idealizar o Plano de Intervenção que atenda às deficiências encontradas e que compreendido no design de serviço, atue a curto, médio Figura 6: Canvas gerado no workshop. Momento ideação Realizado o diagnóstico, no momento Inspiração, e definida a tela do modelo de negócios da Casa da Alfândega, no momento Ideação, chegou-se à última etapa da metodologia GODP, o momento Implementação. Nesta fase foram analisados todos os aspectos envolvidos na implementação do projeto, conforme se observa na tabela 3. Anais do 9º Congresso Internacional de Design da Informação \| CIDI 2019 Proceedings of the 9th Information Design International Conference Anais do 9o Congresso Nacional de Iniciação Científica em Design da Informação \| CONGIC 2019 Proceedings of the 9th Information Design Student Conference 1808 Tabela 3: Momento Implementação. ETAPA AÇÃO REALIZADA (4) Execução Definido o Plano de Intervenção, será o momento de testar as alternativas escolhidas, a fim de se desenvolver modelos e/ou protótipos finais. (5) Viabilização Para que se possa verificar a viabilidade deste estudo, serão necessários testes de simulação real, bem como apresentar e entregar documentos com orientações gerais ao cliente (Casa da Alfândega). (6) Verificação final Posteriormente ao desenvolvimento do Plano de Intervenção, serão realizadas verificações junto à FCC e à Casa da Alfândega, onde serão analisados os impactos deste estudo durante toda a sua cadeia (sustentabilidade e ciclo de vida), bem como apontar novas oportunidades, por meio de melhorias ou demanda por novos projetos. 1808 Anais do 9º Congresso Internacional de Design da Informação \| CIDI 2019 Proceedings of the 9th Information Design International Conference Anais do 9o Congresso Nacional de Iniciação Científica em Design da Informação \| CONGIC 2019 Proceedings of the 9th Information Design Student Conference Silva, D. B. & Triska, R. \| A cultura do design aplicada ao desenvolvimento de negócios Silva, D. B. & Triska, R. \| A cultura do design aplicada ao desenvolvimento de negócios Silva, D. B. & Triska, R. \| A cultura do design aplicada ao desenvolvimento de negócios Implementar uma ideia de design de serviço não significa necessariamente criar uma aplicação, ferramenta, interface ou roteiro, mas sim entender quais mudanças gerenciais são necessárias para que o novo serviço seja operacionalizado. É nesta etapa que a visualização do serviço deve ser a mais clara possível, para que todos os stakeholders que o implementarão consigam entender com facilidade o antes e o depois da cultura de design na Casa da Alfândega. 5 Resultados da aplicação do GODP Com o desenvolvimento deste estudo foi possível verificar a atuação do design de serviço, aplicado a modelo de negócios, para a promoção de serviços. Trabalhou-se no posicionamento da marca e na imagem da Casa da Alfândega, transparecendo nesta, a identidade local e cultural do estado de Santa Catarina. Assim, foi materializado o documento denominado Plano de Intervenção, idealizado durante a aplicação da metodologia GODP, que olhou para a reestruturação da estratégia de negócios na Casa da Alfândega. Plano de Intervenção Foi estabelecida uma abordagem para melhorar a interface de serviço desenvolvida na Casa da Alfândega. Neste sentido, se propôs um Plano de Intervenção como instrumento de gestão orientado para a tomada de decisões internas e necessariamente para a realização de ações. O Plano de Intervenção teve como objetivo, não só sistematizar o processo, mas ser, sobretudo, uma ferramenta que dê garantias de sucesso nas mudanças em curso e a implementar a capacidade de resposta às ameaças e às oportunidades. O Plano de Intervenção buscou prevenir ou minimizar uma situação-problema identificada neste diagnóstico, objetivando atingir determinadas metas, beneficiando direta ou indiretamente colaboradores, artesãos e usuários. Esse foi entregue à FCC (Tabela 4), a partir das hipóteses levantadas durante este estudo. Sua estrutura será fundamentada em três níveis de implantação: curto prazo; médio prazo; longo prazo, que norteará a execução das estratégias de aplicação. Anais do 9º Congresso Internacional de Design da Informação \| CIDI 2019 Proceedings of the 9th Information Design International Conference Anais do 9o Congresso Nacional de Iniciação Científica em Design da Informação \| CONGIC 2019 Proceedings of the 9th Information Design Student Conference 1809 Tabela 4: Plano de Intervenção. Aspectos-chave Recomendações Priorização Acessibilidade Adequar as instalações conforme norma de acessibilidade a edificações (ABNT NRB 9050) e do Iphan. Longo prazo Agenda da semana Ampliar o número de artesãos que participam da demonstração do fabrico. Médio prazo Agentes de turismo Ampliar as parcerias com agentes e agências de turismo. Curto prazo Artesãos de outras regiões Trazer mais artesãos/artistas para representar as diversas culturas de Santa Catarina. Médio prazo Atuação no mercado Realizar estudo de mercado, para o fortalecimento da marca. Médio prazo Cadastro dos artesãos Organizar e ampliar o cadastro dos artesãos, bem como a técnica desenvolvida por cada um. Médio prazo Capacitação Realizar cursos e treinamentos com os colaboradores, no sentido de qualificar o atendimento. Médio prazo Comportamento organizacional Planejar ações para o fortalecimento da organização. Longo prazo Estatuto interno Definir missão, visão e valores Médio prazo Identidade visual Elaboração da identidade visual Curto prazo Informações bilíngues Aplicação de informações bilíngues aos produtos comercializados neste espaço. Médio prazo Integração entre os artesãos Capacitar e propor a integração entre os artesãos, no sentido de ampliar e agregar valor aos produtos. Médio prazo Silva, D. B. & Triska, R. \| A cultura do design aplicada ao desenvolvimento de negócios Silva, D. B. & Triska, R. Plano de Intervenção \| A cultura do design aplicada ao desenvolvimento de negócios Organização do espaço interno Contratar um profissional de arte para organizar e melhorar a distribuição dos produtos e peças em exposição. Médio prazo Pesquisa de público Realizar pesquisa de público, a fim de propor estratégias para a captação de novos clientes. Médio prazo Posicionamento de marca Desenvolver métodos para criar ou fortalecer a imagem da marca, incorporando além das qualidades técnicas, as qualidades que enfatizem a relação da marca com os usuários. Médio prazo Prédio da Alfândega Resgatar e divulgar a importância histórica do prédio da Alfândega, tombado em 1975 pelo Iphan. Médio prazo Redes sociais Fortalecer a divulgação das ações da Casa e dos artesãos em redes sociais, como Facebook e Instagram. Curto prazo Sinalização Projeto de sinalização interna e externa. Médio prazo Site Projeto de redesign do site, deixando mais moderno, dinâmico e organizado. Médio prazo TAGs Criar um padrão visual para as etiquetas que acompanham as peças comercializadas na Casa da Alfândega. Médio prazo Tenda do artesanato Divulgar o projeto Casa da Alfândega nos bairros e praias de Florianópolis, com a participação de artesãos. Médio prazo Uniformes Implantar o uso de uniformes entre os colaboradores. Além do reforço da imagem da marca, mostrará organização, credibilidade, segurança e profissionalismo. Curto prazo Vídeo institucional Produzir um vídeo institucional para divulgação da Casa da Alfândega em eventos e ações da FCC. Longo prazo Como resultado final deste estudo, conclui-se que a disciplina de design está se tornando cada vez mais popular em diversos segmentos do mercado, da engenharia a outras áreas da tecnologia e da gestão, se aproximando cada vez do conceito social e humano do design. Assim, para um serviço que demanda visão holística se sugere investir na contratação de um profissional de design. Nosso papel, enquanto profissionais que querem fazer a diferença, seja na academia ou no mercado, é quebrar esse paradigma de que só grandes organizações podem contar com designers em suas equipes. Anais do 9º Congresso Internacional de Design da Informação \| CIDI 201 g g Anais do 9o Congresso Nacional de Iniciação Científica em Design da Informação \| CONGIC 2019 h g g Anais do 9o Congresso Nacional de Iniciação Científica em Design da Informação \| CONGIC 2019 Proceedings of the 9th Information Design Student Conference g ç g Proceedings of the 9th Information Design Student Conference g g ç eedings of the 9th Information Design International Conferen Anais do 9 Congresso Internacional de Design da Informação \| CIDI 2019 Proceedings of the 9th Information Design International Conference Anais do 9o Congresso Nacional de Iniciação Científica em Design da Informação \| CON g g ais do 9o Congresso Nacional de Iniciação Científica em Des Anais do 9º Congresso Internacional de Design da Informação \| CIDI 2019 Proceedings of the 9th Information Design International Conference Anais do 9o Congresso Nacional de Iniciação Científica em Design da Informação \| CONGIC 2019 Anais do 9 Congresso Nacional de Iniciação Científica em Design da Informa Proceedings of the 9th Information Design Student Conference Anais do 9o Congresso Nacional de Iniciação Científica em Design da Informação \| CONGIC 2019 Proceedings of the 9th Information Design Student Conference Anais do 9º Congresso Internacional de Design da Informação \| CIDI 2019 Proceedings of the 9th Information Design International Conference Anais do 9o Congresso Nacional de Iniciação Científica em Design da Informação \| CONGIC 2019 Proceedings of the 9th Information Design Student Conference 5 Conclusão Este estudo teve como mérito uma aplicação real, que residiu em investigar como a abordagem interdisciplinar do design de serviço, capaz de combinar diferentes métodos e ferramentas oriundos de diversas disciplinas, pode ser inserida em uma organização do terceiro setor, de pequeno porte, onde, até então, não se discutia uma cultura de design. Desta maneira, cumpriu seu objetivo, de indagar como a cultura de design pode contribuir para as questões do serviço na Casa da Alfândega de Santa Catarina. Embora os resultados contribuam claramente para a literatura, o estudo possui certas limitações que devem ser consideradas na avaliação dos resultados. Apresentou um esboço para facilitar a busca por novos modelos de negócios, no contexto de uma organização de pequeno porte, a fim de facilitar a adoção de práticas sustentáveis, dentro do negócio. Por meio de um estudo exploratório e descritivo, foi possível avaliar os níveis de compreensão de cada momento da metodologia GODP, dividida em três momentos: inspiração, ideação e implementação. No momento inspiração foi realizado um diagnóstico preliminar sobre a situação da Casa da Alfândega, a fim de levantar fragilidades e potencialidades, em que se avaliou, por meio dos blocos de referência, o produto (artesanato), o usuário (artesão) e o contexto (Casa da Alfândega). No segundo momento, ideação, estratégias para a resolução do problema de design começaram a ser idealizadas. A partir do workshop, com o auxílio da ferramenta Canvas, permitiu-se visualizar a tela de modelo de negócios, colocando no papel o atual momento da organização. Ainda, essa aplicação trouxe a lógica projetual do design para 1810 Silva, D. B. & Triska, R. \| A cultura do design aplicada ao desenvolvimento de negócios transformar a cultura da organização em espaços de solução de problemas e acesso a conhecimentos que transcendem as fronteiras da Casa da Alfândega. E, no momento implementação, já com a devida apropriação sobre o negócio, foram recomendados possíveis caminhos para implementar as competências do design de serviço. Foi gerado um Plano de Intervenção que foi entregue à FCC com recomendações, a partir das hipóteses levantadas neste estudo, para que sejam avaliados e, talvez implantados no contexto da Casa da Alfândega. 5 Conclusão Essa abordagem pelo design serviço foi um processo que buscou entender qual é a experiência dos usuários ao interagirem com determinado serviço, a experiência do usuário ao consumir esse serviço, a experiência do colaborador ao prestar esse serviço e como uma organização se mantém fiel à sua missão. Foi o que aconteceu em diversas etapas deste estudo. Estas modificações fazem parte do processo de iteração do design e que não são rígidas, porém adaptáveis a cada circunstância. Desta maneira, este pode ser considerado um primeiro indício para a melhor aplicabilidade do design de serviço com a ferramenta do Canvas, em organizações de pequeno porte que buscam inovar em seu modelo de negócios. Como recomendação para trabalhos futuros, sugere-se investigar a atmosfera intangível da Casa da Alfândega – cultura, cor, cheiro, entre outros –, uma vez que a Instituição abrange a identidade cultural de Santa Catarina e o usuário carrega essa experienciação cultural. Assim, o artefato torna-se o fio condutor para as lembranças desses aspectos intangíveis que o usuário leva dessa experiência vivenciada. Anais do 9º Congresso Internacional de Design da Informação \| CIDI 2019 Proceedings of the 9th Information Design International Conference Anais do 9o Congresso Nacional de Iniciação Científica em Design da Informação \| CONGIC 2019 Proceedings of the 9th Information Design Student Conference 1811 Referências Brown, T. (2018). Design thinking: uma metodologia poderosa para decretar o fim das velhas ideias. Alta Books Editora. Girotra, K., & Netessine, S. (2013). OM forum-business model innovation for sustainability. Manufacturing & Service Operations Management, 15(4), 537-544. Kindström, D., & Kowalkowski, C. (2014). Service innovation in product-centric firms: A multidimensional business model perspective. Journal of Business & Industrial Marketing, 29(2), 96-111. Merino, G. S. A. D. (2016). GODP – Guia de Orientação para o Desenvolvimento de Projetos: uma metodologia centrada no usuário. NGD/LDU – UFSC. Mozota, B. B., Klöpsch, C., & Costa, F. C. X. (2009). Gestão do design: usando o design para construir valor de marca e inovação corporativa. Bookman editora. Osterwalder, A., & Pigneur, Y. (2013). Business model generation: inovação em modelos de negócios. Alta Books Editora. PAB (2012). Base conceitual do artesanato brasileiro. [Web page] Disponível em: http://www.desenvolvimento.gov.br/arquivos/dwnl_1347644592.pdf Santa Catarina. (2015). Ofício n. 212/2015. Melhoria na qualidade do artesanato catarinense. Secretaria de Estado do Turismo, Cultura e Esporte, Fundação Catarinense de Cultura. SEBRAE. (2013). SC em números. [Web page]. Disponível em http://www.sebrae.com.br/Sebrae/Portal%20Sebrae/Anexos/Relatorio%20Estadu al.pdf Stickdorn, M., & Schneider, J. (2014). Isto é design thinking de serviços: fundamentos, ferramentas, casos. Bookman Editora. Anais do 9º Congresso Internacional de Design da Informação \| CIDI 2019 Proceedings of the 9th Information Design International Conference Anais do 9o Congresso Nacional de Iniciação Científica em Design da Informação \| CONGIC 2019 Proceedings of the 9th Information Design Student Conference 1811 1811 Silva, D. B. & Triska, R. \| A cultura do design aplicada ao desenvolvimento de negócios Anais do 9º Congresso Internacional de Design da Informação \| CIDI 2019 Proceedings of the 9th Information Design International Conference Anais do 9o Congresso Nacional de Iniciação Científica em Design da Informação \| CONGIC 2019 Proceedings of the 9th Information Design Student Conference Sobre os autores Diego Borges da Silva, Me., UFSC, Brasil <bordiego@gmail.com> Ricardo Triska, Dr., UFSC, Brasil <ricardo.triska@gmail.com> 1812
W1971871735.txt	https://zenodo.org/records/2504598/files/article.pdf	de		Zur Frage der Brüchigkeit der Gerste — eine Berichtigung	Molecular genetics and genomics	1,919	public-domain	398	Kleinere Mitteilungen. 53 Zur Frage der Briichigkeit der Gerste - - eine Berichtig~mg. In meinem Aufsatz: ,Ergebnisse der Bastardierungsversuche bei Gerste" in Heft 6, 1917 der Sitzungsberichte der Gesellschaft Naturf. Freunde ist auf S. 395 leider ein sinnentstellender Fehler im Druck stehen geblieben. Es mug dort auf Zeile 15 briichig statt niclitbriichig heiflen, wodurch allein auch die nachfolgenden Zahlen verst~ndlich sind. In Heft II dieser Zeitschr., S. 75 kniipft G. v. U b i s c h an diesen ,,Irrtum ~ einige zahlenkritische Bemerkungen. Der Teil yon G. v. U b i s c h s Kritik, der sich auf diesen Druckfehler bezieht, ist demnach hinf~llig. Ich fiige die aus meinen Aufzeichnungen entnommenen Daten bei: F 2aus Hordeum spontaneum )< H. 1 (6 zeilige Nacktgerste) 290 briichig 290 briichig .~ 13 schwach briichig ~ 303 briichig 63 nichtbrfichig L 50 nichtbriichig 50 nichtbriichig. Welche yon beiden Zi~hlungen die richtige is t, laflt sich a priori nicht entscheiden. Fiir die Zusammenfassung der schwach br~ichigen mit den nichtbriichigen (Verb. 290:63) spricht der Umstand, das die als schwach briichig bezeichneten sich als volle Ahren abernten lassen; erst bei starkerer Beriihrung bei cler naheren Untersuchung fallen sie in die einzelnen Spindelglieder, mindestens im oberen Drittel, auseinander. Fiir die Zusammenfassung der schwach briichigen mit den briichigen spricht der Umstand, da~ auch innerhalb eines F2-Beetes ein Unterschied zu machen ist zwischen deutlich nichtbriichigen mit bis zur S p i t z e z~her Spindel und den eben beschriebenen schwach briichigen. Dann wiirde sich das Verh~ltnis 3 0 3 : 5 3 - - 6 : 1 ergeben. Es bestehen also, wenn man, worauf G. v. U b i s c h hinweisS, zweckm~fliger das Verh~ltnis 3 : 1 nach J o h a n n s e n auf vier Individuen umrechnet, im ersten Falle 290 br. : (50 nbr. -~- 13 schw. br. - - ) 63 nbr. - - 3,28 : 0,72 theor. Fehler -4-0,09 exp. Fehier ~__ 0,28, im zweiten Falle (290 br. -~ 13 schw. br. --) 353 br. : 50 nbr. ~ 3,43 : 0,57 exp. Fehler ~ q- 43. D i e s e Abweichung liege sich allerdings nicht vernachl~ssigen. Es liegt daher die Notwendigkeit vor, die Frage, ob es sieh um ein genotypisch bedingtes VerhMtnis handelt, welter zu verfolgen; ich hoffe demnachst darauf zurtiekkommen zu kSnnen. E. S c h i e m a n n .
https://openalex.org/W4361852701	https://periodicos.ufpb.br/index.php/rec/article/download/63644/36932	Portuguese	null	CURRÍCULO	Revista Espaço do Currículo	2,023	cc-by	7,484	CURRÍCULO: diversidade e tensões no contexto educacional CURRICULUM: diversity and tensions in the educational context CURRICULUM: diversidad y tensiones en el contexto educativo Revista Espaço do Currículo ISSN 1983-1579 Doi: 10.15687/rec.v16i1.63644 http://periodicos.ufpb.br/ojs2/index.php Quenizia Vieira Lopes Mestra em Linguística Doutoranda em Educação pela Universidade Estadual de Londrina e Pedagoga no Instituto Federal do Tocantins, Brasil. E-mail: quenizia@gmail.com Orcid: https://orcid.org/0000-0001-6199- 0616 Luiz Gustavo Tiroli Mestre em Educação Doutorando pela Universidade Estadual de Londrina, Brasil. E-mail: luiz.gustavo.tiroli@uel.br Orcid: https://orcid.org/0000-0002-7912- 8319 Adriana Regina de Jesus Santos Doutora em Educação Professora da Universidade Estadual de Londrina, Brasil. E-mail: adrianatecnologia@yahoo.com.br Orcid: https://orcid.org/0000-0002-9346- 5311 Como citar este artigo: LOPES, Quenizia Vieira.; TIROLI, Luiz Gustavo.; SANTOS, Adriana Regina de Jesus. CURRÍCULO: diversidade e tensões no contexto educacional. Revista Espaço do Currículo, v. 16, n. 1, p. 1-12, 2023 ISSN2177 2886 DOI: https://doi org/ Resumo: O objetivo deste artigo consiste em discorrer sobre as principais concepções de currículo no cenário educacional a partir de quatro correntes teóricas-metodológicas distintas, quais sejam, o currículo Humanista, o currículo Acadêmico, o currículo Tecnológico e o currículo Reconstrucionista Social. Como percurso metodológico, trata-se de uma abordagem qualitativa, em que se adotou a pesquisa documental. Nas considerações finais, tem-se que o currículo não é campo de neutralidade epistemológica, ao contrário, é espaço de embate e tensões. Da análise, constatou-se que o currículo Reconstrucionista Social se revela mais comprometido com uma educação crítica e emancipatória, tendo em vista que visa instrumentalizar o sujeito com o acervo teórico-científico-cultural elaborado pela humanidade ao longo das quadras históricas, um currículo sócio-historicamente situado com vistas à humanização do sujeito. Orcid: https://orcid.org/0000-0001-6199- 0616 Luiz Gustavo Tiroli Mestre em Educação Doutorando pela Universidade Estadual de Londrina, Brasil. E-mail: luiz.gustavo.tiroli@uel.br Orcid: https://orcid.org/0000-0002-7912- 8319 Adriana Regina de Jesus Santos Doutora em Educação Professora da Universidade Estadual de Londrina, Brasil. E-mail: adrianatecnologia@yahoo.com.br Orcid: https://orcid.org/0000-0002-9346- 5311 Palavras-chave: Currículo. Contexto Educacional. Diversidade. Contexto Como citar este artigo: LOPES, Quenizia Vieira.; TIROLI, Luiz Gustavo.; SANTOS, Adriana Regina de Jesus. CURRÍCULO: diversidade e tensões no contexto educacional. Revista Espaço do Currículo, v. 16, n. 1, p. 1-12, 2023 ISSN2177-2886. DOI: https://doi.org/ 10.15687/rec.v16i1.63644. Recebido em: 07/07/2022 Aceito em: 28/07/2022 Publicação em: 25/01/2023 REV. ESPAÇO DO CURRÍCULO (ONLINE), JOÃO PESSOA, V.16, N.1, P. 1-12, JAN/ABR. 2023. Palabras clave: Currículum. Contexto educativo. Diversidad. Palabras clave: Currículum. Contexto educativo. Diversidad. Keywords: Curriculum. Educational Context. Diversity. Resumem: El objetivo de este artículo es discutir las principales concepciones del currículo en el escenario educativo a partir de cuatro corrientes teórico-metodológicas distintas, como el currículo humanista, el currículo académico, el currículo tecnológico y el currículo reconstruccionista social. Como camino metodológico, se trata de un enfoque cualitativo, en el que se adoptó la investigación documental. En las consideraciones finales, se señala que el currículo no es un campo de neutralidad epistemológica, al contrario, es un espacio de choque y tensiones. Del análisis, se encontró que el currículo Reconstruccionista Social está más comprometido con una educación crítica y emancipadora, considerando que pretende instrumentalizar la asignatura con la colección teórico-científico-cultural elaborada por la humanidad a lo largo de los tribunales históricos, un currículo socio-históricamente situado con miras a la humanización del tema. REV. ESPAÇO DO CURRÍCULO (ONLINE), JOÃO PESSOA, V.16, N.1, P. 1-12, JAN/ABR. 2023. 1 LOPES, Q. V.; TIROLI, L. G.; SANTOS, A. R. J. Abstract: The objective of this article is to discuss the main conceptions of curriculum in the educational scenario based on four distinct theoretical-methodological currents, such as the Humanist curriculum, the Academic curriculum, the Technological curriculum and the Social Reconstructionist curriculum. As a methodological path, this is a qualitative approach, in which documentary research was adopted. In the final considerations, it is noted that the curriculum is not a field of epistemological neutrality, on the contrary, it is a space of clash and tensions. From the analysis, it was found that the Social Reconstructionist curriculum is more committed to a critical and emancipatory education, considering that it aims to instrumentalize the subject with the theoretical-scientific-cultural collection elaborated by humanity along the historical courts, a socio-historically situated curriculum with a view to the humanization of the subject. Keywords: Curriculum. Educational Context. Diversity. 2 CURRÍCULO: espaço de tensões e relações de poder Em uma discussão sobre a diversidade a respeito do currículo no contexto educacional, em um primeiro momento, faz-se necessário um entendimento sobre a definição de currículo. Contudo, para essa primeira leitura, a procura dessa compreensão em um dicionário, por exemplo, pode não atender totalmente ao contexto da pesquisa em tela, visto que, de acordo com o dicionário Aulete1 Digital, o currículo é definido como “o conjunto das matérias de um curso”, contudo, entende-se que esse termo tem muito mais amplitude em sua definição. Nesse sentido, a seguir, apresenta-se algumas das definições de currículo. Sacristán (2013) argumenta que o termo currículo vem do latim, mais especificamente da palavra curriculum, que é derivada de currere, que significa correr, percurso, trajeto. Para o autor, o termo traz consigo um processo histórico de significados, que vêm se transformando, ou até mesmo se ampliando, com o passar dos tempos, de acordo com os contextos. Inicialmente, o termo currículo consistia em “território demarcado e regrado do conhecimento correspondente aos conteúdos que professores e centros de educação deveriam cobrir” (SACRISTÁN, 2013, p. 17), o que não difere totalmente do seu significado atualmente, uma vez que o currículo ainda é considerado norteador dos conhecimentos a serem trabalhados no contexto escolar, além de ser um elemento direcionador da atividade do professor. Contudo, o currículo abrange uma dimensão maior, uma vez que, O currículo há muito tempo deixou de ser apenas uma área meramente técnica, voltada para questões relativas a procedimentos, técnicas, métodos. Já se pode falar agora de uma tradição crítica do currículo, guiada por questões sociológicas, políticas, epistemológicas. Embora questões relativas ao “como” do currículo continuem importantes, elas só adquirem sentido dentro de uma perspectiva que as considere em sua relação com questões que perguntem pelo “por quê” das formas de organização do conhecimento escolar (MOREIRA; TADEU, 2013, p.13). Portanto, o currículo carrega consigo uma bagagem que contempla ideias culturais, sociais e políticas. Desse modo, admite-se que esse seja influenciado por contextos (sociais, culturais, políticos e econômicos, por exemplo), assim como pode servir de influência aos sujeitos subordinados a ele, haja vista a sua capacidade de contribuir para a construção da formação do indivíduo. Llavador (1994, p. 370) lembra que “a palavra currículo engana-nos porque nos faz pensar numa só coisa, quando se trata de muitas simultaneamente e todas elas interrelacionadas”. 1 Disponível em: https://aulete.com.br. Acesso em: 12 jun. 2022. CURRÍCULO: diversidade e tensões no contexto educacional CURRÍCULO: diversidade e tensões no contexto educacional 1 INTRODUÇÃO O contexto educacional envolve múltiplas atividades e diferentes agentes, sendo um lugar propício para desenvolvimento de ações políticas, sociais e culturais, em razão de se tratar de um ambiente em que os conhecimentos são compartilhados e a aprendizagem pode ser realizada. Por consequência, torna-se uma zona de interesse de grande magnitude para aqueles que objetivam obter algo que se relaciona a este contexto, bem como para aqueles que pretendem entender o que ocorre neste cenário, tornando-se, portanto, um campo de estudos diversos. Assim, estudos sobre currículo, como norteador de atividades educacionais, tornam-se significativos. O currículo tem despertado em estudiosos do contexto educacional o interesse por sua análise há muitos anos atrás, contudo, só especificamente no final do século XIX para início do século XX que se tornou objeto de estudos nos Estados Unidos da América (EUA), prosseguindo este trabalho posteriormente em outros países (MOREIRA; TADEU, 2013). A fim de investigar a respeito desse campo de investigação, e sua diversidade no contexto educacional, adota-se o seguinte problema de pesquisa: quais seriam as principais correntes teórico- metodológicas que direcionam a concepção de currículo e suas implicações no âmbito educacional? Isto posto, o objetivo geral consiste em discorrer sobre o currículo como campo de embate de visões de mundo e sobre as principais concepções de currículo que podem ser identificadas a partir de revisão documental. A problemática da pesquisa é desenvolvida por meio da abordagem qualitativa, com intuito de analisar os objetivos em uma perspectiva descritiva e exploratória. A técnica de pesquisa é a revisão documental. A estratégia de coleta de dados adotada é o fichamento da bibliografia. Na consecução do objetivo acima delineado, o artigo está estruturado em duas seções. Inicialmente, busca-se discorrer a respeito do currículo como espaço de relações de poder e tensões, pois tem-se como premissa que o currículo não é naturalmente posto ou providencialmente estabelecido, mas antes é fruto de embates de visões de mundo e sofre influências políticas, econômicas, sociais, ideológicas, culturais e teórico-conceituais, não se revelando um espaço de neutralidade, mas sim de disputas e conflitos. Na sequência, aborda-se quatro perspectivas teóricas que norteiam a concepção de currículo: o currículo Humanista, o currículo Acadêmico, o currículo Tecnológico e o currículo Reconstrucionista Social, cujo intuito consiste em analisar essas vertentes e suas implicações no campo educacional. REV. ESPAÇO DO CURRÍCULO (ONLINE), JOÃO PESSOA, V.16, N.1, P. 1-12, JAN/ABR. 2023. 2 2 CURRÍCULO: espaço de tensões e relações de poder Nesse ponto, é interessante assinalar que o currículo é multidimensional e assume, ao menos, uma dimensão ideológica, cultural e política. Para Moreira e Silva (1997, p. 23), o aspecto ideológico relaciona-se a transmissão de um conjunto de ideias que representa uma visão de mundo, porém, não neutra, uma “visão social vinculada aos interesses dos grupos situados em uma posição de vantagem na organização social”; o aspecto cultural, por sua vez, representa o “[...] o terreno em que se enfrentam diferentes e conflitantes concepções de vida social” (MOREIRA; SILVA, 1997, p. 27); e por fim, o aspecto político, que é relativo às relações de poder, expressando os interesses das classes dominantes na constituição do currículo ‘oficial’, bem como garantindo, por meio desse currículo oficial, a manutenção do status quo, isto é, a perpetuação da estrutura de dominação (MOREIRA; SILVA, 1997). Assim, por exemplo, não prever determinado conteúdo nas diretrizes curriculares não é somente uma negligência, é, antes, uma opção, tendo em vista que o campo curricular é espaço de embate e disputa cultural, ideológica e política, “[...] um dos locais privilegiados onde se entrecruzam saber e poder, representação e domínio, discurso e regulação” (SANTOS; CASALI, 2009, p. 208). Conforme ensina Apple (2017, p. 909), o currículo é “[...] um foco primário de lutas educacionais”. Logo, é manifestação de poder e de intencionalidade formativa. Destaca-se, por conseguinte, que o currículo viabiliza reflexões sobre relações de poder, visto que é controlado por seus detentores, como, por exemplos, um Governo que o institui e o direciona ou uma gestão de unidade escolar que gerencia as atividades cotidianas do currículo, levando-se em conta REV. ESPAÇO DO CURRÍCULO (ONLINE), JOÃO PESSOA, V.16, N.1, P. 1-12, JAN/ABR. 2023. 3 LOPES, Q. V.; TIROLI, L. G.; SANTOS, A. R. J. que, conforme disposto por Sacristán (2013, p. 17), “o currículo recebeu o papel decisivo de ordenar os conteúdos a ensinar”, além de que “o currículo, como campo cultural, como campo de construção e produção de significações e sentido, torna-se, assim, um terreno central dessa luta de transformação das relações de poder” (MOREIRA; TADEU, 2013, p. 39). Nessa linha, uma vez que por meio do currículo é possível direcionar objetivos a serem alcançados, por meio dos conteúdos a serem abordados, já que este se situa no campo das ideias, este também tem um papel de agente influenciador. Assim sendo, quem tem o controle do currículo detém o poder. 2 CURRÍCULO: espaço de tensões e relações de poder Logo, deve-se estar atento aos aspectos do currículo, o que se propõe e o que se pretende, principalmente porque esse pode vir carregado de ideologias daqueles que o controla e que nem sempre estarão dispostas ao benefício da sociedade em detrimento das particularidades, dado que, O currículo não é um elemento inocente e neutro de transmissão desinteressada do conhecimento social. O currículo está implicado em relações de poder, o currículo transmite visões sociais particulares e interessadas, o currículo produz identidades individuais e sociais particulares. O currículo não é um elemento transcendente e atemporal - ele tem uma história, vinculada a formas específicas e contingentes de organização da sociedade e da educação. (MOREIRA; TADEU, 2013, p.14). Percebe-se que o currículo é um ambiente de batalhas das divergentes ideias existentes, prevalecendo aquelas oriundas dos detentores de poder do currículo, pois por meio dele possibilita-se que seja reproduzida as ideias daqueles que o direciona, em conformidade com seus aspectos ideológicos e culturais. Assim, este também é um lugar propício para as transformações sociais, uma vez que possibilita a formação do sujeito em sua perspectiva, o que pode possibilitar a formação de um agente norteador de mudanças na sociedade, posto que, “o currículo proporciona uma ordem por meio da regulação do conteúdo da aprendizagem e ensino na escolarização moderna, uma construção útil para organizar aquilo do qual deve se ocupar a escolarização e aquilo que deverá ser aprendido.”. (SACRISTÁN, 2013, p. 19). Nesta perspectiva, por ser o “[...] núcleo organizacional e intelectual da escola” (MOREIRA, 2021, p. 45), acredita-se que o ideal é que o currículo seja autônomo, construído pelos envolvidos no processo educacional, de acordo com as necessidades que surgem da sociedade, tendo como parâmetros a historicidade e a cultura existente, com vistas a contribuir para a humanização do sujeito, não sendo guiado pelos detentores de poder. Dessa maneira, a discussão sobre currículo deve ser ampla, deve abranger toda a sociedade envolvida com o contexto educacional. Seguindo essa linha, entende-se que todos os cursos iniciais de formação de professores deveriam ter, em suas matrizes curriculares, momentos específicos para discussão de currículo, uma vez que o currículo é uma peça fundamental no contexto educacional e não é somente um documento que traz discussões de carga horária de disciplinas que devem ser utilizadas nos cursos. 2 CURRÍCULO: espaço de tensões e relações de poder Este é muito mais abrangente, até mesmo, como já dito antes, pode influenciar a formação do indivíduo e as suas concepções e visões de mundo e de identidade própria. O currículo pode ser observado por meio de teorias, ou seja, quando você consegue explicar algo a partir do momento em que se observa. Conforme disposto por Moreira e Tadeu (2013), as teorias explicitam procedimentos e concepções de homem que se deseja formar para determinada sociedade. Nessa perspectiva, pode-se identificar três diferentes abordagens: Teorias Tradicionais (baseadas no Positivismo); Teorias Críticas (baseadas no Materialismo Histórico-Dialético), e Teorias Pós-críticas (baseadas no Estruturalismo). “É a temática do poder que separa as teorias tradicionais das teorias crítica e pós-críticas” (MOREIRA; TADEU, 2013, p.7). A seguir, discorre-se sobre alguns currículos dispostos no ponto de vista dessas teorias no processo educacional, categorizados de acordo com McNeil (1981): o currículo Humanista, o currículo Acadêmico, o currículo Tecnológico e o currículo Reconstrucionista Social. 3 CURRÍCULO EDUCACIONAL: diversidade de perspectivas REV. ESPAÇO DO CURRÍCULO (ONLINE), JOÃO PESSOA, V.16, N.1, P. 1-12, JAN/ABR. 2023. 4 Santos e Casali (2009) sistematizam o currículo educacional em três perspectivas: o currículo Santos e Casali (2009) sistematizam o currículo educacional em três perspectivas: o currículo e Casali (2009) sistematizam o currículo educacional em três perspectivas: o currículo REV. ESPAÇO DO CURRÍCULO (ONLINE), JOÃO PESSOA, V.16, N.1, P. 1-12, JAN/ABR. 2023. 4 CURRÍCULO: diversidade e tensões no contexto educacional Formal, aquele expresso em documentos, tais como matrizes ou ementas, isto é, a natureza prescritiva e balizadora do currículo; tem-se ainda o currículo Real, aquele que se materializa na prática cotidiana por meios dos processos de ensino e aprendizagem; e o currículo Oculto, aquele que não está previsto e representa os conhecimentos que são apropriados por meio de práticas e comportamentos que permeiam o ambiente sociocultural em que se estão inseridos professores e alunos. Especificando essas perspectivas de currículos, Moreira e Tadeu (2013) referenciam que, currículo Oculto é aquele cujas perspectivas trabalhadas não estão discriminadas no currículo Oficial (currículo Formal), ou seja, no currículo que está explicitado e registrado na instituição de ensino. Por conseguinte, o currículo Real é aquele trabalhado efetivamente na práxis escolar. Isto significa, a compreensão de currículo como elemento em constante escrita, nunca acabado, haja vista a relação permanente e dinâmica entre a dimensão instituinte, que engloba as forças que implicam na criação e transformação do instituído, aquilo que se encontra consagrado e institucionalizado, mas em constante mudança (GUEDES; ARAUJO, 2017). Como visto, o currículo é terreno de colisão e disputa (MOREIRA; SILVA, 1997; APPLE, 2017), um espaço “[...] propício para a transformação – ou manutenção – das relações de poder e, portanto, das mudanças sociais” (SANTOS; CASALI, 2009, p. 210). Nesse sentido, a análise sobre currículo não se limita a discussão da previsão ou não de determinados elementos formativos, mas permite desvelar as relações de poder que se travam no processo constitutivo institucional-prescritivo (currículo Formal), como também possibilita perceber a realidade material, quando da sua aplicação (currículo Real), bem como auxilia na revelação das influências externas que circunscrevem o processo de ensino e aprendizagem, tais como posicionamentos, atitudes, valores, princípios e paradigmas que implicam nas atividades formativas curriculares (currículo Oculto). 3 CURRÍCULO EDUCACIONAL: diversidade de perspectivas Ao se considerar a multidimensionalidade do currículo, busca-se demonstrar as principais perspectivas teórico-conceituais de currículo e suas implicações no contexto educacional, tendo como premissa o currículo Humanista, o currículo Acadêmico, o currículo Tecnológico e o currículo Reconstrucionista Social. 3.2 Currículo Acadêmico A corrente curriculista acadêmica é também conhecida como vertente tradicional, cujo foco centra-se nos conteúdos que foram consagrados pela humanidade ao longo da história. Esse conhecimento é organizado na forma de conteúdos disciplinados por meio de matérias que são ministradas de maneira transmissiva, sem possibilidade de participação do sujeito que aprende, tendo em vista que o dogmatismo impera nessa perspectiva, de modo a não haver espaço para criticar ou analisar, mas tão somente assimilar e reproduzir (MCNEIL, 1981). Ao invés de comunicar-se, o educador faz comunicações por intermédio do discurso, cabendo aos educandos o arquivamento desses comunicados. “Eis aí a concepção “bancária” da educação, em que a única margem de ação que se oferece aos educandos é a de receberem os depósitos, guardá-los e arquivá-los” (FREIRE, 2020, p. 33). Para Luckesi (2005), os alunos não são concebidos como sujeitos históricos inseridos em um contexto e, consequentemente, mais se classifica do que se ensina. Isto porque parte do princípio que a ciência definiu como o mundo deve ser e, portanto, o conhecimento é tido como verdade absoluta, assim, cabe ao professor somente transferir o conhecimento em uma via única, daquele que sabe (professor) para aquele que nada sabe (aluno). O método de pesquisa dessa vertente é o positivismo, com a questão empírica. No currículo acadêmico, o homem é considerado cidadão a partir do momento que possui acesso ao conhecimento científico, contribuindo assim com a sociedade (ordem e progresso). Logo, o conhecimento tem finalidade pragmática, de modo que a sua função primordial é ensinar os conteúdos tido como fundamentais pela ciência para que o indivíduo possa ser útil à sociedade e se tornar um cidadão. Ao contrário dessa visão de conteúdo dogmático, inquestionável e tido como verdade absoluta, Adorno (1996) apregoava que a tarefa do professor não se resume a transmitir conteúdos, o educador tem uma função política, uma vez que a realidade extrapedagógica influi sobre a realidade da sala de aula e não pode ser desconsiderada (PEREIRA, 2018), isto é, a dimensão formal, real e oculta estão interligadas, por isso não pode haver uma dissociação entre os conteúdos e os processos de ensino e aprendizagem. Ademais, essa perspectiva tradicional naturaliza os aspectos de classe sociais e as relações de poder, não articulando educação e ideologia, incorrendo, assim, na manutenção da ideologia dominante (SANTOS; LOPES; COSTA, 2017). LOPES, Q. V.; TIROLI, L. G.; SANTOS, A. R. J. Além disso, os adeptos do currículo Humanista sustentam que deve haver uma valorização da inteligência emocional do educando, e o processo avaliativo deve contemplar uma autoavaliação, em que o aluno tenha condições de mensurar e controlar o seu próprio desenvolvimento, pois valoriza-se a autonomia. O erro é visto como parte constituinte do processo de aprendizagem, por isso não é descartado. Como demonstrado, embora reconheça a importância dos conhecimentos prévios dos alunos, essa vertente curricular sobrepõe o indivíduo em detrimento do coletivo, pois a base alicerçada no liberalismo distorce a realidade e desconsidera os elementos políticos, econômicos, sociais, culturais, ideológicos e teóricos que atravessam o campo da educação. Para Saviani (2011, p. 118), trata-se de uma proposta atenta às determinantes sociais, que permite articular o conhecimento pedagógico e as relações sociais. Essa centralização do indivíduo ignora os elementos ocultos que permeiam o campo curricular, e para Silva (2006), reconhecer as conexões persistentes entre o currículo, o mundo do trabalho, as relações de poder e de cultura dominante, bem como a tensões e conflitos existentes é o primeiro passo para a construção de uma perspectiva curricular que seja atenta a necessidade da emancipação do homem, e nisto, não contribui o currículo Humanista. 3.1 Currículo Humanista A perspectiva teórica do currículo Humanista leva em consideração tudo aquilo que o sujeito já traz de sua vivência, o indivíduo precisa ser ouvido. O currículo não separa aquilo que o aluno sabe do senso comum daquilo que aprende do conhecimento científico. Portanto, há um deslocamento do foco centrado tradicionalmente no ensino (professor) para a aprendizagem (aluno). No Brasil, o currículo humanista emerge no contexto da Escola Nova, na década de 1930, com o Manifesto dos Pioneiros da Educação Nova (SANTOS; CASALI, 2009), que criticavam o ensino tradicional, tendo em vista ser este marcado pelo domínio da técnica e carente de reflexões teóricas que pudessem aprimorar os processos de ensino e aprendizagem. O currículo, na visão escolanovista, deve considerar as necessidades e potencialidades dos indivíduos-educandos. Conforme ensinamento de McNeil (1981), o currículo deve privilegiar conteúdos disciplinares que possam ir ao encontro das necessidades do indivíduo, com foco na autorrealização do sujeito no processo de ensino e aprendizagem. Isso se deve ao fato de que é o liberalismo a base dessa perspectiva curricular, o problema é que o paradigma ideológico do liberalismo promoveu uma perspectiva individualista, que privilegia a ação individual em detrimento da coletiva, que garante a primazia da liberdade em face da igualdade (ADORNO, 2019) e que recusa projetos coletivos e sociais de desenvolvimento nacional, já que a educação deixa de ser um direito no campo político-social e passa a ser um produto no campo econômico, ora como serviço, ora como catalisador das relações mercadológicas (MACHADO, 2009). A finalidade do currículo Humanista é o indivíduo e sua autorrealização. Assim, valoriza-se a percepção de mundo a partir das experiências; autonomia - liberdade - responsabilidade; professor não deve dar a resposta, sua função é levar o aluno a questionar. Esse objetivo é alcançado quando o educando tem a possibilidade de vivenciar “[...] situações que lhe possibilitem descobrir e realizar sua própria individualidade, agindo, experimentando, errando, avaliando, reordenando e expressando. Tais situações ajudam os educandos a integrar emoções, pensamentos e ações.” (SANTOS; CASALI, 2009, p. 220). REV. ESPAÇO DO CURRÍCULO (ONLINE), JOÃO PESSOA, V.16, N.1, P. 1-12, JAN/ABR. 2023. 5 LOPES, Q. V.; TIROLI, L. G.; SANTOS, A. R. J. definem os conteúdos e aqueles que ministram os mesmos conteúdos. definem os conteúdos e aqueles que ministram os mesmos conteúdos. Quando se ensina, é preciso se questionar o que se ensina, contra quem se ensina e a favor de quem se ensina, quem são os educandos, quem escolhe o currículo, porque escolhe e assim por diante (PIMENTA, 1999; FREIRE, 2002; FREIRE; SHOR, 2021). Essa visão de currículo Acadêmico inviabiliza uma prática pedagógica crítica e propositiva, em que o docente não seja mero repetidor do conhecimento dogmático, posto e acabado, mas uma ação docente em que o “professor internaliza uma postura crítica de interpretação da realidade e assume o comprometimento social como mola de reconstrução dessa realidade” (RIBEIRO JÚNIOR, 2001, p. 14). Assim, o professor não deve se tornar um eterno repetidor, mas precisa estar aberto a manifestação da alteridade no relacionamento com os alunos, construindo uma prática pedagógica alicerçada no respeito às diferenças, de modo a centralizar o discente no processo de ensino e aprendizagem. Não na perspectiva de abandono do aluno na hercúlea missão de reter o conteúdo ministrado, mas na valorização dos conhecimentos prévios, da realidade política, econômica, cultural e social e ainda das necessidades específicas relativas às habilidades e dificuldades em compreender, contextualizar e analisar criticamente o conteúdo. Na prática, a postura libertadora do professor pressupõe a escuta do aluno, pois “escutar o aluno é saber ser professor, é saber se curvar ao novo” (DUSSEL, 1980, p. 50). A postura dialógica consiste em ouvir e ser ouvido, em dar voz aos silenciados historicamente e permitir sua participação na construção coletiva do conhecimento, respeitando os conhecimentos prévios dos alunos que tem origem na experiência cotidiana e reconhecendo que eles têm condições de contribuir para o aprendizado coletivo. Consequentemente, a experiência do ensino dialógico é essencialmente uma experiência de escuta de qualidade (HOOKS, 2017), sendo responsabilidade do professor criar um ambiente em que os alunos aprendam a falar e a ouvir o outro com seriedade e respeito. Entretanto, essa postura não pode incorrer no equívoco de privilegiar o indivíduo em detrimento do coletivo, como ocorre no currículo Humanista, essa alteridade e abertura para a manifestação do sujeito em aprendizagem deve ser concebida no contexto sócio-histórico-cultural em que se está inserido, não pode ser focado na individualidade, mas em perspectivas coletivas e sociais. definem os conteúdos e aqueles que ministram os mesmos conteúdos. Isto é, a experiência cotidiana do aluno e sua manifestação como possibilidade de contextualização dos conhecimentos historicamente produzidos pelo trabalho e acumulados pela humanidade, que não são absolutos, mas passíveis de crítica e contextualização, inclusive pela escola, que não pode ser vista como mera reprodutora, mas também criadora do conhecimento acumulado. 3.2 Currículo Acadêmico Nesse ponto, um dos problemas que podem ser destacados a respeito dessa vertente está assentando na neutralidade epistemológica em que se reveste os conteúdos tidos como os ‘úteis’ ou ‘adequados’ para a formação dos cidadãos. Nas palavras de Apple (1982, p. 49), “[...] o conhecimento introduzido nas escolas é geralmente aceito como dado, como neutro, de maneira que se podem fazer comparações entre grupos sociais, escolas, crianças, etc.” Essa pseudoneutralidade afasta o caráter político, social e cultural do currículo e aparta os agentes envolvidos nos processos educacionais da definição dos conteúdos trabalhados, refletindo uma relação de contraposição entre aqueles que REV. ESPAÇO DO CURRÍCULO (ONLINE), JOÃO PESSOA, V.16, N.1, P. 1-12, JAN/ABR. 2023. 6 CURRÍCULO: diversidade e tensões no contexto educacional 3.3 Currículo Tecnológico O Currículo Tecnológico possui como método de pesquisa o positivismo, empregando o empirismo. De acordo com McNeil (1981), como o próprio nome do currículo indica, tem como ponto chave a tecnologia, sendo esta utilizada como fonte de ensino, visando a eficácia dos programas, métodos e materiais para atingir finalidades pré-determinadas. Esta concepção tem como referência o estímulo-resposta, em que a motivação do sujeito é extrínseca e não intrínseca, ou seja, o sujeito só se envolve, só realiza determinada ação, se for recompensado, em outras palavras, trabalha com o estilo permuta: agi por algo em troca. A Tecnologia, como perspectiva curricular, considera que o homem é uma consequência das influências existentes no meio, o homem se constitui a partir de suas heranças genética e ambiental. Portanto, este se adapta ao meio. De acordo com McNeil (1981), a tecnologia é empregada no currículo como um plano para o uso sistemático de vários dispositivos e mídia, e com uma sequência de instrução inventada com base em princípios da ciência comportamental. Por conseguinte, a tecnologia também é encontrada em modelos e procedimentos para construção ou desenvolvimento e avaliação de materiais curriculares e sistemas instrucionais. O processo de desenvolvimento pode ser declarado como regras que, se seguidas, resultarão em produtos mais previsíveis. Conforme disposto por Santos e Casali (2009, p. 221), “o desenvolvimento do sistema de ensino- aprendizagem, segundo hierarquia de tarefas, constitui o eixo central do planejamento do ensino REV. ESPAÇO DO CURRÍCULO (ONLINE), JOÃO PESSOA, V.16, N.1, P. 1-12, JAN/ABR. 2023. 7 LOPES, Q. V.; TIROLI, L. G.; SANTOS, A. R. J. proposto em termos de uma linguagem objetiva, esquematizadora e concisa”. Agentes educacionais estão acostumados como a utilização dessa corrente do currículo com o uso de seus instrumentos no ensino, como computador, sistemas de aprendizagem individualizados, entre outras ferramentas tecnológicas, porém, nem todos possuem a real convicção que a tecnologia também é um processo de análise de problemas e elaboração, implementação, avaliação e gerenciamento de soluções (McNeil, 1981). Percebe-se que as instituições educacionais fazem uso constante da tecnologia em muitas das atividades educacionais, todavia, nem sempre essas são programadas, constando de um documento norteador que planifique e direcione o trabalho a ser desenvolvido. Esta lacuna pode afastar do atingimento de fins pedagógicos pré-estabelecidos, que caracteriza o currículo Tecnológico. De maneira sucinta, verifica-se que a educação é vista com uma perspectiva de interesse social. 3.4 Currículo Reconstrucionista Social O currículo Reconstrucionista Social tem como base metodológica o Materialismo Histórico Dialético, na perspectiva da Teoria Histórico-cultural, tendo como matriz pedagógica a Pedagogia Histórico-Crítica e como Teoria da Aprendizagem a Teoria da Atividade. De acordo com McNeil (1981), concepções do reconstrucionismo surgiram no currículo americano nas décadas de 1920 e 1930, voltando-se para valores e questões sociais que as escolas precisavam explanar. A gênese do trabalho dos primeiros reconstrucionistas sociais educacionais situam-se na premissa de que as escolas exercem poder na transformação e/ou manutenção sobre as diversificadas camadas da sociedade, ou seja, toma- se por base que as escolas são agentes efetivos que possibilitam o equilíbrio das camadas sociais, visão considerada por muitos autores como bastante otimistas (APPLE, 1982). Enfatiza-se, nesta vertente curricular, a necessidade de manutenção dos interesses coletivos sobre os individuais, ou seja, os interesses da sociedade sobre os interesses de particulares. Nessa lógica, recai sobre o currículo a incumbência de garantir uma sociedade mais crítica e participativa, que se envolva e tente solucionar os problemas existentes na coletividade. Desta forma, o desenvolvimento de valores sociais e do pensamento crítico são partes integrantes e imprescindíveis deste currículo (MCNEIL, 1981). É importante ressaltar que há interesses individuais que são levados em conta quando da formação do sujeito, porém, esse sujeito é estimulado a refletir sobre necessidades sociais, ações democráticas, em suma, sobre a transformação da sociedade visando um mundo melhor para a coletividade, e assim auxiliar nessa mudança à medida que realiza os seus objetivos, uma vez que esse é dotado de capacidade de alcançar por si próprio os seus propósitos. Com relação à constituição do homem, conjectura que se desenvolve a partir do seu meio, da sua realidade, do meio social, sendo concebido de muitas identidades, uma vez que o homem é um ser histórico e social. Outrossim, o homem é percebido vinculado no e com o mundo, sendo que o homem e mundo são considerados conjuntamente, pois o homem influencia e é influenciado. Ademais, considera- se que o homem cria a cultura à medida que reflete seu contexto de vida. Portanto, a cultura é transmitida, porém, pode ser modificada, uma vez que existe a ação do homem no contexto social, visto que este é um ser ativo. 3.3 Currículo Tecnológico Trabalha com alcance de metas, tendo como foco habilidades e competências. A escola é vista como agência educacional, preparando o sujeito para o mercado de trabalho. O ensino, por conseguinte, é possível ser programado, desde que se possa prever o resultado. O aluno, por sua vez, geralmente trabalha sozinho, é conduzido a possuir comportamentos apropriados e a dominar os objetivos propostos pelo currículo, com intuito de aprimorar suas habilidades e competências, dividindo essa responsabilidade com o contexto educacional. entendimento da função da escola neste cenário. Desse modo, expõe-se que, entendimento da função da escola neste cenário. Desse modo, expõe-se que, O que caracteriza a ideologia não é a falsidade ou verdade das ideias que veicula, mas o fato de que as ideias são interessadas, transmitem uma visão do mundo social vinculada aos interesses dos grupos situados, em uma posição de vantagem na organização social. A ideologia é essencial na luta desses grupos pela manutenção das vantagens relativas que lhes advêm dessa posição privilegiada. É muito menos importante saber se as ideias envolvidas na ideologia são falsas ou verdadeiras e muito mais importante saber que vantagens relativas e que relações de poder elas justificam ou legitimam. (MOREIRA; TADEU, 2013, p. 31). Ressalta-se, conforme disposto por Althusser apud Moreira e Tadeu (2013) que a ideologia não está constante somente no campo das ideias, visto que também se consubstancia em práticas materiais e, nesse sentido, o seu significado não tem terminalidade tamanho é a amplitude de sua análise e sua consecução, principalmente no campo educacional. Nessa lógica, o professor deve levar o aluno a refletir sobre o contexto para assim alcançar as respostas necessárias às suas inquietações, sem interferências divergentes daquilo que se almeja. Os estudos, por esse ângulo, partem da realidade do aluno, da realidade concreta, permeada pelos contextos histórico e social. Portanto, conhecimento e educação são construídos na historicidade, em processo de promover o sujeito e não de ajustá-lo à sociedade. Quanto à avaliação do processo educacional nesta corrente curricular, registra-se que ocorre de forma diagnóstica e contínua, realizada individualmente, sendo um momento de reflexão. Santos e Casali definem o papel pedagógico do currículo Reconstrucionista Social no contexto educacional afirmando que, Na perspectiva de reconstrução social, agrupam-se as posições que consideram o ensino uma atividade crítica, cujo processo de ensino-aprendizagem deve se constituir em uma prática social com posturas e opções de caráter ético que levem à emancipação do cidadão e à transformação da realidade. (SANTOS; CASELI, 2009, p. 221). Capta-se que para a este currículo é necessário trabalhar conteúdos escolares, porém, de forma contextualizada, ou seja, entende-se que é necessário ensinar conhecimentos científicos, mas, contextualizando esses conhecimentos, diferente do currículo Acadêmico, que trabalha com conhecimentos científicos, contudo, não expressa essa contextualização. Segundo Apple, Ao contrário do modelo de desempenho acadêmico, a abordagem baseada na sociabilização não deixa necessariamente de examinar o conhecimento escolar. entendimento da função da escola neste cenário. Desse modo, expõe-se que, De fato, um de seus interesses fundamentais está em explorar as normas e valores sociais transmitidos pelas escolas. Contudo, devido a esse interesse, restringe-se ao estudo do que poderia se chamar de “conhecimento moral”. Estabelece como dado o conjunto de valores sociais e investiga como a escola, enquanto agente da sociedade, sociabiliza os estudantes com seu conjunto “compartilhado” de regras e tendências normativas. (APPLE, 1982, p. 49). Entretanto, quanto aos conhecimentos científicos, uma das críticas dos reconstrucionistas, em conformidade com McNeil (1981), é voltada para o fato de que as descobertas científicas permitem interpretações variadas. Ademais, por possuir a visão de transformação da sociedade, também possui olhares que o consideram otimista demais. Em suma, o currículo Reconstrucionista Social entende que por meio do processo educacional torna-se viável a transformação da sociedade, visualizando a escola como um agente valoroso no processo de mudanças na coletividade, diminuindo o desequilíbrio existentes nas camadas sociais, à medida que correlaciona o ensino no contexto histórico e social, levando os indivíduos à reflexão e, consequentemente, à ação crítica sobre a realidade, com o intuito de tonar o mundo mais justo. CURRÍCULO: diversidade e tensões no contexto educacional CURRÍCULO: diversidade e tensões no contexto educacional 3.4 Currículo Reconstrucionista Social Destarte, como este homem é desenvolvido para refletir sobre a realidade, e de forma colaborativa contribuir para a sua transformação, pode considerar a necessidade de mudanças continuamente, em conformidade com as reflexões realizadas a partir das possibilidades dispostas por este currículo no contexto educacional. Sustenta-se a ideia de uma educação popular tendo como parâmetro todos os sujeitos, e não somente uma classe, que geralmente se classifica em classe dominante. A escola é vista como um aparelho ideológico, sendo necessário buscar uma transformação, mudar a atitude do sujeito diante das questões sociais. Importante aqui tecer uma reflexão sobre o significado de ideologia, para um melhor REV. ESPAÇO DO CURRÍCULO (ONLINE), JOÃO PESSOA, V.16, N.1, P. 1-12, JAN/ABR. 2023. 8 CONSIDERAÇÕES FINAIS REV. ESPAÇO DO CURRÍCULO (ONLINE), JOÃO PESSOA, V.16, N.1, P. 1-12, JAN/ABR. 2023. 9 O campo de investigação curricular não pode ser resumido à determinação legal-institucional dos O campo de investigação curricular não pode ser resumido à determinação legal-institucional dos REV. ESPAÇO DO CURRÍCULO (ONLINE), JOÃO PESSOA, V.16, N.1, P. 1-12, JAN/ABR. 2023. 9 LOPES, Q. V.; TIROLI, L. G.; SANTOS, A. R. J. conteúdos que devem ser trabalhados no contexto educacional. O currículo é campo de tensões, embates e conflitos políticos, ideológicos, sociais, culturais, teóricos e de visões de mundo, e o processo de investigação precisa considerar esses condicionantes. Isto é, o currículo implica e é implicado na realidade posta, e por meio da pesquisa, torna-se possível analisar e problematizar esse processo dialético. Na consecução de responder ao problema de pesquisa, foi possível discorrer e analisar quatro perspectivas teóricas-metodológicas que norteiam a concepção de currículo e suas implicações para o campo educacional, organizadas por McNeil, tem-se o currículo humanista, o currículo acadêmico, o currículo tecnológico e o currículo reconstrucionista. Embora seja relevante a perspectiva humanista, sobretudo por considerar as necessidades dos alunos e focar na autonomia e autoavaliação, essa vertente esbarra na dimensão sócio-histórico-cultural que atravessa o campo educacional, portanto, não considera que o aluno é sujeito contextualmente inserido em uma determinada realidade. Logo, o currículo humanista carece de complementação por não observar essas dimensões, pois mantém o foco no sujeito em uma perspectiva liberalista-subjetivista-individualista. Quanto ao currículo acadêmico, não obstante privilegie os conteúdos clássicos, desconsidera a participação dos alunos no processo de produção e apropriação desse conhecimento. Assim, a ciência é a entidade responsável por elaborar e legitimar o conteúdo que será ministrado de maneira transmissiva, do professor, que internalizou o conteúdo e, portanto, sabe, para o aluno, que nada sabe. Desta maneira, o currículo acadêmico inviabiliza que a escola possa se tornar um ambiente de elaboração do conhecimento, e não somente reprodutora do conhecimento. O currículo Tecnológico trabalha com a ciência comportamental, tendo como base a tecnologia, que pode possibilitar a melhoria do processo educacional, por meio de instrumentos tecnológicos, com programas, métodos e materiais hábeis, para melhor eficácia de seus resultados. Contudo, para atingimento dos propósitos de ensino, faz-se necessário a planificação prévia dos objetivos a serem alcançados, bem como a programação das ações a serem realizadas. CONSIDERAÇÕES FINAIS Nesta corrente curricular, a escola é como uma agência educacional, professores são agentes guias no processo de ensino e os alunos são agentes principais, detentores de responsabilidades coletivas com os demais entes do contexto escolar, que devem aprimorar suas habilidades e competências para o mundo do trabalho. Com intuito de contribuir para a transformação da sociedade, em uma perspectiva de resolução de conflitos entre interesses coletivos em detrimento de interesses individuais, na assertiva da conversão de indivíduos alienados em sujeitos mais críticos e participativos, o currículo Reconstrucionista Social requer uma estreita relação da escola com a sociedade. Essa premissa se dar ao fato de que é por meio do currículo que se desenvolve na escola que se torna possível promover a reflexão dos sujeitos, revelando, por meio do ensino, valores e ideologias culturais, sociais e econômicas, que podem contribuir para mudanças necessárias, com vistas a melhoria da comunidade, suprindo as necessidades sociais, uma vez o ensino é voltado para a aplicabilidade prática dos conhecimentos que devem transcender o espaço escolar. Sintetizando, verifica-se que o currículo é marcado pelas relações políticas, econômicas, culturais e sociais, desencadeando, portanto, um processo de relações de poder, e por isso mesmo não pode ser considerado neutro. Assim, em um contexto educacional, deve-se adotar o currículo que seja o mais adequado para o atendimento dos anseios da comunidade o qual este está inserido. Portanto, para que o currículo seja satisfatório às pretensões dispostas, faz-se necessário o conhecimento de cada currículo para determinar a escolha pelo mais pertinente às exigências da contemporaneidade. REFERÊNCIAS ADORNO, Sérgio. Os aprendizes do poder: o bacharelismo liberal na política brasileira. 2. ed. rev. São Paulo: Edusp, 2019. BRASIL. Lei n.º 9.394, de 20 de dezembro de 1996. Estabelece as diretrizes e bases da educação nacional. Disponível em: http://www.planalto.gov.br/ccivil_03/leis/l9394.htm. Acesso em: 17 jan. 2021. DUSSEL, Enrique. Pedagogica latinoamericana. Bogotá: Nueva America, 1980. FREIRE, Paulo. Pedagogia da esperança: um reencontro com a pedagogia do oprimido. 9. ed. São Paulo: Paz e Terra, 2002. FREIRE, Paulo. Pedagogia do oprimido. 74. ed. São Paulo: Paz e Terra, 2020. REIRE, Paulo; SHOR, Ira. Medo e ousadia: o cotidiano do professor. 14. ed. São Paulo: Paz e T GUEDES, Neide Cavalcante; ARAUJO, Hilda Mara Lopes. Fios se unem nas tramas do currículo e da formação. Revista e-Curriculum, São Paulo, v. 15, n. 3, p. 764–786, jul./set.2017. HOOKS, Bell. Ensinando a transgredir: a educação como prática da liberdade. São Paulo: Martins Fontes, 2017. LLAVADOR, F. Beltrán. Las determinaciones y el cambio del currículo. In: ANGULO, José Félix; BLANCO, Nieves (Coord.). Teoría y desarrollo del currículo. Málaga: Ediciones Aljibe, 1994. p. 369-383. LUCKESI, Cipriano Carlos. Tendências pedagógicas na prática escolar. In: LUCKESI, Cipriano Carlos. Filosofia da Educação. São Paulo: Cortez, 2005. p. 53- 75. MACHADO, Antônio Alberto. Ensino jurídico e mudança social. 2. ed. São Paulo: Expressão Popular, 2009. MCNEILL, John. D. Curriculum: A Comprehensive Introduction. 2. ed. Boston, Toronto: Little Brown and Company, 1981. MOREIRA, Antonio Flavio Barbosa; SILVA, Tomaz Tadeu da (org.). Currículo, cultura e sociedade. 2. ed. São Paulo: Cortez, 1997. p. 7-31. MOREIRA, Antonio Flavio. TADEU, Tomaz (org.). Currículo, cultura e sociedade. 12. ed. São Paulo: Cortez, 2013. MOREIRA, Antonio Flavio Barbosa. Formação de professores e currículo: questões em debate. Revista Ensaio: avaliação em políticas públicas em educação, Rio de Janeiro, v. 29, n.110, p. 35-50, jan./mar. 2021. PEREIRA, Luiz Ismael. Adorno e o direito: para uma crítica do capitalismo e do sujeito de direito. São Paulo: Ideias e Letras, 2018. PIMENTA, Selma Garrido. Formação de professores: identidade e saberes. In: PIMENTA, Selma Garrido (org). Saberes pedagógicos e atividade docente. São Paulo: Cortez, 1999. p. 15-34. RIBEIRO JÚNIOR, João. A formação pedagógica do professor de direito: conteúdos e alternativas metodológicas para a qualidade do ensino do direito. Campinas: Papirus, 2001. SACRISTÁN, José Gimeno (org.). Saberes e incertezas sobre o currículo. Tradução por Alexandre Salvaterra. Revisão Técnica por Miguel González Arroyo. Porto Alegre: Penso, 2013. SANTOS, Adriana Regina de Jesus; CASALI, Alipio Marcio Dias. Currículo e educação: origens, tendências e perspectivas na sociedade contemporânea. Olhar de Professor, Ponta Grossa, v. 12, n. 2, p. 207-231, 2009. SANTOS, Adriana Regina de Jesus; LOPES, Rosana Pereira; COSTA, Rogerio da. APPLE, Michael Whitman. Ideologia e currículo. Tradução de Carlos Educardo Ferreira de Carvalho. São Paulo: Brasiliense, 1982. APPLE, Michael Whitman. Ideologia e currículo. Tradução de Carlos Educardo Ferreira de Carvalho. São Paulo: Brasiliense, 1982. BRASIL. Lei n.º 9.394, de 20 de dezembro de 1996. Estabelece as diretrizes e bases da educação nacional. Disponível em: http://www.planalto.gov.br/ccivil_03/leis/l9394.htm. Acesso em: 17 jan. 2021. ADORNO, Sérgio. Os aprendizes do poder: o bacharelismo liberal na política brasileira. 2. ed. rev. São Paulo: Edusp, 2019. ADORNO, Theodor Ludwig Wiesengrund. Teoria da semicultura. Tradução Newton Ramos de Oliveira. Educação e Sociedade, Campinas, v. 15, n. 56, p. 338-411, dez. 1996. ADORNO, Theodor Ludwig Wiesengrund. Teoria da semicultura. Tradução Newton Ramos de Oliveira. Educação e Sociedade, Campinas, v. 15, n. 56, p. 338-411, dez. 1996. APPLE, Michael Whitman. A luta pela democracia na educação crítica. Revista e-Curriculum, São Paulo, v. 15, n. 4, p. 894-926, out./dez. 2017. Disponível em: APPLE, Michael Whitman. A luta pela democracia na educação crítica. Revista e-Curriculum, São Paulo, v. 15, n. 4, p. 894-926, out./dez. 2017. Disponível em: REV. ESPAÇO DO CURRÍCULO (ONLINE), JOÃO PESSOA, V.16, N.1, P. 1-12, JAN/ABR. 2023. 10 https://revistas.pucsp.br/index.php/curriculum/article/view/35530. Acesso em: 25 mar. 2022. BRASIL. Lei n.º 9.394, de 20 de dezembro de 1996. Estabelece as diretrizes e bases da educação nacional. Disponível em: http://www.planalto.gov.br/ccivil_03/leis/l9394.htm. Acesso em: 17 jan. 2021. Os sentidos referentes à classe social e relações de poder presentes no contexto das teorias curriculistas tradicionais e críticas. Revista e-Curriculum, São Paulo, v.15, n.2, p. 325 – 344 abr./jun.2017. SAVIANI, Dermeval. Pedagogia histórico-crítica: primeiras aproximações. 11. ed. Campinas: Autores Associados, 2011. SAVIANI, Dermeval. Pedagogia histórico-crítica: primeiras aproximações. 11. ed. Campinas: Autores Associados, 2011. REV. ESPAÇO DO CURRÍCULO (ONLINE), JOÃO PESSOA, V.16, N.1, P. 1-12, JAN/ABR. 2023. 11 LOPES, Q. V.; TIROLI, L. G.; SANTOS, A. R. J. SILVA, Maria Aparecida da. História do Currículo e Currículo como construção histórico-cultural. In: VI Congresso luso-brasileiro de História da Educação, 2006, Uberlândia. Anais do VI Congreso Luso- Brasileiro de História da Educação. Uberlândia: EDUFU, 2006. v. 1. p. 4820-4828. SILVA, Maria Aparecida da. História do Currículo e Currículo como construção histórico-cultural. In: VI Congresso luso-brasileiro de História da Educação, 2006, Uberlândia. Anais do VI Congreso Luso- Brasileiro de História da Educação. Uberlândia: EDUFU, 2006. v. 1. p. 4820-4828. SILVA, Maria Aparecida da. História do Currículo e Currículo como construção histórico-cultural. In: VI Congresso luso-brasileiro de História da Educação, 2006, Uberlândia. Anais do VI Congreso Luso- Brasileiro de História da Educação. Uberlândia: EDUFU, 2006. v. 1. p. 4820-4828. Esta obra está licenciada com uma Licença Creative Commons Atribuição-NãoComercial 4.0 Internacional. REV. ESPAÇO DO CURRÍCULO (ONLINE), JOÃO PESSOA, V.16, N.1, P. 1-12, JAN/ABR. 2023. 12
https://openalex.org/W2578230457	https://dro.deakin.edu.au/articles/journal_contribution/Students_Attitudes_to_Universal_Design_in_Architecture_Education/20865337/1/files/37128682.pdf	English	null	Students’ Attitudes to Universal Design in Architecture Education	Journal of social inclusion	2,016	cc-by	8,266	Students’ Attitudes to Universal Design in Architecture Education AUTHOR(S) Helen Larkin, K Dell, Danielle Hitch PUBLICATION DATE 01-01-2016 HANDLE 10536/DRO/DU:30090248 Downloaded from Deakin University’s Figshare repository Deakin University CRICOS Provider Code: 00113B Students’ Attitudes to Universal Design in Architecture Education AUTHOR(S) Helen Larkin, K Dell, Danielle Hitch Helen Larkin, K Dell, Danielle Hitch 10536/DRO/DU:30090248 Downloaded from Deakin University’s Figshare repository Deakin University CRICOS Provider Code: 00113B Reproduced by Deakin University under the terms of the Creative Commons Attribution Non- Commercial No-Derivatives Licence Students’ attitudes to universal design in architecture education Citation: Citation: Larkin, Helen, Dell, Kelsey and Hitch, Danielle 2016, Students’ attitudes to universal design in architecture education, Journal of social inclusion, vol. 7, no. 2, pp. 18-34. URL: https://josi.journals.griffith.edu.au/index.php/inclusion/article/view/697 ©2016, The Authors ©2016, The Authors Reproduced by Deakin University under the terms of the Creative Commons Attribution Non- Commercial No-Derivatives Licence Helen Larkin Deakin University Kelsey Dell Deakin University Danielle Hitch Deakin University Students’ Attitudes to Universal Design in Architecture Education Helen Larkin Deakin University Downloaded from DRO: Deakin University CRICOS Provider Code: 00113B Students’ Attitudes to Universal Design in Architecture Education Abstract It is widely recognised that the built environment can dramatically impact on the participation and engagement of people with disability and diverse needs. It has therefore become necessary for architects and designers to consider these needs when working within their profession. The implementation of universal design teaching into architecture and design curriculum has been recognised as an important step in facilitating and enhancing the uptake of universal design during the design process. Using a quantitative approach, this study aimed to compare, contrast and explore the attitudes of two groups of architecture students to the universal design of built environments. One group had received education relating to diversity and universal design as part of a prior project while the other group had not received this content. Findings from this study demonstrated that while there were no significant differences between the two groups, interesting insights and perspectives emerged regarding student attitudes to universal design and the potential influencers of these attitudes. ords: Universal Design, Architecture, Curriculum, Higher Education, Built nment ywords: Universal Design, Architecture, Curriculum, Higher Education, Bu vironment The International Classification of Functioning, Disability and Health (ICF) “acknowledges that every human being can experience a decrement in health and thereby experience some disability” (World Health Organization [WHO] 2002, p. 3). It is now widely recognised that the physical or built environment can dramatically impact the participation and engagement of people with diverse needs, via the inclusion in the ICF of environmental contextual factors (WHO, 2013). This issue has been acted upon through the United Nations Convention on the Rights of Persons with Disabilities where Article 9 specifically requires signatories to eliminate environmental obstacles and barriers (United Nations, 2006), however the implementation of this convention has not been complete. For example; more than 27% of respondents in an Australian study stated the built environment was a barrier to their full participation in the community (National People with Disabilities and Carer Council, 2009). orresponding author: Helen Larkin (helen.larkin@deakin.edu.au) Corresponding author: Helen Larkin (helen.larkin@deakin.edu.au) Journal of Social Inclusion, 7(2), 2016 Despite this global shift in attitudes and policy in recent years, it is yet to transfer to a major shift in the education programs of architects and designers. Abstract Ostroff (2011) argued that there is “only a handful of universities around the world where universal design, or inclusive design, or design for all is even an elective within the professional curriculum” (p.1.9), and there remains a prevalence of architecturally designed environments around the world that continue to create disadvantage for many (Jones, 2014). It is therefore necessary to explore the attitudes of architects and designers to these global challenges. However, as noted by Hitch, Dell & Larkin (2016), there are relatively few published studies around inclusive design curriculum delivered to architecture and design students during their professional training. Professional training is a key site for socialisation into a profession’s values and practices, and previous research suggests that architects’ anticipatory socialisation prior to entering the profession emphasises creativity over administrative or regulatory tasks (Sang et al., 2009). An understanding of the attitudes of architecture students towards inclusive and universal design can inform this aspect of the curriculum, by highlighting any misconceptions or incongruencies between their perspectives and desired professional values. A handful of authors (for example, Altay & Demirkan, 2013; Bernardi & Kowaltowski, 2010; Chang et al., 2000; Evcil, 2012; Helvacioglu & Karamanoglu, 2012; Larkin et al., 2013) have investigated the implementation of universal design into architecture and design curriculum in higher education as a means to encourage new graduates to utilise Universal Design Principles and to enhance the implementation in architectural practice. With an increasing demand on being sensitive to diverse user needs, it is important for upcoming architects to develop new abilities and attitudes towards design (Bernardi & Kowaltowski, 2010). For this to occur, adequate training and education within the design curriculum is required. Many authors have supported this and argued the importance of universal design implementation in design curriculum (Bernardi and Kowaltowski, 2010; Evcil, 2012; Helvacioglu & Karamanoglu, 2012; Olgunturk & Demirkan, 2009; Paulsson, 2005; Watchorn et al., 2013). Bernardi and Kowaltowski (2010) provided further support, stating the curricula must adapt its focus in order to maintain architecture education that is relevant to current public and political debates. However little evidence exists demonstrating how to do this or the most effective methods to enhance attitudes to universal design. The concept of universal design developed from the field of architecture, when Ronald Mace challenged conventional design approaches and provided a design foundation for products and environments that were more usable and accessible (Burgstahler, 2012). Abstract In 1989, Mace established the Centre for Universal Design at North Carolina State University (The Centre for Universal Design, 2008). Here, he worked with others to define universal design as the “design of products and environments to be usable by all people, to the greatest extent possible, without the need for adaption or specialized equipment” (Case 2003, p. 2). From the Centre, seven principles of universal design were established for application to product development and are displayed in Table 1. 19 19 Journal of Social Inclusion, 7(2), 2016 Table 1: Principles of Universal Design Principle Descriptor (1) Equitable use The design is useful and marketable to people with diverse abilities (2) Flexibility in use The design accommodates a wide range of individual preferences and abilities (3) Simple and intuitive use Use of the design is easy to understand, regardless of the user’s experience, knowledge, language skills or current concentration level (4) Perceptible information The design communicates necessary information effectively to the user, regardless of ambient conditions or the user’s sensory abilities (5) Tolerance for error The design minimizes hazards and the adverse consequences of accidental or unintended actions (6) Low physical effort The design can be used efficiently and comfortably and with a minimum of fatigue (7) Size and space for approach and use Appropriate size and space is provided for approach, reach, manipulation, and use regardless of user’s body size, posture or mobility Note. Source: Adapted from Connell et al., 1997. Copyright 1997 by the Centre for Universal Design. Table 1: Principles of Universal Design There remains limited research that explores architects’ application of universal design during the design process and discussions predominantly revolve around issues and barriers to this application. Identified issues relating to universal design application include a lack of knowledge regarding universal design itself and a lack of an efficient and transparent manner for the dissemination of current research and knowledge (Heylighen, 2008), lack of an assessment or evaluation tool (Preiser, 2008) and lack of systematic procedures and priority recommendations (Afacan & Demirkan, 2010; Preiser, 2008). Karol (2008) also briefly discussed the lack of universal design application due to limited consumer requests. The translation of universal design into architectural and design practice, therefore remains a vexed question (Jones, 2014) In 2010, the Design 4 Diversity: Enhancing inter-professional learning for architecture and occupational therapy students initiative was undertaken at Deakin University, Geelong (Larkin et al., 2010a). Abstract The project focused on inter-professional education for architecture and occupational therapy students in relation to universal design practice. Key aims of the project were to; explore the perspective of key industry stakeholders; develop and trial flexible blended teaching and learning resources; provide rich experiences regarding inclusive design practice; and, to explore and identify sustainable opportunities for architecture and occupational therapy students to learn about inclusive design practice within a framework of inter-professional education (Larkin et al., 2010a). Findings from this 20 Journal of Social Inclusion, 7(2), 2016 project included the identification of major enablers and barriers to the uptake of universal design (Watchorn, Hitch, Larkin and Ang, 2014). In a pre and post-test study design, results of the evaluation of the Design 4 Diversity initiative showed that occupational therapy students were significantly more positive initially about the prospect of an inter-professional practice experience compared to their architecture student colleagues and while this trend continued at the end of the project, findings indicated that occupational therapy students actually became less positive in some areas (Larkin, Hitch, Watchorn & Ang, (2013). Additional findings included that architecture and occupational therapy students when provided with ‘real-life’ versus Second Life™ simulations of disability, found the ‘real-life’ simulations more authentic and helpful (Watchorn, Larkin, Ang & Hitch, 2013) The aim of the study in this paper is to report on a follow-up study to the evaluations that were undertaken at the time as part of the 2010 Design 4 Diversity initiative. Given that the intent of the project was to attempt to influence the attitudes of architecture students both locally and internationally, this quantitative study aimed to measure the attitudes regarding universal design of those architecture students who participated in Design 4 Diversity and to compare and contrast them to the attitudes of a group of architecture students who were not part of the project. The specific research questions addressed in this study were: 1) What are the attitudes of architecture students to universal design; and 2) Is there a statistically significant difference in attitudes to the universal design of built environments between Group One and Group Two? Method This study used a quantitative quasi-experimental design, comparing two naturally occurring groups (Punch, 2005). Group One received the intervention (access to an inclusive design specific curriculum), while Group Two was a control group. Using this method the researcher was provided with descriptive data about the sample and a numerical representation of the attitudes of architecture students to the universal design of built environments. Procedure Ethics approval Ethics approval Prior to commencing this study, ethics approval was obtained from Deakin University (HEAG-H 79_2013) Participant recruitment Architecture students in the targeted classes received a brief oral presentation from the researcher, 20 minutes prior to the conclusion of the class. This presentation outlined the research project and invited architecture students to participate. The researcher had no prior relationship with the architecture students, which increased the likelihood of participation being completely voluntary by avoiding power-dependencies or coercion (Portney & Watkins, 2009). Sample p The sample was recruited via purposive sampling from two separate population groups. Third and fourth year architecture students who had participated in the Design 4 Diversity program (Group One) and fifth year architecture students who had not participated in the program (Group Two). This method of sampling ensured participants were a representative sample for the study (Portney & Watkins, 2009). Inclusion criteria for the study were:  Students undertaking a Bachelor of Design (Architecture) at a large regional Australian University in their third year of study in 2013;  Students undertaking a Bachelor of Design (Architecture) at a large regi g g ( ) g g Australian University in their third year of study in 2013;  Students undertaking a Master of Architecture and in their fourth or fifth year of study at a large regional Australian University in 2013;  Students who began a Bachelor of Design (Architecture) at a large regional Australian university and have continued through the degree without a break and without studying at any other educational institution; y g y  Students aged 18 years and over; and, Students aged 18 years and over; and,  Students who are able to speak English to a level that supports study at university. Participants were excluded from the study if they did not meet the above inclusion criteria. A total of 370 architecture students were enrolled in the targeted classes. Of these students, 147 (39.7%) returned questionnaires. Another 13 were excluded as the participants did not meet the inclusion criteria or questionnaires were returned incomplete. This left a total of 134 21 Journal of Social Inclusion, 7(2), 2016 completed questionnaires and a response rate of 36.2%. Of these, 82 (64.9%) were completed by students in their third or fourth years of study (Group One) and 52 (38.8%) were completed by students in their fifth year of study (Group Two). Respondents in Group One (students who had participated in Design 4 Diversity) comprised 53 males (64.6%) and 29 females (35.4%). Participant age ranged from 20 years to 58 years, with a mean age of 24.5 years (SD=6.156). Participants in Group Two (students who had not participated) comprised of 34 males (65.4%) and 18 females (34.6%). Participant age ranged from 22 years to 48 years, with a mean age of 26.5 years (SD=5.573). Instrument A questionnaire entitled An exploration of the attitudes of architecture students to the universal design of built environments was developed for the purpose of this study as none were in existence that met our needs. The questionnaire included a compilation of closed (51) and open (2) questions (both of which invited further comments in relation to the quantitative questions). It incorporated: the collection of demographic information; a standardised assessment (the Interaction with Disabled Persons Scale (IDP) (Gething, 1991)); an adaption of a published assessment in relation to universal design (the Universal Design Performance Measure for Products (The Centre for Universal Design 2000)); and specifically developed questions in relation to the importance of universal design, familiarity with the Australian Standards for Design for Access and Mobility (Standards Australia, 2009) and support for proposed visitability requirements. Results from the IDP from this sample have been published by Hitch, Dell and Larkin (2016). Pilot testing of the questionnaire with a group of nine research students from occupational therapy was conducted to identify any issues prior to the implementation of the questionnaire (Portney & Watkins, 2009) and this did not reveal the need for any changes to be made. During the development of the questionnaire, double-barrelled questions were avoided to ensure each question was confined to a single idea (Portney & Watkins, 2009). Please see Appendix 1 for a full copy of the questionnaire. Procedure Ethics approval Prior to commencing this study, ethics approval was obtained from Deakin University (HEAG-H 79_2013) Data analysis y Following data collection, each questionnaire response was coded, the IDP scale was scored according to its manual and the sum of responses on the universal design scales were calculated. Demographic data obtained from the participants was also coded and entered. The second author visually checked 10% of the data to review accuracy of entry. Data were then analysed using the Statistical Package for Social Sciences (SPSS) Version 21.0. Demographic Data g p Quantitative demographic data obtained from the questionnaire were analysed using descriptive statistics to summarise the characteristics of each sample. Categorical data (e.g. sex and group) were analysed by total number and percentage of a given response, while continuous data (eg. age) were analysed by calculating the mean, standard deviation and range of the data. Data collection Following the researcher’s oral presentation to invite participant recruitment, a Plain Language Statement and a copy of the questionnaire were distributed to all students attending the class. The architecture students were allocated fifteen minutes of class time to complete the questionnaire. The second author was available to answer any questions relating to the study. A box to collect completed questionnaires was positioned by the exit of the room. Consent was assumed if participants returned their questionnaires, all information provided was anonymous, with only year of study provided by students. 22 22 Journal of Social Inclusion, 7(2), 2016 Hypotheses yp Descriptive statistics were used to analyse responses to questions relating to universal design to provide an overview of the scores obtained for each group. The distribution of the data was inspected using a histogram and did not conform to normality. Data were first analysed descriptively to gain an overview of how each group responsed to every specific question. A two-tailed Mann-Whitney U-Test was then performed to determine if there was a significant difference in attitudes to universal design between groups. This is a common form of analysis when parametric assumptions are not met and the sample contains two independent groups of subjects with a non-directional hypothesis (Portney & Watkins, 2009). The Mann-Whitney U-Test variables are in Table 2. The significance level was set to p=<.05. Table 2: Variables utilised in Mann-Whitney U-Test Table 2: Variables utilised in Mann-Whitney U-Test Test Variables 1 Group One and Group Two overall scores of attitudes to universal design 2 Group One and Group Two total scores of attitudes to the universal design of public buildings 3 Group One and Group Two total scores of attitudes to the universal design of private buildings 4 Group One and Group Two scores to the importance of universal design to their future professional career 5 Group One and Group Two scores to the proposed visitability requirements Variables Group One and Group Two scores to the importance of universal design to their uture professional career Data collected from one open question relating to the proposal of building visitability requirements were also transcribed. The process of analysis for this data involved printing a hard copy of the two separate groups’ open responses and reading and re-reading the data to become familiar with it. Responses were then coded to identify common answers within the data. 23 23 Journal of Social Inclusion, 7(2), 2016 Attitudes to universal design g On each of the scales measuring attitudes to the universal design of buildings and built environments, possible scores could range from 10-50. Tables 3 and 4 display the responses provided by each participant group – Group One and Group Two. The possible range of total scores of both universal design scales was 20-100 with Group One scores ranging from 33-100 (Mean = 77.29, SD=13.87) and Group Two scores ranging from 43-100 (Mean = 78.10, SD = 11.35). The statement around buildings being usable without sight attracted the lowest mean score from both Groups, while Group Two gave an equally low score to the statement that buildings and built environments should appeal to all potential users. The mean scores was highest for the statement that potential users of a private building should have access to all of its features regardless of their capability for both Groups, while Group Two also rated the statement regarding use of public buildings without stigma highly. In regards to attitudes to universal design, higher mean scores were given for statements related to buildings and built environments in the public space, than for those for private use. This suggests that architecture students viewed the universal design of public spaces more positively than the universal design of private spaces. Table 3: Group One participant raw scores- Attitudes to universal design Statement Public Private M SD M SD a. All potential users should be able to use a building or built environment in essentially the same way, regardless of differences in abilities 4.15 0.8 0 3.68 1.0 4 b. Potential users should be able to use a building or built environment without feeling stigmatised because of differences in personal capabilities 4.07 0.9 8 3.77 1.0 2 c. Potential users of a building should have access to all features of privacy, security, and safety regardless of capabilities 4.23* 0.8 2 3.94* 0.8 8 d. A building or built environment should appeal to all potential users 4.18 0.8 5 3.74 1.0 8 e. An untrained person should be able to use a building or built environment without instructions 3.88 1.0 3.74 0.9 7 f. A building or built environment should be able to be used without hearing 3.92 0.8 4 3.75 1.0 3 Table 3: Group One participant raw scores- Attitudes to universal design Statement Public Private M SD M SD a. Attitudes to universal design All potential users should be able to use a building or built environment in essentially the same way, regardless of differences in abilities 4.15 0.8 0 3.68 1.0 4 b. Potential users should be able to use a building or built environment without feeling stigmatised because of differences in personal capabilities 4.07 0.9 8 3.77 1.0 2 c. Potential users of a building should have access to all features of privacy, security, and safety regardless of capabilities 4.23* 0.8 2 3.94* 0.8 8 d. A building or built environment should appeal to all potential users 4.18 0.8 5 3.74 1.0 8 e. An untrained person should be able to use a building or built environment without instructions 3.88 1.0 3.74 0.9 7 f. A building or built environment should be able to be used without hearing 3.92 0.8 4 3.75 1.0 3 Table 3: Group One participant raw scores- Attitudes to universal design 24 24 Journal of Social Inclusion, 7(2), 2016 g. A building or built environment should be able to be used without sight 3.73 0.9 0 3.64 0.9 7 h. A building or built environment should be able to be used by someone who is weak or tired or easily fatigues 3.77 0.9 2 3.71 0.9 6 i. It should be easy for a person of any size to see all the important elements of a building or built environment from an position (e.g. standing or seated) 3.84 0.8 7 3.73 0.9 3 j. There should be enough space to use a building or built environment with devices or assistance (e.g. wheelchair, oxygen tank, or service dog). 4.07 0.9 7 3.85 0.9 7 Total Score 39.85 6.6 7 37.44 8.1 9 N t M Total Score Note. M=mean SD=standard deviation * = highest mean score ** = lowest mean score Table 4: Group Two participant raw scores- Attitudes to universal design Statement Public Private M SD M SD a. All potential users should be able to use a building or built environment in essentially the same way, regardless of differences in abilities 4.25 0.88 3.71 1.07 b. Potential users should be able to use a building or built environment without feeling stigmatised because of differences in personal capabilities 4.46* 0.64 3.94 0.83 c. Potential users of a building should have access to all features of privacy, security, and safety regardless of capabilities 4.21 0.78 4.00* 0.79 d. Attitudes to universal design A building or built environment should appeal to all potential users 4.42 0.78 3.62 1.17 e. An untrained person should be able to use a building or built environment without instructions 3.90 1.14 3.83 1.04 f. A building or built environment should be able to be used without hearing 3.96 0.84 3.90 0.89 g. A building or built environment should be able to be used without sight 3.57 0.98 3.62** 0.95 h. A building or built environment should be able to be used by someone who is weak or tired or easily fatigues 3.85 0.89 3.90 0.90 i. It should be easy for a person of any size to see all the important elements of a building or built environment from an position (e.g. standing or seated) 3.63 1.12 3.63 1.11 j. There should be enough space to use a building or built environment with devices or assistance (e.g. wheelchair, 4.15 0.85 3.75 1.02 25 Journal of Social Inclusion, 7(2), 2016 oxygen tank, or service dog). Total Score Note. M=mean SD=standard deviation * = highest mean score ** = lowest mean score 40.42 5.75 37.67 7.49 40.42 5.75 37.67 7.49 Relevance to future professional practice p p Descriptive statistics were calculated for one question relating to the level of valued importance of universal design knowledge for their future professional practice. This question was based on a 5-point Likert Scale with 1 being Not at all important to 5 being Extremely important. Group One sored a mean of 4.33 (SD=0.8) indicating most participants considered it to be Very important or Extremely important to their future professional career. Group Two scored a mean of 4.51 (SD=0.59) also indicating most participants considered it to be Very important or Extremely important. Familiarity with Australian Standards y Descriptive statistics were calculated for one question relating to reported familiarity with the Australian Standards AS 1428.1-2009 Design for Access and Mobility (Australian Standards, 2009). This question was based on a 5-point Likert Scale with 1 being I have never heard of the Standards to 5 being I am very confident in my knowledge of the Standards. Frequencies of scores for this question are presented in Table 5. Response 4; “I am familiar with some aspects of the Standards” was the most frequently chosen response with 29 of 78 Group One participants. Frequencies of scores for Group Two are also presented in Table 5. Response 4; “I am familiar with some aspects of the Standards” was again, the most frequently chosen response with 17 of 51 participants who responded to this question selecting this option. Table 5: Participant raw scores- Familiarity with Australian Standards Response Group One Group Two n % n % 1 I have never heard of the standards 7 9.0 4 7.8 2 I have heard of the standards but have never seen them 12 15.4 10 19.6 3 I have looked at or read the standards but am able to recall very little of the information 25 32.1 14 27.5 4 I am familiar with some aspects of the Standards 29 37.2 17 33.3 5 I am very confident in my knowledge of the Standards 5 6.4 6 11.8 Total 78 100 51 100 Attitudes to visitability These results relate to one question relating to support for the implementation of a number of visitability requirements for the design of all new private homes and residences. Participants were required to select one of three options with 1 being Mandatory, 2 being Aspirational target and 3 being Should not be considered. Familiarity with Australian Standards Group One scored a mean of 1.81 Table 5: Participant raw scores- Familiarity with Australian Standards Response Group One Group Two n % n % 1 I have never heard of the standards 7 9.0 4 7.8 2 I have heard of the standards but have never seen them 12 15.4 10 19.6 3 I have looked at or read the standards but am able to recall very little of the information 25 32.1 14 27.5 4 I am familiar with some aspects of the Standards 29 37.2 17 33.3 5 I am very confident in my knowledge of the Standards 5 6.4 6 11.8 Total 78 100 51 100 Table 5: Participant raw scores- Familiarity with Australian Standards G O G Attitudes to visitability Attitudes to visitability These results relate to one question relating to support for the implementation of a number of visitability requirements for the design of all new private homes and residences. Participants were required to select one of three options with 1 being Mandatory, 2 being Aspirational target and 3 being Should not be considered. Group One scored a mean of 1.81 26 Journal of Social Inclusion, 7(2), 2016 (SD=0.69) indicating most participants thought the visitability requirements should be implemented as mandatory or aspirational targets. Group Two scored a mean of 1.81 (SD=1.74) also indicating most participants thought the visitability requirements should be implemented as mandatory or aspirational targets. (SD=0.69) indicating most participants thought the visitability requirements should be implemented as mandatory or aspirational targets. Group Two scored a mean of 1.81 (SD=1.74) also indicating most participants thought the visitability requirements should be implemented as mandatory or aspirational targets. Analysis of responses provided via further comments to the question was also completed to provide greater insight into architecture student attitudes to universal design of buildings and built environments. Due to the limited number of comments, this review was undertaken as an entire sample, rather than separating the two groups. Content analysis was used to understand the contents of these comments, with the responses coded individually and these codes then grouped into themes. Forty-five (35.4%) participants suggested these requirements should be mandatory, 67 (52.8%) suggested they should be utilised as an aspirational target and 13 (10.2%) participants stated their implementation should not be considered. Analysis of comments indicated four key themes. These were: emerging: concerns relating to cost and expenses; importance of tailoring to the client; general suggestions why it may not work; and, general support for the implementation. In relation to cost and expenses, one participant stated: “Overdesigning for the sake of making the residence accessible in the future, just in case, is an unnecessary cost” (Group One participant). Examples of comments relating to the need to tailor designs to the client included: “Private homes should be designed to the individual” (Group Two participant) and “It essentially boils down to catering for the client” (Group One participant). In relation to the ineffectiveness of visitability requirements, one participant stated: “Legislation restricts design, resulting in negative impacts the ‘requirements’ did not intend” (Group Two participant). Attitudes to visitability Finally, the following comment captures the general support found within this data for the implementation of visitability requirements: “I think it is best for these items to be mandatory as it makes it a lot easier when people move into new houses or change homes for particular reasons. This would allow the home to be adapted appropriately to suit the person moving In” (Group One participant). Discussion Results of data analysis have provided initial insights into the attitudes of architecture students to universal design, an area that has not been explored in previous studies. Overall, relatively positive attitudes were measured across both groups of architecture students. Notably, Item G “A building or built environment should be able to be used without sight” had the lowest mean score indicating less positive attitudes to this item, despite one of the simulation activities used within the universal design curriculum being the use of glasses to imitate a vision impairment (Larkin et al., 2010a). Empirical literature relating to teaching universal design in Brazil has supported this strategy to enhance student awareness and sensitivity (Bernardi & Kowaltowski, 2010). A focus on vision impairment as a basis for education about universal design also appears to be a theme within the empirical literature. Authors from Belgium and Sweden report on the inclusion of users as experts, commonly with some form of vision impairment, to assist students’ understanding and evaluation of designs (Heylighen et al., 2009; Paulsson, 2005). Considering this consistent inclusion of vision impairment throughout the empirical universal design teaching literature, it is unclear why participants viewed this item less positively. Overall, architecture students had positive attitudes to the universal design of buildings and built environments. This was highlighted through responses to the valued importance of universal design knowledge to their future professional careers in which students considered it to be either “Very” or “Extremely important”. Positive attitudes were also noted on specific items of the scale used to measure attitudes to universal design. These items included; “Potential users of a building or built environment should have access to all features of privacy, security, and safety regardless of personal capabilities”, “A building or built environment should appeal to all users”, “There should be enough space to use a building or built environment with devices or assistance (e.g. wheelchair, oxygen tank, or service dog)” and “Potential users should be able to use a building or built environment without feeling stigmatised because of differences in personal capabilities”. This finding supports the continued inclusion of universal design teaching as a valuable aspect of curriculum within architecture higher education, despite there being no quantitative difference between the groups. Architecture students were also noted to view the universal design of public buildings and built environments more positively than the universal design of private buildings or built environments. Comparison of Group One and Group Two p p p A two-tailed Mann-Whitney U-Test was performed to determine if there was a statistically significant difference between the two groups. The results of these analyses are reported in Table 6. As demonstrated by this table, there were no significant differences identified between the two groups on: total scores for the attitudes towards the universal design of public buildings and built environments scale; the attitudes towards the universal design of private buildings and built environments scale; overall scores on the universal design scales; scores on the importance of universal design to participant future professional career; scores to the proposed visitability requirements; or, scores on familiarity with the Australian Standards. Table 6: Comparison of Group one and Group Two- Universal Design UD-Public UD-Private UD-Total UD- Importance Visitability Requirements AS Familiarity p value .909 .865 .951 .278 .621 .827 Note. UD-Public=Attitudes to universal design of public buildings and built environments UD-Private=Attitudes to universal design of private buildings and built environments UD-Total=Total of both attitudes to universal design scales ote. UD-Public=Attitudes to universal design of public buildings and built environments D-Private=Attitudes to universal design of private buildings and built environments D-Total=Total of both attitudes to universal design scales 27 Journal of Social Inclusion, 7(2), 2016 D-Importance=Valued importance of universal design to participant future professional p p g p p p career Visitability Requirements=Support for implementation of proposed visitability requirements AS Familiarity=Familiarity with the Australian Standards 1428.1 – 2009 Design for Access and Mobility career Visitability Requirements=Support for implementation of proposed visitability requirements AS Familiarity=Familiarity with the Australian Standards 1428.1 – 2009 Design for Access and Mobility However, a statistically significant difference was identified on Question B of the attitudes to the universal design of public buildings and built environments measure; potential users should be able to use a building or built environment without feeling segregated or stigmatised because of differences in personal capabilities (U-Test = .019, p<.05). Group Two scored higher on the raw scores of this item (4.46) than Group One (4.07) indicating Group Two had more positive attitudes to universal design than Group One on this item. An Analysis of Covariance (ANCOVA) was also completed for the entire sample to identify any confounding variables. Age, gender, personal experience of a temporary or permanent health condition and knowing someone with a permanent or temporary health condition were not identified as factors influencing attitudes to universal design. Discussion Due to the confines of this study, it was not possible to explore this finding further. However it could be hypothesised that students believe they are 28 Journal of Social Inclusion, 7(2), 2016 ‘designing for all’ when working on public buildings and built environments, however when working on private buildings they are designing specifically for the needs of the client and the client’s requests. ‘designing for all’ when working on public buildings and built environments, however when working on private buildings they are designing specifically for the needs of the client and the client’s requests. Further comments relating to four proposed visitability requirements for the new building work of private buildings also provide interesting insights into architecture student perspectives on the matter. While there was some general support for their implementation, a number of students discussed their disagreement, highlighting that the requirements would provide unnecessary or unwarranted costs during the building process and that clients should have the opportunity to decide for themselves. It is clear at this current time, that architecture students view the priority of visitability as a significant expense, indicating a need for further research into the cost comparison of designing with universal design principles or later retrofitting. These findings also suggest further education is required to enhance architecture student attitudes and the profession in general, to universal design and visitability and to highlight universal design as a key selling point of future designs, in much the same way as energy and sustainability codes. While architecture students in general expressed an overall degree of familiarity with the local standards, it is well recognised that such standards are complex in their implementation and only represent a starting point for a design process that attempts to be inclusive. Larkin, Hitch, Watchorn and Ang (2015) emphasised that visitabiliy or accessibility requirements do not equate with design that is truly inclusive and at the heart of universal design. On the whole, however, this study did not find statistically significant differences between the two groups of students on the measure of attitudes to universal design that was used. Extensive empirical literature exists that identifies important features when teaching architecture students about universal design. Bernardi and Kowaltowski (2010) suggested that role-playing of individuals with motor, vision and hearing difficulties was essential to increasing architecture student’s perspectives and awareness of universal design. Discussion Further to this, Paulsson (2005) discussed the need for collaborative projects with users, experts, partners and other schools as well as the need for staff who are champions and leaders of change. Olgunturk and Demirkan (2009) also recommended universal design implementation as both a separate course as well as within design studios in the Turkish context. These recommendations are consistent with the content and strategies of the Design 4 Diversity program suggesting it was based on the best available current evidence, regardless of the fact that significant differences were not generally found. Pre- and post- results from the Design 4 Diversity project reported by Larkin et al. (2010b) also suggested a trend toward improved architecture student familiarity with the Principles of Universal Design. These results are consistent with findings from a six-week Amercian teaching program in which junior-level interior design students were asked to incorporate the universal design concept when redesigning housing units for students with disabilities (Chang et al., 2000). Similar teaching methods to those recommended within the literature were utilised. Chang et al. (2000) found a significant difference in knowledge of the Universal Design Principles from the overall means obtained from the pre and post- tests. These findings suggest that both programs, which implemented strategies according with the current evidence, were effective in enhancing architecture students’ knowledge of universal design in the short term. However, this literature does not investigate important features of programs to change or enhance the attitudes of architecture students to universal design. Bernardi and Kowaltowski (2010) recognised “this is needed to develop teaching methods to change future design professionals’ attitudes” (p. 387). This lack of evidence and understanding may explain why no significant difference on attitudes to universal design was found between the two groups, as strategies to increase the knowledge of universal design 29 Journal of Social Inclusion, 7(2), 2016 may not be the same as strategies needed to improve or enhance student attitudes to this design concept. Paulsson (2005) also identified that learning Universal Design Principles is a process and requires some time. However, the literature currently lacks any longitudinal or follow-up studies to determine the long-term retention and application of this knowledge. As discussed earlier, Chang et al. (2000) reported on a six-week pre- and post-test research design. Bernardi and Kowaltowski (2010) reported on a semester-long program that aimed to develop the awareness and sensitivity of students. Recommendations for further research It would be valuable for future studies to include a qualitative aspect allowing for greater in- depth understanding of architecture student attitudes to universal design in the longer term, particularly post-graduation and in the context of professional practice. It would also be beneficial to complete a longitudinal or follow-up study to investigate the retention and application of knowledge long-term. A study comparing the associated costs of designing with universal design in mind and the costs associated with retrofitting buildings at a later stage would also be beneficial. Discussion Olgunturk and Demirkan (2009) also reported on a semester-long, 14-week elective course with content relating to diversity and the Universal Design Principles. Finally, as part of the Design 4 Diversity initiative, Watchorn et al. (2013) also reported on the semester-long program in which data were collected in a pre- and post-test design. While a number of studies from around the world have been identified evaluating the effectiveness of their education programs, this current study appears to be the first to collect data at a period of time after completion of the program. It is therefore difficult to confirm ensuring long-term retention and application of universal design knowledge. However, regardless of the similarity in response between the two groups, there was a generally positive view by architecture students towards the universal design of buildings and built environments. Limitations There were a number of limitations associated with this study. As the questionnaire was voluntary, there was the potential for a self-selection bias where students particularly interested in the topic may be more likely to participate. However, due to ethical issues, the questionnaire could not be made mandatory. Due to the nature of the study and the setting in which it was conducted, it was not possible to ensure all participants completed the questionnaire independently without discussing with their peers. Therefore it is not possible to determine if leakage may have occurred and how this may have impacted on results. The generalisability of the study is also limited. Purposive sampling was utilised to ensure a representative sample for the study. However, this also meant that the sample, which was from one university in Australia, is not generalisable to the architecture student population. Given that this was a relatively small-scale study conducted within the context of an undergraduate Honours study, this was accepted as appropriate. References Afacan, Y. & Demirkan H. (2010). A priority-based approach for satisfying the diverse users’ needs, capabilities and expectations: A universal kitchen design case. Journal of Engineering Design, 21(2-3), 315-343. g g g ( ) Altay, B. & Demirkan, H. (2014). Inclusive design: Developing students’ knowledge and attitude through empathic modelling. International Journal of Inclusive Education, 18(2), 196-217. ( ), Bernardi, N. & Kowaltowski, D. (2010). When role playing is not enough: Improved universal design education. International Journal of Architectural Research, 4(2-3), 376-390. Burgstahler, S. (2012). Universal design in education: Principles and applications, Retrieved from http://www.washington.edu/doit/universal-design-education-principles-and- applications Case, J. (2003), Universal Design, Retrieved from http://images.pearsonassessments.com/images/tmrs/tmrs_rg/UniversalDesign.pdf?W T.mc_id=TMRS_Universal_Design Case, J. (2003), Universal Design, Retrieved from http://images.pearsonassessments.com/images/tmrs/tmrs_rg/UniversalDesign.pdf?W T id TMRS U i l D i _ _ _ g Chang, B.V., Tremblay, K.R., & Dunbar, B.H. (2000). An experiential approach to teaching universal design. Education, 121(1), 153-158. Connell, B.R. (1997). Principles of Universal Design, Retrieved from http://www.ncsu.edu/ncsu/design/cud/about_ud/udprinciplestex p g _ p p Evcil, A. (2012). Raising awareness about accessibility. Procedia – Social and Behavioural Sciences, 47, 490-494. Gething, L. (1991). Interaction with Disabled Persons Scale: Manual and kit, University of Sydney: Sydney. Helvacioglu, E. & Karamanoglu, N. (2012). Awareness of the concept of universal design in design education. Procedia – Social and Behavioural Sciences,51, 99-103. Helvacioglu, E. & Karamanoglu, N. (2012). Awareness of the concept of universal design in design education. Procedia – Social and Behavioural Sciences,51, 99-103. Heylighen, A. (2008). Sustainable and inclusive design: A matter of knowledge? Local g Heylighen, A. (2008). Sustainable and inclusive design: A matter of knowledge? Loca Environment, 13(6), 531-540. Heylighen, A., Herssens, J., & Froyen, H. (2009), Architecture criticism blindfolded, Retrieved from https://lirias.kuleuven.be/bitstream/123456789/206651/2/09Include- Heylighen.pdf.pdf. y g p p Hitch, D., Dell, K. & Larkin. H. (2016). Does universal design education impact on the attitudes of architecture students towards people with a disability? Journal of Accessibility and Design for All, 6(1), 26-48. y g ( ) Jones, P. (2014). Situating universal design architecture: Designing with whom? Disability and Rehabilitation, 36(16), 1369-1374. doi: 10.3109/09638288.2014.944274 ( ) Karol, E. (2008). Inclusive design and the new home market: The West Australian situation. Architectural Science Review, 51(1), 80-83. Larkin, H., Ang, S., Watchorn, V., Segrave, S., McCracken, M., Elkadi, H., & Holt, D. (2010a). Design 4 Diversity enhancing inter-professional learning for architecture and occupational therapy students: Literature review. Deakin University: Geelong, Australia. Larkin, H. Conclusion While existing research has evaluated the implementation of universal design teaching programs within architecture higher education, this study is the first to assess architecture student attitudes to universal design at a period in time post-completion of the program. Despite the lack of significant differences between the two groups of architecture students, this study still provides valuable insights into their attitudes to universal design, an area not 30 Journal of Social Inclusion, 7(2), 2016 previously investigated. These findings have contributed to the current body of knowledge relating to universal design in higher education and have identified important areas for future research in order to enhance the training and education of future architects. 31 31 Journal of Social Inclusion, 7(2), 2016 p py p , ( ), arkin, H., Hitch, D., Watchorn, V., & Ang, S. (2015). Working with policy and regulatory References Ang, S., Watchorn, V., Segrave, S., McCracken, M., Elkadi, H., & Holt, D. (2010b). Design 4 Diversity enhancing inter-professional learning for architecture and occupational therapy students: Study findings. Deakin University: Geelong, Australia. p py y g y g Larkin, H., Hitch, D., Watchorn, V., Ang, S., & Stagnitti, K. (2013). Readiness for interprofessional learning: A cross-faculty comparison between architecture and occupational therapy students. Journal of Interprofessional Care, 27(5), 413-419. p py p , ( ), Larkin, H., Hitch, D., Watchorn, V., & Ang, S. (2015). Working with policy and regulatory 32 Journal of Social Inclusion, 7(2), 2016 factors to implement universal design in the built environment: The Australian experience. International Journal of Environmental Research and Public Health, 12, 8157-8171. factors to implement universal design in the built environment: The Australian experience. International Journal of Environmental Research and Public Health, 12, 8157-8171. National People with Disabilities and Carer Council. (2009). Shut Out: The experience of people with disabilities and their families in Australia. Canberra: Commonwealth of Australia. Olgunturk, N. & Demirkan, H. (2009).Ergonomics and universal design in interior architecture education. Middle Eastern Technical University Journal of the Faculty of Architecture, 26(2), 123-138. Ostroff, E. (2011). Universal design: An evolving paradigm. In W Preiser & K.H. Smith (eds.). Universal design handbook (2nd ed.) (pp. 1.3-1.11). NY: McGraw-Hill Education. Paulsson, J. (2005). Universal design education project – Sweden. Retrieved from http://www.accessible- Ostroff, E. (2011). Universal design: An evolving paradigm. In W Preiser & K.H. Smith (eds.). Universal design handbook (2nd ed.) (pp. 1.3-1.11). NY: McGraw-Hill Education. Paulsson, J. (2005). Universal design education project – Sweden. Retrieved from http://www.accessible- buildings.eu/UserFiles/File/Budapest/Paper.2006.08.23.Jan.P.Budapest.pdf g p p p p Portney, L. & Watkins, M. (2009). Foundations of clinical research applications to practice. Pearson: New Jersey. Preiser, W. (2008). Universal design: From policy to assessment research and practice. International Journal of Architectural Research, 2(2), 78-93. Punch, K. (2005). Introduction to social research quantitative and qualitative approaches. SAGE: London. Sang, K., Ison, S., Dainty, A., & Powell, A. (2009). Anticipatory socialisation amongst architects: A qualitative examination. Education + Training, 51(4), 309-321. Standards Australia. (2009). AS 1428.1 - 2009 Design for access and mobility, Part 1: General requirements for access – New building work. Retrieved from http://infostore.saiglobal.com/store/details.aspx?ProductID=1378917 The Centre for Universal Design. (2000). Universal Design Performance Measures for Products. The Centre for Universal Design: North Carolina. The Centre for Universal Design. (2008). References The Centre for Universal Design environments and products for all people. Retrieved from The Centre for Universal Design. (2008). The Centre for Universal Design environments and products for all people. Retrieved from http://www.ncsu.edu/ncsu/design/cud/index.htm http://www.ncsu.edu/ncsu/design/cud/index.htm United Nations (2006). Convention on the Rights of Persons with Disabilities. New York: United Nations. Retrieved from https://www.un.org/development/desa/disabilities/convention-on-the-rights-of- persons-with-disabilities/the-convention-in-brief.html Watchorn, V., Larkin, H., Hitch, D., & Ang, S. (2014). Promoting participation through the universal design of built environments: Making it happen. Journal of Social Inclusion, 5(2), 65-88. Watchorn, V., Larkin, H., Hitch, D., & Ang, S. (2014). Promoting participation through the universal design of built environments: Making it happen. Journal of Social Inclusion, 5(2), 65-88. ( ) Watchorn, V., Larkin, H., Ang, A., & Hitch, D. (2013). Strategies and effectiveness of teaching universal design in a cross-faculty setting. Teaching in Higher Education, 18(5), 1-14. Watchorn, V., Larkin, H., Ang, A., & Hitch, D. (2013). Strategies and effectiveness of teaching universal design in a cross-faculty setting. Teaching in Higher Education, 18(5), 1-14. ( ) World Health Organization. (2002). Towards a common language for functioning, disability and health: ICF The International Classification of Functioning, Disability and Health. World Health Organization: Geneva. ( ) World Health Organization. (2002). Towards a common language for functioning, disability and health: ICF The International Classification of Functioning, Disability and Health. World Health Organization: Geneva. g World Health Organization. (2013). International Classification of Functioning, Disability and Health (ICF), Retrieved from http://www.who.int/classifications/icf/en/ 33 Journal of Social Inclusion, 7(2), 2016 Biographical Notes Helen Larkin brings to academic position as senior lecturer in the occupational therapy program, over 30 years’ experience in the disability sector in service provision, management and research. Her teaching and research interests includes curriculum development, inter- professional education, reflective practice and work integrated learning. Helen’s contribution has been recognised through both university and national teaching awards. Kelsey Dell was formerly an honours student in the Bachelor of Occupational Therapy, achieving a high distinction for her final thesis. She is now employed as an occupation therapist in Tasmania. Dr. Danielle Hitch is a senior lecturer in the Deakin University occupational therapy program. She worked clinically in both the United Kingdom and Australia for over a decade. Her areas of interest include mental health, knowledge brokerage, inclusive education and social justice. 34 34
W4230184639.txt	https://revistas.ufpr.br/revistax/article/download/31842/22626	pt		HERMES E BONECAS RUSSAS: PARA COMPREENDER A FORMAÇÃO DOCENTE	Revista X	2,013	cc-by	2,249	REVISTA X, VOLUME 2, 2013 P á g i n a \| 75 OLIVEIRA, A. L. A. M. Hermes e bonecas russas: um estudo colaborativo para compreender a relação teoria-prática na formação docente. 2006, 247f. Tese de doutorado. Programa de Pós-graduação em Linguística Aplicada. Universidade Federal de Minas Gerais. Belo Horizonte, 2006. HERMES E BONECAS RUSSAS: PARA COMPREENDER A FORMAÇÃO DOCENTE Hermes and Russian dolls: understanding teacher development Resenhado por: Katia Bruginski MULIK, UFPR-PG1 Em sua tese de doutoramento defendida no ano de 2006, Ana Larissa A. M. Oliveira, professora da Faculdade de Letras da Universidade Federal de Minas Gerais (UFMG), propõe, em uma pesquisa-ação de caráter etnográfico, a discussão sobre a açãocolaborativa entre professores de língua inglesa. Pelo viés da linguagem, seu objetivo foi analisar os processos de construção e reconstrução do saber docente em uma situação de formação continuada com três professoras de escolas públicas. As referidas professoras eram participantes do Projeto de Educação Continuada de Professores de Línguas Estrangeiras na UFMG. O estudo de Oliveira divide-se em seis capítulos que serão explanados a seguir. No primeiro capítulo a autora apresenta seu tema de estudo inserindo-o no paradigma pós-moderno. Este último é entendido por ela como uma forma de rever “dicotomias modernas, tais como a relação sujeito/objeto, pesquisador/pesquisado, saber científico/saber comum e a relação: teoria/prática” (OLIVEIRA, 2006, p. 15). Partindo do ideário freireano de crítica pela linguagem, a autora optou por verificar se a açãocolaborativa pode ser uma aliada na (re)construção do conhecimento, bem como na mudança da práxis de professores. O distanciamento da relação entre teoria e prática tem colaborado para um cenário de inconsistências no ensino de línguas, no sentido de que ambas são entendidas como elementos opositivos e não como partes integrantes de um todo. Por isso, Oliveira defende em seu trabalho que a escola precisa ser co-autora e coparticipante da produção do conhecimento vendo esse espaço, e os que nele circulam, não como objetos “descartáveis” para coletas de dados da pesquisas, mas como protagonistas. 1 Mestranda em Estudos Linguísticos pela UFPR. MULIK REVISTA X, VOLUME 2, 2013 P á g i n a \| 76 No capítulo dois, Oliveira apresenta os pilares teóricos de seu estudo: o paradigma crítico-reflexivo na formação de professores e a pesquisa-ação colaborativa. Inicialmente, a pesquisadora discorre sobre a importância do pensamento de Freire e a sua influência nos estudos da Linguística Aplicada. A autora coloca que os estudos da pedagogia crítica auxiliaram no desvelar do multifacetado fenômeno linguístico atentando para as implicações nos âmbitos sociais, culturais, políticos, econômicos, ideológicos e não apenas do pedagógico. A neutralidade da língua é posta em xeque e traz consigo a idealização do “falar como um nativo” reconhecendo assim outras formas de expressão como legítimas. Oliveira (2006, p. 29) advoga de forma assertiva que “a formação que parte dos professores de línguas recebe não lhes permite emancipação crítica suficiente para poderem escolher o que, o como e o porquê ensinar de modo crítico e teoricamente fundamentado”. Diante dessa lacuna na formação, o professor passa a ser mero reprodutor de “modismos metodológicos” executados sem reflexão e sem mensuração dos impactos educacionais. Ainda nesse capítulo, a autora comenta sobre a pedagogia pós-método proposta por Kumaravadivelu (2001). Diante dos princípios do pós-método de Kumaravadivelu e do que Oliveira defende em seu trabalho, a escola pode ser um espaço onde não apenas se reproduz a sociedade, mas um lugar onde é possível projetar a sociedade e a escola que se quer através de ações de autonomia dos agentes envolvidos no processo educativo. Ao longo da pesquisa, Oliveira utiliza o termo formação crítico-reflexiva de professores que incorpora os princípios da pedagogia crítica e do pós-método entendendo-os como complementares. O terceiro capítulo é destinado à apresentação da metodologia utilizada. Primeiramente a autora delineia a caracterização da pesquisa etnográfica e, em seguida, estabelece a discussão sobre as noções da Análise do Discurso Anglo Saxônica e sua relação com o estudo. Posterior a isso, tem-se a descrição dos procedimentos de coletas de dados, os perfis das participantes, o cenário da pesquisa: o projeto de formação e as quatro escolas públicas. A autora ressalta que “não foi um trabalho neutro e nem tampouco de mão única” (OLIVEIRA, 2006, p. 50-51). Exemplo disso é a própria participação das pesquisadas na análise preliminar dos dados, que modificaram hipóteses iniciais interpretadas pela autora sobre os significados produzidos. O construto tensão colaborativa, entendido como um movimento discursivo, é um termo utilizado por Oliveira para desenvolver o debate colaborativo provocado ao longo das seções. A análise discursiva dos dados recorreu a tal construto, o qual compreende MULIK REVISTA X, VOLUME 2, 2013 P á g i n a \| 77 quatro movimentos interacionais distintos, porém conectados entre si: enunciado gerador de tensão – enunciado gerador de debate produzido pela pesquisadora ou pelas participantes; tensão colaborativa – caracterizada pela recuperação por enunciados da fala do outro, na conciliação, no confronto e justificativa do que e porque foi dito; reflexão – explicitação de rupturas e tomadas de consciência pelas professoras e por último, a ação- transformadora – ações de mudança que se concretizaram e que foram descritas pelas professoras. Em relação à coleta dos dados, esta se deu em duas etapas visando responder a questionamentos distintos. As perguntas norteadoras da etapa 1 foram:  A teoria aprendida no projeto integra as práticas das professoras? Em que medida?;  Como a linguagem é utilizada para debater criticamente a teoria divulgada no projeto nas interlocuções com a pesquisadora? Já em relação à fase 2, a pergunta norteadora foi:  A ação-colaborativa pode ser observável mesmo após as professoras perderem o contato com a rede de colaboração? Dessa forma, a coleta se deu em dois momentos com instrumentos distintos. Na primeira etapa, Oliveira utilizou-se de observações em sala de aula, observações das professoras no curso de formação, sessões colaborativas entre professoras e pesquisadora, história de vida das participantes e diários feitos pelas participantes. Para a segunda etapa, denominada coleta tardia de dados, realizada dois anos depois da primeira, foram utilizados os procedimentos de observação de aulas das participantes e entrevista semiestruturada. O quarto capítulo, no qual a análise é apresentada, divide-se em duas partes. Na primeira, a autora vale-se das questões norteadoras iniciais para descrever e interpretar três aulas observadas por ela tendo como referente o ensino da gramática como tema da ação-colaborativa (TAC). A segunda parte do capítulo descreve o percurso de reflexão das professoras em relação a certas concepções de ensinar tais como a produção oral dos alunos e o uso da língua inglesa na sala de aula que eram problemas comumente salientados no contexto de ensino onde atuavam. As professoras mostraram opiniões divergentes das iniciais em relação à própria noção de contexto e de uso significativo da língua, ou seja, o ensino voltado para práticas relacionadas ao cotidiano do aluno. Oliveira comenta que o processo de integração teoria-prática ocorreu disformemente entre as três professoras em um processo lento influenciado por aspectos econômicos, MULIK REVISTA X, VOLUME 2, 2013 P á g i n a \| 78 sociais, culturais e institucionais (principalmente em relação ao contexto distinto de escola pública e particular). O quinto capítulo descreve a análise da segunda etapa de estudo feita com as professoras a qual é nomeada pela pesquisadora de coleta tardia dos dados, uma vez que estes foram gerados cerca de dois anos depois da primeira etapa realizada. Essa parte da investigação teve o intuito de verificar se havia ocorrido ou não mudança na postura crítica das pesquisadas mesmo após a perda de contato efetivo entre pesquisadora e participantes. Oliveira registra os aspectos significativos na mudança da práxis2 como, por exemplo, a mudança de rota no planejamento, os temas das aulas sempre recorrendo ao cotidiano dos alunos, o reconhecimento da identidade e das necessidades desses últimos, bem como uma maior conexão entre teoria-prática, pois “com base na prática a professora consegue teorizar e produzir heurística, encontrando soluções particulares para problemas específicos” (OLIVEIRA, 2006, p. 171). Nessa coleta tardia, a pesquisadora identificou, dentro de uma perspectiva crítica de mudança, três aspectos: disposição para mudar; a reaculturação, e a capacidade crítica para apropriar-se da teoria e produzir heurística. O capítulo seis é destinado à discussão das considerações finais do estudo. Oliveira inicia fazendo uma síntese dos resultados, retomando o contexto em que se insere a pesquisa e seu propósito e a descrição do perfil de seus participantes. Em seguida, a autora inicia uma seção intitulada Hermes e bonecas russas: para compreender a formação docente na qual ela finalmente explica o título de sua tese. Oliveira argumenta que a essência do trabalho colaborativo, que visa à quebra de níveis hierárquicos e a manutenção do respeito à diferença dos participantes, pode ser traduzido pelo mito de Hermes o qual era visto na Mitologia Grega como um mediador, já que sabia falar tanto a língua dos deuses como a dos homens e dessa forma reduzia a distância entre o mundo divino e o mundano. O mito também pode ser utilizado para representar a postura que pesquisadores e professores precisam assumir na transmissão do conhecimento, quer seja na sala de aula, quer seja na formação de profissionais que atuam ou que irão atuar nela. A aproximação dos dois universos por meio da linguagem deve se dar sem que sejam alteradas suas fronteiras e seus circulantes sejam reconhecidos com equidade. Oliveira advoga que com o trabalho colaborativo o objetivo não é que, as teorias sejam confirmadas na práxis, mas 2 Apenas uma das professoras teve sua aula observada nessa coleta tardia, pois uma das professoras havia pedido afastamento temporário e a outra licença médica por problemas de saúde. MULIK REVISTA X, VOLUME 2, 2013 P á g i n a \| 79 que sejam ressignificadas, ou seja, modificadas e readaptadas aos contextos e a realidade de professores e alunos. Já as bonecas russas são utilizadas, metaforicamente, para explicar que a tensão colaborativa é uma ação interdependente e que as bonecas que se encaixam umas nas outras possuem partes diferentes e, ao mesmo tempo iguais, formando um todo, fazem alusão ao crescimento tanto pessoal quanto profissional. Tal incorporação e transformação só são possíveis através da aprendizagem com o outro que, no contexto da pesquisa, se deu pela ação-colaborativa. O estudo de Oliveira elucida a forte conexão entre teoria e prática que por muitos é vista de forma dicotômica. A autora mostra o quão importante é pensar como a teoria acontece na prática e como pode (e deve) ser adaptada para os contextos e realidades em que os sujeitos se encontram. A pesquisadora mostra também que o professor pode ser construtor de teoria através da própria práxis. Ela apresenta a ação-colaborativa como uma possibilidade de fazer com que, aliada à participação de cursos de formação, o professor reflita sobre a forma como e porque ensina. Além disso, as sessões colaborativas são reveladoras das crenças das professoras que, de forma crítica, foram postas em xeque apontando possíveis caminhos para a mudança. A troca de experiências acaba sendo o ponto alto do trabalho colaborativo, já que o objetivo não era identificar os possíveis problemas na prática das professoras, mas sim propor de forma conjunta, e com a participação ativa delas, possíveis formas de se repensar o ensino nas suas nuances de contexto, de conteúdo, etc. Em relação à coleta tardia dos dados, penso que foi um dos pontos em que a pesquisa apresentou uma lacuna, já que só foi possível revisitar a prática de apenas uma das professoras. No entanto, mesmo que de forma restrita, a pesquisadora conseguiu mostrar que houve mudança e que a ação colaborativa tem caráter positivo. O trabalho de Oliveira pode ser comparado ao de Zellermayer e Morgolin (2005) que discutem o processo de mudança de professores trabalhando com a noção de comunidade de aprendizado colaborativo (collaborative learning community) através das lentes da teoria da complexidade. Outra noção interessante para se discutir a formação de professores é o conceito de agência que parte da noção freireana de práxis entendida como “uma prática reflexiva que desafia constantemente, em vários campos ideológicos, seus próprios pressupostos e desdobramentos, enfocando a transformação e constante reavaliação de procedimentos interpretativos de construção de sentidos” (JORDÃO, 2010, p. 438). MULIK REVISTA X, VOLUME 2, 2013 P á g i n a \| 80 É inquestionável que os professores precisam refletir sobre suas práticas, porém para que essa reflexão seja válida é preciso que seja direcionada. No trabalho em questão foi possível verificar como a tensão colaborativa pode gerar reflexão de forma “natural” e acabar contribuindo para que novas formas de ensinar sejam incorporadas. Outro ponto importante é fazer com que os professores reflitam em conjunto para que um possa auxiliar na resolução de conflitos e problemas do outro. Talvez momentos de tensão colaborativa possam ser incorporados a outros contextos de formação como, por exemplo, as semanas de formação pedagógica nas escolas fazendo desse momento não apenas um espaço para que os problemas sejam identificados ou para que o corpo docente possa “desabafar” as angústias vividas na sala de aula, mas sim, chegar próximo a, pelo menos, uma nova forma de lidar com elas. REFERÊNCIAS JORDÃO, C. M. A posição de professor de inglês no Brasil: hibridismo, identidade e agência. Let. & Let. Uberlândia – MG. v. 26 n. 2, p. 427- 442 . jul/dez. 2010. KUMARAVADIVELU, B. Toward a Postmethod Pedagogy. TESOL Quarterly. v. 35, nº 4, p. 537-560, 2001. ZELLERMAYER, M.; MARGOLIN, I. Teachers Educator’s Professional Learning described through the Lens of Complexity Theory. Teachers College Record. Columbia University. v. 107, n. 6, June, 2005, p. 1275-1304. MULIK
https://openalex.org/W3217332367	https://publishing.mpda.ru/index.php/praxis/article/download/683/570	Russian	null	Religious Policy of the King Alexander Karadjordjevic	Praksis	2,019	cc-by-sa	7,592	109 109 1 Святосавская речь 27.01.1935 на Юридическом факультете в г. Суботица (королевство Югославия). Перевод с сербского М. Д. Жоголевой. Перевод подготовлен при финансовой поддержке РГНФ в рамках научного проекта № 10-03-00401а. Сергей Викторович Троицкий Сергей Викторович Троицкий доктор церковного права профессор Московской духовной академии Для цитирования: Троицкий С. В. Религиозная политика короля Александра Объединителя // Праксис. 2019. Т. 2. № 2. С. 109–127 2 «От всей души благодарен такому тёплому приему по случаю празднования десятилетия существования факультета в Суботице и с искренним удовольствием отдаю должное проведенному до сегодняшнего дня плодотворному труду и твёрдо убежден, что будет сохранено вдохновение молодого поколения, которое будет служить во благо своему народу и отечеству. Александр». 3 Речи короля витязя Александра I Объединителя Подготовили и опубликовали Р Паре 3 Речи короля-витязя Александра I Объединителя. Подготовили и опубликовали Р. Паре жанин, Дж. Гаврилович. Београд, 1934. С. 56 Аннотация С темой автокефалии Сербской Православной Церкви неразрывно связана тема вос становления её патриаршества после Первой мировой войны. В данной речи, произне сенной в день памяти святого Саввы в 1935 году, профессор С. В. Троицкий анализирует конфессиональную политику королевства Югославия и отмечает роль в этой политике короля Александра Карагеоргиевича (1888–1934).1 Ключевые слова: конфессиональная политика, государственно-церковные отношения, коро левство Югославия, Сербская Православная Церковь, Александр I Карагеоргиевич. 110 СЕРГЕЙ ВИКТОРОВИЧ ТРОИЦКИЙ У У же в пятнадцатый раз мы собираемся в стенах этого учреж дения, чтобы почтить память крупнейшего сербского про светителя. И сейчас, когда подходит к концу пятнадцатый год существования нашего факультета, вполне естествен но, что наша благодарность обращена к основателю факультета королю-витязю, объединителю, королю-мученику, чьи тёплые слова, произнесённые в Суботице по случаю десятилетия существования факультета, высеченные на этой стене золотыми буквами,2 всегда будут его гордостью. Многие уже говорили и писали о неизмеримых заслугах короля- объединителя перед отечеством и европейским миром, а сегодня я буду говорить о той сфере его деятельности, которая, мне, как канонисту, знакома гораздо лучше, т. е. о его религиозной политике. До сегод няшнего дня об этой стороне деятельности Александра мало гово рилось и писалось, а между тем сам король-освободитель придавал религиозной политике огромное значение. В своей политической деятельности он полностью выполнил завет царя Душана — претво рил в жизнь тысячелетний народный идеал о государственном един стве. Однако он понимал, что без духовного единства не может быть твёрдого политического единства, и в достижении этого духовного единства он видел «вторую часть завета царя Душана», как гласит его грамота об учреждении Сербской Патриархии.3 Александр знал и то, что тонкая проблема установления религиозного единства отнюдь не легче задачи установления государственного единства, и трудился над осуществлением этой цели очень энергично, но в то же время с большой осторожностью. А сейчас обратимся к источникам, из которых возникла его поли тика, а затем к основополагающим принципам, которыми он руковод ствовался при проведении этой политики. Нет сомнений, что первым и главным источником его политики было его личное религиозное чувство. РЕЛИГИОЗНАЯ ПОЛИТИКА КОРОЛЯ АЛЕКСАНДРА ОБЪЕДИНИТЕЛЯ 111 Эту черту духовного облика отмечали ещё его коллеги и ученики русского Училища правоведения и Пажеского корпуса в Петербурге.4 Над кроватью Александра всегда висела большая икона Спасителя, которой он каждый день молился. Этот обычай, к которому он привык в России, продолжал соблюдать и на родине. 5 См.: Љ. Митривић-Мирољуб. О камену Темељцу Христове Цркве. Београд, 1931. 4 Его коллега по Училищу правоведения г. Немирович-Данченко пишет в статье «Друг России»: «Александр с малолетства был очень религиозным человеком. В изголовье его кровати всегда висела кона с образом Спасителя, которой он молился Богу, когда приходил в спальню. Он постоянно ходил с нами церковь, набожно стоял все время Богослужения» // Иллюстрированная Россия. 1934. № 43. С. 13. Аннотация Его дом в Белграде укра шен многочисленными большими и старинными иконами церковных праздников и святителей, а в королевской спальне на отдельном сто лике всегда лежало Евангелие, именно то, которое король Петр пронёс через Албанию в день народного несчастья. Александр свято соблюдал религиозные каноны. В Королевском доме регулярно совершались обряды освящения воды и празднования в честь святого апостола Андрея, каждый год он постился, исповедовался и причащался, любое новое здание освящали, а народные религиозные обычаи венчания, крещения детей, сочельник и Рождество он соблюдал с глубоким почте нием. Он был кумом для многих крестьянских семейств. «Если работать без веры в Бога, пользы не будет», — говорил король, когда украшали мозаикой церковь на Опленаце.5 И эти слова служили напутствием всей его деятельности. В тяжёлые минуты, а таких моментов в многострадальной жизни короля было много, и, прежде всего, когда нужно было принять важное решение, он отправлялся в Опленац или какой-нибудь монастырь и про водил там несколько часов в молитвах и размышлениях. Так он находил успокоение и новые силы для великих свершений и решительных шагов. Особенно часто он посещал маленькую церковь на вершине Каймакчалана, где взывал к Богу и молился за победу своего народа. И перед своей последней роковой поездкой во Францию 6. октября, всего за три дня до гибели, он пришёл в монастырь Саввы, полчаса провёл в молитве, взял четыре восковые свечи, затем зажёг одну, пере крестился, поцеловал свечу и поставил на подсвечник, а затем лично позвонил в монастырский колокол, как будто объявляя своему народу, что идёт на свою марсельскую Голгофу. О глубине его религиозности свидетельствуют его многочислен ные пожертвования как православным, так и католическим храмам: 4 Его коллега по Училищу правоведения г. Немирович-Данченко пишет в статье «Друг России»: «Александр с малолетства был очень религиозным человеком. В изголовье его кровати всегда висела кона с образом Спасителя, которой он молился Богу, когда приходил в спальню. Он постоянно ходил с нами церковь, набожно стоял все время Богослужения» // Иллюстрированная Россия. 1934. № 43. С. 13. СЕРГЕЙ ВИКТОРОВИЧ ТРОИЦКИЙ 112 колокола, церковные одеяния, кадила, кресты и книги; а также бес покойство о сохранении и восстановлении старых церквей и мона стырей. Вечным памятником религиозности короля будет место его упокоения — Опленац. К Опленацу король всегда имел особый интерес, гораздо больший, чем ко двору своего деда в Белграде. Он не щадил ни средств, ни времени на Опленац, вникал в мельчайшие детали этого музея средневекового сербского искусства. 6 Речи... С. 6, 11, 12, 13, 15, 17, 18, 20, 22–26 и др. 7 Речи... С. 22–25. 6 Речи... С. 6, 11, 12, 13, 15, 17, 18, 20, 22–26 и др. 7 Речи... С. 22–25. Аннотация Он знал здесь не просто каждую картину, а, можно сказать, каждый камешек в мозаике, как буд то предчувствовал, что в скором времени Опленац станет его вечной резиденцией. О его глубокой религиозности свидетельствуют его речи, собранные и изданные в конце прошлого 1934 года: все они пронизаны религиозным чувством, повсюду 6 король говорит о своей молитве Богу, о вере в Него и о твёрдой надежде на Его помощь. А некоторые слова короля, которые он говорил своей армии в тяжёлые минуты мировой войны, например его Пасхальные и Рождественские речи 1916 года на о. Корфу, речь на Пасху в 1918 году в Солуни,7 очень похожи на настоящую проповедь. И в завершение, как ещё одно важное доказательство его внимательного отношения к вере, служит тот факт, что практически все нормы закона о различных вероисповеданиях были приняты именно в то время, когда законодательной властью обладал сам король, поэтому можно сказать, что югославянское религиозное законодательство — королевское законодательство. Многие вопросы религиозной политики были решены в пользу вероисповедания только благодаря личному вмешательству короля. Загребский архиепископ Бауэр поясняет, что каждый раз все труд ности, которые возникали перед югославским католиками, можно было уладить, только обратившись лично к Его Величеству. Например, вопрос о теологической науке, которою антиклерикалы хотели полностью ликвидировать. Религиозность короля была не просто сильной и глубокой, но и носила здравый и природный характер. Он не страдал от каких-либо заболеваний в своей религиозной жизни, от которых страдали многие правители. В его гармоничной природе не видно никаких болезнен ных преувеличений или извращённого религиозного чувства. Он был хорошим христианином, но не был ни фанатиком, ни ригористом, ни клерикалом, ни суевером. Он был представителем здорового направ ления религиозной жизни, которое американские психологи называют РЕЛИГИОЗНАЯ ПОЛИТИКА КОРОЛЯ АЛЕКСАНДРА ОБЪЕДИНИТЕЛЯ 113 «религиозным прагматизмом». Он хорошо усвоил христианскую истину о том, что без дела, вера сама по себе мертва (Иак. 2, 17) и всегда ставил на первое место осуществление той жизненной цели, которая была возложена на него свыше, — спасение и объединение своего народа. Он знал слова из Евангелия: Нет больше той любви, как если кто поло жит душу свою за друзей своих (Ин. 15, 13), и вся его жизнь была одним большим подвигом непрерывного жертвования своей жизни во имя своего народа. Народная молва наградила его титулом «король-витязь», но его героизм взрос на христианском поприще и без этой постоянной самоотверженности, без самопожертвования не может быть настоящего героизма. Одним словом, его верой была религия долга перед Оте чеством и своим народом. 8 Речи… С. 70. 9 Речи... С. 77. 10 Речь епископа М. Калоджере // «Старокатолик» 1934. № 10. С. 5–6. 11 По данным переписи населения от 31.01.1921, православные составляли 46,6 %, католи ки — 39,4 %, мусульмане — 11 %, евангелисты — 1,8 %, иудеи — 0,5 %, греко-католики — 0,4 %, другие — 0,2 %. Новая перепись населения от 31.01.1931 даёт следующие цифры: православные — 48,7 %, римо-католики — 34,45 %, греко- и армяно-католики — 0,32 %, старокатолики — 0,05 %, евангелисты, лютеране — 1,26 %, мусульмане — 11,20 %, евреи — 0,49 %, атеисты — 0,01 %. Аннотация Свой судьбоносный государственный акт 6 января 1929 года 8 он закончил такими словами: «Служба народу — самая святая служба». Он говорил то, что думал, и работал всегда так же. Несмотря на то, что Александр на первое место ставил свою про светительскую миссию, он никогда не забывал ни о национальном вопросе, ни о нищете, голодающих и жаждущих. Его благотворительные пожертвования, фонды, стипендии и взносы в большинстве своём расходились по всем уголкам нашей родины. Он всегда был там, где были страдания и горе, где он мог утереть слёзы несчастных. Религиозность короля носила не только здравый и глубокий харак тер, но и была очень широка. Она была так широка, как широко истинное христианство без проявлений его клерикальных искажений. Он знал, что стены, разделяющие вероисповедания, не достигают небес. Он знал, что точки в радиусе круга, чем ближе к центру, тем ближе межу собой. Так и люди: чем ближе они к Богу, тем ближе они друг к другу, и поэтому истинная религиозность может только сближать, а не разделять людей. И будучи настоящим чадом Церкви своих предков, он не терпел религи озной борьбы, соперничества между верами, которые под маской рели гиозных интересов всегда скрывают человеческий эгоизм и людские страсти. В этом смысле король был до мозга костей верен традициям православия старых сербских правителей, покровительствовавших другим вероисповеданиям. Великий сербский князь Неманя высылал средства церкви св. Николы в Баре и св. Петра в Риме, Стефан Перво венчанный поддерживал связи с Римом, а Милутин передал огромные территории католическому монастырю и приглашал в качестве свиде телей православных и католических епископов и даже старца-богомила. В то же время король был лучшим последователем великого югослава СЕРГЕЙ ВИКТОРОВИЧ ТРОИЦКИЙ 114 епископа Штросмайера, который в 1877 году писал в своём послании: «Вера, которая бы сеяла раздор между братьями, не была бы верой, а суеверием; не была бы Божьей истиной, а развратом людским». «Пусть Господь Бог... вдохновит Вашу Светлость на мудрость, дабы вести вперёд Сербскую Православную Церковь в братской любви и согласии с другими вероисповеданиями в Королевстве Югославия в её служении Богу и моральном просвещении народа», — говорил король Патриарху Варнаве 13 апреля 1930 года по случаю его рукоположения.9 Один из церковных деятелей абсолютно правильно писал о короле, обращаясь к его словам, что «как настоящий сын Православной Церк ви, он с полным правом мог называть “своей” или “нашей” любую христианскую церковь.10 Двигаться в том же направлении его заставляли как объективные, так и один субъективный фактор, а именно характер его личной религи озности. которые, сейчас можно сказать, витали в воздухе всего культурного мира и которые без сомнения влияли на наши законодательные факторы.12 Если перейти от этих фактов к идеям, которые пытался воплотить король в своей церковной политике, то мы увидим, что эти идеи пол ностью соответствуют как характеру его личной религиозности, так и основным государственным потребностям. Это такие идеи, как: а) идея свободы вероисповедания, как отдельного человека, так и религиозных организаций, б) право открыто исповедовать религию, в) право на при вилегии и, наконец, г) равноправие всех религиозных организаций. А) Древняя христианская Церковь высоко ценила свободу верои споведания. Писатель III века Лактанций говорит, что религия может существовать только там, где существует свобода, и она исчезает, как только исчезает свобода, и что истину нужно защищать verbis, nоn verberibus, (словами, а не пощёчинами). По сути, Церковь, кото рая находится в рабстве у государства, а значит и в моральном пара личе, не может служить хранителем моральной силы ни общества, ни государства и тогда оказывается не нужной ни первому, ни второму. К сожалению, история Церкви свидетельствует, что этот принцип посто янно попирается и было бы тяжело точно сказать, кто в этом больше виноват — представители государственной власти или представители самой Церкви. Ограничения свободы вероисповедания существовали и в довоенной Сербии, когда действовала система с государственной православной верой и с запретом прозелитизма других вероисповеда ний. Однако югославское законодательство осуществляет идею свободы вероисповедания в очень широком объёме. А) Древняя христианская Церковь высоко ценила свободу верои споведания. Писатель III века Лактанций говорит, что религия может существовать только там, где существует свобода, и она исчезает, как только исчезает свобода, и что истину нужно защищать verbis, nоn verberibus, (словами, а не пощёчинами). По сути, Церковь, кото рая находится в рабстве у государства, а значит и в моральном пара личе, не может служить хранителем моральной силы ни общества, ни государства и тогда оказывается не нужной ни первому, ни второму. К сожалению, история Церкви свидетельствует, что этот принцип посто янно попирается и было бы тяжело точно сказать, кто в этом больше виноват — представители государственной власти или представители самой Церкви. Ограничения свободы вероисповедания существовали и в довоенной Сербии, когда действовала система с государственной православной верой и с запретом прозелитизма других вероисповеда ний. Однако югославское законодательство осуществляет идею свободы вероисповедания в очень широком объёме. Эта идея, провозглашённая ещё в декларации регента Александра в 1919 г., красной нитью проходит через все наши законодательные акты, связанные с вероисповеданием, а особенно чётко отражена в Видовданской и Сентябрьской Конституциях, которые узаконили полную свободу совести и вероисповедания для всех граждан. 12 Чтобы доказать это, достаточно сравнить 12 ст. Видовданской конституции и 11 ст. Сентябрьской конституции с ст. 14 австрийского Основного государственного закона от 21.12.1867 (ном. 142), или со статьей 111 и 112 Конституции Польши от 04.03.1921 и наконец с § 137–141 Веймарской конституции 1919 г. Аннотация В религиозном плане созданное им Королевство Югославия уже не было тем, чем была старая патриархальная Сербия, в которой были практически только православные жители, где неправославного насе ления было настолько мало, что оно долгое время даже не учитывалось в государственном законодательстве и в расчёт принимались только православные. Уже упоминавшийся мирный договор, подписанием которого завершилась Балканская война, изменил ту религиозную гомогенность Сербии и вынудил сербское правительство подписать конкордат в 1914 году. Договоры, подписанные после завершения Мировой войны, обеспечили большинство неправославного населени я.11 В результате сербская система государственной религии, которая основывалась частично на идее византийского двуединства, а частично на западном принципе государственной религии, должна была быть уничтожена. Помимо этих двух факторов — характера личной религиозно сти короля и религиозный состав Югославии — необходимо обратить внимание на третий, на влияние со стороны, на либеральные идеи, РЕЛИГИОЗНАЯ ПОЛИТИКА КОРОЛЯ АЛЕКСАНДРА ОБЪЕДИНИТЕЛЯ 115 которые, сейчас можно сказать, витали в воздухе всего культурного мира и которые без сомнения влияли на наши законодательные факторы.12 которые, сейчас можно сказать, витали в воздухе всего культурного мира и которые без сомнения влияли на наши законодательные факторы.12 Другими словами, все граждане имеют право не только исповедовать любую религию, но также обладают свободой не придерживаться никакой веры, а в случае когда кто-то принадлежит к какой-либо вере, то никто не может его заставлять открыто подтверждать своё вероисповеданье или участвовать в актах, церемониях и обрядах, а равно подпадать под власть другого лица. 116 СЕРГЕЙ ВИКТОРОВИЧ ТРОИЦКИЙ Государственные конституции запрещают не только любое кос венное принуждение граждан совершать вероисповедальные обря ды, но и предоставление, а равно и ущемление политических прав, поскольку по Конституции «осуществление этого права не зависит от вероисповедания». Эта свобода личности подразумевает под собой лишь одно ограничение. Государство не признаёт свободу исповедания той веры, чьи законы противоречат общественному порядку и морали. Из чего следует статья Конституции о том, что никто не может быть освобождён от своих гражданских обязанностей в связи с законами своей веры. Но это ограничение лишь на первый взгляд ограничи вает свободу вероисповедания, так как это ограничение возникает из самой сути религии, которая всегда служит основой общественного порядка и морали. И если некая религиозная организация противоре чит общественному порядку и морали, то становится ясно, что речь идёт об извращённой форме религиозной жизни, которая не может пользоваться теми правами, которыми наделены вероисповедания. По крайней мере, так учил Основатель христианской религии, Кото рый и сам платил налог и заповедовал, что нужно отдавать кесарю то, что ему принадлежит. В нашей стране свобода вероисповедания гарантирована не только каждому человеку в отдельности, но и религиозным организациям. «Принятые и признанные религии самостоятельно устраивают свою внутреннюю деятельность и распоряжаются своими средствами и фон дами в рамках закона» — гласит конституция, её дублируют другие законы об отдельных вероисповеданиях. Закреплённая здесь свобода вероисповедания называется автономией. В качестве автономных религиозные организации осу ществляют свою деятельность в соответствии с законом, и над ними осуществляется формальный надзор со стороны государства, но они не находятся в распоряжении государственных органов. Разумеется, автономия — это ещё не полная самостоятельность, а лишь самостоя тельность в границах закона, которые принимает государство. В этом отношении положение Церкви в нашей стране похоже на положение локальных политических областей с самоуправлением (например, бановина, општина). Однако здесь существует существенное отли чие. Религиозные организации, в отличие от политических областей с самоуправлением, созданы не государством и не выполняют госу дарственных задач. Они возникли сами по себе и имеют свои опре делённые цели и поэтому не соединяются с государством, как другие РЕЛИГИОЗНАЯ ПОЛИТИКА КОРОЛЯ АЛЕКСАНДРА ОБЪЕДИНИТЕЛЯ 117 образования с самоуправлением. 13 Например, 18.5.1926 г. такой письменный административный «конкордат» из 15 пара графов был составлен министром по религиозным вопросам и синодом Сербской Православной Церкви (Син. № 997). Некоторые пункты этого «конкордата» вошли позднее в 1929 г. в Закон о Сербской Православной Церкви. которые, сейчас можно сказать, витали в воздухе всего культурного мира и которые без сомнения влияли на наши законодательные факторы.12 Ещё одно отличие заключается в том, что религиозные общины организованы на своей отдельной базе, в то время как бановины и општины основываются на терри ториальном принципе. Те границы, которые наше государство установило для вероиспове даний своим законодательством, не лишают их свободы, прежде всего потому, что эти законы были установлены не только по воле государства, но и основывались на договоре с представителями религий и выражали волю не только правительства, но и самих церквей. Перед установ лением законодательства в отношении вероисповедания в нашей стране, в 1921 году среди представителей разных религий по всей стране был проведён опрос, и таким образом государство заложило в основу своего законодательства мнения и просьбы самих верующих. И позднее, когда в стране готовился закон о свободе вероисповедания, то уже де-юре или де-факто всегда принималась во внимание воля самих церквей. Отношения с Римско-католической Церковью должны регулироваться односторонним государственным законодательным актом, а не договором с Ватиканом, т. е. конкордатом, — этот принцип несколько раз был официально объявлен государственным факто ром. Что касается других религий, то здесь перед изданием законов о вероисповедании были проведены переговоры с представителями других религий, а в отдельных случаях на основе этих переговоров были собраны sui generis административные конкордаты, в большинстве случаев устные, а иногда даже и письменные.13 Таким образом, наше законодательство в области регулирования религиозных отношений похоже на законодательство некоторых европейских стран, например Германии или Польши,14 и фактически принят принцип договорённости 13 Например, 18.5.1926 г. такой письменный административный «конкордат» из 15 пара графов был составлен министром по религиозным вопросам и синодом Сербской Православной Церкви (Син. № 997). Некоторые пункты этого «конкордата» вошли позднее в 1929 г. в Закон о Сербской Православной Церкви. 14 Австро-Венгрия 31.3.1880 г. подписала конвенцию (по сути, конкордат) с Константи нопольской церковью о православных епархиях в Боснии и Герцеговине. Бавария заключила договор 15.10.1924 г. с евангелистско-лютеранской церковью на правом берегу Рейна и с объединённой евангелистской церковью в Пфальце. 11.5.1931 г. Пруссия заключила договор с прусской Евангелическо-лютеранской церковью, который был ратифицирован 26.6.1831 г. Статья 115 Конституции Польши гласит: «Церкви религиозных меньшинств и другие, признанные законом организации, управляются по своим законам власти, которые государство признаёт, если они не будут содержать в себе никаких правил, противоречащих государственным законам. Отношения государства с такими церквами будут определены путём установления закона по договору с его законными СЕРГЕЙ ВИКТОРОВИЧ ТРОИЦКИЙ 118 в регулировании государственных отношений не только с Римско- католической Церковью, но и с другими вероисповеданиями. Живот” 1926, сентябрь. С. 344–356 о приниципе церковно-государственного договора в Сербии в „Проблем црквеног законодавства“. Ниш, 1926, с. 11-13). 15 Это: а) Закон о Сербской православной церкви от 8.11.1929 г. (Служб. Новине од 16.11.1929 г № 269), б) Закон о религиозной общине Евреев от 29.12.1929 г. (Служб. Новине от 1919, № 301-LXXII), Закон об исламской религиозной общине от 31.1.1930 г.(Служб. Новине от 4.11.1930, № 29-X) и г) Закон о евангелическо-христианских церквах и реформиро ванной христианской церкви т 16.4.1930 г. (Служб. Новине от 26.4.1930 г, № 95-XXXVII). 16 Это следующие уставы: а) Исламской религиозной общины от 9.7.1930 г, (Служб. Новине от 25.7.1930 г., № 161), б) Немецкие евангелистские христианские церкви аугсбургской традиции от 19.11.1930 г., (Служб. Новине от 22.12.1930 г., № 293-CI), в) Сербской Православной Церкви от 16.11.1931 г., (Служб. Новине от 24.11.1931 г. № 244), г) Сло вацких евангелистских христианской церкви аугсбургской традиции от 24.6.1932 г., (Служб. Новине от 16.7.1932 г., № 161-LXXII), д) Реформированной христианской Церкви от 1.5.1933 г. (Служб. Новине от 8.6.1933, № 126-XXVI) 17 С У С б й П й Ц 278 У ф Ц § 19 которые, сейчас можно сказать, витали в воздухе всего культурного мира и которые без сомнения влияли на наши законодательные факторы.12 Наши государственные законы об отдельных конфессиях 15 практи чески не касаются их внутреннего устройства и в большинстве случаев, регулируют лишь их внешние связи с государством. Между тем, их внутреннее устройство не регулировалось ни государственными пра вовыми актами, ни собственными религиозными уставами,16 которые были разработаны представителями вероисповеданий в соответствии с религиозными учениями, которые правительство лишь одобрило. Эти религиозные уставы ни в коей мере не могут быть изменены пра вительством в одностороннем порядке. Более того, государство даже не обладает правом инициативы в этих вопросах, так что устройство религиозных организаций не зависит от обычных партийных изменений государственной политики.17 Что касается утверждения религиозных уставов со стороны государства, то это даже полезно для самих верои споведаний, потому как эти уставы показывают бессилие властей с точки зрения государственных законов, в то время как сами уставы могут быть изменены только по инициативе компетентной религиозной власти. Эти уставы гарантируют религиозным организациям полную независимость и широкие законные права, и на основе этих прав представители каждой религии могут свободно устанавливать нормы, представителями (см.: проф. М. Зызикин. О церковној самостојатељности в новђјших постројенијах церковно-государственних отношениј. Варшава, 1932, подробнее мои статьи: “Међународна заштита религијских права” (Архив за правнеи друштвене науке, 1925, фебр›март. особенно с. 199–209) и “Међународна правна способност цркве”, “Хришћ. Живот” 1926, сентябрь. С. 344–356 о приниципе церковно-государственного договора в Сербии в „Проблем црквеног законодавства“. Ниш, 1926, с. 11-13). представителями (см.: проф. М. Зызикин. О церковној самостојатељности в новђјших постројенијах церковно-государственних отношениј. Варшава, 1932, подробнее мои статьи: “Међународна заштита религијских права” (Архив за правнеи друштвене науке, 1925, фебр›март. особенно с. 199–209) и “Међународна правна способност цркве”, “Хришћ. Живот” 1926, сентябрь. С. 344–356 о приниципе церковно-государственного договора в Сербии в „Проблем црквеног законодавства“. Ниш, 1926, с. 11-13). представителями (см.: проф. М. Зызикин. О церковној самостојатељности в новђјших постројенијах церковно-государственних отношениј. Варшава, 1932, подробнее мои статьи: “Међународна заштита религијских права” (Архив за правнеи друштвене науке, 1925, фебр›март. особенно с. 199–209) и “Међународна правна способност цркве”, “Хришћ. Живот” 1926, сентябрь. С. 344–356 о приниципе церковно-государственного договора в Сербии в „Проблем црквеног законодавства“. Ниш, 1926, с. 11-13). 15 Это: а) Закон о Сербской православной церкви от 8.11.1929 г. (Служб. Новине од 16.11.1929 г № 269), б) Закон о религиозной общине Евреев от 29.12.1929 г. (Служб. Новине от 1919, № 301-LXXII), Закон об исламской религиозной общине от 31.1.1930 г.(Служб. Новине от 4.11.1930, № 29-X) и г) Закон о евангелическо-христианских церквах и реформиро ванной христианской церкви т 16.4.1930 г. (Служб. Новине от 26.4.1930 г, № 95-XXXVII). представителями (см.: проф. М. Зызикин. О церковној самостојатељности в новђјших постројенијах церковно-государственних отношениј. Варшава, 1932, подробнее мои статьи: “Међународна заштита религијских права” (Архив за правнеи друштвене науке, 1925, фебр›март. особенно с. 199–209) и “Међународна правна способност цркве”, “Хришћ. Живот” 1926, сентябрь. С. 344–356 о приниципе церковно-государственного договора в Сербии в „Проблем црквеног законодавства“. Ниш, 1926, с. 11-13). которые, сейчас можно сказать, витали в воздухе всего культурного мира и которые без сомнения влияли на наши законодательные факторы.12 16 Это следующие уставы: а) Исламской религиозной общины от 9.7.1930 г, (Служб. Новине от 25.7.1930 г., № 161), б) Немецкие евангелистские христианские церкви аугсбургской традиции от 19.11.1930 г., (Служб. Новине от 22.12.1930 г., № 293-CI), в) Сербской Православной Церкви от 16.11.1931 г., (Служб. Новине от 24.11.1931 г. № 244), г) Сло вацких евангелистских христианской церкви аугсбургской традиции от 24.6.1932 г., (Служб. Новине от 16.7.1932 г., № 161-LXXII), д) Реформированной христианской Церкви от 1.5.1933 г. (Служб. Новине от 8.6.1933, № 126-XXVI) 15 Это: а) Закон о Сербской православной церкви от 8.11.1929 г. (Служб. Новине од 16.11.1929 г № 269), б) Закон о религиозной общине Евреев от 29.12.1929 г. (Служб. Новине от 1919, № 301-LXXII), Закон об исламской религиозной общине от 31.1.1930 г.(Служб. Новине от 4.11.1930, № 29-X) и г) Закон о евангелическо-христианских церквах и реформиро ванной христианской церкви т 16.4.1930 г. (Служб. Новине от 26.4.1930 г, № 95-XXXVII). 16 16 Это следующие уставы: а) Исламской религиозной общины от 9.7.1930 г, (Служб. Новине от 25.7.1930 г., № 161), б) Немецкие евангелистские христианские церкви аугсбургской традиции от 19.11.1930 г., (Служб. Новине от 22.12.1930 г., № 293-CI), в) Сербской Православной Церкви от 16.11.1931 г., (Служб. Новине от 24.11.1931 г. № 244), г) Сло вацких евангелистских христианской церкви аугсбургской традиции от 24.6.1932 г., (Служб. Новине от 16.7.1932 г., № 161-LXXII), д) Реформированной христианской Церкви от 1.5.1933 г. (Служб. Новине от 8.6.1933, № 126-XXVI) 17 См. например, Устав Сербской Православной Церкви, ст. 278; Устав реформ. Церкви § 19; Устав Словацкой церкви § 124; став Немецкой церкви § 107; Устав Исламской общины § 118. РЕЛИГИОЗНАЯ ПОЛИТИКА КОРОЛЯ АЛЕКСАНДРА ОБЪЕДИНИТЕЛЯ 119 правила, издавать своды законов, которые не должны никем одобряться. О широте свободы действий свидетельствует и тот факт, что на её осно ве некоторые конфессии (сначала немецкая Евангелистская, а затем и Православная Церковь) издают свои статуты собственного брачного права, которые и в настоящее время имеют силу для государства. Рели гиозные организации также имеют право создавать и открывать свои внутренние объединения и учреждения, лишь с одним ограничением: эти внутренние религиозные объединения не могут иметь ни партийно- политических целей, ни целей физического воспитания.18 С точки зрения управления и суда автономия религиозных органи заций ещё шире, чем в плане законодательства. Государство абсолютно не вмешивается в самую важную для религии область — в область религиозного и морального учения и богослужения. 18 Сентябрьская конституция, ст. 13. которые, сейчас можно сказать, витали в воздухе всего культурного мира и которые без сомнения влияли на наши законодательные факторы.12 И если государ ство оставило за собой право утверждать глав религиозных общин, то не стоит забывать, что эти деятели в то же время поддерживают отношения с государством и в некотором смысле похожи на диплома тических представителей, для назначения которых чаще всего необхо димо согласие государства, для работы в котором они аккредитованы. Широкая имущественная религиозная автономия ограничивается только формальным надзором, надзором по соблюдению законности, а не целесообразности. Кроме того, это небольшое ограничение рели гиозной автономии у нас вызвано не негативным отношением к вере, а необходимостью, связанной с другой важной привилегией, которой у нас пользуются религии, — их свободой общественной деятельности. Б) Наше государство не воспринимает религиозные организации индифферентно как частные объединения, а видит в них полезные для общества и государства факторы, влияющие на духовную куль туру, и поэтому предоставляет им права публичных государственных учреждений и оказывает им свою помощь. У нас религиозные орга низации не только имеют право открытого вероисповедования, т. е. право открыто совершать богослужение, религиозные процессии, звон в колокола и т. д., но и право распоряжения, на основе которого официальные религиозные акты имеют характер общественных актов, так что они переплетены с правовыми последствиями, не зависящими от воли частных лиц. А поскольку сами религии не имеют внешних сил, своей полиции, своей армии, отсюда возникает их право в своей официальной деятельности опираться на помощь со стороны админи стративных и судебных властей, например: при сборе дополнительных Б) Наше государство не воспринимает религиозные организации индифферентно как частные объединения, а видит в них полезные для общества и государства факторы, влияющие на духовную куль туру, и поэтому предоставляет им права публичных государственных учреждений и оказывает им свою помощь. У нас религиозные орга низации не только имеют право открытого вероисповедования, т. е. право открыто совершать богослужение, религиозные процессии, звон в колокола и т. д., но и право распоряжения, на основе которого официальные религиозные акты имеют характер общественных актов, так что они переплетены с правовыми последствиями, не зависящими от воли частных лиц. А поскольку сами религии не имеют внешних сил, своей полиции, своей армии, отсюда возникает их право в своей официальной деятельности опираться на помощь со стороны админи стративных и судебных властей, например: при сборе дополнительных СЕРГЕЙ ВИКТОРОВИЧ ТРОИЦКИЙ 120 налогов, вынесении вердикта суда при исполнении приговора рели гиозного суда и т. д. С правом открытого вероисповедания связано также привилеги рованное положение лиц, несущих на себе общественные функции, т. е. священнослужителей. Священнослужители обладают усиленной уголовно-правовой защитой. которые, сейчас можно сказать, витали в воздухе всего культурного мира и которые без сомнения влияли на наши законодательные факторы.12 Во многих своих личных и имущественных привилегиях (или с точки зрения личной и имущественной неприкос новенности) они приравниваются к государственным чиновникам, а в отношении военной службы имеют перед ними преимущество. Для конфессий ещё более важное значение, чем эта административно- полицейская поддержка и привилегии для священнослужителей, имеет помощь, которую они получают от государства при проведении своей образовательной миссии в школах и в своём собственном обеспечении. Хоть Видовданская Конституция и говорит только о факультативном преподавании Закона Божьего («по желанию родителей»), а в Сентябрь ской конституции вообще не упоминается о религиозном образовании, новый закон об основных, педагогических, средних и специальных школах обеспечивает обязательное преподавание Закона Божьего в этих учебных заведениях. Кроме того, государство берёт на себя расходы по содержанию учителей по религиозным дисциплинам, в основных школах частично, а во всех остальных полностью. Государство также взяло на себя расходы по содержанию религиозных богословских школ, которые служат для подготовки будущих священнослужителей. При этом начальная и средняя школы полностью подчинены религиозной власти, а на высших богословских факультетах свобода университетской науки ограничена ради конфессиональных интересов правом надзора цер ковной власти и её правом veto в отношении преподавателей, которые могли бы отступить от религиозных догм. Особенно в этом отношении в благоприятной ситуации находилась Римско-католическая Церковь, которая кроме восьми богословий имела три богословских факультета, в то время как Православная Церковь имела только один. Свобода пре подавания во всех религиозных учебных заведениях ограничивается государством лишь государственным надзором и одним единственным требованием государственной конституции, что «все школы должны давать моральное воспитание и развивать гражданскую сознательность в духе народного единства и веротерпимости» (ст. 16., пар. 3, 4). Особенное внимание в высшей государственной школе уделено изучению устройства религиозных организаций. Церковное право преподаётся не только на богословском и философском факультетах, РЕЛИГИОЗНАЯ ПОЛИТИКА КОРОЛЯ АЛЕКСАНДРА ОБЪЕДИНИТЕЛЯ 121 но и на юридическом факультете. А на юридическом факультете Бел градского университета преподаются и законы шариата.19 В) Государство содержит не только богословские школы, но и пре доставляет значительную финансовую помощь на содержание других церковных учреждений. Каждый год в государственном бюджете пред усматривается сумма на помощь религиозным организациям в размере около 100 000 000 динар.20 Ни по старой, ни по новой конституции государство не обязано помогать церквам и оказывает содействие лишь на своё усмотрение, если бюджет это позволяет. Однако в то же время в конституции прописано, что если бюджет предусматривает подобную помощь, то она должна быть распределена между конфесси ями сообразно числу их верующих и их потребностям. Однако законы об отдельных религиях изменили это конституционное правило в пользу Церкви. 20 Из бюджета за 1934–1935 г. на помощь религиозным конфессиям всего было выделено 100 168 813 динаров, в том числе 83 244 127 дин. на содержание религиозных учреж дений и 16 894 068 дин. на содержание религиозных школ. 19 См.: Троицки С. Место Црквеного права на правним факултетима. Ср. Карловци, 1933. которые, сейчас можно сказать, витали в воздухе всего культурного мира и которые без сомнения влияли на наши законодательные факторы.12 А именно они признали за отдельными церквами постоянное право на всю сумму расходов, одобренных последним государственным бюджетом на содержание церковных органов, организаций и лиц. Эти средства переходили от государства соответствующей религиозной общине. В итоге одновременно с тем, что в последние годы все расходы на государственные учреждения урезаются, размер помощи церквам остаётся тем же и даже увеличивается. Только те церкви, чьё положение ещё не урегулировано законодательными актами (Римско-католическая и Старокатолическая) не получают полной государственной помощи, но не страдают от этого, поскольку государство выделяет необходимы суммы на содержание школ и духовных лиц. Государственная материальная поддержка религиозных организа ций не ограничивается бюджетом. Государство оказывает значительную помощь и другими методами — путём сбора налогов и церковных поборов своими финансовыми органами, путём узаконивания рас ценок на официальные услуги церковных властей, путём содержания священнослужителей в армии, на казённых предприятиях и в других государственных организациях, путём законодательного регулирова ния помощи на религиозные цели от политических общин и регионов с самоуправлением, путём освобождения части церковного имущества от прямых налогов (иммунитет на собственность) и т. д. СЕРГЕЙ ВИКТОРОВИЧ ТРОИЦКИЙ 122 Г) Все эти права и привилегии религиозные организации получают на основе строгих принципов справедливости. Система, при которой в привилегированном положении находится только одна Православная Церковь, имеет под собой некую основу, которая вытекает из её исто рических заслуг, а также из числа её паствы, их гораздо больше, чем людей, исповедующих другие религии, и из религиозной принадлеж ности главы государства. Такая система, которая и сейчас существует во многих странах, например в Польше, Болгарии и Румынии, была упразднена в нашей стране одним из первых государственных актов нового королевства — прокламацией регента Александра от 6 января 1919 года,21 и заменена системой равноправности религий и паритета. Закон о переходе на единый календарь от 01.01.1919 заменил православ ный юлианский календарь на общий для всех западный григорианский. Распоряжение от 31.07.1919 об учреждении министерства по рели гиозным делам поставило перед этим министерством задачу обеспечить последовательное выполнение основного принципа равноправия всех признанных законом вероисповеданий в королевстве. Наконец, государственная конституция гласит: «Все признанные вероисповедания равны перед законом». Отсюда следует, что в кон ституции не говорится о равноправии вообще всех вероисповеданий, а только тех, которые признаны государством. 22 По статистическим данным за 31.01.1931 представители других вероисповеданий составляли лишь 0,12 %. 21 Это будет... гарантировать свободу и равноправие религий» // Речи… С. 40. 23 Если условно взять размер помощи, которую получают старокатолики, за 100, условными цифрами по числу верующих и сторонников остальных вероисповеданий были бы следующими: евреи — 29 %, мусульмане — 21 %, православные — 12 %, католики — 11 %, протестанты — 11 %. которые, сейчас можно сказать, витали в воздухе всего культурного мира и которые без сомнения влияли на наши законодательные факторы.12 Однако на практике это ограничение не имеет значения, поскольку всё население нашего королевства составляют представители всего шести официальных религий (православные, римо-католики, мусульмане, евангелисты, иудеи и старокатолики), в то время как представители других веро исповеданий составляют незначительное меньшинство (всего одну восьмую часть процента 22) и большинство из них рассеяны по всей территории королевства без какой-либо организации. Однако эти рели гиозные течения могут быть признаны государством путём принятия соответствующего закона. В государственных конституциях говорится не о конкретных церк вах, а обо всех религиях в целом. Единственное исключение составляет ислам. В связи с особенностями семейных отношений у мусульман условия Сен-Жерменского договора гарантируют шариатскому суду право регулировать семейно-родственные отношения и вопросы наследования (ст. 100) — привилегия, которой мусульмане сейчас РЕЛИГИОЗНАЯ ПОЛИТИКА КОРОЛЯ АЛЕКСАНДРА ОБЪЕДИНИТЕЛЯ 123 не обладают даже в Турции. Это равноправие существует не просто на бумаге, а осуществляется на деле. Ярким примером этого может служить распределение государственной помощи между конфессиями. Как уже говорилось, средства должны распределяться сообразно числу верующих и по необходимости. Если рассмотреть первый из критериев, по которому государственные средства между конфессиями распре деляются в соответствии с числом верующих, то видно, что суммы, выделяемые для наиболее многочисленных конфессий, Православной и Римско-католической церквей, соответствуют числу их верующих. Между тем религии с меньшим числом верующих получают относи тельно крупную государственную помощь 23 и находятся в привилеги рованном положении. Однако здесь стоит принять во внимание общий экономический закон, что чем меньше организация, тем всегда больше организационные затраты. Кроме того, необходимо учитывать и особенности в устройстве отдельных вероисповеданий, например целибат католического клира, имущество самих церквей, и признать, что фактическое распределение средств между конфессиями близко к идеальной модели справедли вости и полностью соответствует требованиям конституции, чтобы государственная помощь распределялась в соответствии с потребно стями церковных организаций. И как это сильно отличается, например, от Польши, где, несмотря на конституцию, Православная Церковь получает в несколько раз меньше государственных субсидий, чем в соответствии с числом верующих ей положено по закону. Точно так же о конфессиональном равноправии свидетельствует закон об отдельных вероисповеданиях — православии, мусульманстве, иудаизме и еванге листской церкви — практически все имеют одинаковые права, и если между ними существует незначительная разница, то она вызвана раз личиями в устройстве самих церковных организаций и нет никаких различий в отношении них со стороны государства. Такое положение религиозных организаций в нашей стране было создано благодаря религиозной политике покойного короля. Разумеется, мы не можем утверждать, что эта политика выполнила всё, что нужно было сделать. 25 «Существующая в различных европейских странах система церковно-государственных отношений является беспринципной смесью фрагментов различных систем» // Суворов. Н. С. Учебник церковного права. М, 1908. С. 435. 24 Об этом см.: Др. Михаило Лановћ. Конкордат Југославије са Ватиканом. Београд, 1925. которые, сейчас можно сказать, витали в воздухе всего культурного мира и которые без сомнения влияли на наши законодательные факторы.12 В нашем религиозном законодатель стве до сих пор существуют большие лакуны, которые часто имеют СЕРГЕЙ ВИКТОРОВИЧ ТРОИЦКИЙ 124 неблагоприятные социальные и государственные последствия. Пре жде всего это то, что до сегодняшнего дня у нас нет общего закона о межрелигиозных отношениях, вследствие чего часто происходят кол лизии между государственными законами и статутами отдельных вероисповеданий, особенно в области брачного права. Однако работа в этом направлении была начата ещё десять лет назад. Кроме того, ещё не существует закона о положении и устройстве Римско-католической Церкви, но поскольку издание таких законов зависит от договора (от конкордата) с центральной властью Римско-католической Церкви, и было бы неправильным возлагать всю ответственность только на госу дарство.24 Даже на подписание договора с Православной Церковью, учитывая её устройство и положение в стране, потребовалось десять лет. Напоследок можно также отметить, что в законодательстве нашей религиозной политики нет полной согласованности в реализации еди ной правовой теории об отношениях между церковью и государством. А в религиозных законах, за исключением паритетно-правовой теории, мы можем найти отголоски других теорий — теории координации и теории государственной Церкви. Однако, как это верно отмечает профессор Н. Суворов,25 подобной согласованности нет ни в одном религиозном законодательстве в Европе. Кроме того, большая часть недостатков и лакун нашего рели гиозного законодательства находит своё объяснение и оправдание в недостатке времени и политических возможностей. Если сравни вать нашу религиозную политику с политикой других стран, то можно с полной уверенностью утверждать, что она предоставляет различным верам гораздо лучшее положение, чем в тех государствах, где отдаётся предпочтение большинству. Нет никаких сомнений, что положение Православной Церкви в нашей стране гораздо лучше, чем в России и даже в Греции, что протестантская церковь находится в лучших условиях, чем в самой Германии, а некоторые представители Про тестантской церкви после смерти короля даже делали официальные заявления, что нигде в мире протестанты не имеют таких хороших условий. Что касается иудеев, то я лично слышал 28.10.1934 г. подобное заявление от их представителей на траурном заседании в Суботице. Без сомнения, и ислам пользуется большими свободами у нас, чем РЕЛИГИОЗНАЯ ПОЛИТИКА КОРОЛЯ АЛЕКСАНДРА ОБЪЕДИНИТЕЛЯ 125 даже в самой Турции. Тяжело найти такую страну, где бы положение Старокатолической церкви было бы лучше, чем у нас. А поскольку, как мы уже видели, в Сербии все вероисповедания равноправны, то мы можем с уверенностью сделать вывод, что условия для всех религи озных течений, созданные покойным королём, очень благоприятны и все вероисповедания обладают всем необходимым для спокойной и плодотворной работы над осуществлением своих возвышенных задач духовного развития своих верующих. 26 Телеграмма от 10.10.1924 г. // «Гласник Петар». № 32–33. С. 501. которые, сейчас можно сказать, витали в воздухе всего культурного мира и которые без сомнения влияли на наши законодательные факторы.12 Деятельность покойного короля как защитника религиозных инте ресов не ограничивалась территорией Югославии или одной Сербии. Он был fidei defensоr — выступал за веру везде, где это было нужно. Его защитой и поддержкой прежде всего пользовалась Русская Православная Церковь в эмиграции. Протокол Лозаннской конференции 1923 г. сви детельствует, что первый иерарх Православной Церкви — Константино польский патриарх, не лишился своего престола в Константинополе, чего требовала Турция, только благодаря заступничеству представителя нашей страны. Король защищал интересы и Иерусалимской церкви в Палестине, и интересы славянских монахов на Святой Горе. В то же время он был крупнейшим идеологом примирения и сближения всех Православных церквей, и прежде всего Сербской и Болгарской. Глава Польской Церкви митрополит Дионисий по праву называл его защитником единства всех Православных церквей.26 Делегация, прибывшая на похороны короля от Константинопольской Церкви, и телеграммы, присланные главами всех Православных церквей, свидетельствуют об огромных заслугах короля-объединителя перед всем православным миром. Вот так широко понимал взятую на себя задачу осуществления второй части завета царя Душана ныне покойный король, а именно — осуществление не только политического, но и духовного единства своего народа и каких грандиозных успехов он достиг на этом поприще. Он те только объединил разрозненные части Сербской Православной Церкви и восстановил Сербский патриархат, не только создал все усло вия для мирного и плодотворного сотрудничества всех вероисповеданий в нашей стране, но и также старался решить эти вопросы и вне границ своего государства. В этом смысле его деятельность напоминает нам великого славянина, со дня смерти которого прошло уже почти 700 лет, о святом Савве, который точно так же всю свою жизнь посвятил при мирению и сближению православных государств и церквей. 126 СЕРГЕЙ ВИКТОРОВИЧ ТРОИЦКИЙ Однако сродни тому, как король в политической деятельности не ограничивался лишь своим народом, а трудился ещё и на благо мира во всей Европе, так же и в своей религиозной политике он не оставался в рамках своего государства и Православной Церкви, а был вдохнов лён святой мечтой об объединении христианских церквей и прежде всего Православной и Римско-католической. Причём это объединение он представлял себе немного иначе, чем сторонники заключения унии с Римом. Об этом свидетельствует одна очень информированная персона Римско-католической церкви в лондонском журнале «The Universe»: «Король Югославии Александр, чья смерть сильно сказалась на всём культурном мире, был набожным и искренним христианином. Воспи танный в традициях Православной Церкви, он всегда оставался верен своей религии. Одним из самых больших его желаний было сблизить его подданных, исповедующих две христианские конфессии. РЕЛИГИОЗНАЯ ПОЛИТИКА КОРОЛЯ АЛЕКСАНДРА ОБЪЕДИНИТЕЛЯ РЕЛИГИОЗНАЯ ПОЛИТИКА КОРОЛЯ АЛЕКСАНДРА ОБЪЕДИНИТЕЛЯ 127 Библиография Речи короля-витязя Александра I Объединителя / подготовили и опубликовали Р. Парежанин, Дж. Гаврилович. Београд, 1934. Немирович-Данченко В. И. Друг России // Иллюстрированная Россия. 1934. № 43. Суворов Н. С. Учебник церковного права. М., 1908. Суворов Н. С. Учебник церковного права. М., 1908. Зизикин М. О. церковној самостојатељности в новђјших постројенијах церковно- государственних отношениј. Варшава, 1932. Лановћ М. Конкордат Југославије са Ватиканом. Београд, 1925. Митривић-Мирољуб Љ. О камену Темељцу Христове Цркве. Београд, 1931. Калоджере М., епископ. Речь // Старокатолик. 1934. № 10. С. 5–6. Троицки С. Међународна заштита религијских права // Архив за правнеи друштвене науке. 1925. Фебр/март. Троицки С. Међународна правна способност цркве // Хришћ. Живот. 1926. Сентябрь. Троицки С. Место Црквеного права на правним факултетима. Ср. Карловци, 1933. которые, сейчас можно сказать, витали в воздухе всего культурного мира и которые без сомнения влияли на наши законодательные факторы.12 Однажды он мне сказал: “Вероятно вы, католики, по-другому рассматриваете новое объединение церквей, нежели я, но перед тем, как мы дойдём до этой стадии дискуссии, обе стороны должны сделать всё, от них зависящее. Необходимо большее понимание между католиками и православными… ”» «Король Югославии Александр, чья смерть сильно сказалась на всём культурном мире, был набожным и искренним христианином. Воспи танный в традициях Православной Церкви, он всегда оставался верен своей религии. Одним из самых больших его желаний было сблизить его подданных, исповедующих две христианские конфессии. Однажды он мне сказал: “Вероятно вы, католики, по-другому рассматриваете новое объединение церквей, нежели я, но перед тем, как мы дойдём до этой стадии дискуссии, обе стороны должны сделать всё, от них зависящее. Необходимо большее понимание между католиками и православными… ”» Разумеется, эти слова короля Александра, взятые из случайно го разговора, не дают нам полного представления о его программе. Но всё же они дают ясные и ценные сведения о том, что благодаря своей высокой духовности, он занимался такими важными и тяжёлыми про блемами широкой религиозной политики, которыми занимались другие великие славяне, начиная с Штросмайера и Владимира Соловьёва. И в этом вопросе король Александр шёл своим путем, можно сказать, путём синтеза идей св. Саввы и Штросмайера. Было бы большой ошибкой и знаком поверхностности и недаль новидности пытаться объяснить стремление короля к объединению церквей его политическими целями. Нет, это стремление исходило из глубины его религиозности, поскольку эта религиозность была выше тех стен, которые разделяют религии, и которые, как мы уже говорили, не достигают небес. И ключ к пониманию его привлека тельной и героической личности, к его деятельности, наполненной самопожертвованием и стремлением к высоким целям, мы можем найти только в религиозности самого короля. Слава нашему государственному и духовному объединителю! Слава крупнейшему последователю св. Саввы! Слава мученику во благо своего народа и всего человечества! Слава королю-витязю Александру, вечная ему память! Religious Policy of the King Alexander Karadjordjevic Sergej V. Troitsky Doctor of Canon Law, Professor of Moscow Theological Academy Sergej V. Troitsky Doctor of Canon Law, Professor of Moscow Theological Academy For citation: Troitsky, Sergej V. “Religious Policy of the King Alexander Karadjordjevic”. Praxis, vol. 2, no. 2, 2019, pp. 109–127 Abstract. The subject of restoration of its patriarchate after World War I is inseparably linked with a subject of autocephaly of the Serbian Orthodox Church. In this speech delivered in day of remembrance of Saint Savva in 1935, professor S. V. Troitsky analyzes confessional policy of the kingdom Yugoslavia and notes a role in this policy of the king Alexander Karadjordjevic (1888–1934). Keywords: confessional policy, State and Church relations, Kingdom Yugoslavia, Serbian Orthodox Church, Alexander I Karadjordjevic.
https://openalex.org/W2957205500	https://hal.science/hal-02383013/file/acp-19-8979-2019.pdf	English	null	Low-level stratiform clouds and dynamical features observed within the southern West African monsoon	Atmospheric chemistry and physics	2,019	cc-by	17,309	Low-level stratiform clouds and dynamical features observed within the southern West African monsoon Cheikh Dione, Fabienne Lohou, Marie Lothon, Bianca Adler, Karmen Babić, Norbert Kalthoff, Xabier Pedruzo-Bagazgoitia, Yannick Bezombes, Omar Gabella To cite this version: Cheikh Dione, Fabienne Lohou, Marie Lothon, Bianca Adler, Karmen Babić, et al.. Low-level strati- form clouds and dynamical features observed within the southern West African monsoon. Atmospheric Chemistry and Physics, 2019, 19 (13), pp.8979-8997. 10.5194/acp-19-8979-2019. hal-02383013 Distributed under a Creative Commons Attribution 4.0 International License Low-level stratiform clouds and dynamical features observed within the southern West African monsoon Cheikh Dione1,4, Fabienne Lohou1, Marie Lothon1, Bianca Adler2, Karmen Babi´c2, Norbert Kalthoff2, Xabier Pedruzo-Bagazgoitia3, Yannick Bezombes1, and Omar Gabella1 1Laboratoire d’Aérologie, Université de Toulouse, CNRS, UPS, Toulouse, France 2Institute of Meteorology and Climate Research, Karlsruhe Institute of Technology (KIT), Karlsruhe, Germany 3Wageningen University and Research, the Netherlands 4African Centre for Meteorological Applications for Development, Niamey, Niger Cheikh Dione1,4, Fabienne Lohou1, Marie Lothon1, Bianca Adler2, Karmen Babi´c2, Norbert Kalthoff2, Xabier Pedruzo-Bagazgoitia3, Yannick Bezombes1, and Omar Gabella1 1Laboratoire d’Aérologie, Université de Toulouse, CNRS, UPS, Toulouse, France 2Institute of Meteorology and Climate Research, Karlsruhe Institute of Technology (KIT), Karlsruhe, Germany 3Wageningen University and Research, the Netherlands 4African Centre for Meteorological Applications for Development, Niamey, Niger Correspondence: Cheikh Dione (cheikh.dione@aero.obs-mip.fr) Received: 30 October 2018 – Discussion started: 5 December 2018 Revised: 15 June 2019 – Accepted: 19 June 2019 – Published: 16 July 2019 Received: 30 October 2018 – Discussion started: 5 December 2018 Revised: 15 June 2019 – Accepted: 19 June 2019 – Published: 16 July 2019 atmospheric conditions, the maritime inﬂow reaches Savè around 18:00–19:00 UTC on average. This timing is corre- lated with the strength of the monsoon ﬂow. This time of ar- rival is close to the time range of the nocturnal low-level jet settlement. As a result, these phenomena are difﬁcult to dis- tinguish at the Savè site. The low-level jet occurs every night, except during rain events, and is associated 65 % of the time with low stratus clouds. Stratus clouds form between 22:00 and 06:00 UTC at an elevation close to the nocturnal low- level jet core height. The cloud base height, 310±30 m above ground level (a.g.l.), is rather stationary during the night and remains below the jet core height. The cloud top height, at 640 ± 100 m a.g.l., is typically found above the jet core. The nocturnal low-level jet, low-level stratiform clouds, monsoon ﬂow, and maritime inﬂow reveal signiﬁcant day-to-day and intra-seasonal variability during the summer given the im- portance of the different monsoon phases and synoptic at- mospheric conditions. Distributions of strength, depth, on- set time, breakup time, etc. are quantiﬁed here. These results contribute to satisfy the main goals of DACCIWA and allow a conceptual model of the dynamical structures in the lowest troposphere over the southern part of West Africa. Abstract. Low-level stratiform clouds and dynamical features observed within the southern West African monsoon During the boreal summer, the monsoon sea- son that takes place in West Africa is accompanied by low stratus clouds over land that stretch from the Guinean coast several hundred kilometers inland. Numerical climate and weather models need ﬁner description and knowledge of cloud macrophysical characteristics and of the dynam- ical and thermodynamical structures occupying the lowest troposphere, in order to be properly evaluated in this re- gion. The Dynamics-Aerosol-Chemistry-Cloud Interactions in West Africa (DACCIWA) ﬁeld experiment, which took place in summer 2016, addresses this knowledge gap. Low- level atmospheric dynamics and stratiform low-level cloud macrophysical properties are analyzed using in situ and remote sensing measurements continuously collected from 20 June to 30 July at Savè, Benin, roughly 180 km from the coast. The macrophysical characteristics of the stratus clouds are deduced from a ceilometer, an infrared cloud camera, and cloud radar. Onset times, evolution, dissipation times, base heights, and thickness are evaluated. The data from an ultra- high-frequency (UHF) wind proﬁler, a microwave radiome- ter, and an energy balance station are used to quantify the occurrence and characteristics of the monsoon ﬂow, the noc- turnal low-level jet, and the cold air mass inﬂow propagating northward from the coast of the Gulf of Guinea. The results show that these dynamical structures are very regularly ob- served during the entire 41 d documented period. Monsoon ﬂow is observed every day during our study period. The so- called “maritime inﬂow” and the nocturnal low-level jet are also systematic features in this area. According to synoptic HAL Id: hal-02383013 https://hal.science/hal-02383013v1 Submitted on 14 Jun 2022 L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés. HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. Distributed under a Creative Commons Attribution 4.0 International License Atmos. Chem. Phys., 19, 8979–8997, 2019 https://doi.org/10.5194/acp-19-8979-2019 © Author(s) 2019. This work is distributed under the Creative Commons Attribution 4.0 License. 1 Introduction Clouds are an important factor of uncertainty in climate change studies. The low-level stratiform clouds (LLSCs) that develop during the West African monsoon (WAM) along the Guinean coast likely contribute to this uncertainty because they modify the Earth’s energy budget in a region where the dynamics are driven by strong thermal and moisture gradi- ents and deep convection activity (Knippertz et al., 2011). However, until recently, very little attention was paid to these clouds. Poorly represented in numerical climate mod- els (Hannak et al., 2017), LLSCs form during the night and can extend from the Guinean coast several hundred kilome- ters inland; they last until midday the following day (Schrage and Fink, 2012; Schrage et al., 2007; Schuster et al., 2013). These authors emphasize the possible link between LLSC formation and persistence and the dynamical features in the region, like the monsoon ﬂow and the nocturnal low-level jet (NLLJ). However, very few observations of low clouds and associated dynamical processes are available, which pre- vents studies of LLSC formation and dissolution, as well as numerical climate and weather model validation. Filling the gap of observations and studying the LLSC life cycle were therefore the primary goals of the Dynamics-Aerosol- Chemistry-Cloud Interactions in West Africa (DACCIWA) project (Knippertz et al., 2015) with aircraft and ground- based campaigns (Flamant et al., 2017; Kalthoff et al., 2018) performed during summer 2016. At three supersites, Kumasi (Ghana), Savè (Benin), and Ile-Ife (Nigeria), ground-based measurements were performed. The corresponding datasets are described in Bessardon et al. (2019). Figure 1. Mean dew point temperature obtained from the ECMWF reanalysis at 2 m over southern West Africa during June 2016 and (solid black line) mean intertropical discontinuity (ITD) position during that month. The ITD position is deduced from the dew point temperature isoline of 15 ◦C. The star indicates the location of the Savè supersite. gion, the monsoon ﬂow is overlaid by the African easterly jet (Kalapureddy et al., 2010) at roughly 600 hPa. The mon- soon ﬂow exhibits seasonal evolution; its northern limit at the surface, called the intertropical discontinuity (ITD), moves with the apparent latitudinal position of the Sun. The onset of the monsoon ﬂow marked by a typical northward shift of the ITD and moist convection, manifested by mesoscale con- vective systems (MCSs), occurs every year around the end of June (Janicot et al., 2008; Sultan and Janicot, 2003). 1 Introduction It cor- responds with the start of an active phase of the monsoon associated with convection over land to Sahelian regions. In 2016, the monsoon onset was determined to be 21 June by Knippertz et al. (2017). During the DACCIWA ﬁeld cam- paign, which took place from 14 June to 31 July 2016, the ITD was located more than 400 km to the north of Savè (ITD at latitude > 12◦N, Savè at 8◦N). The mean ITD location in June 2016 is indicated in Fig. 1 and estimated using the 15 ◦C dew point temperature from ERA-Interim reanalysis (Buckle, 1996; Dee et al., 2011). According to Kalthoff et al. (2018), LLSCs form most nights at the three supersites instrumented during the DAC- CIWA ﬁeld campaign. Their bases are roughly around 300 m above ground level (a.g.l.) when they form and typically rise up to 800 m a.g.l. at noon on the following day. Deﬁning LLSCs using criteria based on a median cloud base fraction of 100 % at Savé and Kumasi, and a net longwave radiation threshold of −10 W m−2 for Ile-Ife, Kalthoff et al. (2018) noted some differences in the onset times of the LLSCs. These authors found that the onset times varied on average among the three sites: 21:00 UTC at Ile-Ife, 00:00 UTC at Kumasi, and 03:00 UTC at Savè. Beyond this general de- scription, large variability of the LLSC characteristics is ob- served from one night to the next. Knippertz et al. (2017) also deﬁned four synoptic phases of the monsoon using synoptic atmospheric circulations based on the precipitation difference between the coast (south) and the Sudanian–Sahelian zones (north): i. Phase 1 (1–21 June) was a pre-onset phase character- ized by the modulation of the winds and rainfall in the Guinea coastal area, due to three westward-propagating coherent cyclonic vortices between 4 and 13◦N. The onset itself was associated with a breakdown of the African easterly jet, Saharan Heat Low, and rainfall due to an extratropical trough and cold surge over northern Africa. To address this situation, it is important to consider the larger-scale context of the WAM and its dynamical features. Such work was the focus of the previous African Monsoon Multidisciplinary Analysis (AMMA) international project (Redelsperger et al., 2006). C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa 8980 Figure 1. Mean dew point temperature obtained from the ECMWF reanalysis at 2 m over southern West Africa during June 2016 and (solid black line) mean intertropical discontinuity (ITD) position during that month. The ITD position is deduced from the dew point temperature isoline of 15 ◦C. The star indicates the location of the Savè supersite. Published by Copernicus Publications on behalf of the European Geosciences Union. C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa 1 Introduction The primary dynamical feature affecting West Africa during half of the year is the monsoon ﬂow, which is due to synoptic-scale forcing associated with a strong thermal gradient between the cold tongue over the Gulf of Guinea and dry and warm air inland in the Saha- ran Heat Low (Lavaysse et al., 2009). In the southern re- ii. Phase 2 (22 June–20 July) was a post-onset phase char- acterized by a signiﬁcant increase in low-level cloudi- iv. Phase 4 (27–31 July) was a phase of monsoon recovery characterized by undisturbed monsoon conditions. A second very important dynamical feature is the NLLJ, which typically forms over land at the end of the day when turbulence in the atmospheric boundary layer has ceased. Blackadar (1957) associated the nocturnal jet with the ceas- ing of turbulence and predominance of the Coriolis force, which accelerates the wind towards low pressure. However, due to the low latitude and the low Coriolis force in the DAC- CIWA region, frictionless inertial oscillations above the noc- turnal inversion layer might not be applicable. Therefore, the formation of the NLLJ may not be fully explained by this classical theory. Also observed during the AMMA experi- ment in the Sahelian region (Parker et al., 2005; Lothon et al., 2008; Abdou et al., 2010), the NLLJ settles almost ev- ery night in West Africa. Parker et al. (2005) suggested that when turbulence rapidly diminishes, the NLLJ is then able to respond to the pressure-gradient force. y The present study aims to describe the day-to-day vari- ability and mean characteristics of the monsoon ﬂow, the NLLJ, the MI, and the LLSC at Savè, Benin (Fig. 1), during the 20 June–30 July 2016 period, based on the DACCIWA dataset. The focus on the Savè supersite for this study is mo- tivated by the fact that only this site was instrumented in such way that continuous proﬁling of the wind up to several kilo- meters and continuous determination of the cloud summit were accessible. This was not the case at Kumasi or Ile-Ife. This investigation is a necessary step forward in ensuring a better understanding of the LLSC life cycle: this work should facilitate future case studies and the model evaluations. The subset of the instrumentation deployed at the Savè supersite and used in this study is described in Sect. 2. Section 3 pro- vides the method and the criteria used to detect the monsoon ﬂow. It overviews the day-to-day and mean characteristics of the monsoon ﬂow. Section 4 presents the results for the NLLJ and MI, and Sect. 5 for the LLSC. Section 6 suggests a link between these three phenomena by presenting the mean di- urnal cycle. A discussion and conclusions appear in Sect. 7. p p g Studies by Schrage and Fink (2012) and Schuster et al. (2013) have suggested that the NLLJ may play an important role in LLSC formation because of the cold air advection and turbulent mixing that it generates. iv. Phase 4 (27–31 July) was a phase of monsoon recovery characterized by undisturbed monsoon conditions. Climate models tend to un- derestimate the strength of the NLLJ (Knippertz et al., 2011; Nam et al., 2012; Hannak et al., 2017) and consequently the advection and turbulent mixing that are associated with its occurrence. The NLLJ contributes differentially to horizon- tal advection as a function of latitude: transport of moisture occurs toward the north at high latitudes (> 10◦N) (Parker et al., 2005; Lothon et al., 2008). From the point of view of the DACCIWA region, transport by the NLLJ may be different. Due to the moisture gradient in SWA, which is character- ized by less moisture over the Gulf of Guinea than inland, the NLLJ may actually transport drier air northwards, as re- vealed by Adler et al. (2019) and Babi´c et al. (2019a). www.atmos-chem-phys.net/19/8979/2019/ www.atmos-chem-phys.net/19/8979/2019/ Atmos. Chem. Phys., 19, 8979–8997, 2019 C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa 8981 ness and unusually dry conditions above. For example, around 12–14 July 2016, Knippertz et al. (2017) high- lighted a rare situation of a cyclonic–anticyclonic vortex couplet that crossed southern West Africa (SWA). Un- usually dry air was observed in SWA at that time, trans- ported with the anticyclonic vortex which had its center in the Southern Hemisphere. pared with the northern more convective boundary layer, reaches several tens of kilometers inland during daytime con- vective conditions. After 15:00 UTC, with the weakening of convection and the associated turbulence, the wind increases and the front propagates further inland. Such a phenomenon was ﬁrst studied along the Mauritanian coast and termed “At- lantic inﬂow” by Grams et al. (2010). Because this late after- noon propagation of the sea-breeze front occurs in a quite different context than that described by Grams et al. (2010), it is called Gulf of Guinea maritime inﬂow (here after MI) in the DACCIWA experiment (Adler et al., 2019) and is one of the processes involved in the LLSC formation during the DACCIWA ﬁeld campaign (Adler et al., 2019; Babi´c et al., 2019a). Depending on the location of the sea-breeze front inland when it starts its late afternoon propagation, between 50 and 150 km must be traveled to reach Savè (Adler et al., 2019). Assuming a mean wind of 6 m s−1, this would mean that the MI should reach Savè between 19:00 and 21:00 UTC. Adler et al. (2019) and Babi´c et al. (2019a) studied the pro- cesses involved in the formation and evolution of LLSCs. They both highlighted the important role played by the MI and NLLJ within the monsoon ﬂow diurnal cycle. iii. Phase 3 (21–26 July) involved a westerly wind regime associated with wet conditions. During this phase, an- other vortex couplet, located a little further north than the previous one, enhanced westerly moisture trans- ported into SWA. iv. Phase 4 (27–31 July) was a phase of monsoon recovery characterized by undisturbed monsoon conditions. iv. Phase 4 (27–31 July) was a phase of monsoon recovery characterized by undisturbed monsoon conditions. 2.4 Cloud monitoring Three co-located devices deployed during the DACCIWA ﬁeld experiment are used to monitor low clouds: a ceilome- ter, cloud radar, and an infrared (IR) camera. The radar ran continuously from 19 June to 30 July 2016 at Savè during the DACCIWA ﬁeld experiment and provided vertical proﬁles of the wind every 2 min. We block averaged the data at 15 min time intervals for our wind analysis for consistency with other instruments. For our study, we con- sidered the acquisition mode that corresponded to the highest radial resolution of 75 m, which enabled good documentation of the low troposphere from 150 m up to 3 km in height. The UHF proﬁler data are used here to characterize the monsoon ﬂow, the NLLJ, and the MI. A CHM15k ceilometer, which is a 1064 nm wavelength li- dar with a 5–7 kHz pulse rate, was installed by the KIT for the continuous monitoring of cloud base height (CBH). The ceilometer was operated from 3 June to 30 July 2016 at a time resolution of 1 min and a vertical resolution of 15 m. Man- ufacturer software automatically provided three estimates of CBH, which allowed us to detect multiple cloud layers. Here, we only use the lowest CBH, which corresponds to the low clouds under focus. We deﬁne “low-level clouds” as clouds with a base height below 1500 m a.g.l. Adler et al. (2019) used a lower altitude threshold, 600 m a.g.l., which is well adapted to the nocturnal stratus clouds. However, a 1500 m height limit allows us to extend our detection of LLSCs dur- ing daytime when the stratus cloud base height rises due to the growing convective boundary layer. 2.1 Wind proﬁling of the lower troposphere An ultra-high-frequency (UHF) wind proﬁler operated by UPS was devoted to the study of the vertical structure of the atmospheric dynamics in the lower and middle troposphere. This 1274 MHz Doppler radar works with ﬁve beams. The three components of the wind are retrieved from these beams. This instrument was previously used for various studies of the planetary boundary layer: turbulence retrieval (Jacoby- Koaly al., 2002), African easterly jet analysis (Kalapureddy et al., 2010), the WAM diurnal cycle (Lothon et al., 2008), the NLLJ (Madougou et al., 2012), and offshore winds in high precipitation events (Said et al., 2016). This station operated continuously from 13 June to 30 July 2016. In this study, we use the sensible heat ﬂux es- timates to characterize the surface layer stability. 2.3 Observation of surface layer conditions et al. (2018) give a comprehensive overview of the observa- tions made at the three ground-based supersites of the DAC- CIWA projet, including Savè. A 7.77 m scaffold was mounted at Savè by UPS to study en- ergy balance and biogenic ﬂuxes. High-frequency measure- ments of air temperature, speciﬁc humidity, and three compo- nents of the wind were obtained (at 0.1 s time interval). The energy ﬂux (latent heat, sensible heat, momentum ﬂux) were calculated for the samples at a time resolution of 30 min. Ad- ditionally, upward and downward components of shortwave and longwave radiation, air pressure, soil temperature, and soil moisture were measured every minute. Below, we describe the instruments used in our statistical analysis, according to the object of study. 2 Experimental data During the DACCIWA ﬁeld campaign, several remote sens- ing and in situ instruments were jointly deployed at Savè by the Karlsruhe Institute of Technology (KIT) and the Toulouse Paul Sabatier University (UPS). These instruments are all described in detail in Bessardon et al. (2019). The col- lected dataset has been presented in four published datasets: Handwerker et al. (2016), Kohler et al. (2016), Wieser et al. (2016) (for the KIT instrumentation), and Derrien et al. (2016) (for the UPS instrumentation). Furthermore, Kalthoff Another aspect of the DACCIWA region is the sea-breeze circulation that may be superimposed on monsoon ﬂow around coastal regions. This circulation could have an im- pact on the transport of maritime air inland. Adler et al. (2017) and Deetz et al. (2018) used COSMO (Consortium for Small-scale MOdeling) simulations to show that the sea- breeze front, south of which the air is relatively cold com- www.atmos-chem-phys.net/19/8979/2019/ Atmos. Chem. Phys., 19, 8979–8997, 2019 C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa 8982 Atmos. Chem. Phys., 19, 8979–8997, 2019 C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa 8983 Finally, UPS installed a MOBOTIX S15 cloud camera that monitored the cloud cover all day and obtained visible and IR pictures every 2 min. The visible image was a full sky image: the aperture angles for the IR channel were 43◦×32◦(which corresponds to 158 m × 114 m in area at a height of 200 m). In this study, we used 5 MP IR images, coded in red, green, and blue (RGB) components over 256 colors. Here, [R,G,B] denotes the relative contributions of red, green, and blue of a given pixel, all deﬁned between 0 and 1 (with accuracy of 1/256). The color of a pixel depends on the emissivity of the corresponding sky area and consequently its brightness tem- perature (uncalibrated). Typically, a low cloud base is seen as red and a clear sky is seen as blue. Therefore, a homogeneous low cloud deck will create a homogeneous red color image; a fragmented stratocumulus will render an image with col- ors ranging from red to blue. This instrument is used here to study the horizontal homogeneity of the cloud deck and to deﬁne the onset and breakup times of the stratus deck, with a newly designed method. As far as we know, it is the ﬁrst time that such methodology is used for the study of stratus cloud deck formation and breaking. discussed in the previous section, the strong variability of the monsoon ﬂow depth observed on some days, like 27– 28 June, is associated with a large wind-sheared layer and a low wind. However, the intra-seasonal variability of the mon- soon depth (from a few hundred meters to 4 km) and strength (from almost zero wind to 10 m s−1) can often be linked to synoptic conditions (Couvreux et al., 2010). Several days in a row, such as 20–23 June and 10–12 July, exhibited consis- tent monsoon ﬂow characteristics over time with a particu- larly large monsoon ﬂow depth and strong wind speed. The period 10–12 July is included in the vortex phase (9–16 July). During this period, an unusual cyclonic circulation devel- oped and slowly propagated from eastern Mali to Cabo Verde along with an anticyclonic vortex in the west–northwesterly direction along the Guinean coast (Knippertz et al., 2017). 3.1 Monsoon ﬂow detection The monsoon ﬂow can be detected according to the wind di- rection (Kalapureddy et al., 2010). In this study, it is deﬁned as the lower layer with a 135◦(SE)–270◦(W) horizontal wind sector. The choice to use a quite large wind direction sector, including some westerly winds, is motivated by weak monsoon ﬂow (below 1 m s−1) observed during the daytime for which the direction is ill deﬁned and can be inﬂuenced by local effects. This criterion on wind direction allows to exclude atmospheric conditions associated with vortex circu- lations, deep convection and Harmattan ﬂow. The top of the monsoon ﬂow is the level above which the horizontal wind direction is out of the 135–270◦sector for more than 225 m (i.e., third UHF wind proﬁler gate). Very often, a large wind- sheared layer between the monsoon ﬂow and the easterlies (African easterly jet) above is observed, which makes it dif- ﬁcult to determine the monsoon ﬂow depth. The strength of the monsoon ﬂow is deﬁned as the mean wind speed within this depth. A composite 24 h evolution of the monsoon ﬂow depth, strength, and direction from 15:00 UTC on day D-1 to 15:00 UTC the day after during the 20 June to 30 July 2016 period enables a discussion of the diurnal evolution of the monsoon ﬂow characteristics (Fig. 3). The median of the monsoon depth shows a weak diurnal evolution from a min- imum value of 1200 m a.g.l. during the night to 2000 m a.g.l. during convective conditions (Fig. 3a), with a day-to-day variability. This ﬁnding is consistent with what was observed during the AMMA experiment during the full monsoon sea- son (Kalapureddy et al., 2010). Unlike the monsoon ﬂow depth, the strength and direction of the monsoon ﬂow indi- cate a clear diurnal cycle (Fig. 3b and c). The median strength of the monsoon ﬂow is roughly 3.5 m s−1 between noon and 17:00 UTC with a 210◦direction. The median strength regu- C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa The wet westerly regime (21–27 July) was characterized by a particularly low monsoon ﬂow strength associated with im- portant oscillations of the monsoon depth due to rain events that shifted back to the coast (Knippertz et al., 2017). Many convective activities occurred at the Savè site during the campaign, which is typical in SWA (Laing and Fritsch, 1993). These phenomena were either detected by rain mea- surements (Fig. 2) or density current effects at the surface. Table 1 lists the days for which rain and/or density currents are detected after 15:00 UTC; such events are likely to have disturbed or interacted with the formation of LLSCs. The im- pact of density currents at the surface, in terms of tempera- ture, relative humidity, speciﬁc humidity, wind speed, and di- rection depends on the intensity of the convective cell and its location. A sudden decrease in temperature and an increase in wind speed are typically observed (Table 2). A shift in wind direction is not a reliable criterion because it depends on the convective cell location with regard to the Savè supersite. As listed in Table 2, density currents are largely associated with drier air, as already discussed by Schwendike et al. (2010). All of the outﬂow cases listed in Table 1 are associated with a convective cell identiﬁed with the rain radar in the supersite surroundings. All of the rainy and density current cases are excluded (16 out of 41 d) from the statistics from which we estimate the primary characteristics of the monsoon ﬂow. 2.2 Lower troposphere temperature proﬁling A scanning microwave radiometer (humidity and tempera- ture proﬁler HATPRO-G4 manufactured by RPG – Radiome- ter Physics GmbH) from the KIT was used to analyze the temperature proﬁles. The radiometer measured brightness temperature from which integrated water vapor, liquid wa- ter path, temperature proﬁles, and humidity proﬁles could be retrieved using the retrieval algorithm provided by the Uni- versity of Cologne (Löhnert and Crewell, 2003; Löhnert et al., 2009). Every 15 min, temperature proﬁles with enhanced accuracy were obtained using low-elevation scans (Crewell and Löhnert, 2007), which are used here to detect the MI with the 302 K isotherm (Deetz et al., 2018) and the fuzzy logic algorithm described below. A systematic comparison of the radiosounding temperature proﬁles with the HATPRO tem- perature proﬁles (not shown) revealed a systematic cold bias of 0.2 K below 550 m, 0.5 K in the 550–1000 m layer, and 2 K in the 1000–2000 m layer. This ﬁnding is consistent with the accuracies noted by Crewell and Löhnert (2007) (< 1 K below 1000 m). Only microwave radiometer measurements below 550 m are used in this study. The available dataset of this instrument covers the 30 June to 30 July time period. More information on cloud characteristics was provided by the KIT 35.5 GHz cloud radar (i.e., cloud top height (CTH) and cloud microphysics (rain, drizzle)). The cloud radar operated continuously from 14 June to 30 July 2016. It was run with vertical pointing every 5 min and horizon- tal scans every 30 min. Here, we use the observations of the vertical proﬁles for the CTH evaluation. After the despiking process, we averaged the reﬂectivity proﬁles of hydromete- ors over 5 min and applied a threshold of −35 dBz to capture the CTH (Adler et al., 2019). Values below this threshold are considered to be related to clear air above the cloud. Addi- tional details about the CTH retrieval technique can be found in Babi´c et al. (2019a). This algorithm enables a good esti- mation of the CTH, particularly when the clouds are uniform, which is true in the case of stratus clouds. However, CTHs are difﬁcult to capture for scattered clouds or rain (e.g., cu- mulus clouds during the daytime). This fact explains some missing CTH estimates during the daytime in our later anal- ysis. Atmos. Chem. Phys., 19, 8979–8997, 2019 www.atmos-chem-phys.net/19/8979/2019/ 3.2 Monsoon ﬂow characteristics Applying the previous criteria to the UHF wind proﬁler dataset, the temporal evolutions of the monsoon ﬂow depth and strength from 15:00 UTC on day D-1 to 15:00 UTC on day D are calculated for day D from 20 June to 30 July 2016. The results are presented in Fig. 2. Except for 13, 16 and 17 July, dates for which the UHF wind proﬁler data are miss- ing, the short periods shown in white in Fig. 2, are for a wind direction not falling within the 135–270◦sector. As www.atmos-chem-phys.net/19/8979/2019/ Atmos. Chem. Phys., 19, 8979–8997, 2019 C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa 8984 Figure 2. Diurnal evolution of the monsoon ﬂow (a) thickness and (b) strength at Savè from 15:00 UTC on day D-1 to 15:00 UTC on day D during the course of the 20 June through 30 July 2016 time period. Solid black vertical lines delimit the different phases of the monsoon described by Knippertz et al. (2017). Dashed vertical lines indicate the vortex phase included in the post-onset phase. White windows indicate periods for which the monsoon ﬂow was not deﬁned (i.e., a wind direction beyond the 135–270◦sector). Grey windows indicate missing data. Red squares indicate rainy conditions. Figure 2. Diurnal evolution of the monsoon ﬂow (a) thickness and (b) strength at Savè from 15:00 UTC on day D-1 to 15:00 UTC on day D during the course of the 20 June through 30 July 2016 time period. Solid black vertical lines delimit the different phases of the monsoon described by Knippertz et al. (2017). Dashed vertical lines indicate the vortex phase included in the post-onset phase. White windows indicate periods for which the monsoon ﬂow was not deﬁned (i.e., a wind direction beyond the 135–270◦sector). Grey windows indicate missing data. Red squares indicate rainy conditions. larly increases between 17:00 and 01:00 UTC up to 5.5 m s−1 with a simultaneous slight shift in the median wind direction (amplitude and standard deviation around 32.62 and 9.46◦, respectively). These same changes are observed in wind sur- face measurements (Kalthoff et al., 2018). below 3 km a.g.l. Stull (1988) and Andreas et al. (2000) de- ﬁned the NLLJ within the ﬁrst 1500 m as having a maximum wind speed at least 2 m s−1 faster than the minimum wind above it. Baas et al. 3.2 Monsoon ﬂow characteristics (2009) deﬁned NLLJs over the Nether- lands as having a maximum wind speed at least 2 m s−1 be- low 500 m a.g.l. and a wind speed 25 % faster than the local minimum wind speed above. 4 Maritime inﬂow and nocturnal low-level jet analysis 4 1 D t ti f th iti i ﬂ d t l The detection of the NLLJ is based, in this study, on the use of dynamical and surface stability criteria: (i) the wind direction in the lowest atmosphere below 1500 m is between the southeast and west–northwest with (ii) a maximum wind speed of at least 5 m s−1 and at least 2 m s−1 larger than the minimum above and (iii) a surface sensible heat ﬂux lower than 10 W m−2. This last criterion ensures stable to neutral conditions at the surface. The onset of the NLLJ is deﬁned when these criteria are satisﬁed for at least 2 h and the height of the maximum wind speed is below 500 m. The breakup time is deﬁned when one of the three criteria mentioned above has not been satisﬁed for at least 1 h. The use of the surface sensible heat ﬂux as a diagnostic of the stability may be a limitation to this method because this measurement is very local and may not represent atmospheric stability on large spatial scales. The NLLJ arrival and breakup times are 4.1 Detection of the maritime inﬂow and nocturnal low-level jet A low-level jet is characterized by a maximum wind speed a few hundred meters above the surface of the Earth and a clear minimum wind speed above. This situation implies sig- niﬁcant shear below and above the jet core. Because of the occurrence of the low-level jets in various environments and forcing, different criteria can be applied to deﬁne and de- tect them based on vertical wind proﬁles. One of the ﬁrst proposed deﬁnitions of the NLLJ based on observations was given by Bonner (1968). This author deﬁned three types of NLLJs based on three different threshold values for the maxi- mum wind speed (12, 16, and 20 m s−1, respectively), and the existence of a minimum wind speed above the maximum and www.atmos-chem-phys.net/19/8979/2019/ Atmos. Chem. Phys., 19, 8979–8997, 2019 C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa 8985 Table 1. Characteristics (onset and breakup time) of the NLLJ and LLSCs and the arrival time of the MI at Savé during the 20 June– 30 July 2016 period. All times are indicated in UTC. For times related to the LLSCs, blanks indicate that LLSCs were not detectable (because of rain or clear sky). For times related to MI or NLLJ, blanks indicate missing data or undetectable features. The FLFmean was used to estimate the MI onset. DC indicates density current. Table 1. Characteristics (onset and breakup time) of the NLLJ and LLSCs and the arrival time of the MI at Savé during the 20 June– 30 July 2016 period. All times are indicated in UTC. For times related to the LLSCs, blanks indicate that LLSCs were not detectable (because of rain or clear sky). For times related to MI or NLLJ, blanks indicate missing data or undetectable features. The FLFmean was used to estimate the MI onset. DC indicates density current. Table 1. Characteristics (onset and breakup time) of the NLLJ and LLSCs and the arrival time of the MI at Savé during the 20 June– 30 July 2016 period. All times are indicated in UTC. For times related to the LLSCs, blanks indicate that LLSCs were not detectable (because of rain or clear sky). For times related to MI or NLLJ, blanks indicate missing data or undetectable features. The FLFmean was used to estimate the MI onset. DC indicates density current. 4.1 Detection of the maritime inﬂow and nocturnal low-level jet June–July Rain onset Total rain DC time MI arrival time NLLJ onset LLSC onset NLLJ breakup time LLSC breakup time nights (UTC) (mm) (UTC) (UTC) (UTC) (UTC) (UTC) (UTC) 19–20 06:40 0.1 19:00 00:46 08:00 09:12 20–21 16:31 19:30 01:02 08:00 11:34 21–22 19:30 02:42 07:30 09:16 22–23 17:30 07:00 23–24 17:00 07:00 24–25 17:00 05:12 07:30 07:36 25–26 18:00 05:24 07:30 12:06 26–27 21:00 01:26 07:30 11:52 27–28 20:20 2.8 18:30 07:30 28–29 00:30 0.1 19:00 03:00 07:30 13:26 29–30 17:30 00:38 04:30 13:46 30–1 17:30 18:30 22:40 03:30 11:42 1–2 00:30 18:30 07:00 2–3 20:11 20:00 21:00 01:46 02:30 14:58 3–4 21:00 22:30 04:52 08:00 11:22 4–5 18:00 17:30 02:52 07:30 12:44 5–6 16:05 18:30 02:08 08:30 13:36 6–7 17:07 19:30 18:30 00:16 08:00 12:28 7–8 20:00 21:00 00:10 07:30 10:36 8–9 16:20 2.6 16:15 23:00 21:00 00:28 08:30 13:44 9–10 18:00 19:30 21:56 07:00 12:18 10–11 16:00 19:30 00:42 07:00 10:54 11–12 16:30 17:30 06:30 12–13 17:30 12.6 16:47 13–14 04:50 0.1 05:00 00:00 07:00 14–15 17:00 17:00 06:30 15–16 17:30 07:00 16–17 23:30 10:44 17–18 17:00 18:30 00:38 06:30 08:26 18–19 21:00 21:00 03:42 07:00 12:08 19–20 00:10 15.9 17:45 19:30 08:30 20–21 15:00 1.6 23:30 17:30 04:48 07:30 10:26 21–22 19:00 00:00 02:58 08:00 15:00 22–23 20:10 17.1 21:00 03:00 23–24 18:20 45.9 18:50 01:00 06:00 24–25 19:50 1.8 16:38 17:30 18:00 07:00 25–26 00:00 1.4 18:00 18:30 08:00 26–27 19:00 18:30 02:12 07:30 09:42 27–28 18:00 18:00 00:44 07:00 07:50 28–29 19:00 18:00 23:46 07:30 08:02 29–30 18:00 19:30 05:46 07:00 09:36 determined with a 30 min temporal resolution, which corre- sponds to the sample duration for the sensible heat ﬂux esti- mation. and mechanical turbulence vanish at the end of the after- noon, two phenomena occur simultaneously: the monsoon ﬂow increases and the MI, characterized by a higher wind speed than the monsoon ﬂow further north, can propagate further inland, possibly up to Savè. Deetz et al. (2018), using COSMO model simulations for the night of 2–3 July, char- acterized the MI arrival by the 302 K potential temperature at a height of 250 m. This criterion was applied to the temper- ature measured locally by the microwave radiometer at the Savè site in order to detect the arrival of the MI. 4.1 Detection of the maritime inﬂow and nocturnal low-level jet Since the The DACCIWA project focused on a region to the south of the AMMA study area that is affected by coastal phe- nomena, such as sea breeze. Unfortunately, no measurements provided evidence for MI formation and propagation inland. But based on simulations conducted by Adler et al. (2017) and Deetz et al. (2018), Adler et al. (2019) hypothesized that the MI penetrated 50–130 km inland. When convection C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa Six other days presenting a similar strong mon- soon ﬂow are observed during the period between 20 June and 30 July. These days correspond to synoptic anticyclonic atmospheric conditions over the Gulf of Guinea that favor a strong penetration of the monsoon ﬂow inland (Knippertz et al., 2017). In such conditions, the NLLJ onset is not easy to determine. Figure 4c shows that the maximum wind is detected above 500 m before 19:30 UTC, which prevents us from considering this strong monsoon ﬂow as a NLLJ un- til that time (criterion (i) of the NLLJ detection). The MI reaches Savè at 18:00 UTC according to the fuzzy logic method, 2 h earlier than the two previous examples. This re- sult appears in accordance with a strong monsoon ﬂow in- hibiting the turbulent mixing and pushing the MI front faster and farther inland. On the other days, the cooling lasts until midnight. The detection of MI arrival time according to the 302 K criterion yields a later MI arrival time (19:30 UTC). FLFx(rx) =    y1, rx ≤rx1, y1 + y2 −y1 rx2 −rx1 (rx −rx1), rx1 < rx < rx2, y2, rx ≥rx2, (1) FLFx(rx) =    y1, rx ≤rx1, y1 + y2 −y1 rx2 −rx1 (rx −rx1), rx1 < rx < rx2, y2, rx ≥rx2, (1) where rx is the rate of change of the variable x, rx1 (respec- tively, rx2) is a constant value below (above) which FLFx is equal to y1 (y2). rT is multiplied by −1 to obtain posi- tive changes for decreasing temperature. As in Coceal et al. (2018), y1 and y2 are set to 0 and 1, respectively, and rx1 is set to 0 (i.e., no increase in wind speed or no decrease in temperature). Instead of using the maximum value of rx di- vided by two for rx2 (Coceal et al., 2018), for each day, we use the value corresponding to the 99th percentile of rx di- vided by two to avoid outliers. Considering the values used in this study for rx1, y1, and y2 in Eq. (1) can be simpliﬁed as follows: FLFx(rx) =    0, rx ≤0, rx rx2 , 0 < rx < rx2, 1, rx ≥rx2. C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa (2) (2) In this study, the mean fuzzy logic function (FLFmean) is computed using equal weights for FLFws and FLFT , and the same threshold of 1 is used to detect combined changes in the dynamic and thermodynamic conditions. The fuzzy logic method only makes sense if the temperature and wind speed changes are combined. However, to better understand how temperature and wind speed changes distinctly impact FLFmean, FLFws and FLFT are also discussed. FLFws would be a full weight applied to wind speed (i.e., considering that MI is only characterized by a dynamical change), and FLFT would be a full weight applied to temperature in the fuzzy logic method (i.e., considering that MI arrival implies only a thermodynamical change). The MI arrival time can then be determined by noting the ﬁrst time at which the fuzzy logic functions of wind speed and temperature are equal to 1. There are, in this method, important constants and thresh- olds to ﬁx, and several sensitivity tests have been performed. The results for individual days can differ, but the average and the conclusions remain the same. The use of the MI and NLLJ criteria is illustrated for three very different days in Fig. 4. Each panel presents the 24 h time–height section of wind as measured by the UHF wind proﬁler and the three fuzzy logic functions (FLFws, FLFT , and FLFmean). The NLLJ onset times and the two estimates of the MI arrival time (302 K potential temperature and the fuzzy logic method) at Savè are also indicated. Based on these three examples, one can note the large vari- ability that can be observed from one day to the other, which makes it challenging to deﬁne solid common criteria for MI and NLLJ detection. www.atmos-chem-phys.net/19/8979/2019/ Atmos. Chem. Phys., 19, 8979–8997, 2019 www.atmos-chem-phys.net/19/8979/2019/ C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa 8986 poral evolution of the distribution of the monsoon ﬂow (a) depth, (b) strength and (c) direction at Savè for the period between July. The black lines in panels (a), (b) and (c) indicate the median monsoon depth, strength, and direction, respectively. Figure 3. Temporal evolution of the distribution of the monsoon ﬂow (a) depth, (b) strength and (c) direction at Savè for the period between 20 June and 30 July. The black lines in panels (a), (b) and (c) indicate the median monsoon depth, strength, and direction, respectively. changes of temperature (T ) and horizontal wind speed (ws) (termed rT and rws, respectively) are calculated using 30 min averaged T and ws in the 200–550 and 150–525 m layers, respectively; (2) fuzzy logic functions for T and ws, termed FLFT and FLFws, respectively, are computed using Eq. (1); and (3) the mean fuzzy logic function, FLFmean, combining both the changes in T and ws, is the mean of FLFws and FLFT . 302 K criterion relies on one simulated case and because no clear justiﬁcation for this 302 K value exists, another crite- rion based on MI characteristics observed at the surface is proposed in this study. MI arrival at Savè should be detected at the surface by a combination of both an increase in hor- izontal wind and a decrease in temperature. Thus, a second method was used in this study based on the combination of both an increase in horizontal wind speed and a decrease in temperature. These changes should be the signature of the MI arrival time. This method, based on the fuzzy logic method, was used by Coceal et al. (2018) to detect the sea-breeze front around London, in the south of England. Our method for MI detection at Savè follows three steps: (1) the rate of www.atmos-chem-phys.net/19/8979/2019/ Atmos. Chem. Phys., 19, 8979–8997, 2019 C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa 8987 C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa The fuzzy logic function FLFx for the variable x can be written as clear MI arrival time. FLFmean values are larger than 0.5 un- til midnight, which demonstrates that the cooling lasts for at least 4.5 h. For that day, the 302 K potential temperature cri- terion is in very good agreement with the fuzzy logic method with a detection of MI arrival time at 20:15 UTC. The max- imum value of the wind vertical proﬁle reaches the thresh- old of 5 m s−1 at 21:00 UTC, which is the onset time of the NLLJ. The second example (Fig. 4b) is for the night of 7– 8 July, which is analyzed in detail by Babi´c et al. (2019a). The fuzzy logic functions are very similar to those of 2– 3 July. The estimated arrival times of the MI are 20:00 UTC and 20:45 UTC using the fuzzy logic method and the 302 K criterion, respectively. Like the night of 2–3 July, the max- imum value of the jet reaches the threshold of 5 m s−1 at 21:00 UTC. The particularity of that day is that the NLLJ vanishes near the surface after sunrise on 8 July. However, the jet shape of the wind persists above unstable condi- tions; the height of the wind speed maximum increases up to 1000 m at 12:00 UTC which leads a strong rotation of the wind between the southwesterly in the monsoon ﬂow and easterly above. This example illustrates the necessity of the atmospheric stability criterion for NLLJ breakup time deter- mination. The third and last example, the night of 9–10 July (Fig. 4c), illustrates a NLLJ that was hardly detected because it develops within a strong monsoon ﬂow observed all day long and weaker easterly wind above compared to 2–3 and 7–8 July. Six other days presenting a similar strong mon- soon ﬂow are observed during the period between 20 June and 30 July. These days correspond to synoptic anticyclonic atmospheric conditions over the Gulf of Guinea that favor a strong penetration of the monsoon ﬂow inland (Knippertz et al., 2017). In such conditions, the NLLJ onset is not easy to determine. Figure 4c shows that the maximum wind is detected above 500 m before 19:30 UTC, which prevents us from considering this strong monsoon ﬂow as a NLLJ un- til that time (criterion (i) of the NLLJ detection). C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa The MI reaches Savè at 18:00 UTC according to the fuzzy logic method, 2 h earlier than the two previous examples. This re- sult appears in accordance with a strong monsoon ﬂow in- hibiting the turbulent mixing and pushing the MI front faster and farther inland. On the other days, the cooling lasts until midnight. The detection of MI arrival time according to the 302 K criterion yields a later MI arrival time (19:30 UTC). Based on these three examples, one can note the large vari- ability that can be observed from one day to the other, which clear MI arrival time. FLFmean values are larger than 0.5 un- til midnight, which demonstrates that the cooling lasts for at least 4.5 h. For that day, the 302 K potential temperature cri- terion is in very good agreement with the fuzzy logic method with a detection of MI arrival time at 20:15 UTC. The max- imum value of the wind vertical proﬁle reaches the thresh- old of 5 m s−1 at 21:00 UTC, which is the onset time of the NLLJ. The second example (Fig. 4b) is for the night of 7– 8 July, which is analyzed in detail by Babi´c et al. (2019a). The fuzzy logic functions are very similar to those of 2– 3 July. The estimated arrival times of the MI are 20:00 UTC and 20:45 UTC using the fuzzy logic method and the 302 K criterion, respectively. Like the night of 2–3 July, the max- imum value of the jet reaches the threshold of 5 m s−1 at 21:00 UTC. The particularity of that day is that the NLLJ vanishes near the surface after sunrise on 8 July. However, the jet shape of the wind persists above unstable condi- tions; the height of the wind speed maximum increases up to 1000 m at 12:00 UTC which leads a strong rotation of the wind between the southwesterly in the monsoon ﬂow and easterly above. This example illustrates the necessity of the atmospheric stability criterion for NLLJ breakup time deter- mination. The third and last example, the night of 9–10 July (Fig. 4c), illustrates a NLLJ that was hardly detected because it develops within a strong monsoon ﬂow observed all day long and weaker easterly wind above compared to 2–3 and 7–8 July. 4.2 MI and NLLJ characteristics A brief and weak NLLJ settled in the night of 2–3 July between 21:00 and 03:00 UTC in a very weak monsoon ﬂow (Fig. 4a) and with a strong easterly wind above 2 km. The ﬁrst simultaneous wind increase and cooling are observed at 20:00 UTC (FLFws, FLFT both reach 1), which indicates a Characteristics of the MI and NLLJ are deduced from a set of days excluding rainy days and density current cases. For the NLLJ, we use the same days that we used for monsoon ﬂow characteristics. For the MI, the 30 June through 30 July www.atmos-chem-phys.net/19/8979/2019/ Atmos. Chem. Phys., 19, 8979–8997, 2019 al.: Low-level stratiform cloud and dynamical features observed over southern West Africa 8988 88 C. Dione et al.: Low level stratiform cloud and dynamical features observed over southern West A gure 4. Time–height sections of (color) wind speed and (arrows) direction from the UHF wind proﬁler on the nights of (a) 2–3 July y me–height sections of (color) wind speed and (arrows) direction from the UHF wind proﬁler on the nights of (a) 2–3 July, (b) 7 ) 9–10 July 2016. A leftward horizontal arrow indicates an easterly wind; an arrow from bottom to top indicates a souther ack open circles indicate the jet core height detected with a maximum wind speed of at least 5 m s−1, the magenta rectangle eight of the minimum wind speed above the jet core, and red open circles indicate the monsoon ﬂow depth. The black, blu in the lower box indicate the three fuzzy logic functions of the wind speed, temperature, and their mean, respectively. The re hed vertical lines indicate the MI arrival time estimated using two different criteria: (1) the FLFmean > 1 criterion and (2) th pe criterion, respectively. The black vertical line indicates the NLLJ onset. Figure 4. Time–height sections of (color) wind speed and (arrows) direction from the UHF wind proﬁler on the nights of (a) 2–3 July, (b) 7– 8 July, and (c) 9–10 July 2016. A leftward horizontal arrow indicates an easterly wind; an arrow from bottom to top indicates a southerly wind. The black open circles indicate the jet core height detected with a maximum wind speed of at least 5 m s−1, the magenta rectangles indicate the height of the minimum wind speed above the jet core, and red open circles indicate the monsoon ﬂow depth. C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa In addition, the continuous cooling can also be explained by the turbulent mixing under the NLLJ core, which mixes the upper layer with lower layers cooled by radiative and sensible ﬂux divergence (Adler et al., 2019; Babi´c et al., 2019a). – As expected, cooling may occur between 18:00 and 00:30 UTC the following day. Contrary to the wind speed, whose fuzzy logic function reaches 1 but rarely remains at that value for several hours during the night, the temperature fuzzy logic function reaches this value many times during the night. This trend implies con- tinuous cooling (Fig. 4). This result is in accordance with the continuous decrease in temperature within the MI from north to the south discussed by Adler et al. (2019). A large abrupt change in temperature at the front passage with constant temperature behind would have implied a very different temporal occurrence of FLFT = 1. In addition, the continuous cooling can also be explained by the turbulent mixing under the NLLJ core, which mixes the upper layer with lower layers cooled by radiative and sensible ﬂux divergence (Adler et al., 2019; Babi´c et al., 2019a). Figure 5. Temporal occurrence of fuzzy logic functions equal to 1 for temperature (FLFT ), wind speed (FLFws), and mean fuzzy logic function (FLFmean) for the period between 1 and 30 July. for the fuzzy logic method yield different MI detection times for each day but quite similar distributions for the period of study. These results suggest that the MI arrival time is dif- ﬁcult to detect with local measurements. However, the MI arrival time as detected by the fuzzy logic function FLFmean is clearly linked to the mean monsoon ﬂow in the afternoon (Fig. 7): the stronger the monsoon ﬂow strength in the af- ternoon between 12:00 and 15:00 UTC, the earlier the MI arrival time. The absolute value of the correlation coefﬁcient between both is 0.61. The two exceptionally early arrivals at 16:00 and 16:30 UTC shown in Fig. 7 and put in Table 1 are associated with unusually strong monsoon ﬂow all day long (e.g., the nights of 10–11 and 11–12 July). – The FLFmean with values equal to 1 combines the wind increase and the cooling, both of which occur simulta- neously during the entire night; the largest number of cases is observed between 18:00 and 21:00 UTC. The MI arrival times are shown in Fig. 6a. 4.2 MI and NLLJ characteristics The black, blue, and red lines in the lower box indicate the three fuzzy logic functions of the wind speed, temperature, and their mean, respectively. The red and black dashed vertical lines indicate the MI arrival time estimated using two different criteria: (1) the FLFmean > 1 criterion and (2) the 302 K isentrope criterion, respectively. The black vertical line indicates the NLLJ onset. – A large increase in wind speed (FLFws = 1) may be ob- served at any time during the day; however, the largest occurrence (≥5) is between 17:00 and 20:00 UTC. This variability is due to the day-to-day variability of the period is used (imposed by radiometer data availability), and 12 d are excluded (listed in Table 1). The temporal occurrence of fuzzy logic functions equal to 1 for the 30 June–30 July period is shown in Fig. 5. One can note interesting results: www.atmos-chem-phys.net/19/8979/2019/ Atmos. Chem. Phys., 19, 8979–8997, 2019 C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa 8989 Table 2. Variations in surface (7.77 m height) temperature (1T ), relative humidity (1RH), speciﬁc humidity (1q), wind speed, and wind direction outﬂows that can disturb the MI arrival time and NLLJ onset and participate in the cooling of the atmosphere detected after 15:00 UTC. For the wind changes, the wind speed and wind direction averaged within the layer (150–525 m) before and after the event are considered. kg−1) Wind speed shift (m s−1) Wind direction shift (◦) a.g.l. 150–525 m a.g.l. 150–525 m a.g.l. −0.90 8.0–15.3 178–196 0.92 1.6–2.4 243–238 −0.33 3.5–4.5 237–278 1.01 3.1–4.2 176–174 0.18 0.7–12.0 14–179 0.43 4.8–11.4 229–300 −0.37 2.9–7.4 190–318 2.46 1.9–4.8 321–335 Figure 5. Temporal occurrence of fuzzy logic functions equal to 1 for temperature (FLFT ), wind speed (FLFws), and mean fuzzy logic function (FLFmean) for the period between 1 and 30 July. Day Time 1T (◦C) 1RH (%) 1q (g kg−1) Wind speed shift (m s−1) Wind direction shift (◦) (UTC) at 7.7 m a.g.l. at 7.7 m a.g.l. at 7.7 m a.g.l. 150–525 m a.g.l. 150–525 m a.g.l. 20 June 16:31 −3.1 10.2 −0.90 8.0–15.3 178–196 2 July 20:11 −2.8 18 0.92 1.6–2.4 243–238 5 July 16:05 −2.3 6.7 −0.33 3.5–4.5 237–278 6 July 17:07 −1.4 8.9 1.01 3.1–4.2 176–174 8 July 16:15 −5.8 25.6 0.18 0.7–12.0 14–179 12 July 16:47 −7.1 32.0 0.43 4.8–11.4 229–300 23 July 18:50 −2.3 10.2 −0.37 2.9–7.4 190–318 24 July 16:38 −4.2 27.5 2.46 1.9–4.8 321–335 monsoon strength and the arrival time of the NLLJ dur- ing this period. – As expected, cooling may occur between 18:00 and 00:30 UTC the following day. Contrary to the wind speed, whose fuzzy logic function reaches 1 but rarely remains at that value for several hours during the night, the temperature fuzzy logic function reaches this value many times during the night. This trend implies con- tinuous cooling (Fig. 4). This result is in accordance with the continuous decrease in temperature within the MI from north to the south discussed by Adler et al. (2019). A large abrupt change in temperature at the front passage with constant temperature behind would have implied a very different temporal occurrence of FLFT = 1. www.atmos-chem-phys.net/19/8979/2019/ C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa Scatter plot of the average monsoon strength between 12:00 and 15:00 UTC and the MI arrival time using FLFmean = 1, for the time period between 1 and 30 July 2016. characteristics from a vertical point of view: radiosoundings, cloud radar (Bouniol et al., 2012), and ceilometer (Schrage and Fink, 2012). Satellite data have also been used to ana- lyze the horizontal variability LLSCs. Van Der Linden et al. (2015) and Bouniol et al. (2012), for example, made a clima- tology of the occurrence of the LLSCs in this region. Here, we use a ceilometer to estimate the CBH (consider- ing only CBHs below 1500 m a.g.l.), cloud radar to estimate Figure 6. (a) Distribution of MI arrival times determined using the 302 K potential temperature criterion and the fuzzy logic function method (FLFT , FLFws, and FLFmean) for the period between 1 and 30 July 2016. (b) Distribution of NLLJ onset and breakup times at Savè during the period from 20 June to 30 July 2016. The MI arrival times detected with FLFmean, already presented in panel (a), are also indicated. Figure 7. Scatter plot of the average monsoon strength between 12:00 and 15:00 UTC and the MI arrival time using FLFmean = 1, for the time period between 1 and 30 July 2016. Figure 7. Scatter plot of the average monsoon strength between 12:00 and 15:00 UTC and the MI arrival time using FLFmean = 1, for the time period between 1 and 30 July 2016. characteristics from a vertical point of view: radiosoundings, cloud radar (Bouniol et al., 2012), and ceilometer (Schrage and Fink, 2012). Satellite data have also been used to ana- lyze the horizontal variability LLSCs. Van Der Linden et al. (2015) and Bouniol et al. (2012), for example, made a clima- tology of the occurrence of the LLSCs in this region. Figure 6. (a) Distribution of MI arrival times determined using the 302 K potential temperature criterion and the fuzzy logic function method (FLFT , FLFws, and FLFmean) for the period between 1 and 30 July 2016. (b) Distribution of NLLJ onset and breakup times at Savè during the period from 20 June to 30 July 2016. The MI arrival times detected with FLFmean, already presented in panel (a), are also indicated. C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa Here, we use a ceilometer to estimate the CBH (consider- ing only CBHs below 1500 m a.g.l.), cloud radar to estimate the CTH and the IR sky camera to deﬁne the start and end of the LLSCs (Sect. 2.4). With the IR sky camera, we have the advantage of retrieving information about the occurrence of LLSCs, as well as their horizontal structure on a local scale. NLLJ onset and breakup times, in addition to the rainy and density current cases noted before. Figure 8 shows a 15 min distribution of the jet core height, strength, and direction for the 20 June through 30 July pe- riod. The median jet core lowest height (350 m) is observed when the NLLJ settles. The jet core rises after 22:00 UTC to reach about 500 m at NLLJ breakup. The median strength of the jet varies between 7 and 11 m s−1 (Fig. 8b), and the median direction shifts slightly from 200◦at NLLJ onset to 230◦at 03:00 UTC (Fig. 8c). The NLLJ core in Niger and Benin observed during the AMMA campaign (Lothon et al., 2008) was roughly at the same height, but the wind speed of the jet core is in contrast more pronounced in Niger, from 10 to 20 m s−1. To estimate the occurrence of stratus clouds at Savè, we consider the average color [R,G,B] deﬁned by the color vec- tor obtained when averaging all [R,G,B] pixels of the IR camera image, as well as the standard deviation σRGB. σRGB quantiﬁes the color variability within each image, which rep- resents the spatial variability of the lowest CBH within the 43◦× 32◦aperture: σRGB = (σR + σG + σB) 3 , (3) (3) where σR, σG, and σB are the standard deviations of the R, G, and B components, respectively, within the image. A large standard deviation implies a large variability of temperature (i.e., a mix of clear sky and clouds) or spatial variability of the CBHs; a small standard deviation corre- sponds to either a homogeneous stratus deck or a fully clear sky. C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa Four estimates of the MI arrival times are displayed, one using the 302 K po- tential temperature criterion and three corresponding to the ﬁrst time when the three fuzzy logic functions, FLFws, FLFT and FLFmean, attain a value of 1. Most of the observations of the MI arrival time at Savè considering only the wind speed increase fell between 16:00 and 18:00 UTC, while they fell between 16:00 and 21:00 UTC when we considered only the cooling. The arrival time of the MI deduced from FLFmean, which couples with an equal weight cooling and wind speed increase, is observed between 16:00 and 00:00 UTC, with a median arrival time at 19:00 UTC. As noted above, the dif- ferent tests performed to select the constants and thresholds The NLLJ onset and breakup cannot be detected prior to 17:00 UTC and after 08:00 UTC, respectively, because of the stability criterion. The most frequent onset and breakup times of the NLLJ are 17:30 and 07:00 UTC, respectively (Fig. 6b). The MI arrival time detected with FLFmean = 1, is also indi- cated in Fig. 6b. One can note that, with the exception of the two exceptionally early MI arrival times, the NLLJ on- set and MI arrival time distributions are quite similar to one another. Table 1 summarizes, for each day of the 20 June through 30 July period, the MI arrival time using FLFmean, www.atmos-chem-phys.net/19/8979/2019/ Atmos. Chem. Phys., 19, 8979–8997, 2019 C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa 8990 nd dynamical features observed over southern West Africa Figure 7. Scatter plot of the average monsoon strength between 12:00 and 15:00 UTC and the MI arrival time using FLFmean = 1, for the time period between 1 and 30 July 2016. 8990 C. o e et a .: ow eve st at o c oud a d dy a ca eatu es obse ved ove sout e West ca Figure 6. (a) Distribution of MI arrival times determined using the 302 K potential temperature criterion and the fuzzy logic function method (FLFT , FLFws, and FLFmean) for the period between 1 and 30 July 2016. (b) Distribution of NLLJ onset and breakup times at Savè during the period from 20 June to 30 July 2016. The MI arrival times detected with FLFmean, already presented in panel (a), are also indicated. Figure 7. 5.1 LLSC detection West African LLSCs have not been extensively studied in the past, generally due to a lack of observations. However, the systematic visual observations by local meteorologists at ev- ery meteorological station made over a time period of years constitute a unique dataset (Schrage and Fink, 2012). Be- sides this historical approach, several instruments have been used in the context of international projects to estimate LLSC Here, we set the occurrence of the low stratus clouds when (i) R > 0.5, R > G, and R > B, (ii) σRGB <= 0.15, for more than 2 h without any “break” longer than 30 min. Further- more, the ﬁrst time when these criteria are satisﬁed should occur before 06:00 UTC the following morning in order to detect only nocturnal LLSCs. The ﬁrst criterion (i) corre- Atmos. Chem. Phys., 19, 8979–8997, 2019 www.atmos-chem-phys.net/19/8979/2019/ C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa 8991 ure 8. Distribution of the NLLJ (a) core height, (b) strength, and (c) direction at the Savè supersite. The color bar indicates the percentage ays (computed using the number of nights with NLLJ onset during the 20 June through 30 July period). The black lines in panels (a), and (c) indicate the median jet core height, its strength, and its direction, respectively. Figure 8. Distribution of the NLLJ (a) core height, (b) strength, and (c) direction at the Savè supersite. The color bar indicates the percentage of days (computed using the number of nights with NLLJ onset during the 20 June through 30 July period). The black lines in panels (a), (b), and (c) indicate the median jet core height, its strength, and its direction, respectively. sponds to the search of a low base, with an average color close to red. The second criterion (ii) corresponds to the search for a relatively homogeneous cloud deck. The added persistence condition helps with deﬁning a consistent noc- turnal stratus cloud life cycle and avoids spurious variability. These criteria allow one to estimate, for each night, the on- set and breakup times of LLSCs and therefore deduce their lifetimes. Note that during rain events, droplets are retained on the dome of the infrared camera and impact the color of the image as if there was a cloud. C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa 8992 Figure 9. Time series of the LLSC macrophysical characteristics observed on the nights of (a) 7–8 July 2016 and (b) 6–7 July at Savè. The colored band indicates the time series of the average image obtained every 2 min with the IR cloud camera. This time series is plotted arbitrarily at the 0 m abscissa for the sake of clarity. The grey line indicates σRGB (with its scale on the right axis) and the dashed grey line indicates for the threshold (0.15) above which the LLSC deck is not considered to be homogeneous. Black dots indicate the CBHs of the LLSCs estimated from the ceilometer. Blue dots indicate the CTHs of the LLSCs estimated from the cloud radar. The dashed black and red vertical lines indicate from left to right the NLLJ and LLSC onset and breakup times. Figure 9. Time series of the LLSC macrophysical characteristics observed on the nights of (a) 7–8 July 2016 and (b) 6–7 July at Sav The colored band indicates the time series of the average image obtained every 2 min with the IR cloud camera. This time series is plotte arbitrarily at the 0 m abscissa for the sake of clarity. The grey line indicates σRGB (with its scale on the right axis) and the dashed grey lin indicates for the threshold (0.15) above which the LLSC deck is not considered to be homogeneous. Black dots indicate the CBHs of th LLSCs estimated from the ceilometer. Blue dots indicate the CTHs of the LLSCs estimated from the cloud radar. The dashed black and re vertical lines indicate from left to right the NLLJ and LLSC onset and breakup times. Figure 9. Time series of the LLSC macrophysical characteristics observed on the nights of (a) 7–8 July 2016 and (b) 6–7 July at Savè. The colored band indicates the time series of the average image obtained every 2 min with the IR cloud camera. This time series is plotted arbitrarily at the 0 m abscissa for the sake of clarity. The grey line indicates σRGB (with its scale on the right axis) and the dashed grey line indicates for the threshold (0.15) above which the LLSC deck is not considered to be homogeneous. C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa Black dots indicate the CBHs of the LLSCs estimated from the ceilometer. Blue dots indicate the CTHs of the LLSCs estimated from the cloud radar. The dashed black and red vertical lines indicate from left to right the NLLJ and LLSC onset and breakup times. 10:30 UTC as being part of the LLSC lifetime (i.e., they meet our criteria for LLSC detection), which is the start of the so- called “convective phase” according to Babi´c et al. (2019a), who focused on the night of 7–8 July 2016 in their case study, and Lohou et al. (2019), who suggested a conceptual model of LLSCs’ life cycle. responds more to a steady low cloud base as seen by the ceilometer (see, for example, the data from 8 July from 03:00 to 08:00 UTC or the data from 7 July from 06:00 to 09:00 UTC). A larger standard deviation corresponds to more variability of the cloud base (e.g., on 8 July around 11:00 UTC or 7 July around 03:00 UTC or after 12:30 UTC). responds more to a steady low cloud base as seen by the ceilometer (see, for example, the data from 8 July from 03:00 to 08:00 UTC or the data from 7 July from 06:00 to 09:00 UTC). A larger standard deviation corresponds to more variability of the cloud base (e.g., on 8 July around 11:00 UTC or 7 July around 03:00 UTC or after 12:30 UTC). For the night of 7–8 July (Fig. 9a), the criteria for LLSC detection are satisﬁed from midnight until 10:30 UTC. This period deﬁnes the LLSC lifetime for this day (Table 1). Be- fore midnight, the sky is clear for a large part of the time (blue color with the IR camera); some clouds passed be- tween 10:00 and 15:00 m a.g.l. The cloud base heights are variable according to the ceilometer; the IR color and σRGB do not satisfy the LLSC detection criteria. After 10:30 UTC, the cloud base rises and the fractionation of the cloud base (less steady red–pink color; large σRGB for a long duration of time) increases with the development of the convective boundary layer. It deﬁnes the end of the stratus. We con- sider the preceding lifting of the stratus deck from 07:30 to For the night of 6–7 July, the criteria for stratus LLSC are satisﬁed from slightly after midnight until 12:30 UTC (Ta- ble 1). 5.1 LLSC detection Therefore, LLSCs cannot be detected during rain events which are thus excluded. As far as we know, it is the ﬁrst time that such methodology is used for the study of the stratus cloud deck formation and breaking. Figure 9 shows two typical examples to illustrate this method using information provided by the ceilometer and the cloud radar. The evolution of the average color and the stan- dard deviation during two nights (the nights of 7–8 July and 6–7 July are shown in Fig. 9a and b, respectively) are shown with the CBH estimated from the ceilometer and the CTH estimated from the cloud radar. Onset and breakup times deduced when applying the detection method are also indi- cated. For both examples, one can see how the smallest steady standard deviation of the IR image color consistently cor- www.atmos-chem-phys.net/19/8979/2019/ Atmos. Chem. Phys., 19, 8979–8997, 2019 C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa 7 July (Fig. 9b). One can see that low clouds occur between 21:00 UTC and midnight; they have a slightly higher base than later in the night, and there are long periods without clouds. This phase corresponds to the stratus–fractus phase deﬁned by Adler et al. (2019), which precedes the “stra- tus phase”. Due to our criteria related to the persistence of LLSCs, this phase is excluded from our detection method. We instead detect the “stratus phase” that corresponds to full- deck LLSCs. Note that a very short stratus–fractus phase can also be seen in the previous example of the night of 7–8 July (Fig. 9a), from 23:30 to 00:10 UTC, which is also consistent with the work of Adler et al. (2019) and Babi´c et al. (2019a). Clouds after 12:00 UTC for both cases correspond to convective shallow cumulus (within the “convective phase” noted by Babi´c et al., 2019a), as revealed by both a larger IR image standard deviation (σRGB) and the very large scatter in the ceilometer-based CBH between 1000 and 3000 m a.g.l. (Fig. 9). The formation of cumulus clouds below the LLSCs is one of the three LLSC breakup scenarios described in Lo- hou et al. (2019), but the shallow cumulus and fair weather cumulus are not detected by our criteria and indeed not under scope here. 7 July (Fig. 9b). One can see that low clouds occur between 21:00 UTC and midnight; they have a slightly higher base than later in the night, and there are long periods without clouds. This phase corresponds to the stratus–fractus phase deﬁned by Adler et al. (2019), which precedes the “stra- tus phase”. Due to our criteria related to the persistence of LLSCs, this phase is excluded from our detection method. We instead detect the “stratus phase” that corresponds to full- deck LLSCs. Note that a very short stratus–fractus phase can also be seen in the previous example of the night of 7–8 July (Fig. 9a), from 23:30 to 00:10 UTC, which is also consistent with the work of Adler et al. (2019) and Babi´c et al. (2019a). time and breakup time over the same set of 25 selected days as the monsoon ﬂow statistics discussed above. Large vari- ability in onset and breakup times of LLSCs is indeed ob- served. C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa Overall, 26 % of LLSC onsets are observed before 00:00 UTC, and 78 % are observed before 03:00 UTC. All breakup times are observed after sunrise: 33 % of LLSC breakup times occurred before 10:00 UTC, and 74 % oc- curred before 12:00 UTC. We found that LLSCs always clear up after the NLLJ breakup time. 5.2 LLSC lifetime statistics Based on the detection criteria explained above, we noted a signiﬁcant occurrence of LLSCs during the DACCIWA cam- paign at Savè: 65 % of the nights exhibit the development of a nocturnal stratus deck. This ﬁnding is consistent with the work of Kalthoff et al. (2018). Figure 12 provides a schematic evolution of the NLLJ and of the LLSCs that we observe during the DACCIWA ﬁeld ex- periment at the Savè site for the 25 d considered in the statis- tics. This prevalent occurrence is revealed in Fig. 10, which shows, from 15:00 UTC one day (D-1) to 18:00 UTC the fol- lowing day, the MI arrival time and onset and breakup times of LLSC and NLLJ. The median NLLJ core height remains lower than 350 m a.g.l. until the onset of the LLSC. Once the LLSC starts to form, the NLLJ core rises within the stratus clouds until it dissipates in the morning with increasing instability. LLSCs of 340±90 m thickness form on average at the NLLJ core height. This average thickness is very similar to that found by Adler et al. (2019) based on all days of the intensive observation period (340±80 m, over 11 IOP (intensive obser- vation period) days). The CBH is usually stationary during the night from 00:00 to 06:00 UTC. It sharply increases after 08:00 UTC when the convective boundary layer develops. Figure 10 shows some intra-seasonal variability. The pro- nounced occurrence of LLSCs from 26 June through 11 July corresponds to the post-onset phase of the WAM. It is inter- rupted from 12 to 17 July by the drier vortex period iden- tiﬁed by Knippertz et al. (2017) and noted above. From 23 to 26 July, the westerly moist regime observed (not shown here but seen in the UHF wind proﬁler data of Kalthoff et al., 2018) is associated with more rain (nearly on a nightly basis). This situation prevents LLSCs from being detected using the sky camera. 6 Diurnal cycles of the NLLJ and LLSCs Clouds after 12:00 UTC for both cases correspond to convective shallow cumulus (within the “convective phase” noted by Babi´c et al., 2019a), as revealed by both a larger IR image standard deviation (σRGB) and the very large scatter in the ceilometer-based CBH between 1000 and 3000 m a.g.l. (Fig. 9). The formation of cumulus clouds below the LLSCs is one of the three LLSC breakup scenarios described in Lo- hou et al. (2019), but the shallow cumulus and fair weather cumulus are not detected by our criteria and indeed not under scope here. Based on the detection of the NLLJ and LLSCs discussed before, we now discuss their composite diurnal cycle. Figure 12 shows the composite diurnal cycle of the NLLJ core height, LLSC base height, and LLSC top height com- puted using the median of those diagnostics over all consid- ered days as indicated previously. The shaded areas corre- spond to the standard deviation of the ensemble. Note that calculations of the median and standard deviation are pos- sible only when more than ﬁve cases can be taken into ac- count. Thus, for the NLLJ core height, between 10 and 25 d are taken into account between 18:00 (D-1) and 06:30 UTC. For the CBH, between 10 and 25 d are taken into account from 00:00 to 11:40 UTC. For the CTH, between 10 and 25 d are taken into account between 04:00 and 11:20 UTC. C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa The variability of cloud base found between 02:00 and 04:00 UTC, and revealed by higher σRGB and smaller R values, does not persist long enough to interrupt the LLSC deﬁnition. For the night of 7–8 July (Fig. 9a), the criteria for LLSC detection are satisﬁed from midnight until 10:30 UTC. This period deﬁnes the LLSC lifetime for this day (Table 1). Be- fore midnight, the sky is clear for a large part of the time (blue color with the IR camera); some clouds passed be- tween 10:00 and 15:00 m a.g.l. The cloud base heights are variable according to the ceilometer; the IR color and σRGB do not satisfy the LLSC detection criteria. After 10:30 UTC, the cloud base rises and the fractionation of the cloud base (less steady red–pink color; large σRGB for a long duration of time) increases with the development of the convective boundary layer. It deﬁnes the end of the stratus. We con- sider the preceding lifting of the stratus deck from 07:30 to Adler et al. (2019) used the ceilometer to estimate noctur- nal LLSC characteristics (CBH, onset, and breakup times). These authors differentiated between LLSC stratus deck and preceding “stratus–fractus” phase using two lower-bound thresholds of 90 % and 50 %, respectively, for the cloud fraction computed within a 60 min time interval. An exam- ple of those two phases can be seen on the night of 6– www.atmos-chem-phys.net/19/8979/2019/ Atmos. Chem. Phys., 19, 8979–8997, 2019 8993 www.atmos-chem-phys.net/19/8979/2019/ 7 Discussion and conclusions Distribution of (blue) onset and (brown) breakup times of LLSCs at Savè from the period of 20 June through 30 July 2016. to detect and could have some more subtle impacts on the monsoon ﬂow or the MI characteristics or propagation. The intra-seasonal variability of the synoptic situation an- alyzed by Knippertz et al. (2017) impacts not only the mon- soon ﬂow depth and intensity but all low troposphere struc- tures. For example, a break of the WAM activity in mid- July with an unusual cyclonic dry air mass circulation weak- ened the monsoon ﬂow and the NLLJ and inhibited LLSC formation at Savè (Babi´c et al., 2019b). Besides the day-to- day variability of the monsoon ﬂow due to large-scale atmo- spheric conditions, the median diurnal cycle of the monsoon ﬂow exhibits a 1500 m depth southwesterly monsoon ﬂow of 6 m s−1 at night, which slightly deepens (2000 m), slows down (4 m s−1), and becomes more southerly during the day. Figure 11. Distribution of (blue) onset and (brown) breakup times of LLSCs at Savè from the period of 20 June through 30 July 2016. y g y The advection of cool air at Savè is due to two processes already noted by Adler et al. (2019) and Babi´c et al. (2019a): MI and the NLLJ. The two processes occur because of the late afternoon decrease of the turbulence due to convection. Considering the MI propagation time to reach Savè, both processes are hypothesized to be observed at Savè between 17:00 and 21:00 UTC. The MI is rather unknown and poorly observed, but its passage at a given location is expected to imply an increase in the wind speed and a decrease in the temperature. The fuzzy logic method applied in this study to detect the MI arrival time leans on these expectations and depends on the weight attributed to the changes in tempera- ture and wind speed. The NLLJ has been extensively studied in different places around the world but the detection of its time settlement is for sure a question of criteria, particularly the threshold for the wind jet core (5 m s−1 in this study). Despite these inherent limitations of criteria-based methods, MI arrival and NLLJ settlement times are estimated at Savè in June and July 2016. 7 Discussion and conclusions The shortest lifetime of the LLSCs was observed on 30 July. On this day, the onset LLSC is observed very late. For some cases, the breakup time is observed after noon, as on 3 July. A large variability in LLSC characteristics is ob- served: some nights exhibit steady LLSCs, and other nights have lots of intermittent LLSCs (ﬁgure not shown) (e.g., 11, 21, 29 July). On the days that LLSCs form, they appear usu- ally more than 3 h after the onset of the NLLJ and clear up after the NLLJ breakup time (Fig. 10). An analysis of a 41 d DACCIWA dataset during the WAM in 2016 in Benin enabled a quantiﬁed documentation of low- level stratiform clouds and dynamical structures in the low troposphere. A key need for observations to compare with numerical weather and climate models motivated the ﬁeld campaign in the DACCIWA project. The months of June and July in West Africa are character- ized by deep convection, and the conditions at Savè supersite are frequently disturbed by MCSs. Among the 41 d period, 16 d have been excluded from the quantiﬁed statistics. How- Figure 11 gives more quantiﬁcation of those time char- acteristics. It displays the distribution of the LLSC onset www.atmos-chem-phys.net/19/8979/2019/ Atmos. Chem. Phys., 19, 8979–8997, 2019 C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa 8994 C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa Figure 10. The time series of the onset (solid lines) and breakup times (dashed lines) of NLLJ and LLSCs are shown in black and blue, respectively. The red colored circles indicate the arrival time of the MI at Savè using the mean fuzzy logical function. Solid black vertical lines delimit the different phases of the monsoon described by Knippertz et al. (2017). Dashed vertical lines indicate the vortex phase included in the post-onset phase. Figure 10. The time series of the onset (solid lines) and breakup times (dashed lines) of NLLJ and LLSCs are shown in black and blue, respectively. The red colored circles indicate the arrival time of the MI at Savè using the mean fuzzy logical function. Solid black vertical lines delimit the different phases of the monsoon described by Knippertz et al. (2017). Dashed vertical lines indicate the vortex phase included in the post-onset phase. Figure 11. Competing interests. The authors declare that they have no conﬂict of interest. Competing interests. The authors declare that they have no conﬂict of interest. Special issue statement. This article is part of the special issue “Re- sults of the project “Dynamics-aerosol-chemistry-cloud interactions in West Africa” (DACCIWA) (ACP/AMT inter-journal SI)”. It is not associated with a conference. Using an infrared cloud camera, an original methodology is deﬁned to detect the onset and breakup time of the low- level stratiform cloud over Savè. Based on this instrument, the results show that the low-level stratus is a persistent phe- nomenon that occurs 65 % of nights during our studied pe- riod. It forms more than 3 h (6 h on average) after the NLLJ onset at the jet core height and persists until noon on 80 % of days with nocturnal stratus formation. Our diagnostics on NLLJ and LLSC characteristics gave a good representation of the diurnal cycle of the lowest atmosphere conditions ob- served over the southern part of West Africa. We have shown that LLSCs form at around 00:00 UTC with their base at the same height as the NLLJ core. The LLSC base and top heights remain constant during the night (340 m thickness) and rise as the cloud deepens during early morning. LLSCs persist until early afternoon. Acknowledgements. The DACCIWA project has received fund- ing from the European Union Seventh Framework Programme (FP7/2007-2013) under grant agreement no. 603502. The authors thank also Laboratoire d’Aérologie, Université de Toulouse, CNRS, UPS, France, and KIT (Karlsruhe Institute of Technology) and UPS (Université Toulouse) for helping to install the equipment, as well as the people from INRAB in Savè for allowing the equipment on their ground. We thank the Aeris data infrastructure for providing the DACCIWA Operation Center during the campaign and the ac- cess to the data used in this study. Financial support. The research leading to these results has re- ceived funding from the European Union Seventh Framework Pro- gramme (FP7/2007-2013) under grand agreement no. 603502. The results of our study reveal how closely related NLLJ, MI, and LLSC phenomena are. The processes involved in the LLSC formation and related to the MI and NLLJ are ad- dressed in Adler et al. (2019) and Babi´c et al. (2019a), and a conceptual model of the LLSC life cycle proposed in Lohou et al. (2019). The present study brings an overall statistical analysis of the key dynamical features of the low troposphere during the WAM. Competing interests. The authors declare that they have no conﬂict of interest. It also intends to show the day-to-day vari- ability and exhaustively gives quantiﬁed diagnostics for each day of the entire period. Therefore, it is an important basis for any future case study. Moreover, these results can be used for model evaluation. A special focus could be given to the NLLJ simulation since the NLLJ has been shown to play an important role in the air cooling and shear-driven turbulence prior to the saturation and LLSC formation. Review statement. This paper was edited by Susan van den Heever and reviewed by Michael Diamond, Christophe Lavaysse, and one anonymous referee. 7 Discussion and conclusions The MI arrival time at the Savè site occurs between 16:00 and 21:00 UTC; the median occur- rence time is at 19:00 UTC. The NLLJ is a systematic feature within the monsoon ﬂow, with a jet core wind speed between 7 and 11 m s−1 within the (200–300◦) southwesterly sector at a height of about 350 m. It usually sets up before 21:00 UTC (median time occurrence is 18:30 UTC) and ceases before 07:00 UTC. Figure 12. Median diurnal cycle of the NLLJ core height (solid red line), cloud base height (solid green line) and cloud top height (solid grey line) from 20 June to 30 July 2016 at the Savè supersite. Shaded areas indicate variability (±σ). Figure 12. Median diurnal cycle of the NLLJ core height (solid red line), cloud base height (solid green line) and cloud top height (solid grey line) from 20 June to 30 July 2016 at the Savè supersite. Shaded areas indicate variability (±σ). ever, many other MCSs may have impacted the ﬂow between the coast and Savè without being detected at Savè. These MCSs are an important part of the monsoon in West Africa and participate in the variability of the low-level cloud and dynamical structures discussed below. When they occurred in the surroundings of the Savè site, MCSs could be detected (rainfall or density current) and excluded from the analyzed days. However, MCSs occurring upstream of Savè were hard www.atmos-chem-phys.net/19/8979/2019/ Atmos. Chem. Phys., 19, 8979–8997, 2019 8995 C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa The large time range for MI arrival time at Savè is due to the large range of monsoon ﬂow strength 1–5 m s−1 with few cases above 6 m s−1. It also depends on the capability of the turbulent mixing to balance monsoon ﬂow, which slows the front progression in land. The MI arrival time is then concomitant with the NLLJ onset; the two processes, as we deﬁned them in this study, are sometimes difﬁcult to dis- tinguish. The acceleration of the monsoon ﬂow through the NLLJ over West Africa causes the ITD to penetrate about 100 km further north every night (Pospichal and Crewell, 2007; Lothon et al., 2008). It is plausible that the MI con- tributes to this process as well; the Gulf of Guinea/continent contrast and breeze front may come into play as well. Author contributions. CD, FL, ML, NK, BA, YB, OG, and XPB participated in the DACCIWA campaign and data processing. CD prepared the manuscript with contributions from all co-authors. References Abdou, K., Parker, D. J., Brooks, B., Kalthoff, N., and Lebel, T.: The diurnal cycle of lower boundary-layer wind in the West African monsoon, Q. J. Roy. Meteor. Soc., 136, 66–76, https://doi.org/10.1002/qj.536, 2010. Adler, B., Kalthoff, N., and Gantner, L.: Nocturnal low-level clouds over southern West Africa analysed using high- resolution simulations, Atmos. Chem. Phys., 17, 899-910, https://doi.org/10.5194/acp-17-899-2017, 2017. Adler, B., Babi´c, K., Kalthoff, N., Lohou, F., Lothon, M., Dione, C., Pedruzo-Bagazgoitia, X., and Andersen, H.: Noc- turnal low-level clouds in the atmospheric boundary layer over southern West Africa: an observation-based analysis of conditions and processes, Atmos. Chem. Phys., 19, 663–681, https://doi.org/10.5194/acp-19-663-2019, 2019. Data availability. The DACCIWA embargo period is over; the data of the Savè supersite are available on the baobab database (https://doi.org/10.6096/dacciwa.1618, Derrien et al., 2016; https://doi.org/10.6096/dacciwa.1686, Handwerker et al., 2016; https://doi.org/10.6096/dacciwa.1690, Kohler et al., 2016; https://doi.org/10.6096/dacciwa.1659, Wieser et al., 2016) for sci- entists interested in boundary-layer studies in southern West Africa (http://baobab.sedoo.fr/DACCIWA, OMP, 2019). Andreas, E. L., Claffey, K. J., and Makshtas, A. P.: Low-level at- mospheric jets and inversions over the Western Weddell Sea, Bound.-Lay. Meteorol., 97, 459–486, 2000. Baas, P., Bosveld, C., and Klein Baltink, H.,and Holtlag, A. A. M.: A climatology of Nocturnal Low-Level Jets at more frequent oc- Atmos. Chem. Phys., 19, 8979–8997, 2019 C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa data., in review, 2019 Fosu-Amankwah, K., Gabella, O., Groves, J., Handwerker, J., Kalthoff, N., Kohler, M., Kunka, M., Jambert, C., Jegede, G., Leclercq, J., Lohou, F., Lothon, M., Medina, P., Pedruzo, X., Grams, C., Jones, S., Marsham, J., Parker, D., Haywood, J., and Heuveline, V.: The Atlantic inﬂow to the Saharan heat low: ob- servations and modelling, Q. J. Roy. Meteor. Soc., 136, 125–140, 2010. Handwerker, J., Scheer, S., and Gamer, T.: DACCIWA ﬁeld cam- paign, Savè super-site, Cloud and precipitation, SEDOO OMP, https://doi.org/10.6096/dacciwa.1686, 2016. Blackadar, A. K.: Boundary layer wind maxima and their signiﬁ- cance for the growth of nocturnal inversions, B. Am. Meteorol. Soc., 38, 283–290, 1957. Hannak, L., Knippertz, P., Fink, A. H., Kniffka, A., and Pante, G.: Why do global climate models struggle to represent low-level clouds in the West African summer monsoon?, J. Climate, 30, 1665–1687, https://doi.org/10.1175/JCLI-D-16-0451.1, 2017. Bonner, W. D.: Climatology of the low level jet, Mon. Weather Rev., 96, 833–850, 1968. Bouniol, D., Couvreux, F., Kamsu-Tamo, P. H., Lepay, M., Guichard, F., and Favot, F.: Diurnal and seasonal cycles of cloud occurrences, types, and radiative impact over West Africa, J. Appl. Meteorol. Clim., 51, 534–553, 2012. Jacoby-Koaly, S., Campistron, B., Bernard, S., Bénech, B., Girard- Ardhuin, F., and Dessens, J.: Turbulent dissipation rate in the boundary layer via UHF wind proﬁler Doppler spectral width measurements, Bound.-Lay. Meteorol., 103, 361–389, 2002. Buckle, C.: Weather and climate in Africa, Addison-Wesley Long- man Ltd, Harlow, UK, 320 pp., 1996. Janicot, S., Thorncroft, C. D., Ali, A., Asencio, N., Berry, G., Bock, O., Bourles, B., Caniaux, G., Chauvin, F., Deme, A., Kergoat, L., Lafore, J.-P., Lavaysse, C., Lebel, T., Marticorena, B., Mounier, Janicot, S., Thorncroft, C. D., Ali, A., Asencio, N., Berry, G., Bock, O., Bourles, B., Caniaux, G., Chauvin, F., Deme, A., Kergoat, L., Lafore, J.-P., Lavaysse, C., Lebel, T., Marticorena, B., Mounier, F., Nedelec, P., Redelsperger, J.-L., Ravegnani, F., Reeves, C. E., Roca, R., de Rosnay, P., Schlager, H., Sultan, B., Tomasini, M., Ulanovsky, A., and ACMAD forecasters team: Large-scale overview of the summer monsoon over West Africa during the AMMA ﬁeld experiment in 2006, Ann. Geophys., 26, 2569– 2595, https://doi.org/10.5194/angeo-26-2569-2008, 2008. Coceal, O., Bohnenstengel, I. S., and Kotthaus, S.: Detection of sea- breeze events around London using a fuzzy-logic algorithm, At- mos. Sci. Lett., 19, e846, https://doi.org/10.1002/asl.846, 2018. F., Nedelec, P., Redelsperger, J.-L., Ravegnani, F., Reeves, C. www.atmos-chem-phys.net/19/8979/2019/ www.atmos-chem-phys.net/19/8979/2019/ C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa 8996 tation, SEDOO OMP, https://doi.org/10.6096/DACCIWA.1618, 2016. currence of the LLC during the monsoon Cabauw, J. Appl. Me- teorol. Clim., 48, 1627–1642, 2009. currence of the LLC during the monsoon Cabauw, J. Appl. Me- teorol. Clim., 48, 1627–1642, 2009. tation, SEDOO OMP, https://doi.org/10.6096/DACCIWA.1618, 2016. Dione, C., Lothon, M., Badiane, D., Campistron, B., Couvreux, F., Guichard, F., and Sall, S. M.: Phenomenology of Sahelian con- vection observed in Niamey during the early monsoon, Q. J. Roy. Meteor. Soc., 140, 500–516, https://doi.org/10.1002/qj.2149, 2014. Babi´c, K., Adler, B., Kalthoff, N., Andersen, H., Dione, C., Lo- hou, F., Lothon, M., and Pedruzo-Bagazgoitia, X.: The ob- served diurnal cycle of low-level stratus clouds over southern West Africa: a case study, Atmos. Chem. Phys., 19, 1281–1299, https://doi.org/10.5194/acp-19-1281-2019, 2019a. Flamant, C., Knippertz, P., Fink, A., Akpo, A., Brooks, B., Chiu, C., Coe, H., Danuor, S., Evans, M., Jegede, O., Kalthoff, N., Konaré, A., Liousse, C., Lohou, F., Mari, C., Schlager, H., Schwarzenboeck, A., Adler, B., Amekudzi, L., Aeyee, J., Ay- oola, M., Bessardon, G., Bower, K., Burnet, F., Catoire, V., Colomb, A., Fossu-Amankwah, K., Lee, J., Lothon, M., Man- aran, M., Marsham, J., Meynadier, R., Ngamini, J.-B., Rosen- berg, P., Sauer, D., Schneider, J., Smith, V., Stratmann, G., Voigt, C., and Yoboue, V.: The Dynamics-Aerosol-Chemistry- Cloud Interactions in West Africa ﬁeld campaigns: Overview and research highlights, B. Am. Meteorol. Soc., 38, L21808, https://doi.org/10.1029/2011GL049278, 2011. Babi´c, K., Kalthoff, N., Adler, B., Quinting, J. F., Lohou, F., Dione, C., and Lothon, M.: What controls the formation of nocturnal low-level stratus clouds over southern West Africa during the monsoon season?, Atmos. Chem. Phys. Discuss., https://doi.org/10.5194/acp-2019-537, in review, 2019b. Bessardon, G., Brooks, B., Abiye, O., Adler, B., Ajao, A., Ajileye, O. Amekudzi, L. K., Armah Aryee, J. N., Atiah, W. A., Ayoola, M., Bret, G., Bezombes, Y., Cayle-Aethelhard, F., Danuor, S., Delon, C., Derrien, S., Dione, C., Durand, P., Brilouet, P. E., Fosu-Amankwah, K., Gabella, O., Groves, J., Handwerker, J., Kalthoff, N., Kohler, M., Kunka, M., Jambert, C., Jegede, G., Leclercq, J., Lohou, F., Lothon, M., Medina, P., Pedruzo, X., Reinares, I., Sharpe, S., Smith, V., Sunmonu, L., Tan, T., and Wieser, A.: A new high-quality dataset of the diurnal cycle of the southern West African atmospheric boundary layer during the Monsoon season – an overview from the DACCIWA campaign, Sci. C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa 8997 African monsoon region, Geophys. Res. Lett., 38, L21808, https://doi.org/10.1029/2011GL049278, 2011. African monsoon region, Geophys. Res. Lett., 38, L21808, https://doi.org/10.1029/2011GL049278, 2011. Parker, D. J., Burton, R. R., Diongue-Niang, A., Ellis, R. J., Fel- ton, M., Taylor, C. M., Thorncroft, C. D., Bessemoulin, P., and Tompkins, A. M.: The diurnal cycle of the West African monsoon circulation, Q. J. Roy. Meteor. Soc., 131, 2839–2860, https://doi.org/10.1256/qj.04.52, 2005. Knippertz, P., Evans, M. J., Field, P. R., Fink, A. H., Liousse, C., and Marsham, J. H.: The possible role of local air pollution in climate change in West Africa, Nat. Clim. Change, 5, 815–822, 2015. Redelsperger, J.-L., Thorncroft, C. D., Diedhiou, A., Lebel, T., Parker, D. J., and Polcher, J.: African Monsoon Multi- disciplinary Analysis: An international research project and ﬁeld campaign, B. Am. Meteorol. Soc., 87, 1739–1746, https://doi.org/10.1175/BAMS-87-12-1739, 2006. Knippertz, P., Fink, A. H., Deroubaix, A., Morris, E., Tocquer, F., Evans, M. J., Flamant, C., Gaetani, M., Lavaysse, C., Mari, C., Marsham, J. H., Meynadier, R., Affo-Dogo, A., Bahaga, T., Brosse, F., Deetz, K., Guebsi, R., Latifou, I., Maranan, M., Rosenberg, P. D., and Schlueter, A.: A meteorological and chem- ical overview of the DACCIWA ﬁeld campaign in West Africa in June–July 2016, Atmos. Chem. Phys., 17, 10893–10918, https://doi.org/10.5194/acp-17-10893-2017, 2017. Pospichal, B. and Crewell, S.: Boundary layer observations in West Africa using a novel microwave radiometer, Meteorol. Z., 16, 513–523, 2007. Saïd, F., Campistron, B., Delbarre, H., Canut, G., Doerenbecher, A., Durand, P., Fourrié, N., Lambert, D., and Legain, D.: Offshore winds obtained from a network of wind-proﬁler radars during HyMeX, Q. J. Roy. Meteor. Soc., 142, 23–42, 2016. Kohler, M., Kalthoff, N., Seringer, J., and Kraut, S.: DACCIWA ﬁeld campaign, Savè super-site, Surface measurements, SEDOO OMP, https://doi.org/10.6096/dacciwa.1690, 2016. Laing, G. L. and Fritsch, J. M.: Mesoscale convective complexes in Africa, Mon. Weather Rev., 121, 2254–2263, 1993. Schrage, J. M. and Fink, A. H.: Nocturnal continental low-level stra- tus over tropical West Africa: observations and possible mecha- nisms controlling its onset, Mon. Weather Rev., 140, 1794–1809, 2012. Lavaysse, C., Janicot, S., Parker, D. J., Lafore, J.-P., Sultan, B., and Pelon, J.: Seasonal evolution of the West African heat low: a Cli- matological perspective, Clim. Dynam., 33, 313–330, 2009. Schrage, J. M., Augustyn, S., and Fink, A. H.: Nocturnal strati- form cloudiness during the West African monsoon, Meteorol. Atmos. Phys., 95, 73–86, https://doi.org/10.1007/s00703-006- 0194-7, 2007. Löhnert, U. C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa Couvreux, F., Guichard, F., Bock, O., Campistron, B., Lafore, J. P., Redelsperger, J. P.: Synoptic variability of the monsoon ﬂux over west africa prior to the onset, Q. J. Roy. Meteor. Soc., 136, 169– 173, 2010. Crewell, S. and Löhnert, U.: Accuracy of boundary layer tempera- ture proﬁles retrieval with multifrequency multiangle microwave radiometer, IEEE T. Geosci. Remote., 45, 2195–2201, 2007. Kalapureddy, M. C. R., Lothon, M., Campistron, B., Lohou, F., and Saïd, F.: Wind proﬁler analysis of the African Easterly Jet in re- lation with the boundary layer and the Saharan heat-low, Q. J. Roy. Meteor. Soc., 136, 77–91, 2010. Dee, D. P., Uppala, S. M., Simmons, A. J., Berrisford, P., Poli, P., Kobayashi, S., Andrae, U., Balmaseda, M. A., Balsamo, G., Bauer, D. P., and Bechtold, P.: The ERA-Interim reanalysis: Con- ﬁguration and performance of the data assimilation system, Q. J. Roy. Meteor. Soc., 137, 553–597, 2011. Kalthoff, N., Lohou, F., Brooks, B., Jegede, G., Adler, B., Babi´c, K., Dione, C., Ajao, A., Amekudzi, L. K., Aryee, J. N. A., Ay- oola, M., Bessardon, G., Danuor, S. K., Handwerker, J., Kohler, M., Lothon, M., Pedruzo-Bagazgoitia, X., Smith, V., Sunmonu, L., Wieser, A., Fink, A. H., and Knippertz, P.: An overview of the diurnal cycle of the atmospheric boundary layer dur- ing the West African monsoon season: results from the 2016 observational campaign, Atmos. Chem. Phys., 18, 2913–2928, https://doi.org/10.5194/acp-18-2913-2018, 2018. Deetz, K., Vogel, H., Haslett, S., Knippertz, P., Coe, H., and Vo- gel, B.: Aerosol liquid water content in the moist southern West African monsoon layer and its radiative impact, Atmos. Chem. Phys., 18, 14271–14295, https://doi.org/10.5194/acp-18-14271- 2018, 2018. Derrien, S., Bezombes, Y., Bret, G., Gabella, O., Jarnot, C., Medina, P., Piques, E., Delon, C., Dione, C., Campistron, B., Durand, P., Jambert, C., Lohou, F., Lothon, M., Paciﬁco, F., and Meyerfeld, Y.: DACCIWA ﬁeld campaign, Savè super-site, UPS instrumen- Knippertz, P., Fink, A. H., Schuster, R., Trentmann, J., Schrage, J. M., and Yorke, C.: Ultra-low clouds over the southern West www.atmos-chem-phys.net/19/8979/2019/ Atmos. Chem. Phys., 19, 8979–8997, 2019 C. Dione et al.: Low-level stratiform cloud and dynamical features observed over southern West Africa and Crewell, S.: Accurracy of cloud liquid water path from ground-based microwave radiometer. Part 1: De- pendency on cloud model statistics, Radio Sci., 38, 8041, https://doi.org/10.1029/2002RS002654, 2003. Schuster, R., Fink, A. H., and Knippertz, P.: Formation and main- tenance of nocturnal low-level stratus over the southern West African monsoon region during AMMA 2006, J. Atmos. Sci., 70, 2337–2355, 2013. Löhnert, U., Turner, D., and Crewell, S.: Ground-based temper- ature and humidity proﬁling using spectral infrared and mi- crowave observations. Part I: Simulated retrieval performance inclear-sky conditions, J. Appl. Meteorol. Clim., 48, 1017–1032, https://doi.org/10.1175/2008JAMC2060.1, 2009 Schwendike, J., Kalthoff, N., and Kohler, M.: The impact of mesoscale convective systems on the surface and boundary-layer structure in West Africa: Case-studies from the AMMA cam- paign 2006, Q. J. Roy. Meteor. Soc., 136, 566–582, 2010. Lohou, F., Kalthoff, N., Adler, B., Babi´c, K., Dione, C., Lothon, M., Pedruzo-Bagazgoitia, X., and Zouzoua, M.: Concep- tual model of diurnal cycle of stratiform low-level clouds over southern West Africa, Atmos. Chem. Phys. Discuss., https://doi.org/10.5194/acp-2019-566, in review, 2019. Stull, R. B.: An Introduction to Boundary Layer Meteorology, Kluwer Academic, 666 pp., 1988. Sultan, B., and Janicot, S.: The West African monsoon dynamics. Part II: The “preonset” and “onset” of the summer monsoon, J. Climate, 16, 3407–3427, 2003. Lothon, M., Saïd, F., Lohou, F., and Campistron, B.: Observation of the diurnal cycle in the low troposphere of West Africa, Mon. Weather Rev., 136, 3477–3500, 2008. Van Der Linden, R., Fink, A. H., and Redl, R.: Satellite-based cli- matology of low-level continental clouds in southern West Africa during the summer monsoon season, J. Geophys. Res.-Atmos., 120, 1186–1201, https://doi.org/10.1002/2014JD022614, 2015. Madougou, S., Saïd, F., Campistron, B., and Kebe, C. F.: Economic study of the wind energy potential in Sahel, Journal of Energy and Power Engineering, 6, 1934–1975, 2012. Nam, C., Bony, S., Dufresne, J.-L., and Chepfer, H.: The “too few, too bright” tropical low-cloud problem in CMIP5 models, Geophys. Res. Lett., 39, L21801, https://doi.org/10.1029/2012GL053421, 2012. Wieser, A., Adler, B., and Deny, B.: DACCIWA ﬁeld cam- paign, Savè super-site, Thermodynamic data sets, SEDOO OMP, https://doi.org/10.6096/DACCIWA.1659, 2016. OMP: DACCIWA database, available at: http://baobab.sedoo.fr/ DACCIWA/, last access: 9 July 2019. Atmos. Chem. Phys., 19, 8979–8997, 2019 www.atmos-chem-phys.net/19/8979/2019/ www.atmos-chem-phys.net/19/8979/2019/
https://openalex.org/W3090300450	https://www.nepjol.info/index.php/bjhs/article/download/31371/24805	English	null	Analysis of Corneal Astigmatism after Clear Corneal Phacoemulsification : A Prospective Study	Birat journal of health sciences	2,020	cc-by	3,205	Agrawal N et al Agrawal N et al Results Eighty paƟents of group A and 99 paƟents of Group B were included in the analysis. PreoperaƟve keratometry and asƟgmaƟsm was comparable between the two groups. Post- operaƟve asƟgmaƟsm was signiﬁcantly lower in temporal incision (0.75± 0.58) group than in superior incision group (1.2±0.71). SIA in-group A was 0.43D and in group B was 0.18D. SE was found to be signiﬁcantly lower in temporal incision (p<0.01) group. * Corresponding Author Dr. Nisha Agrawal Consultant, Catract and Phaco Specialist Taparia Eye Care Pvt. Ltd., Nepal Email ID: doctor.nisha.agrawal@gmail.com ORCID ID: hƩps://orcid.org/0000-0001-7626-9506 ObjecƟves To calculate and compare SIA and postoperaƟve spherical equivalent (SE) among eyes undergoing phacoemulsiﬁcaƟon performed via superior and temporal approach. A R T I C L E I N F O Received : 28 August, 2019 Accepted : 27 August, 2020 Published : 05 October, 2020 Methodology A longitudinal prospecƟve observaƟonal study was conducted in a terƟary eye care center. 200 consecuƟve paƟents who underwent 2.8mm clear corneal phacoemulsiﬁcaƟon were allocated in Group A (Superior) and Group B (Temporal) in equal numbers by random loƩery method. PaƟents with any corneal pathology, glaucoma, reƟna or macular diseases, with previous history of trauma or ocular surgery were excluded. PreoperaƟve asƟgmaƟsm, postoperaƟve asƟgmaƟsm and postoperaƟve SE were calculated at 6weeks follow-up. SIA was calculated using vector analysis soŌware version 2.1 given by Dr Sawhney. © Authors retain copyright and grant the journal right of ﬁrst publicaƟon with the work simultaneously licensed under CreaƟve Commons AƩribuƟon License CC - BY 4.0 that allows others to share the work with an acknowledgment of the work's authorship and iniƟal publicaƟon in this journal. ORA 171 IntroducƟon 1. Consultant, Cataract and Phaco Specialist, Taparia Eye Care Pvt. Ltd., Nepal 1. Consultant, Cataract and Phaco Specialist, Taparia Eye Care Pvt. Ltd., Nepal With the evoluƟon of cataract surgery over the past years and increasing paƟent demand for spectacle independence, today cataract surgery is moreover a refracƟve surgery. Surgically induced asƟgmaƟsm (SIA) hinders post-operaƟve refracƟve precision by changing the refracƟve power of cornea. 2. Assistant Prof NAMS, Head, Department of Vitreo-reƟna, Biratnagar Eye Hospital 3. Department of Cornea and External Diseases, Sagarmatha Chaudhary Eye Hospital 4. Vitreo-reƟna consultant, Prasad Eye Hospital, Ramgarh 5. Consultant Cataract and Phaco Surgeon, Vasan Eye Care 6. Associate Professor, School of Public Health and Community Medicine, B P Koirala InsƟtue of Health Sciences Conclusion Temporal incision is asƟgmaƟcally more neutral and has beƩer refracƟve precision than superior incision clear corneal phacoemulsiﬁcaƟon. ANALYSIS OF CORNEAL ASTIGMATISM AFTER CLEAR CORNEAL PHACOEMULSIFICATION : A PROSPECTIVE STUDY 1* 2 3 4 5 6 Agrawal N , Agarwal L , Yadav R , Kumari A , Singh K , Baral D 1* 2 3 4 5 6 Agrawal N , Agarwal L , Yadav R , Kumari A , Singh K , Baral D INTRODUCTION acuity were recorded. PreoperaƟve verƟcal and horizontal keratometry (corneal curvature: KV and KH respecƟvely), 6weeks postoperaƟve KV, KH were recorded in mm by autorefractometer (Nidek US-500 ECHOSCAN). The millimeter keratometer readings were converted to Diopteric value with the help of conversion table. IOL power calculaƟon was done by non-contact method (IOL master) and cross checked with manual keratometry method. RefracƟve error was ascertained using manual reƟnoscopy. StaƟsƟcal analysis: Collected data were entered in the Micro-SoŌ EXCEL sheet and converted into SPSS 11.5 Version for staƟsƟcal analysis. For descripƟve staƟsƟcs percentage, the mean and standard deviaƟon were calculated for categorical and numerical data respecƟvely. For inferenƟal staƟsƟcs, a comparison was done between surgically induced asƟgmaƟsm and refracƟve error of the two groups and other selected variables at 95% CI where the level of signiﬁcance was considered at p<0.05. Surgically induced asƟgmaƟsm and refracƟve error of the two groups were compared. acuity were recorded. PreoperaƟve verƟcal and horizontal keratometry (corneal curvature: KV and KH respecƟvely), 6weeks postoperaƟve KV, KH were recorded in mm by autorefractometer (Nidek US-500 ECHOSCAN). The millimeter keratometer readings were converted to Diopteric value with the help of conversion table. IOL power calculaƟon was done by non-contact method (IOL master) and cross checked with manual keratometry method. RefracƟve error was ascertained using manual reƟnoscopy. Cataract surgery is one of the long pracƟced ophthalmic surgical procedures. It has evolved tremendously from the era of couching to intracapsular cataract extracƟon, small incision cataract surgery and the latest technique being phacoemulsiﬁcaƟon. In this technique ultrasound probe emulsiﬁes cataractous lens intraocularly through an incision as small as 3mm (or lesser). These advancements in surgical technique and the use of toric or mulƟ focal IOL have raised cataract surgery to refracƟve surgery plaƞorm. At the same Ɵme paƟents' expectaƟons have increased for fast visual recovery and spectacle independence. StaƟsƟcal analysis: Collected data were entered in the Micro-SoŌ EXCEL sheet and converted into SPSS 11.5 Version for staƟsƟcal analysis. For descripƟve staƟsƟcs percentage, the mean and standard deviaƟon were calculated for categorical and numerical data respecƟvely. For inferenƟal staƟsƟcs, a comparison was done between surgically induced asƟgmaƟsm and refracƟve error of the two groups and other selected variables at 95% CI where the level of signiﬁcance was considered at p<0.05. Surgically induced asƟgmaƟsm and refracƟve error of the two groups were compared. RESULTS Of the total 100 paƟents in each group, 80 paƟents from group A and 99 paƟents from group B completed the follow- up protocol and were included in the analysis. Mean age of the paƟents in group A was 57.62 ± 10.502 years and in group B was 56.46 ± 11.381 years (p=0.484). Equal numbers of men (M) and women (F) were there in Group A (Figure 1). Gender distribuƟon raƟo was non-signiﬁcantly diﬀerent (M: F=60:39) among paƟents of Group B (Figure 1). Majority of paƟent's Right eye (Group A=58.7%, Group B=57.5%) was operated in both the groups (Figure 2). The two groups were comparable in terms of non-signiﬁcant diﬀerences in preoperaƟve vision, horizontal keratometry (KH), verƟcal keratometry (KV) and preoperaƟve asƟgmaƟsm(Table1). Post-operaƟve asƟgmaƟsm was signiﬁcantly lower (p<0.001) in temporal incision group (0.7453D) than in superior incision group (1.2D) (Table 1). INTRODUCTION Cornea being a refracƟve surface, the moment a keratome passes through the cornea, it changes its surface curvature and power. Incision length, locaƟon, architecture and wound suturing inculcates surgically induced asƟgmaƟsm 1 (SIA). SIA aﬀects visual outcome and is sƟll a hurdle in acquiring spectacle independence. This study was thus carried out to calculate SIA, postoperaƟve spherical equivalent and to compare them amongst superior and temporal clear corneal incision. In this study, measures to reduce SIA have been discussed, which can help surgeons to improve surgical outcome. KEYWORD Agrawal N, Agarwal L, Yadav R, Kumari A, Singh K, Baral D. Analysis of Corneal AsƟgmaƟsm aŌer Clear Corneal PhacoemulsiﬁcaƟon: A ProspecƟve Study . BJHS 2020;5(2)12: 1001-1005. AsƟgmaƟsm, cataract surgery, phacoemulsiﬁcaƟon, surgically induced asƟgmaƟsm, spherical equivalent Original Research ArƟcle Agrawal N et al INTRODUCTION DISCUSSION In the present era, due to increasing desire of spectacle independence, cataract surgery has become moreover a refracƟve surgery requiring high refracƟve precision. PostoperaƟve asƟgmaƟsm is governed by two factors: pre-operaƟve asƟgmaƟsm and surgically induced asƟgmaƟsm 2,3 (SIA). Pre-operaƟve asƟgmaƟsm is a non-modiﬁable factor whereas SIA is a modiﬁable factor. SIA is in return inﬂuenced by many surgeon related factors like incision length, locaƟon, site, architecture and use of sutures which have 4 varied role to play in determining SIA. High SIA with high residual refracƟve error could lead to dissaƟsfacƟon among surgeons and unhappy paƟents. With superior incision, there was ﬂaƩening of verƟcal keratometry (p<0.01) and steepening of horizontal keratometry (p=0.192). In case of temporal incision group, there was non-signiﬁcant ﬂaƩening of both the verƟcal and horizontal meridians. Following surgery, visual acuity signiﬁcantly improved in both the groups. There was signiﬁcant increase in postoperaƟve asƟgmaƟsm in the superior incision group (p<0.01). Whereas there was non-signiﬁcant decrease in asƟgmaƟsm in the temporal incision group (p=0.424)(Table 2). Surgicallly induced asƟgmaƟsm as calculated by vector analysis was found to be lower in temporal incision group (Table 3).Emmetropia was achieved in 25% cases of group A and 27.27% cases of group B (Table 4). PostoperaƟve spherical equivalent was signiﬁcantly lower in group B than in group A (Table 1) concluding that temporal incision was temporally more neutral with lesser residual asƟgmaƟsm and more propensiƟes to achieve emmetropia. LocaƟon of incision Surgically induced asƟgmaƟsm is also aﬀected by locaƟon of 4 incision. The farther the incision is from the centre of visual axis the lesser is the asƟgmaƟsm. Scleral incisions induce least asƟgmaƟsm followed by limbal incision, maximum SIA being induced by clear corneal incision. In a study by Kagnici and colleagues comparing superior limbal incision with superior clear corneal incision, lesser asƟgmaƟsm was 5 found in limbal group. However, this ﬁnding could not aƩain staƟsƟcal signiﬁcance in their study. Inspite of having higher asƟgmaƟc eﬀect and delayed wound healing as a result of its avascularity, clear corneal incisions are widely pracƟced. This is because they have advantage of sparing the conjuncƟva from scarring which can be reserved for future 6 possible glaucoma surgery. Table 3: Details of Surgically Induced AsƟgmaƟsm in the two groups Figure 2: Operated eye distribuƟon details in between the two groups. Table 1. PreoperaƟve and postoperaƟve study parameters (6 weeks post surgery) in Group A and Group B. Table 4: Details of postoperaƟve spherical equivalent of the Superior and inferior incision groups METHODOLOGY A longitudinal prospecƟve observaƟonal study was carried out at cataract department of Biratnagar Eye Hospital. Two hundred paƟents who underwent phacoemulsiﬁcaƟon in immature cataract were enrolled in the study. InsƟtuƟon review board approved this study and wriƩen informed consent was obtained from all the study parƟcipants. The paƟents were assigned into two groups (100 in superior incision group and 100 in temporal incision group) by random loƩery method. PaƟents with corneal opacity and other corneal pathologies, glaucoma, reƟna or macular diseases, previous history of trauma or ocular surgery were excluded. All the cases were operated by a single surgeon (NA) under peribulbar block. 0.5ml of bupivacaive mixed with adrenalin was administered with 26Ga needle. They underwent phacoemulsiﬁcaƟon (CataRhex easy, VC820100CE, Oertli) through a 2.8mm clear corneal incision made superiorly (Group A) or temporally (Group B). Two side port incisions were made 90° away from the main incision. This was followed by implantaƟon of a single piece foldable IOL with injector-cartridge system. IOL power calculaƟon was done by SRK T formula. All the incisions were closed by stromal hydraƟon and leŌ sutureless. PostoperaƟve regimen consisted of steroid anƟbioƟc combinaƟon (ciproﬂoxacin and dexamethasone) used 6 Ɵmes a day for the ﬁrst week and gradually tapered over 6weeks. Female Male Figure 1: Gender distribuƟon details in between the two groups n = 80 n = 99 Figure 1: Gender distribuƟon details in between the two groups PaƟent's demographic details, slit lamp examinaƟon ﬁndings, operated eye (whether right or leŌ eye), preoperaƟve visual Original Research ArƟcle Agrawal N et al Figure 2: Operated eye distribuƟon details in between the two groups. Table 2: Comparison of PreoperaƟve and postoperaƟve parameters (6 weeks post-surgery) in superior and inferior incision group. RECOMMENDATIONS It is recommended that cataract surgeons should operate temporally for lesser surgically induced asƟgmaƟsm and residual postoperaƟve refracƟve error. Size of incision The size of incision has a great role to play in terms of SIA. The size of incision has a great role to play in terms of SIA. Original Research ArƟcle Agrawal N et al Agrawal N et al spherical equivalent in group A was found to be 0.27±0.46D which was signiﬁcantly higher than that of group B (0.06 ± 0.25D). This implies that there was lesser residual refracƟve error in temporal incision group than in superior incision group. spherical equivalent in group A was found to be 0.27±0.46D which was signiﬁcantly higher than that of group B (0.06 ± 0.25D). This implies that there was lesser residual refracƟve error in temporal incision group than in superior incision group. Smaller incisions have least asƟgmaƟsm as compared to large incisions. They have added advantage of beƩer intra 7 operaƟve chamber stability and faster wound healing. Many authors have studied the eﬀect of size of incision over asƟgmaƟsm in the past. Masket and his friends compared 2.2mm coxial phacoemulsiﬁcaƟon with 3mm incision. They 8 found that 2.2mm group had signiﬁcantly low SIA. Some authors also debate on the fact that too small incisions cause irreversible stretching of wound leading to compromise of wound architecture. Moon SC et al supported this ﬁnding in their study comparing SIA in incision sizes of 2.5mm, 3mm 4 and 3.5mm. They concluded that 3.0mm incision was asƟgmaƟcally most stable in comparison to that on 2.5mm 4 and 3.5mm. Studies have also compared inter-surgeon changes in postoperaƟve asƟgmaƟsm. A study compared the SIA of four surgeons who operated with same locaƟon, site, size and architecture of incision. It was found that there was no signiﬁcant diﬀerence in asƟgmaƟsm amongst the four surgeons concluding that SIA was more dependent on 16 incision related factors than surgeon related factors. Scholars have suggested that those who invariably operate on temporal axis will have variable results. Surgeons should move around the eye in accessible ﬁeld to operate on the steep axis to get beƩer postoperaƟve refracƟve outcome. If preoperaƟve against the rule asƟgmaƟsm is <1D then it is beƩer to operate on temporal (horizontal) meridian. CONCLUSION Temporal incisions in phacoemulsiﬁcaƟon are asƟgmaƟcally more neutral than superior incisions. They also have lesser postoperaƟve residual refracƟve error, which have higher propensity of spectacle independence. In this study, clear corneal superior incision was found to have signiﬁcantly higher postoperaƟve asƟgmaƟsm in comparison to preoperaƟve asƟgmaƟsm (p<0.01). The temporal group had less asƟgmaƟsm postoperaƟvely but this ﬁnding couldn't aƩain staƟsƟcal signiﬁcance (p=0.424). SIA was found to be higher in superior group than that of temporal group. Yaycıoğlu and his colleagues found that asƟgmaƟsm was maximum with nasal incision followed by superior, superior-temporal and least with temporal 10 incision. With nasal incision maximum stress was induced over the incision because of perpendicular entrance of 11 phaco probe. Superior incisions induce greater asƟgmaƟsm due to its proximity to visual centre of the cornea and variable pressure eﬀect on wound architecture by the 5 eyelid. Many study ﬁndings match our results concluding that SIA was maximum with superior incision than with 12-14 temporal incision. Susic and his friends studied the eﬀect of oblique incision on asƟgmaƟsm. No signiﬁcant diﬀerence was noted by them while comparing superonasal incision in 1 right eye with supero-temporal incision in leŌ eye. Temporal incision was compared with incision on steep axis by Borasio et al and they concluded that temporal incision 15 group was associated with lesser SIA. Site of incision Incision on cornea causes ﬂaƩening of that meridian due to wound gaping, Ɵssue addiƟon at that site and increase in 9 radius of curvature. Similar ﬂaƩening of verƟcal meridian in superior incision group and horizontal meridian in temporal incision group was found in our study. In a study by Susic they observed ﬂaƩening of verƟcal meridian and not much 1 change in horizontal meridian. FINANCIAL DISCLOSURE REFERENCES 1. SušićN,Brajković J, Kalauz-Surać I. Analysis of PostoperaƟve Corneal AsƟgmaƟsm aŌer PhacoemulsiﬁcaƟon Through a Clear Corneal Incisions. ActaClin Croat 2007;46(1):37-40. 2. Albert D.M., Miller J.W., AzarD.T,Blodi B.A. (2008) Principles and PracƟce of Ophthalmology. 3rd EdiƟon, Saunders Company, Philadelphia, Chapter 120 3. Henderson B., Pineda R., Ament C., Chen S, Kim, J. (2007) EssenƟals of Cataract Surgery. Slack Incorporated, Online Library, Chapter 7. 4. Moon SC, Mohamed T, Fine IH. Comparison of Surgically Induced AsƟgmaƟsms aŌer Clear Corneal Incisions of Diﬀerent Sizes. Korean Journal of Ophthalmology 2007; 21(1):1-5 .DOI: 10.3341/ kjo.2007. 21.1.1. ISSN: 2542-2758 (Print) 2542-2804 (Online) 1004 Birat Journal of Health Sciences Vol.5/No.2/Issue 12/ May-August 2020 LIMITATION OF STUDY 1. Only temporal and superior incisions were considered. Oblique and nasal incisions were not studied. 2. All the incisions were of 2.8mm and other incision sizes were not taken into consideraƟon. 3. All the incisions were clear corneal and other incision locaƟons were not taken into consideraƟon. 4. Single surgeon operated all the cases and inter surgeon variability was not analysed. CONFLICT OF INTEREST None In the present study, post-operaƟve visual acuity improved signiﬁcantly in all the paƟents of both groups. Mean (±SD) REFERENCES 3. Henderson B., Pineda R., Ament C., Chen S, Kim, J. (2007) EssenƟals of Cataract Surgery. Slack Incorporated, Online Library, Chapter 7. 3. Henderson B., Pineda R., Ament C., Chen S, Kim, J. (2007) EssenƟals of Cataract Surgery. Slack Incorporated, Online Library, Chapter 7. 1. SušićN,Brajković J, Kalauz-Surać I. Analysis of PostoperaƟve Corneal AsƟgmaƟsm aŌer PhacoemulsiﬁcaƟon Through a Clear Corneal Incisions. ActaClin Croat 2007;46(1):37-40. 4. Moon SC, Mohamed T, Fine IH. Comparison of Surgically Induced AsƟgmaƟsms aŌer Clear Corneal Incisions of Diﬀerent Sizes. Korean Journal of Ophthalmology 2007; 21(1):1-5 .DOI: 10.3341/ kjo.2007. 21.1.1. 2. Albert D.M., Miller J.W., AzarD.T,Blodi B.A. (2008) Principles and PracƟce of Ophthalmology. 3rd EdiƟon, Saunders Company, Philadelphia, Chapter 120 Original Research ArƟcle Agrawal N et al 5. Kağnici, DB, Kocatürk, T, Çakmak, H, Dündar, SO. Surgically Induced Corneal AsƟgmaƟsm Following Cataract Surgery. Open Journal of Ophthalmology 2015; 5: 47-53. DOI:10.4236/ojoph.2015.52008 5. Kağnici, DB, Kocatürk, T, Çakmak, H, Dündar, SO. Surgically Induced 11. Kohnen, S, Neuber R, Kohnen, T. Eﬀect of Temporal and Nasal UnsuturedLimbal Tunnel Incisions on Induced AsƟgmaƟsm aŌer PhacoemusiﬁcaƟon. Journal of Cataract RefracƟve Surgery 2002; 28: 821-825. DOI: 10.1016/S0886-3350(01)01215-9. 6. Barequet IS, Yu E, Vitale S, Cassard S, Azar DT, Stark WJ. AsƟgmaƟsm Outcomes of Horizontal Temporal versus Nasal Clear Corneal Incision Cataract Surgery. Journal of Cataract RefracƟve Surgery 2004; 30(2): 418-423.DOI:10.1016/S0886-3350(03)00492-9 12. TejedorJ,Pérez-Rodríguez JA. AsƟgmaƟc Change Induced by 2.8mm Corneal Incisions for Cataract Surgery. InvesƟgaƟve Ophthalmology & Visual Science March 2009; 50: 989-994. DOI:10.1167/iovs.08- 2778. 7. Phleger T, Scholz U, Skorpik C. PostoperaƟve asƟgmaƟsm aŌer no- sƟtch, small incision cataract surgery with 3.5 mm and 4.5 mm incision. J Cataract Refract Surg1994;20:400-5. PMID:7932128 DOI:10.1016/s0886-3350(13)80174-5. 13. Nikose AS, Saha D, Laddha PM, PaƟl M. Surgically induced asƟgmaƟsm aŌer phacoemulsiﬁcaƟon by temporal clear corneal and superior clear corneal approach: a comparison. ClinOphthalmol2018 Jan 3;12:65-70. DOI:10.2147/OPTH.S149709. 8. Masket E, Wang L, Belani S. Induced asƟgmaƟsm with 2.2- and 3.0- mm coaxial phacoemulsiﬁcaƟon incisions. J Refract Surg. 2009;25(1):21-4. PMID:19244949. 14. MerriamJC,Zheng L, Cheng B. Eﬀect of Incision Size and LocaƟon on Surgically-Induced AsƟgmaƟsm (SIA). InvesƟgaƟve Ophthalmology & Visual Science 2009,50(13), 55-67. PMID-2367933 9. Freitas GO, Boteon JE, Carvalho MJ, Pinto RMC. AsƟgmaƟsm treatment during phacoemulsiﬁcaƟon: a review of current surgical strategies and their raƟonale. Rev Bras Ophtalmol 2013; 72 (6): 419- 23. DOI:10.1590/S0034-72802013000600013 15. Borasio E, Mehta JS, Maurino V. REFERENCES Surgically induced asƟgmaƟsm aŌer phacoemulsiﬁcaƟon in eyes with mild to moderate corneal asƟgmaƟsm: temporal versus on-axis clear corneal incisions. J Cataract Refract Surg. 2006 Apr;32(4):565-72. PMID: 16698473. DOI:10.1016/j.jcrs.2005.12.104 10. Yaycioglu, A., Akova, Y.A., Akca, S., Gur, S. and Oktem, C. Eﬀect on AsƟgmaƟsm of the LocaƟon of Clear Corneal Incision in PhacoemulsiﬁcaƟon of Cataract. Journal of RefracƟve Surgery 2007; 23: 515-518. PMID:17523515 16. Theodoulidou S, Asproudis I, Athanasiadis A, Kokkinos M, AspioƟs M. Comparison of surgically induced asƟgmaƟsm among diﬀerent surgeons performing the same incision. Int J Ophthalmol. 2017; 10(6): 1004–1007. DOI: 10.18240/ijo.2017.06.26 16. Theodoulidou S, Asproudis I, Athanasiadis A, Kokkinos M, AspioƟs M. Comparison of surgically induced asƟgmaƟsm among diﬀerent surgeons performing the same incision. Int J Ophthalmol. 2017; 10(6): 1004–1007. DOI: 10.18240/ijo.2017.06.26
https://openalex.org/W2791585216	https://wnus.usz.edu.pl/frfu/file/article/view/11170.pdf	Polish	null	Sector Indicators in the Assessment of Public Companies Based on Financial Statements Presented in Accordance With IAS	Finanse, Rynki Finansowe, Ubezpieczenia	2,016	cc-by-sa	5,028	Edyta Mioduchowska-Jaroszewicz, Katarzyna Hawran, Joanna Kowalik* Streszczenie: Celem artykułu jest przedstawienie sytuacji spółek publicznych z wykorzystaniem wskaźni- ków sektorowych obliczonych na podstawie sprawozdań finansowych sporządzonych według Międzynaro- dowych Standardów Rachunkowości. Słowa kluczowe: spółki publiczne, wskaźniki sektorowe * dr Edyta Mioduchowska-Jaroszewicz, Uniwersytet Szczeciński, Wydział Nauk Ekonomicznych i Zarządzania, Katedra Analizy i Strategii Przedsiębiorstw, e-mail: emioduch@wneiz.pl; mgr Katarzyna Hawran, doktorantka, Uniwersytet Szczeciński, Wydział Nauk Ekonomicznych i Zarządzania, Instytut Ekonomii; mgr Joanna Kowalik, doktorantka, Uniwersytet Szczeciński, Wydział Nauk Ekonomicznych i Zarządzania, Instytut Ekonomii. 1 https://rachunkowosc.com.pl/wskazniki-i-stawki/wskazniki_sektorowe.html (dostęp 29.04.2016). 2 Na dzień 29.04.2016 na GPW jest notowanych 484 spółek. Finanse, Rynki Finansowe, Ubezpieczenia nr 4/2016 (82), cz. 1 DOI: 10.18276/frfu.2016.4.82/1-14 s. 167–17 Finanse, Rynki Finansowe, Ubezpieczenia nr 4/2016 (82), cz. 1 DOI: 10.18276/frfu.2016.4.82/1-14 s. 167–178 2 Na dzień 29.04.2016 na GPW jest notowanych 484 spółek. Wskaźniki sektorowe w ocenie spółek publicznych na podstawie sprawozdań finansowych przedstawionych według MSR Edyta Mioduchowska-Jaroszewicz, Katarzyna Hawran, Joanna Kowalik* 1 https://rachunkowosc.com.pl/wskazniki-i-stawki/wskazniki_sektorowe.html (dostęp 29.04.2016). 2 Na dzień 29.04.2016 na GPW jest notowanych 484 spółek. * dr Edyta Mioduchowska-Jaroszewicz, Uniwersytet Szczeciński, Wydział Nauk Ekonomicznych i Zarządzania, Katedra Analizy i Strategii Przedsiębiorstw, e-mail: emioduch@wneiz.pl; mgr Katarzyna Hawran, doktorantka, Uniwersytet Szczeciński, Wydział Nauk Ekonomicznych i Zarządzania, Instytut Ekonomii; mgr Joanna Kowalik, doktorantka, Uniwersytet Szczeciński, Wydział Nauk Ekonomicznych i Zarządzania, Instytut Ekonomii. 1 https://rachunkowosc.com.pl/wskazniki-i-stawki/wskazniki_sektorowe.html (dostęp 29.04.2016). 2 Na dzień 29.04.2016 na GPW jest notowanych 484 spółek. Wprowadzenie W 2002 roku T. Dudycz i W. Skoczylas rozpoczęli publikowanie wskaźników sektorowych obliczanych dla działów gospodarki według klasyfikacji PKD na łamach „Rachunkowości”. Wyróżniającą cechą tej publikacji jest prezentacja, obok średnich, także innych miar staty- styki opisowej, charakteryzujących rozkład badanej zmiennej oraz jej zmienność. Miary te, jak się wydaje, mają większą wartość poznawczą, ponieważ umożliwiają „pozycjonowanie” badanego przedsiębiorstwa w grupie podmiotów zaliczanych do danego działu. Jest to zu- pełnie inna jakościowo informacja niż tylko średnia arytmetyczna czy mediana1. Wskaźniki sektorowe publikowane na łamach czasopisma „Rachunkowość” liczone są dla szerokiego grona podmiotów gospodarczych sporządzających sprawozdania finanso- we według Ustawy o rachunkowości (Dudycz, Hamrol, Skoczylas, Niemiec 2005; Dudycz, Skoczylas 2005, 2006, 2007, 2008, 2009, 2010, 2011). Artykuł jest odpowiedzią na istniejące zapotrzebowanie na średnie sektorowe oblicza- ne na podstawie MSR. Do badania wybrano wszystkie spółki publiczne funkcjonujące na Giełdzie Papierów Wartościowych w Polsce z sektorów niefinansowych. Grupa badawcza objęła około 500 spółek publicznych2. 168 Edyta Mioduchowska-Jaroszewicz, Katarzyna Hawran, Joanna Kowalik 1. Charakterystyka wskaźników sektorowych Wskaźniki sektorowe wybrano do badania ze względu na cztery kryteria oceny sytuacji finansowej: rentowność, płynność finansową, zadłużenie i ryzyko. Pierwsza grupa wskaźników zawiera: Pierwsza grupa wskaźników zawiera: –– rentowność operacyjną aktywów, –– rentowność kapitału własnego, –– rentowność netto sprzedaży, –– rentowność sprzedaży działalności podstawowej, –– rentowność ekonomiczną sprzedaży. Druga grupa wskaźników oceniająca płynność finansową w ujęciu statycznym i dyna- micznym i poziom zadłużenia dotyczy:i –– bieżącej płynności finansowej, –– płynności szybkiej, –– płynności gotówkowej, –– okresu spłaty należności krótkoterminowych, –– okresu spłaty zobowiązań, –– szybkości obrotu zapasami, –– stopnia pokrycia aktywów trwałych kapitałem własnym i rezerwami długotermino- wymi,i –– stopnia pokrycia aktywów trwałych kapitałem własnym i rezerwami długotermino- wymi, –– trwałości struktury finansowania, –– wskaźników wydajności gotówkowej sprzedaży, –– wydajności gotówkowej majątku, –– ogólnego zadłużenia. Ryzyko finansowe i operacyjne ocenia grupa wskaźników rentowności, płynności fi- nansowej i zadłużenia. Wszystkie formuły wskaźników zaprezentowano w tabeli 1. Mimo że są powszechnie znane, istnieje potrzeba przedstawiania w każdym badaniu formuł wskaźników, aby otrzymane wyniki mogły być jednoznacznie interpretowane. Wskaźniki w tabeli 1 są dobrane w ten sposób, aby oceniać wszystkie obszary działal- ności przedsiębiorstwa, przede wszystkim efektywność gospodarowania w ujęciu memo- riałowym (wskaźniki rentowności) i kasowym (wskaźniki wydajności gotówkowej), dłu- go- i krótkoterminowej płynności finansowej w ujęciu statycznym (wskaźniki płynności finansowej, stopień pokrycia aktywów trwałych kapitałem własnym i rezerwami długo- terminowymi, wskaźnik trwałości struktury finansowania, okres spłaty należności krót- koterminowych, okres spłaty zobowiązań oraz szybkość obrotu zapasami) i dynamicznym (wskaźniki wydajności gotówkowej) oraz poziom zadłużenia (wskaźnik ogólnego zadłuże- nia). 169 Wskaźniki sektorowe w ocenie spółek publicznych na podstawie sprawozdań finansowych... Tabela 1 Formuły wskaźników sektorowych Nazwa wskaźnika Formuła Wskaźnika Rentowność operacyjna aktywów zysk operacyjny/średnie aktywa. Rentowność kapitału własnego zysk netto/średni kapitał własny. Rentowność netto sprzedaży zysk netto/przychody ogółem. Rentowność sprzedaży działalności podstawowej zysk na sprzedaży/przychody ze sprzedaży. Rentowność ekonomiczna sprzedaży (zysk operacyjny + amortyzacja)/(przychody ze sprzedaży + pozostałe przychody operacyjne). Bieżąca płynność finansowej (aktywa obrotowe – należności z tytułu dostaw i usług pow. 12 m-cy)/zadłużenie krótkoterminowe. Płynność szybka (inwestycje krótkoterminowe + należności krótkoterminowe – należności z tytułu dostaw i usług pow. 12 m-cy)/zadłuże- nie krótkoterminowe. Płynność gotówkowa inwestycje krótkoterminowe/zadłużenie krótkoterminowe. Okres spłaty należności krótkoterminowych (średnie należności krótkoterminowe × 365)/przychody ze sprzedaży. Okres spłaty zobowiązań (średnie zobowiązania krótkoterminowe × 365)/przychody ze sprzedaży. Szybkość obrotu zapasami (średnie zapasy × 365)/przychody ze sprzedaży. Stopień pokrycia aktywów trwałych kapitałem własnym i rezerwami długoterminowymi (kapitał własny + rezerwy długoterminowe)/(aktywa trwałe + należności z tytułu dostaw i usług pow. Tabela 1 (średnie zapasy × 365)/przychody ze sprzedaży. (kapitał własny + rezerwy długoterminowe)/(aktywa trwałe + należności z tytułu dostaw i usług pow. 12 m-cy). (kapitał własny + zobowiązania długoterminowe + rezerwy długoterminowe)/(aktywa trwałe + należności z tytułu dostaw i usług pow. 12 m-cy). zadłużenie ogółem/pasywa razem. przepływy operacyjne/przychody ze sprzedaży. Wydajność gotówkowa sprzedaży przepływy operacyjne/przychody ze sprzedaży. przepływy operacyjne/średnia aktywa. Wydajność gotówkowa majątku 1. Charakterystyka wskaźników sektorowych 12 m-cy). Trwałość struktury finansowania (kapitał własny + zobowiązania długoterminowe + rezerwy długoterminowe)/(aktywa trwałe + należności z tytułu dostaw i usług pow. 12 m-cy). Ogólne zadłużenie zadłużenie ogółem/pasywa razem. Wydajność gotówkowa sprzedaży przepływy operacyjne/przychody ze sprzedaży. Wydajność gotówkowa majątku przepływy operacyjne/średnia aktywa. Źródło: opracowanie własne na podstawie Waśniewski, Skoczylas (2005), s. 73–76. 2. Międzynarodowe Standardy Rachunkowości jako źródło informacji finansowej o sytuacji finansowej przedsiębiorstw Międzynarodowe Standardy Rachunkowości (MSR) wyznaczają ramy rachunkowości, umożliwiając stosowanie jednolitych zasad jej prowadzenia na skalę ogólnoświatową, za- pewniając porównywalność sprawozdań finansowych sporządzonych na ich podstawie. MSR stosują emitenci papierów wartościowych dopuszczonych do obrotu na regulowanych rynkach oraz banki. Należy też wspomnieć, że MSR stosowane są do sporządzania skonso- lidowanego sprawozdania (Gos 2007, s. 481–502). Nie można porównywać sprawozdań finansowych sporządzanych według Ustawy o Ra- chunkowości (UoR) i według Międzynarodowych Standardów Sprawozdawczości Finan- sowej (MSSF), gdyż sporządzane są one według dwóch różnych regulacji prawnych. Za- wartość bilansu nie jest tożsama z zawartością sprawozdania z sytuacji finansowej, treść 170 Edyta Mioduchowska-Jaroszewicz, Katarzyna Hawran, Joanna Kowalik rachunku zysków i strat nie jest tożsama z treścią sprawozdania z dochodów. Rachunek przepływów pieniężnych nie jest sprawozdaniem z przepływów pieniężnych, a zestawienie zmian w kapitale własnym nie jest sprawozdaniem ze zmian kapitale własnym. W tabeli 2 zestawiono bilans i sprawozdanie z sytuacji finansowej w celu porównania ich treści. Źródło: opracowanie własne na podstawie przykładowych sprawozdań finansowych przedsiębiorstw zgodnie z przepisami UoR obowiązującymi do dnia 31.12.2015. Tabela 2 Tabela 2 Bilans a sprawozdanie z sytuacji finansowej Bilans a sprawozdanie z sytuacji finansowej Bilans Sprawozdanie z sytuacji finansowej Aktywa A. Aktywa trwałe –– wartości niematerialne i prawne –– rzeczowe aktywa trwale –– należności długoterminowe –– inwestycje długoterminowe –– długoterminowe rozliczenia międzyokresowe Aktywa trwałe –– rzeczowe aktywa trwałe –– środki trwałe w budowie –– wartości niematerialne i prawne oraz wartość firmy –– inwestycje długoterminowe –– rozliczenia międzyokresowe B. Aktywa obrotowe –– zapasy –– należności krótkoterminowe –– inwestycje krótkoterminowe –– krótkoterminowe rozliczenia międzyokresowe Aktywa obrotowe –– zapasy –– należności handlowe i inne –– zaliczki, rozliczenia międzyokresowe i pozostałe aktywa –– gotówka i ekwiwalenty Pasywa A. Kapitał własny –– kapitał (fundusz) podstawowy –– należne lecz niewniesione, wkłady na poczet kapitału podstawowego(wielkość ujemna) –– udziały (akcje) własne (wielkość ujemna) –– kapitał (fundusz) zapasowy –– kapitał (fundusz) z aktualizacji wyceny –– pozostałe kapitały (fundusze) rezerwowe –– zysk (strata) z lat ubiegłych –– zysk (strata) netto –– odpisy z zysku netto w ciągu roku obrotowego (wielkość ujemna) Kapitał własny –– wyemitowany kapitał –– pozostałe składowe kapitału własnego –– rezerwa z tytułu aktualizacji wyceny –– zyski zatrzymane –– wynik finansowy (zysk/strata) B. Zobowiązania i rezerwy na zobowiązania –– rezerwy na zobowiązania –– zobowiązania długoterminowe –– zobowiązania krótkoterminowe –– rozliczenia międzyokresowe Kapitał obcy zobowiązania długoterminowe –– pozostałe długoterminowe zobowiązania finansowe –– rezerwy z tytułu pozostałych zobowiązań i obciążeń –– pozostałe zobowiązania długoterminowe zobowiązania krótkoterminowe –– pozostałe finansowe zobowiązania krótkookresowe –– zobowiązania handlowe i inne –– krótkoterminowe rozliczenia międzyokresowe –– rezerwy z tytułu pozostałych zobowiązań i obciążeń –– pozostałe zobowiązania krótkoterminowe –– zobowiązania z tytułu bieżącego podatku dochodowego Źródło: opracowanie własne na podstawie przykładowych sprawozdań finansowych przedsiębiorstw zgodnie z przepisami UoR obowiązującymi do dnia 31.12.2015. p y zobowiązania długoterminowe –– pozostałe długoterminowe zobowiązania finansowe –– rezerwy z tytułu pozostałych zobowiązań i obciążeń –– pozostałe zobowiązania długoterminowe zobowiązania krótkoterminowe –– pozostałe finansowe zobowiązania krótkookresowe –– zobowiązania handlowe i inne –– krótkoterminowe rozliczenia międzyokresowe –– rezerwy z tytułu pozostałych zobowiązań i obciążeń –– pozostałe zobowiązania krótkoterminowe –– zobowiązania z tytułu bieżącego podatku dochodowego Źródło: opracowanie własne na podstawie przykładowych sprawozdań finansowych przedsiębiorstw zgodnie z przepisami UoR obowiązującymi do dnia 31.12.2015. 171 Wskaźniki sektorowe w ocenie spółek publicznych na podstawie sprawozdań finansowych... Porównanie pozycji z podstawowego sprawozdania finansowego, czyli bilansu, i spra- wozdania z sytuacji finansowej sporządzanego według UoR i MSSF wskazuje na brak moż- liwości przenoszenia i przyporządkowania pozycji bilansu utworzonego według MSSR do bilansu utworzonego według UoR. Tabela 2 Powodów braku porównywalności tak sporządzanych sprawozdań jest wiele. Międzynarodowe Standardy Sprawozdawczości Finansowej nie dają w pełni jedno- znacznych wytycznych co do sposobów wyceny czy ujmowania transakcji. Mimo że od dawna trwają prace nad standaryzacją rachunkowości, to nadal w wielu obszarach dopusz- czane są różne rozwiązania. Nawet układ sprawozdań finansowych nie jest ściśle określony, co stanowi regres w stosunku do ustawy o rachunkowości, a w szczególności do przepisów obowiązujących polskie spółki publiczne. Standardy formułują jedynie minimalne wyma- gania, pozostawiając kierownictwu jednostki swobodę w zakresie kształtowania polityki rachunkowości oraz dostosowania struktury sprawozdań finansowych do indywidualnych potrzeb. Natomiast zakres informacji dodatkowej jest podporządkowany wymogom po- szczególnych standardów. Dokonując porównań regulacji w zakresie Ustawy o Rachunkowości i Międzynarodo- wych Standardów Sprawozdawczości Finansowej można stwierdzić, że istnieje potrzeba ujednolicenia formy i nazw pozycji występujących w sprawozdaniach finansowych w obu regulacjach oraz stworzenia średnich sektorowych na podstawie danych publikowanych sprawozdań finansowych według MSFF, bo te liczone na podstawie UoR już istnieją. Uwagi końcowe Porównania w stosowaniu analizy wskaźnikowej są podstawą wyprowadzania wniosków dotyczących ocenianego podmiotu gospodarczego, dlatego ważne jest, aby istniały bazy porównawcze, do których użytkownicy sprawozdań finansowych mogliby odnieść się do- konując oceny sytuacji finansowej przedsiębiorstwa. Celem artykułu było przedstawienie wyników z pomiarów średnich sektorowych dla spółek publicznych funkcjonujących na GPW w Polsce i ten cel został zrealizowany. Na koniec wnioski dotyczące porównywalności i znaczeniu innych miar statystycznych niż sama średnia wykorzystywanych w pomiarze średnich sektorowych. M. Hamrol (2013) stwierdził, że „wartość poznawcza wyników, uzyskanych w rezultacie przeprowadzonych porównań, jest uzależniona od stopnia zachowania porównywalności wskaźników, która może być zakłócana różnicami natury metodologicznej. Dotyczy to przede wszystkim po- równań z innym przedsiębiorstwem (przedsiębiorstwami), zwłaszcza w sytuacji, kiedy ko- rzystamy z «gotowych» wskaźników, bez wystarczającej wiedzy co do sposobu ich liczenia. Pamiętać bowiem należy, że brak jest standardów, tak krajowych, jak i międzynarodowych, w zakresie budowy wskaźników finansowych. Z kolei w przypadku porównań sektoro- wych trzeba mieć na uwadze fakt, że rozkład wartości obserwacji (wskaźników) rzadko ma charakter normalny. Operowanie zatem średnią arytmetyczną (lub ważoną), która jest 172 Edyta Mioduchowska-Jaroszewicz, Katarzyna Hawran, Joanna Kowalik podstawowym narzędziem służącym do obliczenia wskaźników sektorowych (branżo- wych), prowadzić może do fałszywych wniosków. Konieczne jest zatem wykorzystanie narzędzi statystyki opisowej, umożliwiających bardziej precyzyjne usytuowanie badanego przedsiębiorstwa w otoczeniu sektorowym”. Dlatego przedstawione w artykule wskaźniki sektorowe i publikowane w „Rachunkowości” uwzględniają nie tylko średnia arytmetycz- ną, ale medianę, odchylenie standardowe, minimalną i maksymalną wartość wskaźnika. W załączniku przedstawiono wskaźniki sektorowe zawarte w 10 tabelach, obliczone dla następujących sektorów: 1. Energetyczny i paliwowy (12 spółek). 2. Hotele i restauracje (4 spółki). 3. Telekomunikacyjny (8 spółek). 4. Handel detaliczny (18 spółek). 5. Deweloperski (20 spółek). 6. Spożywczy (13 spółek). 7. Tworzyw sztucznych (4 spółki). 8. Motoryzacyjny (6 spółek). 9. Informatyczny (25 spółek). 10. Budownictwo (12 spółek). Pozostałe sektory w opracowaniu. Pozostałe sektory w opracowaniu. Załącznik Średnie sektorowe obliczone dla spółek giełdowych sektora hotele i restauracje Nazwa wskaźnika Średnia Mediana Max wartość Min wartość Odchylenie standardowe 2013 2014 2013 2014 2013 2014 2013 2014 2013 2014 Rentowność operacyjna aktywów 0,03 0,03 0,02 0,00 0,15 0,12 –0,04 –0,06 0,06 0,05 Rentowność kapitału własnego 0,07 0,07 0,08 0,07 0,14 0,15 –0,04 –0,10 0,06 0,07 Rentowność netto sprzedaży 0,15 0,16 0,13 0,13 0,48 0,60 –0,03 –0,08 0,14 0,18 Rentowność sprzedaży działalności podstawowej –0,07 –0,15 0,02 0,02 0,25 0,21 –1,25 –1,80 0,45 0,63 Rentowność ekonomiczna sprzedaży 0,04 0,01 0,09 0,11 0,61 0,54 –1,03 –1,24 0,44 0,49 Bieżąca płynności finansowej 1,56 2,14 1,47 1,33 2,58 5,42 0,68 0,49 0,73 1,70 Płynność szybka 1,25 1,90 1,16 0,72 2,20 5,42 0,46 0,41 0,71 1,85 Płynność gotówkowa 0,70 1,18 0,73 0,39 2,05 5,28 0,00 0,00 0,65 1,68 Okres spłaty należności krótkoter- minowych 62 83 47 51 144 281 30 32 35 77 Okres spłaty zobowiązań 505 414 84 113 3834 2949 40 44 1250 952 Szybkość obrotu zapasami 28 30 16 22 83 93 2 4 27 29 Stopień pokrycia aktywów trwałych kapitałem własnym i rezerwami długoterminowymi 0,93 0,84 0,91 0,90 1,38 1,11 0,58 0,58 0,22 0,18 Trwałość struktury finansowania 1,06 1,04 1,06 1,04 1,38 1,19 0,92 0,82 0,14 0,10 Ogólne zadłużenie 0,29 0,31 0,30 0,28 0,48 0,55 0,11 0,03 0,12 0,16 Wydajność gotówkowa sprzedaży 0,00 –0,10 0,06 0,04 0,37 0,35 –1,19 –1,85 0,46 0,67 Wydajność gotówkowa majątku 0,07 0,06 0,05 0,02 0,19 0,17 –0,01 –0,01 0,07 0,07 Źródło: opracowanie własne. 2. Średnie sektorowe obliczone dla spółek giełdowych sektora hotele i restauracje bela Z3. Średnie sektorowe obliczone dla spółek giełdowych sektora telekomunikacyjnego Tabela Z3. Załącznik Tabela Z1. Średnie sektorowe obliczone dla spółek giełdowych sektora energetycznego i paliwowego Nazwa wskaźnika Średnia Mediana Max wartość Min wartość Odchylenie standardowe 2013 2014 2013 2014 2013 2014 2013 2014 2013 2014 Rentowność operacyjna aktywów 0,10 0,11 0,08 0,09 0,24 0,27 0,00 0,00 0,09 0,10 Rentowność kapitału własnego 0,03 –0,03 0,03 0,03 0,07 0,05 0,01 –0,23 0,02 0,12 Rentowność netto sprzedaży 0,04 0,04 0,03 0,04 0,09 0,12 0,00 –0,02 0,04 0,05 Rentowność sprzedaży działalności podstawowej 0,16 0,16 0,17 0,18 0,29 0,30 0,00 0,00 0,11 0,11 Rentowność ekonomiczna sprzedaży 0,15 0,14 0,12 0,12 0,28 0,29 0,08 0,01 0,08 0,10 Bieżąca płynności finansowej 0,92 0,95 0,60 0,71 2,30 2,27 0,15 0,13 0,82 0,82 Płynność szybka 0,80 0,34 0,45 0,26 2,18 0,75 0,12 0,09 0,81 0,25 Płynność gotówkowa 0,73 0,28 0,36 0,17 2,11 0,73 0,09 0,05 0,80 0,27 Okres spłaty należności krótkoterminowych 32 26 25 22 57 46 20 14 15 13 Okres spłaty zobowiązań 78 64 75 56 100 92 61 53 15 16 Szybkość obrotu zapasami 13 4 4 5 40 6 2 2 16 2 Stopień pokrycia aktywów trwałych kapitałem własnym i rezerwami długoterminowymi 0,73 0,74 0,69 0,70 1,05 1,10 0,48 0,44 0,23 0,26 Trwałość struktury finansowania 0,96 1,01 0,96 1,00 1,07 1,12 0,84 0,94 0,08 0,07 Ogólne zadłużenie 0,37 0,39 0,40 0,41 0,61 0,67 0,07 0,08 0,24 0,26 Wydajność gotówkowa sprzedaży 0,16 0,14 0,14 0,10 0,27 0,31 0,09 0,06 0,07 0,10 Wydajność gotówkowa majątku 0,10 0,10 0,09 0,11 0,18 0,14 0,05 0,03 0,05 0,04 Źródło: opracowanie własne. 173 Wskaźniki sektorowe w ocenie spółek publicznych na podstawie sprawozdań finansowych... Tabela Z2. Załącznik Średnie sektorowe obliczone dla spółek giełdowych sektora – handel detaliczny Nazwa wskaźnika Średnia Mediana Max wartość Min wartość Odchylenie standardowe 2013 2014 2013 2014 2013 2014 2013 2014 2013 2014 Rentowność operacyjna aktywów 0,05 0,08 0,05 0,04 0,06 0,04 0,05 –0,27 0,01 0,05 Rentowność kapitału własnego 0,09 0,06 0,09 0,09 0,09 0,09 0,08 –0,46 0,00 0,04 Rentowność netto sprzedaży 0,01 –0,01 0,01 0,03 0,01 0,03 0,01 –0,07 0,00 0,08 Rentowność sprzedaży działalności podstawowej 0,07 0,36 0,07 0,24 0,08 0,24 0,07 –0,06 0,01 0,64 Rentowność ekonomiczna sprzedaży 0,01 –0,06 0,01 0,07 0,01 0,07 0,01 –4,93 0,00 0,84 Bieżąca płynności finansowej 1,00 1,51 1,00 1,20 1,00 1,20 0,99 0,56 0,01 12,20 Płynność szybka 0,50 0,64 0,50 0,50 0,53 0,50 0,48 0,01 0,03 7,53 Płynność gotówkowa 0,00 0,25 0,00 0,09 0,00 0,09 0,00 0,00 0,00 9,50 Okres spłaty należności krótkoter- minowych 26,15 1051,4 26,15 33,98 28,91 33,98 23,39 2,62 3,90 3825,33 Okres spłaty zobowiązań 48,38 1116,9 48,38 91,07 51,38 91,07 45,38 31,67 4,25 3781,52 Szybkość obrotu zapasami 18,73 395,17 18,73 57,03 18,91 57,03 18,55 0,00 0,25 1337,40 Stopień pokrycia aktywów trwałych kapitałem własnym i rezerwami długoterminowymi 1,02 3,10 1,02 1,04 1,03 1,04 1,01 0,25 0,01 1,24 Trwałość struktury finansowania 1,06 3,46 1,06 1,18 1,07 1,18 1,05 0,25 0,01 0,32 Ogólne zadłużenie 1,90 1,70 1,90 0,81 2,01 0,81 1,78 0,01 0,16 1,86 Wydajność gotówkowa sprzedaży 0,01 –0,16 0,01 0,03 0,01 0,03 0,01 –2,32 0,00 1,63 Wydajność gotówkowa majątku 0,05 0,03 0,05 0,04 0,06 0,04 0,05 –0,13 0,01 0,07 Źródło: opracowanie własne. 4. Średnie sektorowe obliczone dla spółek giełdowych sektora – handel detaliczny Tabela Z5. Załącznik Średnie sektorowe obliczone dla spółek giełdowych sektora telekomunikacyjnego Nazwa wskaźnika Średnia Mediana Max wartość Min wartość Odchylenie standardowe 2013 2014 2013 2014 2013 2014 2013 2014 2013 2014 Rentowność operacyjna aktywów –0,17 0,00 0,00 0,01 0,03 0,03 –1,39 –0,03 0,49 0,02 Rentowność kapitału własnego –1,20 –0,14 0,01 0,01 0,12 0,12 –9,85 –1,39 3,49 0,51 Rentowność netto sprzedaży –0,06 –0,39 0,01 0,10 0,52 0,47 –0,61 –4,25 0,36 1,57 Rentowność sprzedaży działalności pod- stawowej –0,11 –0,66 0,01 –0,01 0,14 0,46 –0,70 –5,49 0,28 1,97 Rentowność ekonomiczna sprzedaży 0,02 0,12 0,15 0,09 0,36 0,69 –0,65 –0,31 0,35 0,31 Bieżąca płynności finansowej 65,6 7,56 0,57 0,85 522 50,72 –1,13 0,23 184,4 17,5 Płynność szybka 65,6 7,48 0,48 0,64 521,9 50,7 –1,14 0,21 184,4 17,5 Płynność gotówkowa 17,3 0,34 0,10 0,25 137,6 0,78 0,01 0,01 48,6 0,32 Okres spłaty należności krótkoterminowych 1148 847 237 221 4352 3020 3 2 1612 1167 Okres spłaty zobowiązań 1149 764 270 203 4867 2327 50 5 1693 946 Szybkość obrotu zapasami 0,98 0,85 0,00 0,02 6,11 5,45 0,00 0,00 2,12 1,88 Stopień pokrycia aktywów trwałych kapita- łem własnym i rezerwami długotermi- nowymi 0,71 0,87 0,81 0,88 1,11 1,33 –0,11 0,23 0,40 0,36 Trwałość struktury finansowania 0,80 0,97 0,95 0,98 1,20 1,33 –0,11 0,23 0,42 0,35 Ogólne zadłużenie 0,41 0,25 0,30 0,22 1,23 0,67 0,08 0,00 0,36 0,21 Wydajność gotówkowa sprzedaży –1,32 0,03 0,00 0,06 0,82 0,62 –10,8 –0,68 3,86 0,41 Wydajność gotówkowa majątku –0,01 0,03 –0,01 0,01 0,13 0,14 –0,17 –0,03 0,10 0,06 Źródło: opracowanie własne. Edyta Mioduchowska-Jaroszewicz, Katarzyna Hawran, Joanna Kowalik 174 Tabela Z4. Źródło: opracowanie własne. Załącznik Średnie sektorowe obliczone dla spółek giełdowych sektora deweloperskiego Nazwa wskaźnika Średnia Mediana Max wartość Min wartość Odchylenie standardowe 2013 2014 2013 2014 2013 2014 2013 2014 2013 2014 Rentowność operacyjna aktywów 0,02 0,02 0,02 0,02 0,07 0,08 –0,02 –0,04 0,03 0,03 Rentowność kapitału własnego –0,01 –0,03 0,02 0,01 0,09 0,07 –0,25 –0,30 0,09 0,12 Rentowność netto sprzedaży –0,23 0,06 0,06 0,04 0,50 4,14 –3,17 –3,05 0,87 1,18 Rentowność sprzedaży działalności podstawowej 0,15 0,10 0,09 0,12 1,00 1,00 –0,36 –0,87 0,32 0,37 Rentowność ekonomiczna sprzedaży 1,23 –1,81 0,11 0,09 19,03 2,42 –0,67 –41,23 4,18 9,07 Bieżąca płynności finansowej 5,05 4,52 2,33 1,98 35,45 33,54 0,24 0,21 7,96 7,15 Płynność szybka 2,60 2,06 1,45 1,28 16,88 9,34 0,00 0,00 3,77 2,34 Płynność gotówkowa 1,08 0,96 0,40 0,54 4,11 4,35 0,00 0,00 1,27 1,11 Okres spłaty należności krótkoterminowych 1 609 3 207 213 209 25 745 56 973 0,00 0 5 548 12 342 Okres spłaty zobowiązań 1 861 3 458 581 576 1 545 26 727 86 91 3 682 7 358 Szybkość obrotu zapasami 676 842 570 667 2242 2830 0,00 0 645 781 Stopień pokrycia aktywów trwałych kapita- łem własnym i rezerwami długotermi- nowymi 5,44 9,15 1,06 1,02 31,81 94,96 0,58 0,58 10,10 22,17 Trwałość struktury finansowania 6,65 10,56 1,27 1,31 48,06 100,52 0,77 0,81 12,75 24,74 Ogólne zadłużenie 0,60 3,00 0,32 0,31 4,23 50,22 0,02 0,01 0,93 10,86 Wydajność gotówkowa sprzedaży –0,19 –0,45 –0,12 0,07 0,53 0,67 –1,89 –5,16 0,58 1,49 Wydajność gotówkowa majątku 0,01 0,01 0,00 0,00 0,11 0,12 –0,03 –0,05 0,03 0,03 Źródło: opracowanie własne. Tabela Z5. Średnie sektorowe obliczone dla spółek giełdowych sektora deweloperskiego Wskaźniki sektorowe w ocenie spółek publicznych na podstawie sprawozdań finansowych... 175 Tabela Z6. Załącznik Średnie sektorowe obliczone dla spółek giełdowych sektora spożywczego Nazwa wskaźnika Średnia Mediana Max wartość Min wartość Odchylenie standardowe 2013 2014 2013 2014 2013 2014 2013 2014 2013 2014 Rentowność operacyjna aktywów 0,04 0,06 0,04 0,05 0,21 0,19 –0,18 0,01 0,09 0,05 Rentowność kapitału własnego 0,08 0,12 0,09 0,08 0,91 0,56 –0,27 0,01 0,29 0,14 Rentowność netto sprzedaży 0,01 0,04 0,01 0,04 0,13 0,15 –0,16 0,00 0,07 0,04 Rentowność sprzedaży działalności podstawowej 0,10 0,10 0,05 0,06 0,38 0,41 –0,01 –0,01 0,12 0,12 Rentowność ekonomiczna sprzedaży 0,05 0,07 0,06 0,06 0,18 0,20 –0,18 –0,01 0,08 0,05 Bieżąca płynności finansowej 1,23 1,22 1,11 1,33 2,57 4,01 0,29 –4,37 0,63 2,00 Płynność szybka 0,67 0,60 0,69 0,95 2,20 2,74 –0,40 –5,63 0,64 2,01 Płynność gotówkowa 0,14 0,23 0,08 0,05 0,58 1,03 0,00 0,01 0,18 0,36 Okres spłaty należności krótkotermino- wych 51 53 47 39 113 116 0 0 42 43 Okres spłaty zobowiązań 104 96 105 98 280 294 3 2 68 72 Szybkość obrotu zapasami 35 35 33 33 76 93 0 0 24 26 Stopień pokrycia aktywów trwałych kapitałem własnym i rezerwami długoterminowymi 0,90 0,92 0,84 0,86 1,90 2,20 0,18 0,27 0,50 0,61 Trwałość struktury finansowania 1,12 1,20 1,02 1,16 1,97 2,22 0,58 0,34 0,39 0,53 Ogólne zadłużenie 0,44 0,39 0,45 0,44 0,96 0,90 –0,08 –0,26 0,27 0,29 Wydajność gotówkowa sprzedaży 0,04 0,05 0,03 0,03 0,14 0,17 –0,04 0,00 0,05 0,06 Wydajność gotówkowa majątku 0,05 0,07 0,06 0,06 0,16 0,19 –0,05 0,00 0,07 0,06 Źródło: opracowanie własne. Tabela Z6. Średnie sektorowe obliczone dla spółek giełdowych sektora spożywczego abela Z7. Średnie sektorowe obliczone dla spółek giełdowych sektora tworzyw sztucznych Tabela Z7. Załącznik Średnie sektorowe obliczone dla spółek giełdowych sektora tworzyw sztucznych Nazwa wskaźnika Średnia Mediana Max wartość Min wartość Odchylenie standardowe 2013 2014 2013 2014 2013 2014 2013 2014 2013 2014 Rentowność operacyjna aktywów 0,06 0,05 0,06 0,06 0,10 0,08 0,00 0,03 0,04 0,02 Rentowność kapitału własnego 0,08 0,30 0,09 0,11 0,16 0,95 –0,02 0,02 0,07 0,44 Rentowność netto sprzedaży 0,06 0,06 0,06 0,03 0,12 0,17 –0,01 0,01 0,05 0,07 Rentowność sprzedaży działalności pod- stawowej 0,11 0,09 0,07 0,04 0,27 0,29 0,00 0,00 0,12 0,13 Rentowność ekonomiczna sprzedaży 0,11 0,09 0,12 0,09 0,17 0,15 0,02 0,04 0,07 0,05 Bieżąca płynności finansowej 1,34 1,31 1,38 1,23 1,46 1,66 1,15 1,13 0,14 0,24 Płynność szybka 0,77 0,76 0,75 0,75 0,94 0,86 0,64 0,67 0,15 0,10 Płynność gotówkowa 0,09 0,07 0,04 0,03 0,30 0,22 0,00 0,00 0,14 0,10 Okres spłaty należności krótkoterminowych 73 71 78 72 100 75 37 65 27 5 Okres spłaty zobowiązań 110 94 112 90 125 113 89 82 15 14 Szybkość obrotu zapasami 55 46 47 42 96 70 31 29 29 19 Stopień pokrycia aktywów trwałych kapita- łem własnym i rezerwami długotermi- nowymi 0,98 0,95 0,96 0,95 1,20 1,11 0,79 0,79 0,21 0,19 Trwałość struktury finansowania 1,26 1,20 1,25 1,18 1,47 1,37 1,09 1,07 0,16 0,14 Ogólne zadłużenie 0,49 0,49 0,49 0,49 0,56 0,53 0,41 0,43 0,06 0,04 Wydajność gotówkowa sprzedaży 0,13 0,06 0,12 0,07 0,23 0,10 0,05 0,03 0,09 0,03 Wydajność gotówkowa majątku 0,11 0,06 0,11 0,06 0,16 0,07 0,06 0,04 0,05 0,02 Źródło: opracowanie własne. Edyta Mioduchowska-Jaroszewicz, Katarzyna Hawran, Joanna Kowalik 176 Tabela Z8. Załącznik Średnie sektorowe obliczone dla spółek giełdowych sektora informatycznego Nazwa wskaźnika Średnia Mediana Max wartość Min wartość Odchylenie standardowe 2013 2014 2013 2014 2013 2014 2013 2014 2013 2014 Rentowność operacyjna aktywów 0,07 0,07 0,06 0,06 0,27 0,22 –0,27 –0,14 0,09 0,07 Rentowność kapitału własnego 0,05 0,08 0,10 0,06 0,44 0,30 –1,74 –0,20 0,37 0,11 Rentowność netto sprzedaży 0,08 0,04 0,05 0,03 0,41 0,19 –0,23 –0,24 0,12 0,08 Rentowność sprzedaży działalności pod- stawowej 0,19 0,17 0,14 0,10 0,88 0,89 –0,18 –0,21 0,25 0,22 Rentowność ekonomiczna sprzedaży 0,14 0,13 0,11 0,10 0,40 0,44 –0,20 –0,08 0,13 0,12 Bieżąca płynności finansowej 3,45 3,74 2,08 2,10 23,12 26,40 0,83 –1,05 4,30 5,12 Płynność szybka 2,96 3,29 1,35 1,68 20,99 22,77 0,16 –0,99 4,04 5,11 Płynność gotówkowa 1,04 1,16 0,58 0,34 5,79 6,85 0,00 –0,26 1,49 2,03 Okres spłaty należności krótkoterminowych 92 92 76 101 200 157 4 2 50 43 Okres spłaty zobowiązań 102 108 88 81 286 481 13 29 60 89 Szybkość obrotu zapasami 35 41 12 12 478 747 1 1 91 142 Stopień pokrycia aktywów trwałych kapita- łem własnym i rezerwami długotermi- nowymi 2,27 1,45 1,25 1,21 25,34 3,98 0,15 0,63 4,68 0,74 Trwałość struktury finansowania 2,37 1,56 1,36 1,34 25,49 3,98 0,16 0,84 4,69 0,73 Ogólne zadłużenie 0,32 0,32 0,27 0,32 0,81 0,58 0,11 0,09 0,18 0,13 Wydajność gotówkowa sprzedaży 0,13 0,07 0,11 0,06 0,46 0,33 –0,21 –0,51 0,13 0,15 Wydajność gotówkowa majątku 0,10 0,09 0,11 0,10 0,36 0,36 –0,27 –0,09 0,12 0,10 Źródło: opracowanie własne. a Z8. Średnie sektorowe obliczone dla spółek giełdowych sektora informatycznego Tabela Z9. Źródło: opracowanie własne. Załącznik Średnie sektorowe obliczone dla spółek giełdowych sektora budowlanego Nazwa wskaźnika Średnia Mediana Max wartość Min wartość Odchylenie standardowe 2013 2014 2013 2014 2013 2014 2013 2014 2013 2014 Rentowność operacyjna aktywów 0,03 0,00 0,02 0,03 0,21 0,06 –0,06 –0,14 0,06 0,06 Rentowność kapitału własnego 0,12 0,04 0,05 0,04 0,49 0,09 –0,15 –0,04 0,17 0,04 Rentowność netto sprzedaży 0,08 0,03 0,06 0,02 0,34 0,11 –0,06 –0,02 0,10 0,03 Rentowność sprzedaży działalności pod- stawowej –0,11 –0,21 0,03 0,03 0,07 0,81 –1,66 –3,51 0,47 1,02 Rentowność ekonomiczna sprzedaży 0,02 –0,14 0,04 0,04 0,21 0,07 –0,41 –1,88 0,14 0,53 Bieżąca płynności finansowej 2,48 2,34 2,47 2,15 5,34 6,04 1,16 0,79 1,13 1,38 Płynność szybka 1,84 2,00 1,48 1,77 5,31 6,03 0,23 0,37 1,32 1,51 Płynność gotówkowa 0,66 0,68 0,38 0,44 2,65 2,70 0,05 0,01 0,72 0,77 Okres spłaty należności krótkoterminowych 492 199 112 118 4600 1174 79 53 1239 296 Okres spłaty zobowiązań 386 389 109 99 3478 3562 21 24 933 958 Szybkość obrotu zapasami 22 25 10 8 131 191 0 0 35 51 Stopień pokrycia aktywów trwałych kapita- łem własnym i rezerwami długotermi- nowymi 1,34 1,40 0,99 1,18 3,31 4,02 0,75 0,50 0,70 0,90 Trwałość struktury finansowania 1,48 1,54 1,18 1,31 3,31 4,23 0,97 0,80 0,65 0,90 Ogólne zadłużenie 0,35 0,39 0,34 0,36 0,64 0,68 0,15 0,13 0,14 0,16 Wydajność gotówkowa sprzedaży –0,06 –0,16 0,06 0,04 0,17 0,14 –1,27 –2,41 0,37 0,68 Wydajność gotówkowa majątku 0,04 0,04 0,04 0,04 0,15 0,13 –0,08 –0,07 0,07 0,07 Źródło: opracowanie własne. Tabela Z9. Średnie sektorowe obliczone dla spółek giełdowych sektora budowlanego Wskaźniki sektorowe w ocenie spółek publicznych na podstawie sprawozdań finansowych... 177 Tabela Z10. Załącznik Średnie sektorowe obliczone dla spółek giełdowych sektora motoryzacyjnego Wskaźniki sektorowe Średnia Mediana Max wartość Min wartość Odchylenie standardowe 2013 2014 2013 2014 2013 2014 2013 2014 2013 2014 Rentowność operacyjna aktywów 0,06 0,04 0,06 0,05 0,02 0,02 0,02 0,01 0,02 0,02 Rentowność kapitału własnego 0,14 0,13 0,15 0,14 0,22 0,24 0,22 0,24 0,07 0,08 Rentowności netto sprzedaży 0,06 0,06 0,06 0,05 0,10 0,11 0,10 0,11 0,02 0,03 Rentowność ekonomiczna sprzedaży 0,08 0,07 0,08 0,06 0,10 0,10 0,10 0,10 0,01 0,02 Bieżąca płynność finansowa 1,52 1,47 1,47 1,47 1,95 1,76 1,95 1,76 0,29 0,22 Płynność szybka 0,31 0,24 0,13 0,04 0,34 0,47 0,61 0,61 0,29 0,29 Płynność gotówkowa 0,16 0,13 0,07 0,02 0,34 0,47 0,34 0,47 0,15 0,23 Okres spłaty zobowiązań 95,05 88,71 88,42 89,16 137,06 89,91 137,06 89,91 21,54 1,36 Szybkość obrotu zapasami 10,21 9,30 9,34 7,97 17,53 15,09 17,53 15,09 5,62 4,74 Stopień pokrycia aktywów trwałych kapitałem własnym i rezerwami długotermino- wymi 1,39 1,37 1,37 1,31 1,78 1,80 1,78 1,80 0,31 0,30 Trwałość struktury finansowania 1,42 1,40 1,38 1,33 1,87 1,88 1,87 1,88 0,34 0,33 Wydajność gotówkowa sprze- daży 0,08 0,06 0,09 0,05 0,13 0,11 0,13 0,11 0,04 0,05 Wydajność gotówkowa majątku 0,09 0,08 0,10 0,07 0,14 0,14 0,14 0,14 0,05 0,06 Źródło: opracowanie własne. 10. Średnie sektorowe obliczone dla spółek giełdowych sektora motoryzacyjnego Sector Indicators In The Assessment Of Public Companies Based On Financial Statements Presented In Accordance With IAS Abstract: The aim of this article is to present the situation of public companies using sectoral indicators cal- culated on the basis of financial statements prepared in accordance with International Accounting Standards. Keywords: public companies, sectoral indicators Literatura Dudycz T., Hamrol M., Skoczylas W., Niemiec A. (2005). Finansowe wskaźniki sektorowe – pomoc przy analizie finansowej i ocenie zdolności przedsiębiorstwa do kontynuacji działalności. Rachunkowość, 3.i Dudycz T., Skoczylas W. (2010). Ocena sytuacji finansowej sektorów w roku 2008. Rachunkowość, 6.i Dudycz T., Skoczylas W. (2011). Sektorowe wskaźniki finansowe za rok 2009 i ich porównywalności. Rachunko- wość, 4. Dudycz T., Skoczylas W. (2005). Wskaźniki sektorowe za lata 2002 i 2003. Rachunkowość, 3. Dudycz T., Skoczylas W. (2006). Wskaźniki sektorowe za rok 2004. Rachunkowość, 5. Dudycz T., Skoczylas W.(2008). Wskaźniki sektorowe za rok 2006. Rachunkowość, 5. Dudycz T., Skoczylas W. (2009). Wskaźniki sektorowe za rok 2007. Rachunkowość, 6. Dudycz T., Skoczylas W. (2007). Wykorzystanie wskaźników sektorowych do oceny przedsiębiorstwa. Rachunko- wość, 7. Mioduchowska-Jaroszewicz E. (2015). Analiza poziomu zadłużenia przedsiębiorstw na przykładzie „dobrych” i „złych” firm w latach 2010–2012. W: A. Zimny (red.), Wybrane problemy i wyzwania gospodarcze. Ujęcie interdyscyplinarne (s. 203–214). Konin: Państwowa Wyższa Szkoła Zawodowa w Koninie. ( ) p p ę p y i „złych” firm w latach 2010–2012. W: A. Zimny (red.), Wybrane problemy i wyzwania gos interdyscyplinarne (s. 203–214). Konin: Państwowa Wyższa Szkoła Zawodowa w Koninie.i Mioduchowska-Jaroszewicz E. (2015). Diagnoza pozycji finansowej spółek giełdowych metodą quick testu. Studia i Prace Wydziału Nauk Ekonomicznych i Zarządzania, 35, 57–70. Mioduchowska-Jaroszewicz E. (2015). Diagnoza pozycji finansowej spółek giełdowych metodą qui i Prace Wydziału Nauk Ekonomicznych i Zarządzania, 35, 57–70. Mioduchowska-Jaroszewicz E. (2005). Metody i kierunki analizy wypłacalności przedsiębiorstw. Szczecin: Wy- dawnictwo Naukowe Uniwersytetu Szczecińskiego.i dawnictwo Naukowe Uniwersytetu Szczecińskiego Mioduchowska-Jaroszewicz E. (2015). Płynność finansowa w spółkach giełdowych w latach 2010–2012, wsp. E. Ziółkowska, T. Bednarz, w: W: A. Zimny (red.), Wybrane problemy i wyzwania gospodarcze. Ujęcie interdy- scyplinarne (s. 187–202). Konin: Państwowa Wyższa Szkoła Zawodowa w Koninie. Mioduchowska-Jaroszewicz E. (2015). The Level of Depreciation in Cash. W: A. Malina, R. Oczkowska, J. Kacz- marek (red.), Knowledge – Economy – Society. In the Development and Challenges of Modern Economies, Fi- nance and Information Technology (s. 213–224). Krakow: Foundation of the Cracow University of Economics. 178 Edyta Mioduchowska-Jaroszewicz, Katarzyna Hawran, Joanna Kowalik Mioduchowska-Jaroszewicz E. (2015). Zakłócenia porównywalności wykorzystaniu wyników analizy finansowej w podejmowaniu decyzji menedżerskich (konferencja). Kraków–Krynica: Uniwersytet Ekonomiczny w Kra- kowie. W. (red) (2007). Vademecum Głównego Księgowego. Warszawa: Polska Akademia Rachunkowości.i Waśniewski T., Skoczylas W. (2004). Teoria i praktyka analizy finansowej w przedsiębiorstwie. Warszawa: FRR w Polsce. Cytowanie Mioduchowska-Jaroszewicz E., Hawran K., Kowalik J. (2016). Wskaźniki sektorowe w ocenie spółek publicznych na podstawie sprawozdań finansowych przedstawionych według MSR. Finanse, Rynki Finansowe, Ubezpiecze- nia, 4 (82/1), 167–178. DOI: 10.18276/frfu.2016.4.82/1-14.
https://openalex.org/W4389979547	https://www.bio-conferences.org/articles/bioconf/pdf/2023/26/bioconf_icesai2023_00044.pdf	English	null	The effect of isoamylase application in cassava root meal on broiler growth performance, feed retention time, and metabolite profile	Bio web of conferences/BIO web of conferences	2,023	cc-by	5,762	* Corresponding author: nahrowi2504@yahoo.com BIO Web of Conferences 81, 00044 (2023) ICESAI 2023 BIO Web of Conferences 81, 00044 (2023) ICESAI 2023 https://doi.org/10.1051/bioconf/20238100044 The effect of isoamylase application in cassava root meal on broiler growth performance, feed retention time, and metabolite profile Poppy Satya Puspita1, Widya Hermana2, Muhammad Pramujo1, and Nahrowi Nahrowi2* 1Faculty of Animal Science, Universitas Brawijaya, Jl. Veteran, Kota Malang 65145, Indonesia 2Department of Nutrition and Feed Technology, Faculty of Animal Science, IPB University, Jl. Agatis, Kampus IPB Darmaga Bogor 16680, Indonesia Abstract. The objectives of this study were to evaluate isoamylase application on the cassava root meal (CRM) and identify its effect on broiler growth performance, feed retention time, and metabolite profile. The experiment used 270 sexed broilers strain Lohman with an initial body weight of 45.88 ± 0.41 g were allotted into 3 treatments and 6 replications with 15 birds in each pen for 35 days. Dietary treatments were T0: Basal Ration + 0% CRM, T1: Basal Ration + 50% CRM, T2: Basal Ration + 50% CRM + 0.05% isoamylase. The results showed that the inclusion 50% of CRM both with and or without isoamylase was increased (P<0.01) feed intake and feed conversion ratio (FCR) and gave the same effect (P>0.05) on body weight, body weight gain, carcass weight, and carcass percentage during the overall period compared to fed T0. The treatment had no significant difference in coefisien total starch digestibility, metabolizable energy, and blood glucose levels. In conclusion, the inclusion of 50% cassava root meal with and or without isoamylase to subtitute corn in ration does not affect the broiler growth performance however FCR was higher than the control. © The Authors, published by EDP Sciences. This is an open access article distributed under the terms of the Creative Commons Attribution License 4.0 (https://creativecommons.org/licenses/by/4.0/). 1 Introduction An increasing hydrolyzed of branch chains in amylopectin is expected to increase the amylose content. Thus it can increase nutrient availability and broiler growth performance. Amylase was able to increase starch digestibility as well as reduce the presence of glucose as a potential substrate for non-beneficial bacteria at the end of the digestive tract [6]. [7] Found that feed containing 50% peeled cassava meal suplemented by cocktail enzymes (Maxigrain) produced the same production as a ration containing 100% maize. [8] Reported that the addition of carbohydrase and phytase significantly increased the body weight and metabolizable energy in poultry fed cassava pellets and mash. Supplementation of α-1.6 isoamilase in broiler feed increased nutrient digestibility, metabolizable energy, and broiler productivity [9]. This study was designed to evaluate isoamylase application on the CRM and identify its effect on broiler growth performance, feed retention time, and blood glucose level. 2.1 Material The main material in this study used cassava root meal from UD. Setia Flour Product (Bogor, Indonesia) and Isoamylase (EC 3.2.1.68) was from Bacillus licheniformis, purchased from Creative Enzymes (USA), Creative Biomart, Inc. This enzyme has optimal activity at pH around 5.6 - 7.0 and at temperatures between 50 - 55 ° C. Enzyme activity ≥ 10,000 Ug-1. The diet was formulated according to the recommendations of Leeson and Summers (2005) for a type of high-density diet for open cages (Table 1). The ration was given in the form of crumble for the starter period (0-21 days) and pellets for the finisher period (22 - 35 days). Feed and drink were given ad libitum. One-day-old sexed Lohmann (270 chicks), obtained from PT Japfa Comfeed Indonesia Hatchery, vaccinated with New castle Disease, Infectious Bursal Disease, and Avian Influenza, and raised for 35 days. The cages (1 m x 1 m x 0.8 m) were randomly alloted of 3 treatments (6 replications with 15 chicks each). Dietary treatments in this study were T0: Basal ration + 0% CRM (control); T1: Basal ration + 50% CRM; T2: Basal ration + 50% CRM + 0.05% Isomylase (500 Ug-1). 1 Introduction Feed costs in poultry production constitute more than 70% of production costs [1], where energy feed sources occupy the largest portion of 70% - 75% in rations [2]. Corn is predominantly used as animal feed raw material, especially poultry feed in Indonesia. However, the availability of maize, especially feed maize, becomes scarce especially in every rainy season and causes an increasing in the price of maize in the market. The increasing in feed raw material costs has accelerated the demand to find alternative feed ingredients that can replace this material with lower production costs. In order to meet the needs of carbohydrates, cassava is the third food crop in Indonesia after rice and maize. Cassava (Manihot spp) is the highest carbohydrate source among staple crops and can potentially replace maize as an energy source in poultry feed. Starch is a major component of carbohydrates which is a source of energy. The main composition of cassava starch generally https://doi.org/10.1051/bioconf/20238100044 BIO Web of Conferences 81, 00044 (2023) ICESAI 2023 consists of amylose (AMS) and amylopectin (AMP). Cassava starch contains 17% amylose and 83% amylopectin, compared with maize starch which has 28% amylose and 72% amylopectin [3,4]. [5] Reported that high levels of resistant starch in cassava probably consisted of 82.85% amylopectin with 5.79% branch associations and 17.25% amylose with 0.48% branch associations. This causes cassava starch difficult to digest and utilize by poultry. One way to increase the amylose content in cassava root meal (CRM) is adding isoamylase. Isoamylase, also known as a debranched enzyme, hydrolyzes the branch chain α-1.6 glucosidic in glucose, amylopectin, and beta-limit dextrin. An increasing hydrolyzed of branch chains in amylopectin is expected to increase the amylose content. Thus it can increase nutrient availability and broiler growth performance. consists of amylose (AMS) and amylopectin (AMP). Cassava starch contains 17% amylose and 83% amylopectin, compared with maize starch which has 28% amylose and 72% amylopectin [3,4]. [5] Reported that high levels of resistant starch in cassava probably consisted of 82.85% amylopectin with 5.79% branch associations and 17.25% amylose with 0.48% branch associations. This causes cassava starch difficult to digest and utilize by poultry. One way to increase the amylose content in cassava root meal (CRM) is adding isoamylase. Isoamylase, also known as a debranched enzyme, hydrolyzes the branch chain α-1.6 glucosidic in glucose, amylopectin, and beta-limit dextrin. 2.2.1 Growth Performance Feed intake (FI), body weight (BW), body weight gain (BWG), feed conversion ratio (FCR) were measured weekly. Weight and percentage of digestive tracts and carcass, and intestinal length were observed at 35 days of age. 2 https://doi.org/10.1051/bioconf/20238100044 BIO Web of Conferences 81, 00044 (2023) ICESAI 2023 2.2.2 Feed Retention Time Observation of retention time at 30th and 90th minutes was carried out at 22 days of age and 60th minutes at 23 days of age. Feed retention time was measured by placing broiler chickens on individual cages and calculating the amount of feed left in the crop from the amount of feed intake for 2 hours before the chicken was fasted for 24 hours, then slaughtered at 30th,60th, and 90th minutes. 𝐹𝑒𝑒𝑑 𝑟𝑒𝑡𝑒𝑛𝑡𝑖𝑜𝑛 𝑡𝑖𝑚𝑒 (%) = 𝑓𝑒𝑒𝑑 𝑖𝑛 𝑡ℎ𝑒 𝑐𝑟𝑜𝑝 (𝑔) 𝑓𝑒𝑒𝑑 𝑖𝑛𝑡𝑎𝑘𝑒 (𝑔) 𝑥 100% (1) 𝐹𝑒𝑒𝑑 𝑟𝑒𝑡𝑒𝑛𝑡𝑖𝑜𝑛 𝑡𝑖𝑚𝑒 (%) = 𝑓𝑒𝑒𝑑 𝑖𝑛 𝑡ℎ𝑒 𝑐𝑟𝑜𝑝 (𝑔) 𝑓𝑒𝑒𝑑 𝑖𝑛𝑡𝑎𝑘𝑒 (𝑔) 𝑥 100% (1) (1) Table 1. Feed rations and nutrient compositions of broilers Ingredients Starter (0-21 days) Finisher (22-35 days) T0 T1/T2 T0 T1/T2 Maize 55.00 27.50 55.00 27.50 CRM 0.00 27.50 0.00 27.50 Soy Bean Meal 28.00 28.00 27.35 27.35 Meat Bone Meal 2.00 2.000 2.820 2.820 Corn Gluten Meal 9.590 9.660 6.420 7.420 Crude palm oil 1.650 1.700 4.320 4.630 CaCO3 0.240 0.070 0.250 0.050 DCP 2.460 2.490 2.900 1.850 NaCl 0.060 0.090 0.040 0.040 L-Methionine 0.220 0.190 0.200 0.180 L-Lysine 0.270 0.300 0.200 0.170 Premixa 0.500 0.500 0.500 0.500 Isoamylase - -/0.05 - -/0.05 Total (%) 100 100 100 100 Nutrient Composition Dry Matter (%) 87.99 88.07 85.72 88.23 Crude Protein (%) 23.04 23.00 21.50 21.50 Crude Fiber (%) 2.41 3.12 2.29 3.08 Extract Ether (%) 4.29 3.59 6.70 6.55 Ash (%) 5.61 5.98 5.31 5.52 Metabolizable Energy (kcalkg-1) 3005.88 3000 3150 3150 Ca (%) 0.96 0.96 0.87 0.87 P available (%) 0.48 0.48 0.44 0.44 P total (%) 0.78 0.75 0.72 0.70 Lysine (%) 1.44 1.44 1.29 1.29 Methionine (%) 0.56 0.56 0.51 0.51 Methionine + Cystine (%) 0.97 0.90 0.83 0.82 Salt (%) 0.23 0.22 0.17 0.17 aPer kg diet: Copper 16 mg, Iodine 1,25 mg, Iron 40 mg, Manganese 120 mg, Selenium 0,3 mg, Zinc 100 mg. Vitamin A 12000 IU, Vitamin D3 5000 IU, Vitamin E 75 IU, Vitamin K3 3 mg, Riboflavin (B2) 8 mg, Nicotin Acid 60 mg, Panthotenic Acid 15 mg, Pyridoxin (B6) 4 mg, Biotin 0,15 mg, Folic acid 2 mg, Vitamin B12 0,016 mg, Choline 1600 ppm. Table 1. Feed rations and nutrient compositions of broilers 3 3 https://doi.org/10.1051/bioconf/20238100044 BIO Web of Conferences 81, 00044 (2023) ICESAI 2023 2.2.6 Statistical Analysis The dietary experimental design used Completely Randomized Design (CRD) 3 treatments 6 replications. Data were analyzed using analysis of variance (ANOVA) and significantly different data continued with Duncan's multiple range test [11] using SPSS software (SPSS® version 16.0). 2.2.4 Feed metabolizable energy Metabolizable energy measurements based on [10] were carried out by placing two groups of broilers on individual cages. The first group was fasted for 48 hours, then excreta was collected for 24 hours. The second group was fasted for 24 hours, then fed for 2 hours and calculated the amount of feed intake. Excreta was collected fot 24 hours after completion of feeding. Excreta was dried and analyzed for gross energy content (Calorimetry Bomb). 𝐴𝑀𝐸 (𝑘𝑐𝑎𝑙𝑘𝑔¯¹) = (𝐺𝐸𝑓−𝐹𝐼)−(𝐺𝐸𝐸𝑝−𝐸) 𝐹𝐼 𝑥 1000 (3) 𝑇𝑀𝐸 (𝑘𝑐𝑎𝑙𝑘𝑔¯1) = (𝐺𝐸𝑓−𝐹𝐼)−[(𝐺𝐸𝐸𝑝−𝐸)−(𝐺𝐸𝐸𝑒 𝑥 𝐸𝑒)] 𝐹𝐼 𝑥 1000 (4) (4) 𝐹𝐼 GEf: Groos energy feed (kkalkg-1), GEEp: gross energy excreta (kkalkg-1), GEEe: gross energy endogenous excreta (kkalkg-1), FI: feed intake (g), E: excreta (g), Ee: endogenous excreta (g). GEf: Groos energy feed (kkalkg-1), GEEp: gross energy excreta (kkalkg-1), GEEe: gross energy endogenous excreta (kkalkg-1), FI: feed intake (g), E: excreta (g), Ee: endogenous excreta (g). 2.2.3 Starch digestibility coefficient in small intestine At the age of 33 days, broiler chickens were placed in individual cages and calculated the amount of feed intake for 2 hours after the chickens were fasted for 24 hours, then slaughtered. Digesta in the small intestine was accommodated in plastic and dried in an oven 60 oC until the weight was constant. The dried samples were analyzed for its starch content. The results of the analysis of starch content in the digesta were going to be compared with the total content of starch consumed by chicken. (2) 2.2.5 Blood Glucose Levels Blood sampling was taken at 34 day of age via branchial vein. Blood sampling was taken as ± 2 ml using 3 ml disposable syringe. The blood was put into a vacuum tube that has been filled with Ethylenediaminetetraacetic acid (EDTA) and homogenized. Serum blood was taken for glucose analysis (Enzymatic Colorimetric AKL Method 20101803460). BIO Web of Conferences 81, 00044 (2023) ICESAI 2023 https://doi.org/10.1051/bioconf/20238100044 The highly significant of FCR was due to BW and BWG in the treatment showed the same results with the control. [9] revealed that 300 and 600 Ukg α-1.6 isoamylase were able to increase BW, decrease FI, and achieve the same effect from controls but FCR was much higher than control. The higher of FI in the treatments compared to control was due to faster feed rate in the digestive tract. The higher inclusion of CRM will increase the content of amylopectin in the ration. Amylopectin which has branch chains tends to be less soluble in water. Components of fiber which were insoluble in water cause a faster feed rate [12,13]. Amylopectin in cassava had a relatively long chain length and caused cassava starch to be difficult to digest by poultry [5]. In addition, amylose has a molecular weight of about 100 kDa, amylopectin has a higher molecular weight of 104 - 106 kDa [14]. Increasing of feed rate in this study were supported by a lower feed retention in the treatments (Figure 1). Higher feed retention rates allow feed to be in contact with digestive tract enzymes more optimally in order to increase feed digestibility. Crop filling was highly dependent on feed volume in the proventriculus and gizzard, whose capacity in chickens was estimated to reach 5-10 g of feed [15]. When these organs were fully filled, feed storage occurred [16]. After a while, when the gizzard was emptied, muscle contraction caused the passage of the digesta moved to a further part of the intestinal tract [17]. Therefore, the function of the anterior digestive tract was closely related, and affected the peristalsis further part of the digestive system. Furthermore, ad libitum feeding caused reducing physical use of crop by chickens [18] because the chance of feed in crop was only about 30 minutes according to digesta rate [19]. Higher feed retention in the control was caused by a slower feed rate compared to the other treatments which can be seen from the highly significant (P <0.01) increasing on gizzard weight (Table 2). Higher gizzard weight indicated a higher process of mechanical digestion. A less developed gizzard would function as a transit organ from the mechanical digestive organs, with the implication of accelerating retention time [20]. Conversely, lower feed retention occured at T1 and T2. BIO Web of Conferences 81, 00044 (2023) ICESAI 2023 This was due to the difference in the AMS / AMP ratio which affected on the starch structure. Starch was arranged in a very complex and large structure, where amorphous and crystalline layers alternately form stiff and semi-crystalline granules, with sizes varying from 1 to 50 μm [21]. The semi-crystalline layer consisted of alternating layers of α-glucan crystals which extended from successive branches of amylopectin and amorf amylopectin branches [22]. The increasing amylopectin content would increase the formation of semi-crystalline layers so that it had the potential to increase the feed rate. [23] stated that the degree of crystallinity was inversely proportional to the stach digestion rate, and [24] proved this by concluding that slow starch digestibility was associated with a regular structure of crystalline and amorphous layers alternating in starch granules. [24] revealed that the digestion of starch granules started from the surface pores and interior channels, which allow amylase to enter the interior and digest granules gradually from the inside. 3.1 Broiler Growth Performance and Feed Retention Time The results on the broiler growth performance showed that the inclusion of CRM up to 50% both with or without isoamylase addition were highly significant (P <0.01) increasing FI, BW, and BWG at starter period. There were no differences on FI, BW, and BWG but FCR highly significant (P <0.01) increased at finisher period. Feed intake and FCR highly significant (P <0.01) increased during the overall period compared to control. The control 4 4 4 BIO Web of Conferences 81, 00044 (2023) ICESAI 2023 https://doi.org/10.1051/bioconf/20238100044 treatment had feed retention which tended to be higher in the range of 30-90 minutes. In contrast, lower feed retention occurred in treatments T1 and T2. Table 2. Effect of dietary treatments on the growth performance of broilers Parameters Treatments P Value T0 T1 T2 BW DOC (g bird-1) 46.29 45.87 45.76 Starter (0-21 days) FI (g bird-1) 957.54 ± 46.22b 1215.08 ± 35.36a 1219.20 ± 58.83a 0.000 BW(g bird-1) 731.89 ± 50.69b 884.84 ± 41.13a 871.11 ± 19.35a 0.000 BWG (g bird-1) 685.60 ± 51.29b 838.97 ± 41.11a 825.35 ± 19.56a 0.000 FCR (FIBWG-1) 1.39 ± 0.06 1.45 ± 0.06 1.47 ± 0.06 0.135 Finisher (22-35 days) FI (g bird-1) 1769.79 ± 87.59 1897.80 ± 93.89 1845.42 ± 114.84 0.115 BW(g bird-1) 1756.38 ± 136.85 1819.82 ± 80.61 1805.66 ± 125.30 0.623 BWG (g bird-1) 1024.49 ± 98.38 934.97 ± 47.47 934.55 ± 118.56 0.190 FCR (FIBWG-1) 1.73 ± 0.06b 2.03 ± 0.06a 1.99 ± 0.10a 0.002 Cumulative (0-35 days) FI (g bird-1) 2725.11±119.71b 3112.89 ± 121.45a 3064.63 ± 120.70a 0.000 BW(g bird-1) 1756.38 ± 136.85 1819.82 ± 80.61 1805.66 ± 125.30 0.623 BWG (g bird-1) 1710.09 ± 137.01 1773.95 ± 79.99 1759.91 ± 125.58 0.619 FCR (FIBWG-1) 1.59 ± 0.05b 1.75 ± 0.05a 1.74 ± 0.05a 0.004 ** The numbers on the same line and followed by different letters indicate a highly significant difference at the test level of 1%. FI: Feed Intake; BW: Body Weight; BWG: Body Weight Gain; FCR: Feed Conversion Ratio Fig. 1. Feed retention time (The amaount of left feed in crop) at 30th, 60th, and 90th minutes 0 10 20 30 40 50 60 30TH MINUTES 60TH MINUTES 90TH MINUTES FEED RETENTION TIME (%) T0 T1 T2 90TH MINUTES Fig. 1. Feed retention time (The amaount of left feed in crop) at 30th, 60th, and 90th minutes 5 5 BIO Web of Conferences 81, 00044 (2023) ICESAI 2023 3.2 Carcass and Visceral Organs The inclusion of CRM up to 50% subtituted maize was highly significant (P <0.01) reducing both the weight and percentage of gizzard. Addition of isoamylase decreased (P <0.05) the 6 6 6 https://doi.org/10.1051/bioconf/20238100044 BIO Web of Conferences 81, 00044 (2023) ICESAI 2023 percentage of liver. The treatments did not affect (P> 0.05) on weight and percentage of small intestines, abdominal fat, and carcass as well as lenght of small intestines. The treatments was highly significant reducing both weight and percentage of gizzard. This was due to the decreasing corn which had a larger particle size than CRM in rations. Thus it reduced mechanical digestive activity in the gizzard and made gizzard less develop. The use of coarse particles or grains had been aimed to stimulate gizzard growth, improve mechanical function, increase digesta peristalsis, and increase gizzard volume, in order to increase retention time [25, 26]. Gizzard would shrink feed particle size to 100-200 μm before entering the small intestine [27, 28]. The higher particle size of the feed in the gizzard, the longer it would take to reach the size of the feed that can enter the small intestine. The liver was the main organ for synthesis and metabolism, and served for glucose absorption of around 25% to 35% [29]. After consuming carbohydrates, 33% of glucose was absorbed by the liver [30]. Glucose derived from starch degradation was transported to the blood to maintain sufficient concentration to supply energy to other tissues [31, 32]. The results of the observations also showed that the treatment had no effect on the weight and percentage of the proventriculus. This is not in accordance with [33, 34]. The effect of the treatment in this study did not affect the weight, percentage, and length of the small intestine (duodenum, jejenum, ileum). This result was in line with the research [35, 36]. Table 3. * The numbers on the same line and followed by different letters indicate a significant difference at the test level of 5%. The numbers on the same line and followed by different letters indicate a highly significant difference at the test level of 1%. 3.3 Broiler Chicken Metabolite Profile The inclusion of 50% CRM in rations both with and without isoamylase did not give a significant difference to the starch digestibility coefficient in the small intestine. The results also showed no significant differences in AME, TME, and blood glucose levels in all treatments. Table 4. Effect of dietary treatments on starch digestibility coeficient in small intestine, metabolizable energy, and blood glucose level on broiler Table 4. Effect of dietary treatments on starch digestibility coeficient in small intestine, metabolizable energy, and blood glucose level on broiler Parameters Treatments P Value T0 T1 T2 Starch digestibility coefficient 0.98 ± 0.010 0.98 ± 0.001 0.97 ± 0.004 0.162 AME (kcal kg-1 feed) 3142.89 ± 220.84 3185.42 ± 168.67 3167.95 ± 172.83 0.950 TME (kcal kg-1 feed) 3317.73 ± 272.95 3292.95 ± 158.63 3313.50 ± 167.63 0.984 Blood glucose (mgdL-1) 169.65 ± 9.78 155.21 ± 16.43 150.54 ± 22.30 0.161 AME: Apparent Metabolizable Energy; TME: True Metabolizable Energy. The inclusion of 50% CRM in rations both with or without isoamilase did not give a significant difference to the coefficient starch digestibility in small intestine. The results of the study on starch digestibility coefficient in small intestine were 0.97 - 0.98. [38] found that the total starch digestibility coefficient had increased by 0.96 at the age of 3 days of broiler chicken. The same authors also found starch digestibility increased (P <0.01) linearly as grew older in fast-growing broilers but not in slow-growing layer chickens. [39] found that 50% of poultry showed ileal starch digestion above 0.94, when fed with pellet diet containing 38.5% of whole wheat. Better results were obtained from [40] revealed ileal starch digestion was 0.98 in a diet containing 44% of whole wheat with other pellet ingredients. The whole process of starch digestion and absorbtion were carried out during short retention times in the duodenum and jejunum, estimated by [41] for about 1 hour. Starch digestibility was able to affect AME in feed [42, 43]. The results showed no significant differences in AME and TME in all treatments. This was due to there was no significant difference in starch digestibility coefficient. [37] reported that metabolizable energy (ME) consumption decreased (P <0.001) in the treatment fed cassava pellet and chips without enzyme supplementation, but increased (P <0.001) in all enzyme-supplemented feeds. BIO Web of Conferences 81, 00044 (2023) ICESAI 2023 https://doi.org/10.1051/bioconf/20238100044 The results did not affect the weight and percentage both of abdominal fat and carcass. Higher abdominal fat in the treatment group was a factor that contributes to the increasing in carcass weight due to higher body weight. The result of this study was in accordance with [37] found that the use of cassava in the form of pellets and chips were able to replace corn in broiler feed by up to 50%. This was supported by the fact that cassava products did not have an adverse effect on carcass weight, abdominal fat, and carcass composition. The opposite results are shown in the [33] although the result did not record a significant effect of feed treatment on carcass percentage, data showed a tendency to decrease with increasing use of cassava leaf concentrates and cassava tubers. 3.2 Carcass and Visceral Organs Effect of dietary treatments on weight and percentage of carcass, visceral organ, and intestinal length of broilers aged 35 days Parameters Treatments P Value T0 T1 T2 Proventriculus (g) 10.66 ± 2.25 10.50 ± 2.94 7.83 ± 1.32 0.084 Proventriculus percentage (%) 0.61 ± 0.15 0.58 ± 0.18 0.43 ± 0.06 0.097 Gizzard (g) 23.33 ± 3.50a 18.00 ± 1.67b 14.50 ± 0.06c 0.000 Gizzard percentage (%)** 1.33 ± 0.15a 0.99 ± 0.13b 0.80 ± 0.10c 0.000 Liver (g)* 41.33 ± 4.22ab 44.00 ± 6.63a 36.33 ± 2.87b 0.044 Liver percentage (%)* 2.37 ± 0.15a 2.41 ± 6.63a 2.00 ± 0.11b 0.015 Duodenum (g) 16.66 ± 4.22 18.33 ± 4.58 15.83 ± 1.60 0.511 Duodenum percentage (%) 0.95 ± 0.20 1.00 ± 0.22 0.87 ± 0.10 0.521 Jejenum (g) 26.66 ± 8.01 30.16 ± 6.11 27.77 ± 4.22 0.622 Jejenum percentage (%) 1.52 ± 0.42 1.65± 0.31 1.53 ± 0.23 0.754 Ileum (g) 20.83 ± 4.53 24.16 ± 6.24 20.30 ± 3.26 0.354 Ileum percentage (%) 1.19 ± 0.22 1.32 ± 0.32 1.12 ± 0.19 0.413 Duodenum (cm) 38.00 ± 1.41 39.00 ± 4.42 36.50 ± 4.76 0.539 Jejenum (cm) 81.91 ± 12.69 91.50 ± 9.97 81.91 ± 12.15 0.289 Ileum (cm) 80.16 ± 12.96 89.25 ± 10.74 81.08 ± 13.41 0.401 Abdominal fat (g) 26.50 ± 7.81 28.00 ± 5.29 33.83 ± 5.38 0.136 Abdominal fat percentage (%) 1.51 ± 0.40 1.54 ± 0.31 1.87 ± 0.30 0.175 Carcass weight (g) 1249.3 ± 111.22 1288.16 ± 91.19 1278.00 ± 62.40 0.747 Carcass percentage (%) 71.75 ± 2.26 70.70 ± 2.80 70.67 ± 1.30 0.636 ** The numbers on the same line and followed by different letters indicate a highly significant Table 3. Effect of dietary treatments on weight and percentage of carcass, visceral organ, and intestinal length of broilers aged 35 days 7 7 BIO Web of Conferences 81, 00044 (2023) ICESAI 2023 4. Conclusion It can be concluded that the application of 0.05% isoamylase in CRM was able to reduce the amylose / amylopectin ratio which was close to the amylose / amylopectin ratio in corn and increase the reducing sugar content. The inclusion of 50% CRM with or without isoamylase to subtitute corn in the ration did not affect growth performance, intestinal starch digestibility coefficient, AME, TME, and blood glucose levels of broiler chickens but had a higher FCR compared to control. This work was financially supported by Indonesia Endowment Fund for Education (LPDP). 3.3 Broiler Chicken Metabolite Profile Supplementation of 200 and 400 Ukg-1 α-1.6 Isoamylase in feed did not have a significant effect on ME compared to control [44]. The results of this study, TME on T1 and T2 tended to be lower compared to control, indicating that to hydrolyze or digest CRM requires higher energy. The results of blood glucose levels ranged from 150.54 to 169.65 mgdL-1. [45] stated that glucose levels in broilers were higher than mammals, ranging from 180 to 250 mgdL-1. 8 https://doi.org/10.1051/bioconf/20238100044 BIO Web of Conferences 81, 00044 (2023) ICESAI 2023 Chicken in a state of hypoglycemia if the blood glucose level was 137 mgdL-1 and hyperglycemia in blood glucose levels 363 mgdL-1 [46]. The results of this study did not show significant differences in all treatments but the blood glucose levels were still in normal standards. The diet containing CRM with or without isoamylase addition tended to have lower blood glucose levels compared to the control ration (155.21 mgdL-1 and 150.54 mgdL- 1). The same result was shown by [44] reported that there was no significant difference in broiler serum glucose concentration between the dietary treatments supplemented with 1500 and 3000 Ukg-1 α-Amylase (A. oryzae), 480 and 960 Ukg-1 α-1.4 Amylase (B. subtilis) and 200 and 400 Ukg-1-1.6 Isoamylase (B. subtilis). Previous research had shown that glucose absorption was not a factor that limits starch utilization [47]. References 1. McNab, J. Advance in poultry nutrition in the world. Proceedings of the seminar and international poultry show, WPSA, Bnagladesh Branch. p52 (1999) 2. Van der Klis JD, Kwakernaak C, Jansman A, Blok M. Energy in poultry diets: Adjusted AME or Net Energy. Proceedings of Australian Poultry Science Symposium. 21: 44- 49 (2010) 3. Gomes E, Souza SR, Grandi RP, Silva RD. Production of thermostable glucoamylase by newly isolated Aspergillus flavus A1.1 and thermomyces Lanuginosus A13.37. Braz J Microbiol 36: 75-82 (2005) 4. Promthong S, Kanto U, TIrawattanawanich C, Tongyai S, Isariyodom S, Markvichitr K. Comparison of nutrient compositions and carbohydrate fractions of corn, cassava chip and cassava pellet ingredients: animals. In: Proceedings of 43rd Kasetsart University Annual Conference; p. 211-20 (2005) 5. Kiatponglarp W, Tongta S. Structural and physical properties of debranched tapioca starch. Suranaree J Sci Tech 2007;14(2): 195-204 (2007) ́ ́ 6. Anguita M, Gasa J, Martı ́n-Oru ́e SM, Perez JF. Study of the effect of technological processes on starch hydrolysis, non-starch polysaccharides solubilization and physicochemical properties of different ingredients using a two-step in vitro system. Animal Feed Science and Technology 129, 99 – 115 (2006) processes on starch hydrolysis, non-starch polysaccharides solubilization a physicochemical properties of different ingredients using a two-step in vitro syste Animal Feed Science and Technology 129, 99 – 115 (2006) p y p p f ff g g Animal Feed Science and Technology 129, 99 – 115 (2006) mal Feed Science and Technology 129, 99 – 7. Midau A, Augustine C, Yakubu B, Yahaya SM, Kibon A, Udoyong AO. Performance of broiler chicken fed enzyme supplemented cassava peel meal based diets. Int J Sust Agric 3(1): 1-4 (2011) 8. Bhuiyan M, Romero LF, Iji P. Maximising the energy value of cassava products in diets for broiler chickens. In: Proceedings of the 23rd Annual Australian Poultry Science Symposium; p. 27-25 (2012) 9. Huang Z. Effects and mechanism of isoamylase additive on growth performance in broilers. Siliao Gongye. 35:1–5 (2014) 9 https://doi.org/10.1051/bioconf/20238100044 BIO Web of Conferences 81, 00044 (2023) ICESAI 2023 10. Farrell DJ. Rapid Determination of Metabolizale Energy of Feed Using Cockrerels. J Poult. Sci. 19: 303-308 (1978) 11. Steel RGD, Torrie JH. Prinsip dan Prosedur Statistika. Terjemahan: B. Sumantri. Jakarta (ID): Gramedia Pr. (1995) 12. Cao B, Kumao T, Karasawa Y. References Effects of dietary cellulose levels on growth, nitrogen utilization and retention time of diets on intestine in chicks fed equal amounts of nutrients. Proceeding of the 6th Asian Pacific Poultry Congress, Nagoya Japan pp. 402- 403 (1998) 13. Langhout DJ. The role of intestinal flora as affected by non-starch polysaccharides in broiler chicks. PhD Thesis. Wageningen University, Wageningen Netherlands pp 162 (1998) 14. Buléon A, Colonna P, Planchot V, Ball S. Starch granules: Structure and biosynthesis. Int. J. Biol. Macromol. 23:85–112 (1998) 15. Svihus B. Function of the digestive system. J. Appl. Poultry Res., 23: 306–314 (2014) 16. Jackson S, Duke GE. Intestine fullness influences feeding behaviour and crop filling in the domestic turkey. Physiol. Behav., 58: 1027–1034 (1995) 17. Langenfeld M.S. Systema digestorium, s. apparatus digestorius. In: Chicken Anatomy (in Polish). Scientific Publishing Company PWN, pp. 91–117 (1992) 18. Svihus B, Sacranie A, Denstadli V, Choct M. Nutrient utilization and functionality of the anterior digestive tract caused by intermittent feeding and inclusion of whole wheat in diets for broiler chickens. Poult. Sci. 89: 2617–2625 (2010) 19. Svihus B, Lund V, Borjgen B, Bedford M, Bakken M. Effect of intermittent feeding, structural components and phytase on performance and behaviour of broiler chickens. Brit. Poultry Sci., 54: 222–230 (2013) 20. Ravindran V. Feed enzyme: The science, practice, and metabolic realities. J. Appl Poult. Res. 22 : 628–636 (2013) 21. Svihus B. Starch digestion capacity of poultry. Poult. Sci. 93: 2394-2399 (2014) 22. Perez S, Bertoft E. The molecular structures of starch components and their contribution to the architecture of starch granules: A comprehensive review. Starch 62:389–420 (2010) 23. Björck I, Liljeberg H, Östman E. Low glycaemic-index foods. Br. J. Nutr. 83(Sup 1):S149–S155 (2000) 24. Zhang G, Ao Z, Hamaker BR. Slow digestion property of native cereal starches. Biomacromolecules 7:3252–3258 (2006) 25. Amerah AM, Ravindran V, Lentle RG, Thomas DG. Feed particle size: Implications on the digestion and performance of poultry. World’s Poult. Sci. J. 63:439–455 (2007) 25. Amerah AM, Ravindran V, Lentle RG, Thomas DG. Feed particle size: Implications on the digestion and performance of poultry. World’s Poult. Sci. J. 63:439–455 (2007) 26. Svihus B. The gizzard function, influence of diet structure and effects on nutrient availability. World’s poult. Sci. J., 67: 207-223 (2011) on the digestion and performance of poultry. World’s Poult. Sci. J. 63:439–455 (2007) 26. Svihus B. References The gizzard function, influence of diet structure and effects on nutrient availability. World’s poult. Sci. J., 67: 207-223 (2011) 26. Svihus B. The gizzard function, influence of diet structure and effects on nutrient availability. World’s poult. Sci. J., 67: 207-223 (2011) 27. Amerah AM, Ravindran V, Lentle RG, Thomas DG. Influenced of feed particle size on the performance, energy utilization, digestive tract development, and digesta parameters of broiler starter fed wheat-and-corn-based diets. Poult. Sci. 87: 2320-2328 (2008) 28. Xu Y, Stark CR, Ferket PR, Williams CM, Pacheco WJ, Brake J. Effect of dietary coarsely ground corn on broiler live performance, gastrointestinal tract development, apparent ileal digestibility of energy and nitrogen, and digesta particle size distribution and retention time. Poult. Sci. 94: 53-60 (2015) 29. Sharabi KC, Tavares D, Rines AK, Puigserver P. Molecular pathophysiology of hepatic glucose production. Mol. Aspects Med. 46: 21–33 (2015) 30. Moore MC, Coate KC, Winnick JJ, An Z, Cherrington AD. Regulation of hepatic glucose uptake and storagein vivo. Adv. Nutr. 3: 286–294 (2012) 10 https://doi.org/10.1051/bioconf/20238100044 BIO Web of Conferences 81, 00044 (2023) ICESAI 2023 31. Denardin CC, Walter M, Silva LPD, Souto GD, Fagundes CA. Effect of amylose content of rice varieties on glycemic metabolism and biological responses in rats. Food Chem. 105:1474–1479 (2007) 32. Denardin CC, Boufleur N, Reckziegel P, Silva LPD, Walter M. Amylose content in rice (Oryza sativa) affects performance, glycemic and lipidic metabolism in rats. Ciˆencia Rural. 42:381–387 (2012) 33. Eruvbetine D, Tajudeen ID, Adeosun AT dan Olojede AA. Cassava (M. esculenta) leaf and tuber concentrate in diets for broiler chickens. Biores. Tech., 86: 277-281 (2003) 34. Adeyemi OA, Eruvbetine D, Ongutona T, Dipeolu M, Agunbiade JA. Feeding broiler chicken with diets containing whole cassava root meal fermented with rumen filtrate. Arch. Zootech. 57 (218): 247-258 (2008) 35. Nwokoro and Ekhosuehi, Effect of replacement of maize with cassava peel in cockerel diets on performance and carcass characteristics. Tropical Animal Health and Production, 37: 495-501 (2005) 36. Khempaka S, Molee W, Guillaume M. Dried cassava pulp as an alternative feedstuff for broilers: Effect on growth performance, carcass traits, digestive organs, and nutrient digestibility. J. Appl. Poult. Res. 18: 487-493 (2009) 37. Bhuiyan MM, Iji PA. Energy Value of Cassava Products in Broiler Chicken Diets with or without Enzyme Supplementation. Asian-Australas. J. Anim. Sci. 28:1317–1326 (2015) 38. Zelenka J, Ceresnakova Z. References Effect of age on digestibility of starch in chickens with different growth rate. Czech J. Anim. Sci. 50:411–415 (2005) 39. Svihus B, Hetland H. Ileal starch digestibility in growing broiler chickens fed on a wheat-based diet is improved by mash feeding, dilution with cellulose or whole wheat inclusion. Br. Poult. Sci. 42:633–637 (2001) ( ) 40. Hetland, H., B. Svihus and M. Choctt. Role of insoluble fiber on gizzard activity in layers. J. Apply. Poultry Res. 14: 38—46 (2005) 41. Rougiere N, Carré B. Comparison of gastrointestinal transit times between chickens from D+ and D− genetic lines selected for divergent digestion efficiency. Animal 4:1861–1872 (2010) ( ) 42. Gracia MI, Arani´bar MJ, L´azaro R, Medel P, Mateos GG. α-Amylase Supplementation of Broiler Diets Based on Corn. Poult. Sci. 82:436–442 (2003) 43. Kaczmarek SA, Rogiewicz A, Mogielnicka M, Rutkowski A, Jones RO, Slominski BA. The effect of protease, amylase, and nonstarch polysaccharide-degrading enzyme supplementation on nutrient utilization and growth performance of broiler chickens fed corn-soybean meal-based diets. Poult. Sci. 93:1745–1753 (2014) 44. Yuan J, Xingyu W, Dafei Y, Maofei W, Xiaonan Y, Zhao L, Yuming G. Effect of different amylases on the utilization of cornstarch in broiler chickens. Poult. Sci. 96:1139–1148 (2017) 45. Hazelwood RL. Pancreatic hormones, insulin, glicagon molar ratios and somatostatin as determinants of avian carbohydrate metabolism. J Exp Zool. 232:647-652 (2000) 46. Goodwin, Denise, John B, McMurray BL, William IR, Danny LM. Blood glucose value and definitions of hypoglicemia and hyperglycemia. Georg Poult Lab (Georgia). The University of Georgia (1994) 47. Suvarna S, Christensen VL, Ort DT, Croom Jr WJ. High levels of dietary carbohydrate increase glucose transport in poult intestine. Comp. Biochem. Physiol. A Mol. Integr. Physiol. 141:257–263 (2005) 11
https://openalex.org/W2014460666	https://bg.copernicus.org/articles/9/4787/2012/bg-9-4787-2012.pdf	English	null	Intense photooxidative degradation of planktonic and bacterial lipids in sinking particles collected with sediment traps across the Canadian Beaufort Shelf (Arctic Ocean)	Biogeosciences	2,012	cc-by	11,578	Received: 31 May 2012 – Published in Biogeosciences Discuss.: 26 June 2012 Revised: 22 October 2012 – Accepted: 29 October 2012 – Published: 23 November 2012 Received: 31 May 2012 – Published in Biogeosciences Discuss.: 26 June 2012 Revised: 22 October 2012 – Accepted: 29 October 2012 – Published: 23 November 2012 Abstract. The lipid content of seven samples of sinking par- ticles collected with sediment traps moored at ∼100 m depth in summer and fall across the Canadian Beaufort Shelf (Arc- tic Ocean) was investigated. Our main goal was to quan- tify and characterize the biotic and abiotic degradation pro- cesses that acted on sinking material during these periods. Diatoms, which dominated the phytoplanktonic assemblage in every trap sample, appeared to be remarkably sensitive to Type II (i.e. involving singlet oxygen) photodegradation processes in summer, but seemed to be relatively unaffected by biotic degradation at the same time. Hence, the relative recalcitrance of phytodetritus towards biodegradation pro- cesses during the Arctic midnight sun period was attributed to the strong photodegradation state of heterotrophic bacte- ria, which likely resulted from the efﬁcient transfer of sin- glet oxygen from photodegraded phytoplanktonic cells to at- tached bacteria. In addition, the detection in trap samples of photoproducts speciﬁc to wax ester components found in her- bivorous copepods demonstrated that zooplanktonic faecal material exported out of the euphotic zone in summer were affected by Type II photodegradation processes as well. By contrast, sinking particles collected during the autumn were not inﬂuenced by any light-driven stress. Further chemical analyses showed that photodegraded sinking particles con- tained an important amount of intact hydroperoxides, which could then induce a strong oxidative stress in underlying sed- iments. Biogeosciences 1 Introduction Abstract. The lipid content of seven samples of sinking par- ticles collected with sediment traps moored at ∼100 m depth in summer and fall across the Canadian Beaufort Shelf (Arc- tic Ocean) was investigated. Our main goal was to quan- tify and characterize the biotic and abiotic degradation pro- cesses that acted on sinking material during these periods. Continental shelves of the Arctic Ocean receive a consider- able amount of terrestrial matter from river runoff, mixed with important autochthonous production from microalgal photosynthesis during spring and summer (Rachold et al., 2004; Wassmann et al., 2004). The ongoing trend of de- clining sea ice extent and thickness in the Arctic Ocean ap- pears to induce a steady increase in pelagic primary produc- tion (Arrigo et al., 2008), whereas permafrost thawing com- bined with enhanced river discharge are currently increasing the seaward ﬂux of terrigenous material (Frey and McClel- land, 2009). In turn, the annual lengthening of the ice-free period and the rise in river run-off could lead to an increase in particulate matter export that could modify the biogeochem- istry and trophic balance of Arctic ecosystems through the coastal-marine realm (Valli`eres et al., 2008; Wassmann and Reigstad, 2011). Therefore, it is essential to understand how abiotic (autoxidation and photooxidation) and biotic (bac- terial degradation) processes affect the dynamic of sinking ﬂuxes of particulate organic matter in such environments. Particles in the water column exist in a continuum of sizes (McCave, 1984), with two classes usually operationally rec- ognized (Bacon et al., 1985; Wakeham and Lee, 1989): (i) suspended particles (≤102 µm diameter) sinking very slowly through the water column and constituting most of the stand- ing stock of particulate matter in the ocean and (ii) sinking 1Aix-Marseille Universit´e, Universit´e du Sud Toulon-Var, CNRS/INSU, UMR7294, IRD, MIO, UM110, 13288, Marseille, Cedex 09, France 2Universit´e Laval, Takuvik CNRS-Laval U. Joint Laboratory, 1045 av. de la M´edecine, Qu´ebec, Quebec G1V 0A6, Canada 3Centre de Formation et de Recherche sur l’Environnement M´editerran´een (CEFREM, UMR CNRS 5110), Bˆat. U, 1Aix-Marseille Universit´e, Universit´e du Sud Toulon-Var, CNRS/INSU, UMR7294, IRD, MIO, UM110, 13288, Marseille, Cedex 09, France 2Universit´e Laval, Takuvik CNRS-Laval U. Joint Laboratory, 1045 av. de la M´edecine, Qu´ebec, Quebec G1V 0A6, Canada 3Centre de Formation et de Recherche sur l’Environnement M´editerran´een (CEFREM, UMR CNRS 5110), Bˆat. U, Universit´e de Perpignan, Via Domitia (UPVD), 66860 Perpignan, France Cedex 09, France 2Universit´e Laval, Takuvik CNRS-Laval U. Joint Laboratory, 1045 av. de la M´edecine, Qu´ebec, Quebec G1V 0A6, Canada 3Centre de Formation et de Recherche sur l’Environnement M´editerran´een (CEFREM, UMR CNRS 5110), Bˆat. U, Universit´e de Perpignan, Via Domitia (UPVD), 66860 Perpignan, France 2Universit´e Laval, Takuvik CNRS-Laval U. Joint Laboratory, 1045 av. de la M´edecine, Qu´ebec, Quebec G1V 0A6, Canada 3Centre de Formation et de Recherche sur l’Environnement M´editerran´een (CEFREM, UMR CNRS 5110), Bˆat. U, Universit´e de Perpignan, Via Domitia (UPVD), 66860 Perpignan, France Correspondence to: J.-F. Rontani (jean-francois.rontani@univ-amu.fr) Received: 31 May 2012 – Published in Biogeosciences Discuss.: 26 June 2012 Revised: 22 October 2012 – Accepted: 29 October 2012 – Published: 23 November 2012 J.-F. Rontani1, B. Charriere1, A. Forest2, S. Heussner3, F. Vaultier1, M. Petit1, N. Delsaut3, L. Fortier2, and R. Semp´er´e1 J.-F. Rontani1, B. Charriere1, A. Forest2, S. Heussner3, F. Vaultier1, M. Petit1, N. Delsaut3, L R. Semp´er´e1 2.1 Study area The Canadian Beaufort Shelf (Fig. 1) represents around 2 % (i.e. 64 000 km2) of the Arctic Ocean. The shelf is delimited on the west by the Mackenzie Canyon and on the east by Amundsen Gulf. The Mackenzie River is the largest river draining into the Arctic in terms of sediment and partic- ulate organic carbon supply (127 × 1012 g y−1 of sediment and 2.1 × 1012 g y−1 of particulate organic carbon respec- tively; Macdonald et al., 1998) and the fourth largest in terms of freshwater discharge (3.3 × 1011 m3 y−1; Milliman and Meade, 1983; Brunskill, 1986; Macdonald et al., 1998). The Mackenzie River supplies about 95–99 % of the sedi- ment to the Beaufort Shelf, with coastal erosion and other rivers (Hill et al., 1991; Rachold et al., 2004). The main river load occurs between end of May and the end of August with considerable interannual variance (O’Brien et al., 2006). Pri- mary productivity over the Mackenzie Delta/Beaufort Shelf (3.3 × 1012 g y−1 of particulate organic carbon) is mainly due to phytoplanktonic blooms during late spring and sum- mer (Macdonald et al., 1998). Production by ice algae ac- counts for less than 10 % of the marine production in this area (Horner and Schrader, 1982). y g In a companion paper (Rontani et al., 2012b), we exam- ined the lipid content of suspended particulate matter (SPM) samples collected in August 2009 in the Mackenzie River and in surface waters of the adjacent Beaufort Sea. Lipid biomarkers, although representing a very minor fraction of the total organic matter (OM), convey important information on the source (terrigenous, marine or bacterial) and degrada- tion state of OM, which is commonly more diagnostic than that provided by bulk parameters (Saliot et al., 2002). Using speciﬁc lipid degradation products that have been proposed for distinguishing biotic from abiotic processes (Rontani et al., 2011; Christodoulou et al., 2009), we showed that ma- rine particulate organic matter (POM) was weakly degraded across the study area, while biodegradation and autoxida- tion processes acted intensely on terrigenous POM present in seawater (Rontani et al., 2012b). This result was unex- pected as POM originating from land is generally considered to be well preserved due to its previous degradation during transit to the sea. 2 Material and methods particles (≥102 µm diameter) (including zooplankton fecal pellets and marine snow aggregates) numerically less abun- dant but responsible for most of the downward ﬂux of mate- rial from the upper ocean to the sea ﬂoor. Suspended particles are typically collected by ﬁltration, whereas sinking material is commonly collected using sediment traps. It may be noted that a continual exchange exists between these two particle classes owed to the complex suite of aggregation and disag- gregation processes that occur in the water column (Wake- ham and Lee, 1989; Hill, 1998). The sum of these processes affects particle settling velocity, residence time, and thus the efﬁciency of organic matter remineralization. 2.2 Sample collection Time series sediment traps (Technicap PPS 4/3; 24 cups; cylindrical-conical shape; collecting area: 0.125 m2) were deployed on 4 mooring lines located in the Amundsen Gulf (CA16, CA05) and on the Mackenzie Shelf (CA10, G09) in the Beaufort Sea (Fig. 1; Table 1). Bafﬂed lids covered the opening of the sediment traps to reduce internal turbulence. Before deployment, sediment traps were thoroughly rinsed with freshwater and seawater following the JGOFS protocol (Knap et al., 1996). Sample cups were ﬁlled with ﬁltered sea- water (GFF 0.7 µm) adjusted to 35 PSU with NaCl. Forma- lin was added to preserve the material collected (5 % v/v, sodium borate buffered). 2.1 Study area In order to explain the speciﬁc induction of autoxidative processes on vascular plant-derived material, a mechanism involving homolytic cleavage of photochemi- cally produced hydroperoxides resulting from the senescence of higher plants on land was proposed (Rontani et al., 2012b). This cleavage could be catalyzed by some redox active metal ions released from SPM in the mixing zone of riverine and marine waters. In contrast, the intense biodegradation of ter- restrial POM observed was attributed to the well known (Bianchi, 2011) high “priming effect potential” of deltaic re- gions. Published by Copernicus Publications on behalf of the European Geosciences Union. Published by Copernicus Publications on behalf of the European Geosciences Union. J.-F. Rontani et al.: Intense photooxidative degradation of planktonic and bacterial lipids 4788 2.3 Sample treatment It was previously observed in the Mediterranean Sea (Rontani et al., 2009; Christodoulou et al., 2009) and the equatorial Paciﬁc Ocean (Rontani et al., 2011) that the mech- anisms of POM degradation vary according to particle size. In the present work, we present biogeochemical data based on speciﬁc lipid biomarkers studies for samples collected by sediment traps deployed at 100 m depth over the shelf of the Canadian Beaufort Sea and Amundsen Gulf during the pe- riod of August 2009 and October 2003. After retrieval, sample cups were checked for salinity and put aside 24 h to allow particles to settle down. Swim- mers were removed from the samples then quantitative split- ting into several fractions was completed using a McLane Wet Samples Divider or peristaltic pump. Subsamples for the determination of particulate organic carbon (POC) and lipids and their degradation products were ﬁltered in tripli- cates through preweighted Whatman glass ﬁber ﬁlters (GFF 0.7 µm, 25 mm, combusted 4 h at 450 ◦C). For POC analysis, ﬁlters were dried for 12 h at 60 ◦C and weighed again for dry weight. After exposure for 12 h to concentrated HCl fumes to remove inorganic carbon fraction, the samples were ana- lyzed with a Perkin Elmer CHNS 2600 Series II. Total and Biogeosciences, 9, 4787–4802, 2012 www.biogeosciences.net/9/4787/2012/ 4789 J.-F. Rontani et al.: Intense photooxidative degradation of planktonic and bacterial lipids 40 Fig. 1. Map of the studied area with locations of the different stations investigated. Fig. 1. Map of the studied area with locations of the different stations investigated. Table 1. Mooring location and deployment information. Table 1. Mooring location and deployment information. Mooring Location Latitude Longitude Deployment period Water depth (m) Trap depth (m) CA05 A22 Amundsen Gulf 71.31245 127.58237 2 August 2009–7 August 2009 204 97.3 CA05 A23 Amundsen Gulf 71.31245 127.58237 8 August 2009–15 August 2009 204 97.3 CA05 A24 Amundsen Gulf 71.31245 127.58237 16 August 2009–31 August 2009 204 97.3 CA16 A23 Amundsen Gulf 71.78677 126.49695 8 August 2009–15 August 2009 314 110.2 G09 A2 Mackenzie Shelf 71.00254 135.47930 31 July 2009–14 August 2009 702 100.5 G09 A3 Mackenzie Shelf 71.00254 135.47930 30 August 2009–30 August 2009 702 100.5 CA10 A1 Mackenzie Shelf 69.95512 138.67373 7 October 2003–31 October 2003 250 105 POC ﬂuxes were expressed as daily ﬂuxes (mg C m−2 d−1) (Heussner et al., 1990; Lalande et al., 2009). 2.3 Sample treatment The collected samples were processed in the laboratory according to the method described by Heussner et al. (1990). The total sam- ple was divided into several aliquots to obtain different sub- samples for analyzing total mass ﬂux, TOC, lipids and their degradation products. Subsamples were ﬁltered through a precombusted quartz ﬁber ﬁlter (Whatman GF/F, 0.7 µm) un- der low vacuum. 2.3.1 Lipid extraction All solvents employed in this study were glass distilled an- alytical grade. Each ﬁlter was extracted four times with CHCl3-MeOH-H2O (1 : 2 : 0.8, v/v/v) using ultrasonication (separation of particles and solvents by centrifugation at 3500 G for 9 min). To initiate phase separation after ultrason- ication, CHCl3 and puriﬁed H2O were added to the combined extracts to give a ﬁnal volume ratio of 1 : 1 : 0.9 (v/v/v). The upper aqueous phase was extracted twice with CHCl3 and the combined CHCl3 extracts were dried over anhydrous 2.3.3 Alkaline hydrolysis Metal ions can promote autoxidation during hot saponiﬁ- cation procedures, so the prior reduction of hydroperox- ides with NaBH4 allowed us to avoid such autoxidative ar- tifacts during the alkaline hydrolysis. After NaBH4 reduc- tion, 15 ml of water and 1.7 g of potassium hydroxide were added and the mixture was directly saponiﬁed by reﬂuxing for 2 h. After cooling, the content of the ﬂask was acidi- ﬁed with HCl (pH 1) and subsequently extracted three times with dichloromethane. The combined dichloromethane ex- tracts were dried over anhydrous Na2SO4, ﬁltered and con- centrated to give the total lipid fraction. J.-F. Rontani et al.: Intense photooxidative degradation of planktonic and bacterial lipids Na2SO4, ﬁltered and the solvent removed via rotary evapo- ration. and subsequently dried over anhydrous Na2SO4, ﬁltered and concentrated. and subsequently dried over anhydrous Na2SO4, ﬁltered and concentrated. and subsequently dried over anhydrous Na2SO4, ﬁltered and concentrated. 2.3.5 Osmium tetroxide oxidation Double bond positions of monounsaturated fatty acids were determined unambiguously after OsO4 treatment. A fraction of total lipid extracts and OsO4 (1 : 2, w/w) were added to a pyridine-dioxane mixture (1 : 8, v/v, 5 ml) and incubated for 1 h at room temperature. Then, 6 ml of Na2SO3 suspension (16 % Na2SO3 in water-methanol, 8.5 : 2.5, v/v) was added and the mixture was again incubated for 1.5 h. The resulting mixture was acidiﬁed (pH 3) with HCl and extracted three times with DCM (5 ml). The DCM extracts were combined 2.3.4 Derivatization 5α (H)-stan-3β-ols, 14-stera-3β, 6α/β-diols and 3β, 5α, 6β- steratriols (deriving from cholest-5-en-3β-ol (cholesterol), 24-ethylcholest-5-en-3β-ol (sitosterol), 24-methylcholesta- 5,22E-dien-3β-ol (brassicasterol) and 24-methylcholesta- 5,24(28)-dien-3β-ol (24-methylenecholesterol)) were used to estimate biodegradation, photooxidation and autoxidation state of the different components of the samples, respec- tively (Fig. 2) (Rontani et al., 2009, 2011; Christodoulou et al., 2009). Abiotic oxidation of the non-speciﬁc 19 mo- nounsaturated fatty acids (oleic and palmitoleic acids) af- fords 8-E, 8-Z, 9-E, 10-E, 11-E and 11-Z hydroxyacids (Frankel, 1998), which were employed to estimate the im- portance of autoxidation and photooxidation of bulk organic matter (Fig. 2) (Marchand and Rontani, 2001; Rontani et al., 2011). In contrast, abiotic oxidation of cis-vaccenic acid (a typical biomarker for Gram-negative bacteria, Sicre et al., 1988; Keweloh and Heipieper, 1996) produces 10-E, 10-Z, 11-E, 12-E, 13-E and 13-Z hydroxyacids, which were use- ful to estimate autoxidation and photooxidation state of bac- teria (Fig. 2; Rontani et al., 2003; Christodoulou et al., 2010). After solvent evaporation, residues were taken up in 300 µl of a mixture of pyridine and N,O- bis(trimethysilyl)triﬂuoroacetamide (BSTFA; Supelco) (2 : 1, v/v) and silylated for 1 h at 50 ◦C to convert OH- containing compounds to their TMSi-ether derivatives. After evaporation to dryness under a stream of N2, the derivatized residues were taken up in a mixture of ethyl acetate and BSTFA (to avoid desilylation of fatty acids) for analysis using GC-EIMS. It should be noted that under these conditions steran-3β, 5α, 6β-triols were only silylated at the 3 and 6 positions and thus need to be analyzed with great care (Rontani et al., 2012b). 2.3.2 Hydroperoxide-reduction NaBH4-reduction of the lipid extracts was carried out to reduce labile hydroperoxides resulting from photooxida- tion and autoxidation to alcohols that are amenable to gas chromatography-electron impact mass spectrometry (GC- EIMS). The ﬁlters were put in methanol (15 ml) and hy- droperoxides were reduced to the corresponding alcohols by excess NaBH4 (70 mg) (30 min at room temperature). During this treatment, ketones are also reduced and the possibility of some ester cleavage cannot be excluded. Compounds were identiﬁed by comparison of retention times and mass spectra with those of standards and quantiﬁed (calibration with external standards) by GC-EIMS. For low concentrations, or in the case of co-elutions, quantiﬁcation was achieved using selected ion monitoring (SIM). GC- EIMS analyses were carried out with an Agilent 6890 gas chromatograph connected to an Agilent 5973 Inert mass spectrometer. The following conditions were employed: 30 m × 0.25 mm (i.d.) fused silica column coated with HP- 1-MS (Agilent; 0.25 µm ﬁlm thickness); oven temperature programmed in three sequential steps: (i) 70 ◦C to 130 ◦C at 20 ◦C min−1; (ii) 130 ◦C to 250 ◦C at 5 ◦C min−1; and (iii) 250 ◦C to 300 ◦C at 3 ◦C min−1; carrier gas (He) maintained at 0.69 bar until the end of the temperature program and then programmed from 0.69 bar to 1.49 bar at 0.04 bar min−1; injector (on column with retention gap) temperature 50 ◦C; electron energy 70 eV; source temperature 190 ◦C; cycle time 1.99 and 8.3 cycles s−1 in SCAN and SIM modes, respec- tively. www.biogeosciences.net/9/4787/2012/ Biogeosciences, 9, 4787–4802, 2012 Biogeosciences, 9, 4787–4802, 2012 4790 J.-F. Rontani et al.: Intense photooxidative degradation of planktonic and bacterial lipids R R R R R R COOH OOH OOH COOH OOH COOH OOH COOH OOH CH2OH OOH COOH COOH OOH Photo- and autoxidation state of organic matter Autoxidation state of organic matter Tracers1 Application for the estimation of: 9 10 11 11 8 8 Hexadec-9-enoic acid (Palmitoleic acid) Octadec-9-enoic acid (Oleic acid) R = H R = CH3-CH2 8Z 11Z 9E 10E 8E 11E Chlorophyll phytyl side-chain Photodegradation state of organic matter Estimation of chlorophyll photodegradation Phytyldiol Application for the estimation of: Photodegradation state of organic matter Estimation of chlorophyll photodegradation Chlorophyll phytyl side-chain Phytyldiol Hexadec-9-enoic acid (Palmitoleic acid) OOH OOH OOH8 Octadec-11-enoic acid (Vaccenic acid) 41 HO RO HO HO OOH HO HO OH Photooxidation state of phytoplankton, higher plants and zooplankton Biodegradation state of phytoplankton, higher plants and zooplankton Autoxidation state of phytoplankton, higher plants and zooplankton ∆5-sterols hν 1O2 Bacteria 3 4 5 6 1 2 3β,5α,6β-steratriols (Triols) ∆4-stera-3β,6α/β-diols 5α(H)-stan-3β-ols (Stanols) Fig. 2. Formulae, vernacular names and potential applications of the main (non-exhaustive list) lipid tracers of degradation processes em- l d i h k 1Q iﬁd f N BH d i f h d id h di l h l d b il l i OOH ∆4-stera-3β,6α/β-diols Bacteria Fig. 2. Formulae, vernacular names and potential applications of the main (non-exhaustive list) lipid tracers of degradation processes em- ployed in the present work. 1Quantiﬁed after NaBH4-reduction of hydroperoxides to the corresponding alcohols and subsequent silylation. The residue obtained after extraction was dissolved in 4 ml of dichloromethane and separated in two equal subsamples. After evaporation of the solvent, degradation products were obtained for the ﬁrst subsample after acetylation (inducing complete conversion of hydroperoxides to the corresponding ketones, Mihara and Tateba, 1986) and saponiﬁcation and for the second after reduction with NaBD4 and saponiﬁcation. Comparison of the amounts of alcohols present after acety- lation and after NaBD4 reduction made it possible to esti- mate the proportion of hydroperoxides and alcohols present in the samples, while after NaBD4-reduction deuterium la- belling allowed to estimate the proportion of ketones really present in the samples (Marchand and Rontani, 2003). Acetylation was carried out in 300 µl of a mixture of pyri- dine and acetic anhydride (2 : 1, v/v), allowed to react at 50 ◦C overnight and then evaporated to dryness under nitro- gen. Biogeosciences, 9, 4787–4802, 2012 2.6 Quantiﬁcation of hydroperoxides and their ketonic and alcoholic degradation products A different treatment was employed to quantify hydroper- oxides and their ketonic and alcoholic degradation products. www.biogeosciences.net/9/4787/2012/ www.biogeosciences.net/9/4787/2012/ Biogeosciences, 9, 4787–4802, 2012 4791 J.-F. Rontani et al.: Intense photooxidative degradation of planktonic and bacterial lipids J.-F. Rontani et al.: Intense photooxidative degradation of planktonic and bacterial lipids . Rontani et al.: Intense photooxidative degradation of planktonic and bacterial lipids OOH COOH OOH COOH OOH COOH OOH COOH OOH R R R COOH COOH OOH R R R COOH OOH OOH COOH OOH COOH OOH COOH OOH CH2OH OOH COOH COOH OOH Photo- and autoxidation state of bacteria Autoxidation state of bacteria Photo- and autoxidation state of organic matter Autoxidation state of organic matter Tracers1 Application for the estimation of: 9 10 11 11 8 8 13 10 12 11 13 10 Hexadec-9-enoic acid (Palmitoleic acid) Octadec-9-enoic acid (Oleic acid) R = H R = CH3-CH2 Octadec-11-enoic acid (Vaccenic acid) 8Z 11Z 9E 10E 8E 11E 10Z 13Z 11E 12E 10E 13E Chlorophyll phytyl side-chain Photodegradation state of organic matter Estimation of chlorophyll photodegradation Phytyldiol 41 HO RO HO HO OOH HO HO OH OOH COOH OOH COOH OOH COOH OOH COOH OOH R R R COOH COOH OOH R R R COOH OOH OOH COOH OOH COOH OOH COOH OOH CH2OH OOH COOH COOH OOH Photo- and autoxidation state of bacteria Autoxidation state of bacteria Photooxidation state of phytoplankton, higher plants and zooplankton Biodegradation state of phytoplankton, higher plants and zooplankton Photo- and autoxidation state of organic matter Autoxidation state of organic matter Tracers1 Application for the estimation of: 9 10 11 11 8 8 13 10 12 11 13 10 Hexadec-9-enoic acid (Palmitoleic acid) Autoxidation state of phytoplankton, higher plants and zooplankton ∆5-sterols hν 1O2 Bacteria 3 4 5 6 1 2 Octadec-9-enoic acid (Oleic acid) R = H R = CH3-CH2 Octadec-11-enoic acid (Vaccenic acid) 8Z 11Z 9E 10E 8E 11E 10Z 13Z 11E 12E 10E 13E Chlorophyll phytyl side-chain Photodegradation state of organic matter Estimation of chlorophyll photodegradation Phytyldiol 3β,5α,6β-steratriols (Triols) ∆4-stera-3β,6α/β-diols 5α(H)-stan-3β-ols (Stanols) 2. Formulae, vernacular names and potential applications of the main (non-exhaustive list) lipid tracers of degradation processes em- ed in the present work. 1Quantiﬁed after NaBH4-reduction of hydroperoxides to the corresponding alcohols and subsequent silylation. 3 Results and discussion principally of diatoms (Boutry et al., 1971). The im- portant contribution of diatoms to these samples is also well supported by the presence of brassicasterol and 24- methylenecholesterol, which are major constituents of sev- eral diatom species (Lee et al., 1980). However, it may be noted that brassicasterol is also present in some dinoﬂagel- lates and in many haptophytes (Volkman, 1986, 2003). If the lack of dinosterol, which is present in some dinoﬂagellates (Mansour et al., 1999), does not support the presence of such organisms, a signiﬁcant contribution of dinoﬂagellates to the samples cannot be totally excluded since they do not system- atically produce dinosterol (Leblond and Chapman, 2002). In contrast, a contribution of haptophytes to the samples is very likely. Indeed, it was previously observed that larger haptophytes and diatoms co-dominated in near-surface as- semblages of the Beaufort Sea in summer (Hill et al., 2005). 3.1.1 Sterols In the different samples investigated the major sterols were cholesterol (10–31 %) and sitosterol (28–45 %) (Table 2). Although a substantial amount of cholesterol may be de- rived from diatoms or Prymnesiophycean algae (Volkman, 1986), its dominance suggests an important contribution of zooplanktonic faecal material to these samples. Indeed, it is well known that zooplankton convert much of the sterols produced by algae into cholesterol (Volkman et al., 1980; Prahl et al., 1984). Sitosterol is commonly associ- ated with terrestrial higher plants inputs (L¨utjohann, 2004), but some microalgae (and notably diatoms) are now known to produce this sterol (Volkman, 1986, 2003). The other sterols, found in lower ranges in this study, include 24- nor-cholesta-5,22E-dien-3β-ol, 27-nor-24-methylcholesta- 5,22E-dien-3β-ol, cholesta-5,22E-dien-3β-ol, brassicast- erol, 24-methylenecholesterol, 24-methylcholest-5-en-3β-ol (campesterol), 24-ethylcholesta-5,22E-dien-3β-ol (stigmas- terol) and 24-ethylcholesta-5,24(28)E-dien-3β-ol (fucos- terol) (Table 2). 24-nor-cholesta-5,22-dien-3β-ol, has been previously identiﬁed in a phytoplankton sample composed www.biogeosciences.net/9/4787/2012/ J.-F. Rontani et al.: Intense photooxidative degradation of planktonic and bacterial lipids terols in the different samples J.-F. Rontani et al.: Intense photooxidative degradation of planktonic and bacterial lipids 4792 Table 2. Relative percentage of sterols in the different samples. Sterols CA05-A22 CA05-A23 CA05-A24 CA16-A23 CA01-A1 G09-A2 G09-A3 24-Nor-cholesta-5,22E-dien-3β-ol 6.9 6.8 6.8 7.0 9.6 7.8 7.1 27-Nor-24-methylcholesta-5,22E-dien-3β-ol 2.9 2.5 3.4 2.8 2.1 3.2 3.7 Cholesta-5,22E-dien-3β-ol 3.2 3.7 3.1 3.2 4.0 5.0 5.9 Cholesterol 9.6 21.0 30.7 17.9 18.6 18.8 29.1 Brassicasterol 5.6 5.9 5.0 6.3 3.9 8.3 5.8 24-Methylenecholesterol 10.1 12.4 9.0 9.1 6.5 11.0 15.2 Campesterol 2.8 2.4 2.0 1.8 1.8 2.2 2.7 Stigmasterol 5.9 5.1 4.3 5.9 5.4 5.6 3.8 Sitosterol 45.0 33.5 28.3 39.2 38.8 31.5 22.5 Fucosterol 8.1 6.7 7.4 6.8 9.3 6.6 4.2 Table 3. Chlorophyll phytyl side-chain degradation products detected in the different samples and CPPI-based estimates of chlorophyll photodegradation. yl side-chain degradation products detected in the different samples and CPPI-based estimates of chlorophy de-chain degradation products detected in the different samples and CPPI-based estimates of chlorophyll Table 3. Chlorophyll phytyl side-chain degradation products detected in the different samples and CPPI-based estimates of chlorophyll photodegradation. Chlorophyll phytyl side-chain degradation products detected in the different samples and CPPI-based estim radation. Mooring 4,8,12-TMTD Pristanic Phytanic Dihydrophytol Pristane CPPI Chlorophyll acid (%)∗ acid (%)∗ acid (%)∗ (%)∗ (%)∗ photodegradation % CA05-A22 18.8 7.9 69.0 8.1 0.4 40.0 ± 2.5 99.8 ± 0.1 CA05-A23 41.6 19.4 108.9 5.8 1.9 53.3 ± 0.4 100 CA05-A24 66.1 39.2 72.7 11.5 3.1 35.6 ± 3.7 99.6 ± 0.2 CA16-A23 55.0 27.4 87.3 15.9 2.4 104.1 ± 0.1 100 G09-A2 37.2 27.9 218.9 5.4 – 120.9 ± 41.7 100 G09-A3 37.6 10.1 110.5 6.7 14.2 36.7 ± 4.9 99.6 ± 0.2 CA10-A1 – – 120.0 – – 22.7 ± 9.9 94.3 ± 5.1 ∗Relative to phytol. 3.1.2 Isoprenoids Despite the ap- parent strong contribution of diatom and zooplankton ma- terial to the samples (see previous sections), polyunsaturated fatty acids (PUFA) were not detected. The lack of these com- pounds could be attributed to their well known, very high re- activity towards photooxidation (Kawamura and Gagosian, 1987) and autoxidation (Frankel, 1998) processes and to the intense abiotic degradation state of the samples investigated (see Sect. 3.2). to the sampling material. Indeed, pelagic crustaceans de- crease the abundance of the chlorophyll phytyl side-chain when feeding herbivorously (Prahl et al., 1984; Harvey et al., 1987; Bradshaw et al., 1990). Any phytol remaining in the faeces of pelagic zooplankton after herbivorous feeding will be quickly removed on subsequent reprocessing of the faecal material (Bradshaw and Eglinton, 1993). Several phy- tol degradation products have been thus identiﬁed in zoo- planktonic faecal pellets during feeding experiments, includ- ing pristane, isomeric pristenes, isomeric phytadienes, di- hydrophytol and phytanic, pristanic, 4,8,12-TMTD and iso- meric phytenic acids (for a review see Rontani and Volkman, 2003). 3.1.2 Isoprenoids Dihydrophytol, pristane, 3,7,11,15-tetramethylhexadecanoic (phytanic), 2,6,10,14-tetramethylpentadecanoic (pristanic) and 4,8,12-trimethyltridecanoic (4,8,12-TMTD) acids could be detected in most of the total lipid extracts analysed (Ta- ble 3). The presence of relatively high proportions of these degradation products of the chlorophyll phytyl side-chain well supports the contribution of zooplanktonic faecal pellets www.biogeosciences.net/9/4787/2012/ Biogeosciences, 9, 4787–4802, 2012 www.biogeosciences.net/9/4787/2012/ J.-F. Rontani et al.: Intense photooxidative degradation of planktonic and bacterial lipids Table 4. Flux (µg m−2 d−1) of fatty acids and n-alkan-1-ols. Compounds CA05-A22 CA05-A23 CA05-A24 CA16-A23 CA10-A1 G09-A2 G09-A3 Fatty acids C14:0 227.2 802.0 317.9 523.6 16.5 275.3 404.2 C15:0 35.4 99.2 86.3 160.3 19.1 32.8 85.3 C16:119 (Palmitoleic) 465.3 1066.5 306.8 570.9 81.8 457.2 507.7 C16:0 971.0 2907.3 1024.7 4572.9 506.6 1581.8 1443.4 C17:1 – 21.8 38.5 – 11.4 – – C17:0 14.1 39.5 36.3 95.3 15.0 23.2 – C18:119 (Oleic) 170.4 438.2 248.3 442.8 183.1 181.6 353.1 C18:1111 (Vaccenic) 73.4 310.5 114.3 183.8 55.1 82.2 167.8 C18:1113 12.0 88.1 34.7 62.0 15.6 18.2 41.7 C18:0 368.6 992.1 172.5 1999.0 131.0 451.1 260.6 C20:1111 34.0 123.3 79.8 217.5 68.9 – – C20:0 6.0 17.4 10.5 49.8 6.7 12.0 9.2 C22:1111 5.7 23.7 16.1 28.6 3.8 5.6 21.7 C22:0 4.7 12.4 4.2 31.4 3.2 6.9 6.6 C24:1111 3.3 37.7 18.3 14.3 2.0 5.7 26.3 C24:0 2.7 16.4 5.8 41.3 4.9 7.2 5.7 n-Alkan-1-ols C14:0 5.3 105.2 73.9 94.3 15.9 24.6 236.2 C16:1 7.8 81.3 46.1 61.5 5.2 29.1 109.8 C16:0 58.9 357.1 273.8 408.1 32.1 122.6 471.0 C18:0 25.1 117.4 46.0 257.3 23.1 62.0 81.7 C20:1111 122.0 204.6 186.2 111.6 66.0 20.4 130.7 C22:1111 605.9 500.4 217.5 120.9 43.8 20.7 213.7 J.-F. Rontani et al.: Intense photooxidative degradation of planktonic and bacterial lipids 4793 nds CA05-A22 CA05-A23 CA05-A24 CA16-A23 CA10-A1 G09-A2 G09-A3 sible biological origins (plants, fungi, yeasts, bacteria, ani- mals or algae) (Harwood and Russell, 1984), while vaccenic acid is a typical biomarker for Gram-negative bacteria (Sicre et al., 1988; Keweloh and Heipieper, 1996). Small amounts of the very unusual octadec-13-enoic acid could be also de- tected (Table 4). The production of this compound was pre- viously observed during linolenic acid biohydrogenation by rumen microorganisms (Ward et al., 1964). In these samples it could thus result from biohydrogenation of phytoplank- tonic linolenic acid in the gut of large calanoid copepods that feed herbivorously and dominate the zooplankton as- semblage in the area (Forest et al., 2012a). 3.1.3 Fatty acids and n-alkan-1-ols Total lipid extracts of the different samples exhibited a dis- tribution of even-carbon number dominated fatty acids rang- ing from C14 to C24 (Table 4), suggesting the presence of a material dominated by marine organisms (plankton and bacteria). Long-chain (> C24) saturated fatty acids, which are characteristic of epicuticular waxes of terrestrial higher plants (Kolattukudy, 1977; Gagosian et al., 1987), could not be detected. The three dominant monounsaturated fatty acids appeared to be hexadec-9cis-enoic (palmitoleic), octadec- 9cis-enoic (oleic) and octadec-11cis-enoic (vaccenic) acids (Table 3). Palmitoleic and oleic acids have numerous pos- Interestingly, although sinking particles are generally con- sidered as the main contributors to the sedimentary record (Wakeham and Lee, 1989), after OsO4 treatment of the dif- ferent trap samples investigated, we failed to detect signif- icant amounts of monounsaturated fatty acids with a trans double bond, which were previously observed in surface 3.2.1 Chlorophyll Although the visible light-dependent degradation rate of the chlorophyll tetrapyrrole ring is three to four times higher than for its phytyl side-chain (Cuny et al., 1999), no spe- ciﬁc and stable tetrapyrrole photodegradation products could be identiﬁed in the literature. Type II photosensitized oxi- dation (i.e., involving 1O2) of the phytyl side-chain, how- ever, leads notably to the production of 3-methylidene- 7,11,15-trimethylhexadecan-1,2-diol (phytyldiol) (Rontani et al., 1994). Phytyldiol is ubiquitous in the marine envi- ronment and constitutes a stable and speciﬁc tracer for pho- todegradation of chlorophyll phytyl side-chain (Rontani et al., 1996; Cuny and Rontani, 1999). Further, the molar ra- tio phytyldiol : phytol (Chlorophyll Phytyl side-chain Pho- todegradation Index, CPPI) was proposed to estimate the extent of chlorophyll photodegraded in natural marine sam- ples (Cuny et al., 2002). 42 period chlorophyll was practically entirely photodegraded in sinking particles (Table 3). In the sample CA10 A1 collected in October, a strong photodegradation state of chlorophyll (> 94 %) was also observed (Table 3). www.biogeosciences.net/9/4787/2012/ Biogeosciences, 9, 4787–4802, 2012 J.-F. Rontani et al.: Intense photooxidative degradation of planktonic and bacterial lipids J.-F. Rontani et al.: Intense photooxidative degradation of planktonic and bacterial lipids J.-F. Rontani et al.: Intense photooxidative degradation of planktonic and bacterial lipids 4794 otooxidative degradation of planktonic and bacterial lipids Fig. 3. Percentages of photooxidation and autoxidation products (relative to the residual parent compound) of monounsaturated fatty acids and alkan-1-ols observed in G09-A2 (A) and G09-A3 (B) samples. sediments of this zone in very high proportions (Rontani et al., 2012a). The isomerization process responsible for the formation of trans monounsaturated fatty acids seems thus to act in sediments and not in sinking particles. Four principal types of storage lipids have been found in marine zooplankton: triacylglycerols, wax esters, phospho- lipids and diacylglycerol ethers (Lee et al., 2006). Wax es- ters are generally the major storage lipids in high latitude species (Lee et al., 2006). The most common alkan-1-ols of the wax esters found in herbivorous zooplankton are C20:1111 and C22:1111, while omnivorous or carnivorous zooplankton have a predominance of C14:0 and C16:0 alkan-1-ols (Lee and Nevenzel, 1979; Albers et al., 1996). C20:1111 and C22:1111 alcohols are only known to occur in copepods that undergo diapause (Graeve et al., 1994), which are largely distributed in the Arctic. The detection of high proportions of these two speciﬁc compounds in most of the total lipid extracts (Ta- ble 4) conﬁrmed the presence of high amounts of herbiv- orous zooplanktonic material in the different samples. The source of this material is probably lipid droplets remaining “trapped” in faecal pellets (Najdek et al., 1994) produced by the large herbivorous copepods Calanus hyperboreus and C. glacialis that undergo diapause. Fig. 3. Percentages of photooxidation and autoxidation products (relative to the residual parent compound) of monounsaturated fatty acids and alkan-1-ols observed in G09-A2 (A) and G09-A3 (B) samples. 3.2.2 Monounsaturated fatty acids and n-alkan-1-ols Percentages of photooxidation and autoxidation products (relative to the residual parent compound) of monounsaturated fatty acids and alkan-1-ols observed in CA16-A23 (A) and CA10-A1 (B) samples. lipopolysaccharide membrane of Gram negative bacteria (the dominant bacteria in the ocean) leads to the formation of re- active secondary products, such as peroxyl radicals, which may in turn accentuate cell death (Dahl et al., 1989). The in- tense oxidative stress resulting from singlet oxygen damages in bacteria should limit their growth (and thus biodegradation planktonic and bacterial lipids 4795 J.-F. Rontani et al.: Intense photooxidative degradation of planktonic and bacterial lipids 4795 -F. Rontani et al.: Intense photooxidative degradation of planktonic and bacterial lipids 4795 43 Fig. 4. Percentages of photooxidation and autoxidation products (relative to the residual parent compound) of monounsaturated fatty acids and alkan-1-ols observed in CA05-A22 (A), CA05-A23 (B) and CA05-A24 (C) samples. 43 Fig. 4. Percentages of photooxidation and autoxidation products (relative to the residual parent compound) of monounsaturated fatty acids and alkan-1-ols observed in CA05-A22 (A), CA05-A23 (B) and CA05-A24 (C) samples. Fig. 5. Percentages of photooxidation and autoxidation products (relative to the residual parent compound) of monounsaturated fatty acids and alkan-1-ols observed in CA16-A23 (A) and CA10-A1 (B) samples. 44 lipopolysaccharide membrane of Gram negative bacteria (the dominant bacteria in the ocean) leads to the formation of re- active secondary products, such as peroxyl radicals, which may in turn accentuate cell death (Dahl et al., 1989). The in- tense oxidative stress resulting from singlet oxygen damages in bacteria should limit their growth (and thus biodegradation processes) during the settling of sinking particles. 43 Fig. 4. Percentages of photooxidation and autoxidation products (relative to the residual parent compound) of monounsaturated fatty acids and alkan-1-ols observed in CA05-A22 (A), CA05-A23 (B) and CA05-A24 (C) samples. p g g g p Important amounts of oxidation products of C20:1111 and C22:1111 alkan-1-ols (values ranging from 10 to 800 % of the residual parent compound), which are speciﬁc components of zooplankton wax esters (Lee and Nevenzel, 1979; Albers et al., 1996), could be also detected in the different sam- ples (Figs. 3–5). The major part of these compounds results from the involvement of Type II (i.e. involving singlet oxy- gen) photoprocesses (Fig. 6). It is important to note that these oxidation products disappeared when the alkaline hydrolysis step was avoided during the treatment. 3.2.2 Monounsaturated fatty acids and n-alkan-1-ols Due to the lack of speciﬁcity of palmitoleic and oleic acids, their oxidation products have been used to assess abiotic degradation of bulk OM. The results obtained are summa- rized in Figs. 2–4. Photooxidation percentages of these two acids appeared to be very high (values ranging from 45 to 270 % relative to the residual parent compound) in summer, but not in fall (values < 20 %). These results, which are in good agreement with the total photodegradation of chloro- phyll observed in summer (Table 3), conﬁrm that during this period, photooxidation processes act very intensely on sink- ing particles of the Beaufort Sea. Autoxidation (free radical oxidation) processes also contributed to the degradation of these two fatty acids (10–30 %), but to a lesser extent than light-driven degradation (Figs. 3–5). The strong spatial vari- ability in the photooxidation stress (Figs. 3–5) could likely be attributed to differences in water clarity at the different sam- pling stations that typically increases from a shelf-edge loca- tion (CA05), to a mid-slope area (CA16), up to a basin-close environment (G09) (Table 1). Alternatively, visual inspection of sediment trap samples revealed that particles collected at CPPI ranges from 36 to 121 in the case of the samples collected in summer (Table 3). These values are particularly high when compared with CPPI previously measured in par- ticulate matter collected in summer in the Equatorial Paciﬁc (Rontani et al., 2011) and in the northwestern Mediterranean Sea (Cuny et al., 2002) (ranging from 1 to 8 and from 1 to 24, respectively). This attests to the exceptional efﬁciency of photooxidation processes in the Arctic Ocean region in summer, most probably because of the midnight sun that per- sists for 3 months (May–July) at 70◦N. On the basis of these very high CPPI values, it could be estimated that during this www.biogeosciences.net/9/4787/2012/ www.biogeosciences.net/9/4787/2012/ Biogeosciences, 9, 4787–4802, 2012 lanktonic and bacterial lipids 4795 Fig. 5. Percentages of photooxidation and autoxidation products (relative to the residual parent compound) of monounsaturated fatty acids and alkan-1-ols observed in CA16-A23 (A) and CA10-A1 (B) samples. J.-F. Rontani et al.: Intense photooxidative degradation of planktonic and bacterial lipids 4795 43 Fig. 4. Percentages of photooxidation and autoxidation products (relative to the residual parent compound) of monounsaturated fatty acids and alkan-1-ols observed in CA05-A22 (A), CA05-A23 (B) and CA05-A24 (C) samples. 44 Fig. 5. 3.2.2 Monounsaturated fatty acids and n-alkan-1-ols These results clearly showed that photooxidation processes acted directly on wax esters and not on the corresponding n-alkan-1-ol after enzy- matic hydrolysis. The high efﬁciency of Type II photooxida- tion processes in such micro-environments may be attributed to: (i) the high concentration of wax esters in the droplets trapped in faecal pellets (as discussed above) favoring the likelihood of interaction between singlet oxygen (produced from chlorophyll and phaeopigments contained in the pel- lets) and their double bonds and (ii) the apolar character of G09 were generally ﬁner and less aggregated than at CA16 and CA05, with the latter being obviously affected by the sinking of large diatom colonies (A. Forest, personal obser- vation, 2009). Organic matter contained in large aggregates could then be relatively more protected against photooxida- tion than in ﬁne particles that might offer a high surface-to- volume ratio. Oxidation products of vaccenic acid allowed us to es- timate photo- and autoxidation state of heterotrophic bac- teria associated to sinking particles. These bacteria were also strongly photodegraded in summer (photodegradation percentage ranging from 45 to 260 %) (Figs. 3–5) and weakly in fall (photodegradation percentage < 10 %). During the summer period, transfer of singlet oxygen from senes- cent phytoplanktonic cells to bacteria (Rontani et al., 2003; Christodoulou et al., 2010) seems thus to have been espe- cially efﬁcient. Vaccenic acid also appeared to be affected by autoxidation but less intensively (Figs. 3–5). Indeed, reaction of singlet oxygen with unsaturated components of the outer www.biogeosciences.net/9/4787/2012/ www.biogeosciences.net/9/4787/2012/ www.biogeosciences.net/9/4787/2012/ Biogeosciences, 9, 4787–4802, 2012 4796 J.-F. Rontani et al.: Intense photooxidative degradation of planktonic and bacterial lipids 796 J.-F. Rontani et al.: Intense photooxidative degradation of planktonic and bacterial lipids R' CH2O C R" O R' CH2O-C-R" O OOH R' CH2O-C-R" O OOH R' CH2O-C-R" O OOH R' CH2O-C-R" O OOH R' CH2OH OH R' CH2OH OH R' CH2OH OH R' CH2OH OH 1O2 hνννν 1. NaBH4 reduction 2. Alkaline hydrolysis Allylic rearrangement Allylic rearrangement R' = CH3- or CH3-(CH2)2- R = Polyunsaturated alkyl chain R" = Highly photooxidized polyunsaturated alkyl chain ig. 6. Type II (i.e. involving singlet oxygen) photooxidation of C20:1111 and C22:1111 alkan-1-ols. R' CH2O C R" O R' CH2O-C-R" O OOH R' OOH R' CH2O-C-R" O OOH R' OOH R' CH2OH OH R' CH2OH OH R' CH2OH OH R' CH2OH OH 1O2 hνννν 1. NaBH4 reduction 2. Alkaline hydrolysis Ally rearran Allylic rearrangement R' = CH3- or CH3-(CH2)2- R = Polyunsaturated alkyl chain R" = Highly photooxidized polyunsaturated alkyl chain Fig. 6. Type II (i.e. involving singlet oxygen) photooxidation of C20:1111 and C22:1111 alkan-1-ols. R' CH2O C R" O R' CH2O-C-R" O OOH R' CH2O-C-R" O OOH R' CH2O-C-R" O OOH R' CH2O-C-R" O OOH 1O2 hνννν Allylic rearrangement Allylic rearrangement OOH Allylic rearrangement Allylic rearrangement OOH 1. NaBH4 reduction 2. Alkaline hydrolysis OH OH OH H R" = Highly photooxidized polyunsaturated alkyl chain Fig. 6. Type II (i.e. involving singlet oxygen) photooxidation of C20:1111 and C22:1111 alkan-1-ols. 45 alkoxyl radicals. During NaBH4-reduction (carried out in or- der to avoid thermal breakdown of hydroperoxides during the treatment), hydroperoxides and ketones were reduced to the corresponding alcohols. The sum of hydroperoxy acids, ke- toacids and hydroxy acids was thus quantiﬁed in the form of hydroxyacids. A different treatment was employed (see Sect. 2.6.) in order to speciﬁcally quantify hydroperoxyacids and their main degradation products: hydroxyacids and ke- toacids. The results obtained in the case of the sample G09- A2 are summarized in Fig. 7a. It appears that in sinking par- ticles, a signiﬁcant proportion (ranging from 12 to 22 % of the sum of hydroperoxides, ketones and alcohols) of oxi- dation products of monounsaturated fatty acids are still un- der the form of hydroperoxides. An important ﬂux of these compounds should thus reach the seaﬂoor, inducing a strong oxidative stress in surface sediments. www.biogeosciences.net/9/4787/2012/ www.biogeosciences.net/9/4787/2012/ A release of sub-Arctic terrestrial POM in two different pools was re- cently proposed (Vonk et al., 2010a, b). A pool composed of mineral-bound POC derived from erosion, which has short initial residence times in the surface water and quickly settles to the sea ﬂoor, and another pool composed of higher plants debris mainly contributing to suspended particulate matter. These two pools probably settle too quickly or too slowly, respectively, to contribute signiﬁcantly to the material col- lected by the different traps deployed. Low proportions of 5α-stanols corresponding to 24- methylenecholesterol and brassicasterol could be observed in the different samples (values ranging from 3 to 15 % of the corresponding sterol) (Fig. 8). These values, which are very close to those generally considered as typical of healthy phytoplanktonic cells (5–10 %, Wakeham et al., 1997), sug- gest that biodegradation processes acted only very weakly on phytoplanktonic material. The lack of the correspond- ing ster-4-en-3-ones, which are classical bacterial metabo- lites of 15-sterols (de Leeuw and Baas, 1986; Wakeham, 1989) often detected in sinking particles (Bayona et al., 1989; Christodoulou et al., 2009), well supports this assumption. This apparently good resistance of phytodetritus against bac- terial degradation might result from the inhibition of bacte- rial growth when 1O2 generated by the photolysis of senes- cent phytoplanktonic cells in the euphotic zone is efﬁciently transferred to attached bacteria, as it was observed in the pre- vious section. Fig. 7. Relative percentages of intact hydroperoxides and their ke- tonic and alcoholic degradation products measured in the case of monounsaturated fatty acids and alkan-1-ols (A) and cholesterol (B) oxidation products. 46 sediments of this zone (Rontani et al., 2012a). Indeed, the formation of the formers was attributed to the involvement of enzymes catalyzing epoxidation of unsaturated fatty acids in the presence of alkylhydroperoxides as co-substrates, and this of the latters to cis/trans isomerization reactions induced by thiyl radicals resulting from the reaction of thiols with hy- droperoxides. www.biogeosciences.net/9/4787/2012/ These results support the mechanisms proposed in our companion paper to explain the presence of unusual very high proportions of epoxyacids and monounsaturated fatty acids with a trans double bond in these droplets. Indeed, the lifetime of singlet oxygen in apo- lar environments is longer, and its potential diffusion distance greater, than under polar conditions (Suwa et al., 1977). To our knowledge, this is the ﬁrst in situ demonstration of pho- todegradation of zooplanktonic faecal material. g Allylic hydroperoxides resulting from photo- and autoxi- dation of monounsaturated fatty acids may undergo: (i) het- erolytic cleavage catalyzed by protons (Frimer, 1979) lead- ing to the formation of ω-oxocarboxylic acids and other volatile products, and (ii) homolytic cleavage induced by transition metal ions (Pokorny, 1987; Schaich, 2005) or UVR (since hydroperoxides absorb in the UVR range; Hor- spool and Armesto, 1992). Homolytic cleavage of hydroper- oxyacids would lead to the formation of alkoxyl radicals, which can then: (i) abstract a hydrogen atom from another molecule to give hydroxyacids, (ii) lose a hydrogen atom to yield ketoacids, or (iii) undergo β-cleavage reaction af- fording volatile products. It may be noted that hydroxyacids and ketoacids may also result from disproportionation of two www.biogeosciences.net/9/4787/2012/ www.biogeosciences.net/9/4787/2012/ Biogeosciences, 9, 4787–4802, 2012 J.-F. Rontani et al.: Intense photooxidative degradation of Fig. 7. Relative percentages of intact hydroperoxides and their ke- tonic and alcoholic degradation products measured in the case of monounsaturated fatty acids and alkan-1-ols (A) and cholesterol (B) oxidation products. J.-F. Rontani et al.: Intense photooxidative degradation of planktonic and bacterial lipids J.-F. Rontani et al.: Intense photooxidative degradation of planktonic and bacterial lipids 4797 40 to 81 % relative to the parent sterol) than on brassicast- erol (arising from diatoms and/or Prymnesiophytes) (values ranging from 18 to 33 %) (Fig. 8). These differences suggest a higher efﬁciency of photodegradation processes in diatoms than in Prymnesiophytes. The similarity observed between the overall behaviors for brassicasterol and sitosterol (Fig. 8) with respect to degradation well supports a major contribu- tion of Prymnesiophytes to sitosterol. Interestingly, the sitos- terol and campesterol contained in suspended particles col- lected in this zone (mainly arising from terrestrial higher plants) were strongly autoxidized (Rontani et al., 2012b). The lack of sitosterol autoxidation products in all the total lipid extracts obtained from sediment trap samples thus con- ﬁrms that terrestrial higher plant material does not contribute signiﬁcantly to this sterol in sinking particles. 3.2.3 Sterols Degradation products of four model 15−sterols (choles- terol, 24-methylenecholesterol, brassicasterol and sitosterol) were quantiﬁed. The results obtained are summarized in Fig. 8. Photooxidation of 15−sterols appears less impor- tant than that of chlorophyll phytyl side-chain or monoun- saturated fatty acids (see Sects. 3.2.1 and 3.2.2). Indeed, degradation rate constants of 1O2-mediated photooxidation (type II photoreactions) are generally lower for 15−sterols than for chlorophyll phytyl side-chain and monounsaturated fatty acids (Rontani et al., 1998), possibly due to steric hin- drance during the attack of the sterol 15 double bond by 1O2 (Beutner et al., 2000). Photodegradation processes acted more intensively on 24-methylenecholesterol (mainly arising from diatoms) (photodegradation percentage ranging from As in the case of monounsaturated fatty acids, the spe- ciﬁc proportion of hydroperoxysterols and their alcoholic and ketonic degradation products have been also deter- mined. The results obtained are summarized in Fig. 7b. Hydroperoxysterols appeared to be more stable than hy- droperoxyacids in sinking particles (Fig. 7a); signiﬁcant amount of these compounds should thus reach the sedi- ments. This observation is in good agreement with the previ- ous detection of intact hydroperoxysterols in recent surﬁcial sediments (Rontani and Marchand, 2000). While ratio 14- 6α/6β-hydroperoxides/ 15-7α/7β-hydroperoxysterols pro- duced during irradiation of senescent phytoplanktonic cells ranged from 0.30 to 0.35 (Rontani et al., 1997), very high www.biogeosciences.net/9/4787/2012/ J.-F. Rontani et al.: Intense photooxidative degradation of planktonic and bacterial lipids J.-F. Rontani et al.: Intense photooxidative degradation of planktonic and bacterial lipids 4798 47 Fig. 8. Percentages of photooxidation and autoxidation products (relative to the residual parent compound) of sitosterol, cholesterol, brassi- casterol and 24-methylenecholesterol in the different samples investigated. ity was tentatively proposed by Christodoulou et al. (2009): 14-6-hydroperoxysterols > 15-7-hydroperoxysterols > 16- 5-hydroperoxysterols. The results obtained in the present study (Fig. 7b) well support this assumption. ity was tentatively proposed by Christodoulou et al. (2009): 14-6-hydroperoxysterols > 15-7-hydroperoxysterols > 16- 5-hydroperoxysterols. The results obtained in the present study (Fig. 7b) well support this assumption. values (ranging from 0.4 to 2.55) of the ratio cholest-4- en-3β,6α/6β-diols/cholest-5-en-3β, 7α/7β-diols were previ- ously measured (after NaBH4-reduction) in sediment trap samples collected in Mediterranean Sea (Christodoulou et al., 2009). These high values were attributed to a faster degradation of 15-7-hydroperoxysterols than 14-6- hydroperoxysterols during settling through the water column (Christodoulou et al., 2009). According to the theoretical sta- bility of the alkyl radicals formed during β-scission of the corresponding alkoxyl radicals, the following order of stabil- www.biogeosciences.net/9/4787/2012/ Biogeosciences, 9, 4787–4802, 2012 4798 J.-F. Rontani et al.: Intense photooxidative degradation of planktonic and bacterial 47 Fig. 8. Percentages of photooxidation and autoxidation products (relative to the residual parent compound) of sitosterol, cholesterol, b casterol and 24-methylenecholesterol in the different samples investigated. 4 Conclusions Lipids and their degradation products were quantiﬁed in seven samples of sinking particles collected with sediment traps in summer and fall across the Canadian Beaufort Shelf. These samples were dominated by diatoms and zooplank- tonic faecal pellets. Terrestrial higher plants resulting from Mackenzie River discharge did not contribute signiﬁcantly to the sinking material. During the summer period, Type II (i.e. involving singlet oxygen) photooxidation processes acted strongly on senescent phytoplanktonic cells, heterotrophic bacteria and zooplanktonic faecal material. Diatoms, which dominated the phytoplanktonic assemblage, appeared to be remarkably sensitive to photodegradation. Singlet oxygen transfer from phytodetritus to attached bacteria was particu- larly efﬁcient inducing strong oxidative damages in these het- erotrophic organisms. The presence of high amounts of pho- toproducts of C20:1111 and C22:1111 alcohols, which are spe- ciﬁc components of wax esters found in herbivorous cope- pods (Lee and Nevenzel, 1979; Albers et al., 1996), allowed us to demonstrate for the ﬁrst time the high efﬁciency of Type II photodegradation processes in zooplanktonic faecal material. In contrast, phytoplanktonic cells seemed to be relatively preserved towards biodegradation processes in sinking POM. As proposed by Rontani et al. (2011), there is a synergy be- tween senescing phytoplanktonic cells and attached bacteria and between photooxidation and biodegradation. Photolysis of chlorophyll in senescing algal cells produces singlet oxy- gen, which if transferred from algal cells to attached bacteria may inhibit bacterial growth and reduce the extent of het- erotrophic degradation. Therefore, it seems that there is a di- rect link between the photooxidation state of lipids of senes- cent phytoplanktonic cells in particles and their resistance to- wards biotic degradation. However, this does not exclude that the ﬂux of labile dissolved organic matter produced through the photo-cleavage of phytodetritus might sustain an active community of free-living bacteria around sinking particles (as proposed by Forest et al., 2012b). Acknowledgements. This study was conducted as part of the MALINA Scientiﬁc Program funded by ANR (Agence Nationale pour la Recherche) and French and European Space Agencies. Sediment trap samples were obtained through the Long-Term Oceanic Observatories Program of the ArcticNet Network of Centres of Excellence (NCE) of Canada. We would like to thank M. Babin, chief scientist of the cruise and coordinator of the MALINA program. This work was carried out in the framework of the transverse axis DEBAT (DEgradation Biotique et Abiotique de la maTi`ere organique en milieu marin: processus et interactions) of the MIO. J.-F. Rontani et al.: Intense photooxidative degradation of planktonic and bacterial lipids ment biomarkers are incomplete without a careful consider- ation of particle alteration and/or preservation during their transport from the euphotic zone to the benthic boundary layer. It is thus critical to understand how biotic and abiotic processes may alter the environmental signal encoded by the biomarker proxy. This is particularly important for Arctic re- gions, which indeed provide the earliest and most dramatic manifestations of global change (Hansen et al., 2012). Here, we have observed an extraordinary efﬁciency of photooxi- dation processes acting on sinking particles collected during summer in the Beaufort Sea. These processes, which should gradually increase as ice-free waters increase in extent and duration (NSIDC, 2012), appear to destroy most of the un- saturated components of organisms initially present in the settling material – thus strongly altering their lipid signa- ture. Although it is generally considered that sinking par- ticles are the main contributors to the sedimentary record (Wakeham and Lee, 1989), it may be noted that zooplank- tonic monounsaturated n-alcohols and their photooxidation products (present in very high proportion in these parti- cles) were totally absent in the underlying surface sediments (Rontani et al., 2012a). The disappearance of those products as they transit through the aphotic layer of the water column might result from the involvement of free radical oxidation processes (induced by the cleavage of photochemically pro- duced hydroperoxides), rather than from biodegradation pro- cesses limited by the photochemical alteration of attached bacteria. In the future, it will thus be essential to take into account the effects of abiotic degradation processes (rela- tively ignored until now in the literature) within sedimentary palaeoenvironmental reconstructions. www.biogeosciences.net/9/4787/2012/ Biogeosciences, 9, 4787–4802, 2012 4799 4 Conclusions Thanks are due to S. Belt and G. J. M. Versteegh for their helpful and useful comments. In addition, we demonstrated that strongly photodegraded sinking particles contained an important amount of intact hy- droperoxides. After sedimentation, these compounds should induce a strong oxidative stress in surface sediments, which could be at the origin of the formation of epoxyacids and monounsaturated fatty acids with a trans double bonds pre- viously detected in unusual high proportion in this zone (Rontani et al., 2012a). Edited by: M. Babin The publication of this article is ﬁnanced by CNRS-INSU. Edited by: M. Babin In order to put the current global warming trend into the context of natural climate variability, it is essential to re- construct sedimentary palaeoenvironments. Lipid biomark- ers preserved in bottom sediments are particularly useful for this purpose, since they represent the ultimate signature of sympagic and pelagic productivity (e.g. Volkman et al., 1998; Wakeham et al., 1997; Rontani and Volkman, 2005; Bianchi and Canuel, 2011). However, reconstructions based on sedi- Biogeosciences, 9, 4787–4802, 2012 www.biogeosciences.net/9/4787/2012/ References Dahl, T. A., Midden, W. R., and Hartman, P. E.: Comparison of killing of Gram-negative and Gram-positive bacteria by pure sin- glet oxygen, J. Bacteriol., 171, 2188–2194, 1989. Albers, C. S., Kattner, G., and Hagen, W.: The compositions of wax esters, triacylglycerols and phospholipids in Arctic and Antarc- tic copepods: evidence of energetic adaptation, Mar. Chem., 55, 347–358, 1996. De Leeuw, J. W. and Baas, M.: Early stage diagenesis of steroids, in: Biological Markers in the Sedimentary Record, edited by: Johns, R. B., Elsevier, 101–123, 1986. Arrigo, K. R., van Dijken, G., and Pabi, S.: Impact of a shrinking Arctic ice cover on marine primary production, Geophys. Res. Lett., 35, L19603, doi:10.1029/2008GL035028, 2008. Forest, A., Stemmann, L., Picheral, M., Burdorf, L., Robert, D., Fortier, L., and Babin, M.: Size distribution of particles and zoo- plankton across the shelf-basin system in southeast Beaufort Sea: combined results from an Underwater Vision Proﬁler and verti- cal net tows, Biogeosciences, 9, 1301–1320, doi:10.5194/bg-9- 1301-2012, 2012a. Bacon, M. P., Huh, C. A., Fleer, A. P., and Deuser, W. G.: Seasonal- ity in the ﬂux of natural radionuclides and plutonium in the deep Sargasso Sea, Deep-Sea Res., 32, 73–286, 1985. Bayona, J. M., Farran, A., and Albaiges, J.: Steroid alcohols and ketones in coastal waters of the western Mediterranean: sources and seasonal variability, Mar. Chem., 27, 79–104, 1989. Forest, A., Babin, M., Stemmann, L., Picheral, M., Sampei, M., Fortier, L., Gratton, Y., B´elanger, S., Devred, E., Sahlin, J., Doxaran, D., Joux, F., Ortega-Retuerta, E., Jeffrey, W. H., Mart´ın, J., Gasser, B., and Miquel, J. C.: Ecosystem function and par- ticle ﬂux dynamics across the Mackenzie Shelf (Beaufort Sea, Arctic Ocean): an integrative analysis of spatial variability and biophysical forcings, Biogeosciences Discuss., 9, 10883–10960, doi:10.5194/bgd-9-10883-2012, 2012b. Beutner, S., Bloedorn, B., Hoffman, T., and Martin, H. D.: Syn- thetic singlet oxygen quenchers, 226-241, edited by: Packer, L. and Sies, H., in: Methods in enzymology, 319, Academic press, 2000. Bianchi, T. S.: The role of terrestrially derived organic carbon in the coastal ocean: A changing paradigm and the priming effect, Proc. Nat. Acad. Sci., 108, 19473–19481, 2011. Frankel, E. N.: Lipid Oxidation, the Oily Press, Dundee, 1–303, 1998. Bianchi, T. S. and Canuel, E. A.: Chemical Biomarkers in Aquatic Ecosystems, Princeton University Press, USA, 19471–19481, 2011. Frey, K. E. and McClelland, J. W.: Impacts of permafrost degra- dation on Arctic river biogeochemistry, Hydrol. Processes, 23, 169–182, 2009. References Boutry, J., Alcaide, A., and Barbier, M.: De la pr´esence d’un st´erol en C26 d’un plancton marin v´eg´etal, C. R. Acad. Sci., Paris, 272, 1022–1023, 1971. Frimer, A. A.: The reaction of singlet oxygen with oleﬁns: the ques- tion of mechanism. Chem. Rev., 79, 359–387, 1979. Gagosian, R. B., Peltzer, E. T., and Merrill, J. T.: Long-range trans- port of terrestrially derived lipids in aerosols from the south Pa- ciﬁc, Nature, 325, 800–803, 1987. Bradshaw, S. A. and Eglinton, G.: Marine invertebrate feeding and the sedimentary lipid record, edited by: Engel, M. H. and Macko, S. A., Org. Geochem., Plenum Press, New York, 225–235, 1993. Graeve, M., Kattner, G., and Hagen, W.: Diet-induced changes in the fatty acid composition of Arctic herbivorous copepods, J. Exp. Mar. Biol. Ecol., 182, 97–110, 1994. Bradshaw, S. A., O’Hara, S. C. M., Corner, E. D. S., and Eglinton, G.: Changes in lipids during simulated herbivorous feeding by the marine crustacean Neomysis integer, J. Mar. Biol. Assoc. UK, 70, 225–243, 1990. Hansen, J., Ruedy, R., Sato, M., and Lo, K.: GISS Surface Tempera- ture Analysis (GISTEMP): Global Temperature in 2011, Trends, and Prospects, http://data.giss.nasa.gov/gistemp/2011/, 2012. Brunskill, G. J.: Environmental features of the Mackenzie system, in: The ecology of river systems, edited by: Davis, B. R. and Walker, W. F., Dordrecht, The Netherlands, Dr. W. Junk Publica- tions, 435–471, 1986. Harvey, H. R., Eglinton, G., O’Hara, S. C. M., and Corner, E. D. S.: Biotransformation and assimilation of dietary lipids by Calanus feeding on a dinoﬂagellate, Geochim. Cosmochim. Acta, 51, 3031–3040, 1987. Christodoulou, S., Marty, J.-C., Miquel, J.-C., Volkman, J. K., and Rontani, J.-F.: Use of lipids and their degradation products as biomarkers for carbon cycling in the northwestern Mediterranean Sea, Mar. Chem., 113, 25–40, 2009. Harwood, J. L. and Russell, N. L.: Lipids in Plants and Microbes, George Allen and Unwin, London, 1–162, 1984. Heussner, S., Ratti, C., and Carbonne, J.: The PPS 3 time series sediment traps and the trap sample processing techniques used during the ECOMARGE experiment, Cont. Shelf Res., 10, 943– 958, 1990. Christodoulou, S., Joux, F., Marty, J.-C., Semp´er´e, R., and Rontani, J.-F.: Comparative study of UV and visible light induced degra- dation of lipids in non-axenic senescent cells of Emiliania hux- leyi, Mar. Chem., 119, 139–152, 2010. Hill, P. S.: Controls on ﬂoc size in the sea, Oceanogr., 11, 13–18, 1998. Cuny, P. J.-F. Rontani et al.: Intense photooxidative degradation of planktonic and bacterial lipids 4800 J.-F. Rontani et al.: Intense photooxidative degradation of planktonic and bacterial lipids W.: Geochemistry of sterols in sediments from Black Sea and the southwest African shelf and slope, Org. Geochem., 2, 103–113, 1980. Rontani, J.-F., Grossi, V., Faure, R., and Aubert, C.: “Bound” 3-methylidene-7,11,15-trimethylhexadecan-1,2-diol: a new iso- prenoid marker for the photodegradation of chlorophyll-a in sea- water, Org. Geochem., 21, 135–142, 1994. Lee, R. F. and Nevenzel, J. C.: Wax esters in the marine environ- ment: origin and composition of the wax from the Bute Inlet, British Columbia, J. Fish. Res. Bd. Can., 36, 1519–1523, 1979. Rontani, J.-F., Raphel, D., and Cuny, P.: Early diagenesis of the in- tact and photooxidized chlorophyll phytyl chain in a recent tem- perate sediment, Org. Geochem., 24, 825–832, 1996. Lee, R. F., Hagen, W., and Kattner, G.: Lipid storage in marine zoo- plankton. Mar. Ecol. Prog. Ser., 307, 273–306, 2006. Rontani, J.-F., Cuny, P., and Aubert, C.: Rates and mechanism of light-dependent degradation of sterols in senescent cells of phy- toplankton, J. Photochem. Photobiol., A: Chem., 111, 139–144, 1997. L¨utjohann, D.: Sterol autoxidation: from phytosterols to oxyphytos- terols, Brit. J. Nutr., 91, 3–4, 2004. Macdonald, R. W., Solomon, S. M., Cranston, R. E., Welch, H. E., Yunker, M. B., and Gobeil, C.: A sediment and organic carbon budget for the Canadian Beaufort Shelf, Mar. Geol., 144, 255– 273, 1998. Rontani, J.-F., Cuny, P., and Grossi, V.: Identiﬁcation of a pool of lipid photoproducts in senescent phytoplanktonic cells, Org. Geochem., 29, 1215–1225, 1998. Mansour, M. P., Volkman, J. K., Jackson, A. E., and Blackburn, S. I.: The fatty acid and sterol composition of ﬁve marine dinoﬂag- ellates, J. Phycol., 35, 710–720, 1999. Rontani, J.-F., Koblizek, M., Beker, B., Bonin, P., and Kolber, Z.: On the origin of cis-vaccenic photodegradation products in the marine environment, Lipids, 38, 1085–1092, 2003. Marchand, D. and Rontani, J.-F.: Characterisation of photooxida- tion and autoxidation products of phytoplanktonic monounsatu- rated fatty acids in particulate matter and recent sediments, Org. Geochem., 32, 287–304, 2001. Rontani, J.-F., Zabeti, N., and Wakeham, S. G.: The fate of marine lipids: Biotic vs. abiotic degradation of particulate sterols and alkenones in the Northwestern Mediterranean Sea, Mar. Chem., 113, 9–18, 2009. Marchand, D. and Rontani, J.-F.: Visible light-induced oxidation of lipid components of purple sulphur bacteria: a signiﬁcant process in microbial mats, Org. Geochem., 34, 61–79, 2003. Rontani, J.-F., Zabeti, N., and Wakeham, S. J.-F. Rontani et al.: Intense photooxidative degradation of planktonic and bacterial lipids 4801 implications for sediment transport and deposition, Cont. Shelf Res., 26, 41–81, 2006. Horspool W. and Armesto D.: Organic Photochemistry: A Compre- hensive Treatment, Ellis Horwood, Chichester, UK, 1–521, 1992. Kawamura, K. and Gagosian, R. B.: Implication of ω-oxocarboxylic acids in the remote marine atmosphere for photo-oxidation of unsaturated fatty acids, Nature, 325, 330–332, 1987. Pokorny, J.: Major factors affecting the autoxidation of lipids, 141– 206, edited by: H. W.-S. Chan, in: Autoxidation of unsaturated lipids, Academic Press, 1987. Keweloh, H. and Heipieper, H. J.: Trans-unsaturated fatty acids in bacteria, Lipids, 31, 129–137, 1996. Prahl, F. G., Eglinton, G., Corner, E. D. S., O’Hara, S. C. M., and Forsberg, T. E. V.: Changes in plant lipids during passage through the gut of Calanus, J. Mar. Biol. Assoc. UK, 64, 317–334, 1984. Knap, A., Michaels, A., Close, A., Ducklow, H., and Dickson, A.: Protocols for the Joint Global Ocean Flux Study (JGOFS) core measurements, JGOFS Report, 19, 155–162, 1996. Rachold, V., Eicken, H., Gordeev, V. V., Grigoriev, M. N., Hub- berten, H. W., Lisitzin, A. P., Shevchenko, V. P., and Schirrmeis- ter, L.: Modern terrigenous organic carbon input to the Arctic Ocean, in: The Organic Carbon Cycle in the Arctic Ocean, edited by: Stein, R. S. and Macdonald, R. W., Springer, New York, 33– 55, 2004. Kolattukudy, P. E.: Lipid polymers and associated phenols, their chemistry, biosynthesis, and role in pathogenesis, Recent Adv. Phytochem., 77, 185–246, 1977. Lalande, C., B´elanger, S., and Fortier, L.: Impact of a decreasing sea ice cover on the vertical export of particulate organic carbon in the northern Laptev Sea, Siberian Arctic Ocean, Geophys. Res. Lett., 36, L21604, doi:10.1029/2009GL040570, 2009. Rontani, J.-F. and Marchand, D.: 15-Stenol photoproducts of phy- toplanktonic origin: a potential source of hydroperoxides in ma- rine sediments?, Org. Geochem., 31, 169–180, 2000. Leblond, J. D. and Chapman, P. J.: A survey of the sterol composi- tion of the marine dinoﬂagellates Karenia brevis, Karenia miki- motoi and Karlodinium micrum: distribution of sterols within other members of the class dinophyceae, J. Phycol., 38, 670–682, 2002. Rontani, J.-F. and Volkman, J. K.: Phytol degradation products as biogeochemical tracers in aquatic environments, Org. Geochem., 34, 1–35, 2003. Rontani, J.-F. and Volkman, J. K.: Lipid characterization of coastal hypersaline cyanobacterial mats from the Camargue (France), Org. Geochem., 36, 251–272, 2005. Lee, C., Gagosian, R. B., and Farrington, J. References and Rontani, J.-F.: On the widespread occurrence of 3- methylidene-7,11,15-trimethylhexadecan-1,2-diol in the marine environment: a speciﬁc isoprenoid marker of chlorophyll pho- todegradation, Mar. Chem., 65, 155–165, 1999. Hill, P. R., Blasco, S. M., Harper, J. R., and Fissel, D. B.: Sedi- mentation on the Canadian Beaufort Shelf, Cont. Shelf Res., 11, 821–842, 1991. Cuny, P., Romano, J.-C., Beker, B., and Rontani, J.-F.: Comparison of the photo degradation rates of chlorophyll chlorin ring and phytol side chain in phytodetritus: is the phytyldiol versus phytol ratio (CPPI) a new biogeochemical index?, J. Exp. Mar. Biol. Ecol., 237, 271–290, 1999. Hill, V., Cota, G., and Stockwell, D.: Spring and summer phyto- plankton communities in the Chukchi and eastern Beaufort Seas, Deep-Sea Res., 52, 3369–3385, 2005. Horner, R. A. and Schrader, G. C.: Relative contributions of ice al- gae, phytoplankton and benthic microalgae to primary produc- tion in nearshore regions of the Beaufort Sea, Arctic, 35, 485– 503, 1982. Cuny, P., Marty, J.-C., Chiaverini, J., Vescovali, I., Raphel, D., and Rontani, J.-F.: One-year seasonal survey of the chlorophyll pho- todegradation process in the Northwestern Mediterranean sea, Deep Sea Res., II 49, 1987–2005, 2002. Biogeosciences, 9, 4787–4802, 2012 www.biogeosciences.net/9/4787/2012/ J.-F. Rontani et al.: Intense photooxidative degradation of planktonic and bacterial lipids 4802 organic matter in the Northwest Black Sea mixing zone, Mar. Chem., 79, 243–259, 2002. organic matter in the Northwest Black Sea mixing zone, Mar. Chem., 79, 243–259, 2002. Vonk, J. E., S´anchez-Garc´ıa, L., Semiletov, I., Dudarev, O., Eglin- ton, T., Andersson, A., and Gustafsson, ¨O.: Molecular and ra- diocarbon constraints on sources and degradation of terres- trial organic carbon along the Kolyma paleoriver transect, East Siberian Sea, Biogeosciences, 7, 3153–3166, doi:10.5194/bg-7- 3153-2010, 2010b. Schaich, K. M.: Lipid oxidation: theoretical aspects, in: Bailey’s Industrial Oil and Fat Products, Sixth edition, edited by: Shahidi, F., John Wiley & Sons, 269–355, 2005. Sicre, M.-A., Paillasseur, J. L., Marty, J.-C., and Saliot, A.: Charac- terization of seawater samples using chemometric methods ap- plied to biomarker fatty acids, Org. Geochem., 12, 281–288, 1988. Wakeham, S. G.: Reduction of stenols to stanols in particulate mat- ter at oxic-anoxic boundaries in seawater, Nature, 342, 787–790, 1989. Suwa, K., Kimura, T., and Schaap, A. P.: Reactivity of singlet molecular oxygen with cholesterol in a phospholipidic mem- brane matrix: a model for oxidative damage of membranes, Biochem. Biophys. Res. Commun., 75, 785–792, 1977. Wakeham, S. G. and Lee, C.: Organic geochemistry of particulate matter in the ocean: The role of particles in oceanic sedimentary cycles, Org. Geochem., 14, 83–96, 1989. Wakeham, S. G., Hedges, J. I., Lee, C., Peterson, M. L., and Hernes, P. J.: Compositions and transport of lipid biomarkers through the water column and surﬁcial sediments of the EqPac Ocean, Deep- Sea Res., 44, 2131–2162, 1997. Valli`eres, C., Retamal, L., Ramlal, P., Osburn, C. L., and Vincent, W. F.: Bacterial production and microbial food web structure in a large Arctic river and the coastal Arctic Ocean, J. Mar. Syst., 74, 756–773, doi:10.1016/j.jmarsys.2007.12.002, 2008. Ward, P. F. V., Scott, T. W., and Dawson, R. M. C.: The hydro- genation of unsaturated fatty acids in the ovine digestive tract, Biochem. J., 92, 60–68, 1964. Volkman, J. K.: A review of sterol markers for marine and terrige- nous organic matter, Org. Geochem., 9, 83–99, 1986. Volkman, J. K.: Sterols in microorganisms, Appl. Microbiol. Biotechnol., 60, 495–506, 2003. Wassmann, P. and Reigstad, M.: Future Arctic Ocean seasonal ice zones and implications for pelagic-benthic coupling, Oceanogra- phy, 24, 220–231, doi:10.5670/oceanog.2011.74, 2011. Volkman, J. K., Corner, E. D. S., and Eglinton, G.: Transformations of biolipids in the marine food web and in underlying bottom sediments, Colloques Int. Cent. Natn. Rech. J.-F. Rontani et al.: Intense photooxidative degradation of planktonic and bacterial lipids G.: Degradation of par- ticulate organic matter in the equatorial Paciﬁc Ocean: biotic or abiotic?, Limnol. Oceanogr., 56, 333–349, 2011. McCave, I. N.: Size spectra and aggregation of suspended particles in the ocean, Deep-Sea Res., 31, 329–352, 1984. Rontani, J.-F., Charriere, B., Petit, M., Vaultier, F., Heipieper, H., Link, H., Chailloux, G., and Semp´er´e, R.: Degradation state of organic matter in surface sediments from the Southern Beaufort Sea: a lipid approach, Biogeosciences, 9, 3513–3530, doi:10.5194/bg-9-3513-2012, 2012a. Mihara, S. and Tateba, H.: Photosensitized oxygenation reactions of phytol and its derivatives, J. Org. Chem., 51, 1142–1144. 1986. Milliman, J. D. and Meade, R. H.: World-wide delivery of river sed- iment to the oceans, J. Geol., 91, 1–21, 1983. Rontani, J.-F., Charriere, B., Semp´er´e, R., Doxaran, D., Vaultier, F., Vonk, J. E., and Volkman, J. K.: Lipid biomarker evidence for ex- tensive biotic and abiotic degradation of terrestrial organic matter in Arctic coastal waters, Geochim. Cosmochim. Acta, submitted, 2012b. Najdek, M., Puskaric, S., and Bochdansky, A.: Contribution of zoo- plankton lipids to the ﬂux of organic matter in the northern Adri- atic Sea, Mar. Ecol. Prog. Ser., 111, 241–249, 1994. NSIDC: National Snow and Ice Data Center, Arctic Sea Ice News and Analysis, http://nsidc.org/arcticseaicenews/, 2012. Saliot, A., Parrish, C. C., Sadouni, N., Bouloubassi, I., Fillaux, J., and Cauwet, G.: Transport and fate of Danube Delta terrestrial O’Brien, M. C., Macdonald, R. W., Melling, H., and Iseki, K.: Par- ticle ﬂuxes and geochemistry on the Canadian Beaufort Shelf: Biogeosciences, 9, 4787–4802, 2012 www.biogeosciences.net/9/4787/2012/ Biogeosciences, 9, 4787–4802, 2012 Biogeosciences, 9, 4787–4802, 2012 J.-F. Rontani et al.: Intense photooxidative degradation of planktonic and bacterial lipids Scient., 293, 185– 197, 1980. Wassmann, P., Bauerfeind, E., Fortier, M., Fukuchi, M., Hargrave, B., Honjo, S., Moran, B., Noji, T., N¨othig, E. M, and Peinert, R.: Particulate organic carbon ﬂux to the sea ﬂoor, edited by: Stein, Wassmann, P., Bauerfeind, E., Fortier, M., Fukuchi, M., Hargrave, B., Honjo, S., Moran, B., Noji, T., N¨othig, E. M, and Peinert, R.: Particulate organic carbon ﬂux to the sea ﬂoor, edited by: Stein, R., MacDonald, R., in: The Organic Carbon Cycle in the Arctic Ocean, Springer-Verlag, New York, 101–138, 2004. Volkman, J. K., Barrett, S. M., Blackburn, S. I., Mansour, M. P., Sikes, E. L., and Gelin, F.: Microalgal biomarkers: A review of recent research developments, Org. Geochem., 29, 1163–1179, 1998. R., MacDonald, R., in: The Organic Carbon Cycle in the Arctic Ocean, Springer-Verlag, New York, 101–138, 2004. Vonk, J. E., van Dongen, B. E., and Gustafsson, O.: Se- lective preservation of old organic carbon ﬂuvially released from sub-Arctic soils, Geophys. Res. Lett., 37, L11605, doi:10.1029/2010gl042909, 2010a. Biogeosciences, 9, 4787–4802, 2012 www.biogeosciences.net/9/4787/2012/
https://openalex.org/W2160678121	https://europepmc.org/articles/pmc6159127?pdf=render	English	null	Intravascular lipoma of the renal vein	BJR case reports	2,015	cc-by	2,533	Cite this article as: Cite this article as: Doyle Z, Wolford B, Morshedi MM, Santillan CS. Intravascular lipoma of the renal vein. BJR Case Rep 2015; 1: 20150072 ABSTRACT Lipomas are benign neoplasms composed of adipocytes encased in a fibrous capsule. Intravascular lipomas are rare and almost always incidental findings. In the published literature, the majority are described within the inferior vena cava (IVC) and less frequently reported in the superior vena cava, brachiocephalic vein, subclavian vein, internal jugular vein, external iliac vein and common femoral vein. We present the case of a 59-year-old male who presented with a symptomatic ureteral calculus and was found to have an intravascular lipoma of the right renal vein with extension into the IVC. To our knowledge, this is the first ever report of an intravascular lipoma in the renal vein. We discuss the imaging characteristics of intravascular lipomas and the differential diagnosis that should be considered. CLINICAL PRESENTATION homogeneous, circumscribed, non-enhancing fatty mass. CT imaging typically shows a well-defined, ovoid, non- enhancing, hypoattenuating mass consistent with fat den- sity. Intravenous contrast demonstrates a filling defect, cor- responding to the intraluminal location of the lipomas, although some published cases report both intra- and extravascular extension.1,2 MRI was obtained in eight pub- lished case reports2–9 to confirm location and fatty compo- sition, and demonstrated non-enhancing, intravascular space occupying, T1 and T2 hyperintense, circumscribed fatty lesions that subtract out on fat-subtraction imaging, similar to our case described above. Others have used angiography to assess the level of obstruction, and in one angiographic study, the obstruction was demonstrated and abnormal venous collaterals were seen, suggesting long- standing disruption of venous flow.2 However, angiogra- phy is not routinely performed if other imaging modalities do not show significant vascular obstruction. A 59-year-old male initially presented to an outside hospi- tal with right flank pain and gross haematuria due to a par- tially obstructing right ureteral calculus. 1Z DOYLE, BA, 2B WOLFORD, MD, 3M M MORSHEDI, MD, PhD and 2C S SANTILLAN 1School of Medicine, University of California San Diego, La Jolla, CA, USA 2Department of Radiology, University of California San Diego, San Diego, CA, USA 3Department of Radiology, University of California Los Angeles, Los Angeles, CA, USA Address correspondence to: Dr Maud M. Morshedi E-mail: maud.morshedi@my.rfums.org Address correspondence to: Dr Maud M. Morshedi E-mail: maud.morshedi@my.rfums.org Address correspondence to: Dr Maud M. Morshedi E-mail: maud.morshedi@my.rfums.org IMAGING FINDINGS A non-contrast CT scan of the abdomen/pelvis performed at the time of diagnosis incidentally revealed a homoge- neous smooth tubular fat-attenuating mass within the anterior right renal vein, extending into the inferior vena cava (IVC) and terminating at the level of the proximal intrahepatic portion of the vena cava (Figure 1a). The patient was referred to our institution for an MRI to fur- ther characterize the mass. The MRI showed that the well-circumscribed mass was hyperintense on T2 imaging, subtracted out on fat-saturated images, showed chemical shift artefact on out-of-phase imaging and did not enhance with contrast (Figure 1b–e). These findings were compatible with a lipoma extending from the segmental branches of the right renal vein to the main renal vein and into the IVC. The mass occupied the majority of the lumen of the main renal vein and approximately 50% of the lumen of the affected IVC. Received: 22 January 2015 Revised: 14 April 2015 Accepted: 26 April 2015 Cite this article as: Doyle Z, Wolford B, Morshedi MM, Santillan CS. Intravascular lipoma of the renal vein. BJR Case Rep 2015; 1: 20150072 BJR\|case reports doi: 10.1259/bjrcr.20150072 Received: 22 January 2015 Revised: 14 April 2015 Accepted: 26 April 2015 Cite this article as: Doyle Z, Wolford B, Morshedi MM, Santillan CS. Intravascular lipoma of the renal vein. BJR Case Rep 2015; 1: 20150072 BJR\|case reports doi: 10.1259/bjrcr.20150072 © 2015 The Authors. Published by the British Institute of Radiology. This is an open access article under the terms of the Creative Commons Attribution 4.0 International License, which permits unrestricted use, distribution and reproduction in any medium, provided the original author and source are credited DIFFERENTIAL DIAGNOSIS In addition to lipomas, the differential diagnosis of fat- containing intravascular masses includes a number of benign and malignant disease processes that may be dif- ferentiated by imaging. In the case of a fatty mass, lipo- sarcoma must be considered. Typically, on imaging, liposarcomas appear as heterogeneous lesions. In con- trast, lipomas are characteristically composed entirely of adipose tissue and therefore demonstrate homogeneous fatty attenuation. However, one will note that 31% of lipomas have non-adipose areas on imaging, a feature A review of the currently available literature on PubMed was performed by using the search terms “lipoma”, “intra- vascular lipoma”, “intravenous lipoma”, “renal vein lipoma”, “inferior vena cava lipoma”, “IVC lipoma” and “renal lipoma” that found 27 reported cases of intravascu- lar lipoma. Imaging characteristics are typical for a lipoma, with nearly all imaging modalities demonstrating a BJR\|case reports Z Doyle et al Figure 1. Coronal CT image (a) demonstrates a well-circumscribed homogeneous fat-attenuating intravascular mass (arrows) extending from the renal veins to the inferior vena cava (IVC). Axial T2 MR image (b) shows a hyperintense round mass in the IVC that subtracts out on post-contrast T1 fat-saturated axial (c) and coronal (d) MR images as well as demonstrates chemical shift arte- fact on out-of-phase coronal MRI (e). The lesion is space filling, non-enhancing and nearly all fat, most compatible with an intravas- cular lipoma. DIFFERENTIAL DIAGNOSIS of leiomyosarcomas arising from the renal vein and the IVC.18,19 They are typically seen on CT and MRI as non-fatty, lobulated soft-tissue masses of intermediate attenuation signal, with areas of low attenuation that correspond to necrosis.19,20 that is more typical of liposarcomas.10 Other features that sug- gest malignancy include thick septa, nodular and/or globular areas of non-adipose tissue, associated non-adipose masses and total non-adipose tissue comprising more than 25% of the lesion.10 Gaskin and Helms found that MRI was 100% specific in diagnosing a simple lipoma when a fatty mass contained no or only a few thin septa and minimal or no areas of enhancement or high T2 signal.11 Haemangioendothelioma refers to a heterogeneous group of low-grade vascular neoplasms, which display numerous histo- pathological characteristics.21 On imaging, these lesions resem- ble non-specific soft-tissue masses and may be associated with calcifications, oedema or haemorrhage.22 They invariably show some degree of enhancement, but are not intrinsically fat-con- taining.22 First described by Masson in 1923,23 intravascular haemangioendothelioma is a lesion characterized by endothelial proliferation within medium-sized veins; however, only rare cases have been reported in the abdomen.24 Another differential diagnosis is renal angiomyolipoma. Angio- myolipomas are nearly always heterogeneous tumours com- posed of fat and soft-tissue material, unlike lipomas, which are primarily fat. Although there have been reported cases of renal angiomyolipomas involving the renal vein or IVC,12 angiomyo- lipomas arise from the renal parenchyma and present with a renal parenchymal defect with a direct juncture between normal renal parenchyma and the tumour.13 TREATMENT A renal cell carcinoma (RCC) with intravascular extension is another consideration. RCCs are almost always solid, enhancing exophytic masses. Their appearance is variable, as they may contain low-density cystic areas. Small tumours may be well marginated, but larger tumours—the ones that are typically associated with intravascular extension—have less of a distinct interface with the renal parenchyma.14 There are rare cases in the published literature of RCCs containing small amounts of fat.15,16 However, the amount of fat is minimal and these tumours usually have associated calcifications, with only a few reports of fat-containing RCCs without calcifications.16,17 Management of intravascular lipoma varies in the literature. Because fat-specific sequences in MRI are able to characterize lesions with high specificity and differentiate fatty tissue from solid components, biopsy has typically been seen as unnecessary when imaging demonstrates a lesion composed entirely of fat and the patient is asymptomatic. However, lesions with large lipomatous components may be more difficult to accurately diagnose on imaging, such as benign lipomas vs well-differenti- ated liposarcomas.25 Biopsy of such lesions would be reasonable, although diagnostic biopsy was not performed in any of the reported cases of intravascular lipomas; cases that elected to obtain histological confirmation did so by complete excision of the lesion instead. It should also be noted that biopsy of Rarely, retroperitoneal leiomyosarcomas may be found in a completely intravascular pattern, with several published reports 2 of 4 birpublications.org/bjrcr BJR Case Rep;1:20150072 2 of 4 BJR\|case reports Case report: Intravascular lipoma of the renal vein heterogeneous lesions also has its own problems of sampling error and may lead to a false-negative result.26 malignant potential on imaging, the primary team elected for conservative management through surveillance. A repeat CT scan at 4 months showed no change in size or character of the lesion and no evidence of vascular obstruction or thrombosis. The patient remained asymptomatic on follow-up. Often, surgical resection is advised only if patients are symptom- atic. Intravascular lipomas are most often asymptomatic, but can rarely cause venous obstructive symptoms, such as superior vena cava syndrome. Bravi et al2 describe a patient who devel- oped a thrombotic complication owing to the occlusive effects of a lipoma preventing adequate venous return. LEARNING POINTS 1. On imaging, an intravascular lipoma is seen as a well- defined, homogeneous, non-enhancing, fat-attenuating mass without areas of soft tissue, haemorrhage or necrosis. A hyperintense signal on T2 imaging that subtracts on fat-saturated images confirms the diagnosis. 2. Intravascular lipomas must be distinguished from their more malignant counterpart— liposarcomas—given that a subset of lipomas can have soft-tissue components and may be indistinguishable on imaging alone. 3. Although benign, intravascular lipomas may require surgery depending on concern for obstructive complications. 46452134 4. Martín-Pedrosa JM, del Blanco I, Carrera S, González-Fajardo JA, Gutiérrez V, Vaquero C. Intravascular lipoma of the external iliac vein and common femoral vein. Eur J Vasc Endovasc Surg 2002; 23: 470–2. 17. Schuster TG, Ferguson MR, Baker DE, Schaldenbrand JD, Solomon MH. Papillary renal cell carcinoma containing fat without calcification mimicking angiomyolipoma on CT. AJR Am J Roentgenol 2004; 183: 1402–4. doi: 10.2214/ajr.183.5.1831402 11. Gaskin CM, Helms CA. Lipomas, lipoma variants, and well-differentiated liposarcomas (atypical lipomas): results of MRI evaluations of 126 consecutive fatty masses. AJR Am J Roentgenol 2004; 182: 733–9. doi: 10.2214/ajr.182.3.1820733 5. Mcclure MJ, Sarrazin J, Kapusta L, Murphy J, Arenson AM, Geerts W. Intravascular femoral vein lipoma: an unusual cause of lower limb venous obstruction. AJR Am J Roentgenol 2001; 176: 463–7. 18. Lipton M, Sprayregen S, Kutcher R, Frost A. Venous invasion in renal vein leiomyosarcoma: case report and review of the literature. Abdom Imaging 1995; 20: 64–7. doi: 10.1007/BF00199648 12. Islam AH, Ehara T, Kato H, Hayama M, Kashiwabara T, Nishizawa O. Angiomyolipoma of kidney involving the inferior vena cava. Int J Urol 2004; 11: 897–902. doi: 10.1111/j.1442-2042.2004. 19. Hemant D, Krantikumar R, Amita J, Chawla A, Ranjeet N. Primary leiomyosarcoma of inferior vena cava, a rare entity: imaging features. Australas Radiol 2001; 45: 448–51. doi: 10.1046/j.1440-1673. 2001.00955.x TREATMENT In 2 of the 27 reported cases that were reviewed, the primary team elected to do surgery despite a lack of symptoms in order to prevent poten- tial obstructive and thromboembolic complications, and to rule out malignant disease by providing a definitive histological diag- nosis.7,27 Although an intravenous lesion may alter flow dynam- ics, there was no evidence of collateral formation or venous thrombosis to suggest a clinically significant flow obstruction in this case. The pedicle of implantation in this patient was not clear and therefore embolization could be a potential concern, but given that these lesions are well-encapsulated and likely arise from the vascular wall, embolization was thought to be unlikely to occur. For these reasons and because there was no evidence of 2009.67.2.154 9. Thorogood SV, Maskell GF. Case report: intravascular lipoma of the superior vena cava–CT and MRI appearances. Br J Radiol 1996; 69: 963–4. radiology.188.2.8327691 2. Bravi MC, Salvadei S, Scarponi P, Loforte A, Musumeci F, Gasbarrone L. Intravascular lipoma of the superior vena cava. Intern Emerg Med 2012; 7: 79–81. 16. Richmond L, Atri M, Sherman C, Sharir S. Renal cell carcinoma containing macroscopic fat on CT mimics an angiomyolipoma due to bone metaplasia without macroscopic calcification. Br J Radiol 2010; 83: e179–e181. doi: 10.1259/bjr/ 10. Kransdorf MJ, Bancroft LW, Peterson JJ, Murphey MD, Foster WC, Temple HT. Imaging of fatty tumors: distinction of lipoma and well-differentiated liposarcoma. Radiology 2002; 224: 99–104. doi: 10.1148/ radiol.2241011113 3. Cabri-Wiltzer M, Danse E. Benign lipoma of the inferior vena cava. JBR-BTR 2013; 96: 212–15. 001.00955.x 7. Mordant P, Mercier O, Fadel E, Muniappan A, Fabre D, Chataigner O, et al. Surgical resection of an intravascular superior vena cava primary lipoma. J Thorac Cardiovasc Surg 2010; 140: 1437–8. doi: 10. 1016/j j 2009 04 072 20. Hartman DS, Hayes WS, Choyke PL, Tibbetts GP. From the archives of the AFIP. Leiomyosarcoma of the retroperitoneum and inferior vena cava: radiologic-pathologic correlation. Radiographics 1992; 12: 1203–20. doi: 10.1148/radiographics.12.6. 1439022 radiol.13121187 14. Zagoria RJ, Wolfman NT, Karstaedt N, Hinn GC, Dyer RB, Chen YM. CT features of renal cell carcinoma with emphasis on relation to tumor size. Invest Radiol 1990; 25: 261–6. doi: 10.1097/00004424- 199003000-00010 REFERENCES 1. Lomeo A, D'Arrigo G, Scolaro A, Mudanò M, Monea MC, Mauceri G, et al. A case of intra and extra-vascular lipoma of the subclavian vein. EJVES Extra 2007; 13: 37–9. doi: 10.1016/j.ejvsextra.2006.08.003 15. Hélénon O, Chrétien Y, Paraf F, Melki P, Denys A, Moreau JF. Renal cell carcinoma containing fat: demonstration with CT. Radiology 1993; 188: 429–30. doi: 10.1148/ radiology.188.2.8327691 lipoma of the right subclavian vein. Tuberc Respir Dis 2009; 67: 154–7. doi: 10. 4046/trd.2009.67.2.154 00909.x 6. Moore FO, , Norwood SH. Intravascular lipoma of the right innominate vein in a trauma patient. J Am Coll Surg 2008; 207: 139. doi: 10.1016/j.jamcollsurg.2007.10.015 13. Yarmish G, Dipoce J. Case 199: aggressive angiomyolipoma with renal vein thrombosis and pulmonary fat embolus. Radiology 2013; 269: 615–8. doi: 10.1148/ di l 13121187 BJR\|case reports Z Doyle et al 21. Requena L, Kutzner H. Hemangioendothelioma. Semin Diagn Pathol 2013; 30: 29–44. doi: 10.1053/j. Intravascular papillary endothelial hyperplasia (Masson's hemangioma) of the liver: a new hepatic lesion. J Korean Med Sci 2004; 19: 305–8. doi: 10.3346/jkms.2004.19. 2.305 26. El Ouni F, Jemni H, Trabelsi A, Ben Maitig M, Arifa N, Ben Rhouma K, et al. Liposarcoma of the extremities: MR imaging features and their correlation with pathologic data. Orthop Traumatol Surg Res 2010; 96: 876–83. doi: 10.1016/j.otsr. 2010.05.010 1016/j.jtcvs.2009.04.072 8. Ryu SR, Park JY, Ryu YS, Yu YH, Yang DJ, Lee BH, et al. Intravascular 3 of 4 BJR Case Rep;1:20150072 3 of 4 birpublications.org/bjrcr semdp.2012.01.003 22.. Hayat MA. Cancer imaging: instrumentation and applications. Vol 2. Burlington, MA: Elsevier Science; 2008. 25. O'Donnell PW, Griffin AM, Eward WC, Sternheim A, White LM, Wunder JS, et al. Can experienced observers differentiate between lipoma and well- differentiated liposarcoma using only MRI? Sarcoma 2013; 2013: 1–6. doi: 10.1155/2010/ 506182 27. Yoon RH, Buzas CJ, Garvin RP, Franklin DP. Intravascular lipoma of the internal jugular vein. J Vasc Surg Venous Lymphat Disord 2013; 1: 406–8. doi: 10. 1016/j.jvsv.2012.12.004 23. Masson MP. L'Hemangioendotheliome intravasculaire. Ann Anat Pathol 1923; 93: 517. 24. Hong S-G, Cho H-M, Chin H-M, Park I-Y, Yoo J-Y, Hwang S-S, et al. 4 of 4 birpublications.org/bjrcr 4 of 4 BJR Case Rep;1:20150072
https://openalex.org/W4212842969	https://link.springer.com/content/pdf/10.1007/JHEP02(2022)145.pdf	English	null	Axions, higher-groups, and emergent symmetry	The Journal of high energy physics/The journal of high energy physics	2,022	cc-by	12,997	Received: October 4, 2021 Accepted: February 2, 2022 Published: February 17, 2022 Received: October 4, 2021 Accepted: February 2, 2022 Published: February 17, 2022 Open Access, c⃝The Authors. Article funded by SCOAP3. Keywords: Higher Spin Symmetry, Renormalization Group, Discrete Symmetries 1 Introduction JHEP02(2022)145 Contents Contents 1 Introduction 1 2 Symmetry structure in models of axions 4 2.1 Higher-form symmetry 4 2.1.1 Higher-group global symmetry 5 2.2 The symmetries of axion-Maxwell theory 7 2.2.1 Coupling to background ﬁelds 9 2.2.2 Extension to axion-QED 11 2.3 The symmetries of axion-Yang-Mills 12 2.3.1 Background ﬁelds for axion Yang-Mills 14 2.3.2 Extension to axion-QCD 15 3 Higher-groups and constraints on emergence 16 3.1 Inequalities for axion-Yang-Mills 16 3.1.1 Pecci-Quinn example 17 3.2 Inequalities for axion-QCD 20 3.3 Inequalities for axion electrodynamics 21 3.3.1 Adjoint Higgsing example 22 Axions, higher-groups, and emergent symmetry JHEP02(2022)145 T. Daniel Brennan and Clay Córdova KadanoﬀCenter for Theoretical Physics & Enrico Fermi Institute, University of Chicago, Michelson Center for Physics, 933 E 56th St, Chicago, IL 60637, U.S.A. E-mail: tdbrennan@uchicago.edu, clayc@uchicago.edu Abstract: Axions, periodic scalar ﬁelds coupled to gauge ﬁelds through the instanton den- sity, have a rich variety of higher-form global symmetries. These include a two-form global symmetry, which measures the charge of axion strings. As we review, these symmetries typically combine into a higher-group, a kind of non-abelian structure where symmetries that act on operators of diﬀerent dimensions, such as points, lines, and strings, are mixed. We use this structure to derive model independent constraints on renormalization group ﬂows that realize theories of axions at long distances. These give universal inequalities on the energy scales where various infrared symmetries emerge. For example, we show that in any UV completion of axion-Yang-Mills, the energy scale at which axion strings can decay is always larger than the mass scale of charged particles. ArXiv ePrint: 2011.09600 Open Access, c⃝The Authors. Article funded by SCOAP3. Open Access, c⃝The Authors. Article funded by SCOAP3. https://doi.org/10.1007/JHEP02(2022)145 Contents 1 Introduction 1 2 Symmetry structure in models of axions 4 2.1 Higher-form symmetry 4 2.1.1 Higher-group global symmetry 5 2.2 The symmetries of axion-Maxwell theory 7 2.2.1 Coupling to background ﬁelds 9 2.2.2 Extension to axion-QED 11 2.3 The symmetries of axion-Yang-Mills 12 2.3.1 Background ﬁelds for axion Yang-Mills 14 2.3.2 Extension to axion-QCD 15 3 Higher-groups and constraints on emergence 16 3.1 Inequalities for axion-Yang-Mills 16 3.1.1 Pecci-Quinn example 17 3.2 Inequalities for axion-QCD 20 3.3 Inequalities for axion electrodynamics 21 3.3.1 Adjoint Higgsing example 22 1 Introduction Axions are ubiquitous in models of beyond the standard model physics and string theory, and have been a central focus of a variety of experiments. (See e.g. [1–3] and references therein.) One of the key phenomena in the physics of axions is the existence of string-like excitations. Let a denote the periodic axion ﬁeld with periodicity 2πf where f is the axion decay constant. Around a string of charge n the axion ﬁeld winds as I da = 2πnf . (1.1) (1.1) In a low energy model of axions, the strings are non-dynamical defects with inﬁnitesimal thickness. At higher energies these theories have a rich array of new degrees of freedom. In particular, the axion is often revealed to be the angular part of a complex ﬁeld ϕ which is liberated in the ultraviolet. Relatedly, at higher energies the strings are brought to life as dynamical excitations of the ﬁeld theory. For instance at the scale set by the string tension T 1/2 string, strings may nucleate from the vacuum. In a low energy model of axions, the strings are non-dynamical defects with inﬁnitesimal thickness. At higher energies these theories have a rich array of new degrees of freedom. In particular, the axion is often revealed to be the angular part of a complex ﬁeld ϕ which is liberated in the ultraviolet. Relatedly, at higher energies the strings are brought to life as dynamical excitations of the ﬁeld theory. For instance at the scale set by the string tension T 1/2 string, strings may nucleate from the vacuum. g Thus, quantum ﬁeld theories with axions in their long distance physics are intrinsically equipped with an approximate energy scale Estring above which the string quantum number is no longer conserved. This scale is less than that set by the string tension, but in – 1 – perturbative models of axions these scales are often far separated. One of our goals in the following will be to derive model independent constraints on the scale Estring applicable to any UV completion of axion physics. perturbative models of axions these scales are often far separated. One of our goals in the following will be to derive model independent constraints on the scale Estring applicable to any UV completion of axion physics. It is fruitful to describe aspects of axion strings in the language of generalized global symmetries [4]. 1A continuous p-form global symmetry has a conserved current which is a (p + 1)-form. The charged objects are extended operators of dimension p. Familiar ordinary global symmetries are described by the special case when p = 0. 2Recall from [6], that in two spacetime dimensions a continuous global symmetry cannot be sponta- neously broken because the associated Goldstone boson would necessarily have an unphysical logarithmic two-point function. The statement above is the direct dimensional uplift to four spacetime dimensions. 1 Introduction In this terminology, the periodic scalar ﬁeld a gives rise to a conserved current of a generalized two-form global symmetry U(1)(2):1 Jµνρ ∼εµνρσ∂σa , (1.2) (1.2) which is tautologically conserved. The axion strings described above carry the charge as- sociated to this current and equation (1.1) expresses the Ward identity of a current in the presence of a charged object. The current deﬁned by (1.2) should be distinguished from the frequently discussed current Jα = ∂αa for the shift symmetry of the axion. The latter is spontaneously broken by the choice of vacuum in the theory (with a the associated Gold- stone mode). Meanwhile the higher-dimensional analog of the Coleman-Mermin-Wagner theorem [4, 5] implies that the symmetry associated to Jµνρ is never spontaneously broken at long distances.2 JHEP02(2022)145 The hallmark of an axion is its coupling to gauge ﬁelds. Consider for instance axion- Yang-Mills with gauge group SU(N) and action: S = 1 2 Z da ∧∗da + 1 g2 Z Tr(F ∧∗F) − i 8π2f Z aTr(F ∧F) . (1.3) (1.3) In addition to the two-form symmetry U(1)(2) described above, this theory also has a discrete one-form symmetry Z(1) N . The objects charged under this discrete symmetry are Wilson lines, and the conserved quantum number is the charge under the ZN center of the gauge group. This is preserved by interactions because, in the absence of additional matter, the dynamical gluons can only screen adjoints and other representations of vanishing charge. This one-form symmetry is also intimately connected to conﬁnement: at long distances it is preserved or broken precisely when the gauge theory is in a conﬁned or deconﬁned phase respectively [4]. (1) Thus, the presence of the one-form symmetry Z(1) N , endows UV completions of axion- Yang-Mills with another important energy scale, Escreen, above which the one-form sym- metry is broken. Physically, this is the energy scale where matter ﬁelds charged under SU(N) appear and hence can screen general Wilson lines. Below in section 3 we will show that, independent of the details of the ultraviolet physics, there is an inequality on these symmetry breaking scales Estring and Escreen: Escreen ≲Estring . (1.4) (1.4) Escreen ≲Estring . 1A continuous p-form global symmetry has a conserved current which is a (p + 1)-form. The charged objects are extended operators of dimension p. Familiar ordinary global symmetries are described by the special case when p = 0. 1 Introduction 1A continuous p-form global symmetry has a conserved current which is a (p + 1)-form. The charged objects are extended operators of dimension p. Familiar ordinary global symmetries are described by the special case when p = 0. 2Recall from [6], that in two spacetime dimensions a continuous global symmetry cannot be sponta- neously broken because the associated Goldstone boson would necessarily have an unphysical logarithmic two-point function. The statement above is the direct dimensional uplift to four spacetime dimensions. – 2 – Of course, since they are deﬁned by emergent symmetries, the energy scales above are in general only approximately meaningful. Far below Ei the corresponding quantum number is preserved up to small eﬀects, while far above Ei the symmetry is badly broken. We thus interpret (1.4) as forbidding a parametric separation of scales violating the inequality. In other words, Escreen ≫Estring is not possible in any renormalization group trajectory, which ﬂows at long distances to axion-Yang-Mills described by the action (1.3). Of course, since they are deﬁned by emergent symmetries, the energy scales above are in general only approximately meaningful. Far below Ei the corresponding quantum number is preserved up to small eﬀects, while far above Ei the symmetry is badly broken. We thus interpret (1.4) as forbidding a parametric separation of scales violating the inequality. In other words, Escreen ≫Estring is not possible in any renormalization group trajectory, which ﬂows at long distances to axion-Yang-Mills described by the action (1.3). To argue for the inequality (1.4) and other similar bounds, we ﬁrst study in detail the symmetry structure of theories of axions in section 2. As we describe, the various symmetries of these models do not factorize into a simple product of the symmetries of each form degree. Instead, a more precise analysis shows that they form a mixed structure known as a “higher group” global symmetry, which is somewhat analogous to a non-abelian symmetry group composed of symmetries of diﬀerent degrees. JHEP02(2022)145 Higher-group global symmetry is an analog of the Green-Schwarz mechanism [7] for global symmetries (as opposed to its more common application when all gauge ﬁelds are dy- namical) [8, 9]. This symmetry structure frequently appears in gauge theories and in topo- logical quantum ﬁeld theories when symmetries of diﬀering form degrees are present [10–12]. 1 Introduction In the context of models of axions, the presence of these interesting symmetry structures has been noted for non-abelian gauge groups in [13, 14], and recently explored in abelian gauge theories in [15, 16], and our presentation in section 2 draws closely from these refer- ences. Our analysis leading to inequalities on energy scales using higher-group symmetry follows closely the logic of [11], and more broadly the spirit of [17]. One straightforward way to make the presence of this higher-group transparent is by coupling the theory to background gauge ﬁelds for the global symmetries. For example, in axion Yang-Mills the U(1)(2) two-form symmetry that measures the charges of strings, has a natural source which is a three-form gauge ﬁeld A(3) with gauge-invariant ﬁeld strength G(4), while the Z(1) N one-form symmetry that measures the charges of Wilson lines is sourced by a two-form gauge ﬁeld B(2) which has periods that are N-th roots of unity. The higher- group structure derived below then implies that the ﬁeld strength G(4) is: G(4) = dA(3) + N(N −1) 4π B(2) ∧B(2) . (1.5) (1.5) Notice in particular that the presence of the one-form symmetry background induces non- trivial (and in general fractional) G(4) ﬂux. This mixing between sources for global sym- metries is the signature of higher-group global symmetry. Notice in particular that the presence of the one-form symmetry background induces non- trivial (and in general fractional) G(4) ﬂux. This mixing between sources for global sym- metries is the signature of higher-group global symmetry. In section 3 we derive the inequality (1.4) using the ﬂux formula (1.5). We then investigate the bound in the context of weakly-coupled KSVZ-type models. In such the- ories axions and the symmetry structure above emerge at low-energies after breaking an anomalous Peccei-Quinn symmetry [18–21]. As we show, the inequality (1.4) implies con- straints on ultraviolet coupling constants which are indeed true in theories that ﬂow to axion-Yang-Mills in the IR. In section 3 we also discuss constraints on axion-Maxwell theories. We consider the action: 1 Z 1 Z iK Z In section 3 we also discuss constraints on axion-Maxwell theories. We consider th S = 1 2 Z da ∧∗da + 1 2g2 Z F ∧∗F −iK 8π2f Z aF ∧F , (1.6) (1.6) – 3 – where above K ∈Z is an integral coupling constant. 1 Introduction In particular, for \|K\| > 1 we ﬁnd a non-trivial higher-group global symmetry and resulting constraints on emergent symmetry scales. The symmetries involved are of two types. First, there is the continuous U(1)(2) string symmetry of the axion and the U(1)(1) m magnetic symmetry of the photon that measures the magnetic charges of ’t Hooft lines. There is also a discrete Z(0) K shift-symmetry of the axion and a discrete Z(1) K one-form symmetry of the photon that measures the charges of Wilson lines modulo K. In addition to results similar to (1.4) we also derive the general constraint min{Eshift, Escreen} ≲Emagnetic , (1.7) (1.7) min{Eshift, Escreen} ≲Emagnetic , JHEP02(2022)145 where above, Eshift is the scale where the shift symmetry of the axion is violated, Escreen is the scale at which Wilson lines can be screened by charged matter, and Emagnetic is the scale of emergence of the U(1) gauge ﬁeld deﬁning the photon, above which ∗F, ceases to deﬁne a conserved current. We verify (1.7) in a simple example and ﬁnd that the implied constraints on coupling constants are indeed necessary for self-consistency of our analysis. Finally, we also discuss analogs of (1.4) and (1.7) in theories with axions and charged matter ﬁelds that have higher-group global symmetry and hence the ensuing consequences. 2 Symmetry structure in models of axions In this section we review the basic deﬁnitions of higher-form global symmetry following [4], and higher-group global symmetry following [11, 12]. We then present the symmetries and couplings to background gauge ﬁelds in models of axions. In the case of axions coupled to abelian gauge ﬁelds these symmetries have been previously studied in [15, 16]. In the case of axions coupled to non-abelian gauge ﬁelds, our analysis follows from results in [13, 14]. 2.1 Higher-form symmetry In general, a p-form global symmetry is a symmetry that acts on extended operators of dimension p [4]. Thus, ordinary global symmetries, which act on point operators, cor- respond to the case p = 0. Meanwhile one-form symmetries act on line operators, and two-form symmetries act on surface operators. We will encounter all of these symmetries below. Higher-form symmetries form a group (necessarily abelian if p > 0) that we denote G(p). In the simplest case of continuous G(p), there is a conserved current with p + 1 antisymmetric indices: ∂µ1Jµ1···µp+1 = 0 . (2.1) (2.1) It is frequently convenient to work with the hodge dual ∗J which is therefore closed, d ∗Jp+1 = 0. As usual when discussing the implications of symmetry it is fruitful to introduce back- ground gauge ﬁelds that act as sources for the conserved currents. For a p-form global symmetry the appropriate background is a (p + 1)-form gauge ﬁeld A(p+1). In the simplest case of a continuous G(p) symmetry, the background gauge ﬁeld couples in a standard way to the conserved current: Z S ⊃i Z A(p+1) ∧∗Jp+1 . (2.2) (2.2) – 4 – Current conservation means that (up to possible ’t Hooft anomalies) the theory is invariant under background gauge transformations of the source: A(p+1) 7−→A(p+1) + dΛ(p) , (2.3) (2.3) where Λ(p) is locally a p-form gauge parameter. The normalization of the coupling (2.2) between the source and the current is ﬁxed by charge quantization. In our conventions G(p) charges, which are measured on any closed surface Σd−p−1 of dimension (d −p −1), are integrally quantized: Z Z Σd−p−1 ∗Jp+1 ∈Z . (2.4) (2.4) JHEP02(2022)145 In particular, this means that the ﬂuxes and gauge parameters of the background ﬁelds are also quantized: ( 1) ( ) Z Σp+2 dA(p+1) 2π ∈Z , Z Σp+1 dΛ(p) 2π ∈Z . (2.5) (2.5) Alternatively, we can also describe symmetry through the symmetry group operators. For an element g ∈G(p), the associated operator is denoted Ug(Σd−p−1) and is deﬁned on any closed Σd−p−1. In the case of a continuous symmetry these are given by exponentiated integrals of the current Ug(Σd−p−1) = e iλR Σd−p−1 ∗Jp+1 , (2.6) (2.6) where g = eiλ. The primary advantage of the symmetry group operators is that the continue to exist even when the symmetry is discrete. 2.1 Higher-form symmetry Mathematically these symmetry defect operators can be thought of as Poincaré dual to the background ﬁelds. They describe ﬂat background gauge ﬁelds, i.e. those with vanishing ﬁeld strength. 2.1.1 Higher-group global symmetry Higher-group global symmetry is a natural possibility in theories with symmetries of dif- fering form degrees. As in the discussion above, there are various related point of view in terms of currents, symmetry group operators, and background ﬁelds. Perhaps the simplest context where this mixing occurs is in situations where all sym- metries in question are continuous and so we can reduce to statements about the conserved currents. In this case we can ﬁnd contact terms in the operator product expansion of the generic form: d ∗Jp1+1(x)Jp2+1(y) ∼∂kδ(d)(x −y)Jp3+1(x) , (2.7) (2.7) where ∂k is a k-th derivative and p3 = p1 + p2 + k. Note in particular that, since k ≥0, the fusion always produces symmetries of form degree at least p1 + p2 on the right-hand- side. Such fusion laws are similar in spirit to the more familiar non-abelian current algebra of ordinary (0-form) global symmetries. In that context, the currents are conserved at separated points, but at coincident points one encounters a contact term in the divergence controlled by the Lie algebra structure constants fabc. Equation (2.7) is similar except that now the fusion involves currents of diﬀering form degrees and there are derivatives on the delta function. For instance, an example of this structure, explored in detail in [11], – 5 – =1 g h k ghk gh hk β(g, h, k) Figure 1. A junction where zero-form symmetry defects of type g, h, k, ghk ∈G meet in codimension three. This conﬁguration is generic in spacetime dimension three and above. The junctions of three codimension-one defects are in red, and their intersection is the black point. At the codimension-three intersection, a one-form symmetry defect β(g, h, k) emanates, signaling the 2-group symmetry. In d dimensions, all objects span the remaining d −3 dimensions. h JHEP02(2022)145 ghk =1 Figure 1. A junction where zero-form symmetry defects of type g, h, k, ghk ∈G meet in codimension three. This conﬁguration is generic in spacetime dimension three and above. The junctions of three codimension-one defects are in red, and their intersection is the black point. At the codimension-three intersection, a one-form symmetry defect β(g, h, k) emanates, signaling the 2-group symmetry. In d dimensions, all objects span the remaining d −3 dimensions. occurs in multiﬂavor massless QED, where the zero-form ﬂavor symmetry forms a higher- group with the one-form magnetic symmetry with current J2 = ∗F. 2.1.1 Higher-group global symmetry Other examples are described in detail in [17]. In the applications below, we will often be interested in higher-groups where the sym- metries involved are discrete. In this case we read oﬀthe structure from the symmetry group operators Ug(Σ) described above following [12]. To begin with, we recall from [4] that the group multiplication laws in G(p) are encoded by an operator product algebra of the symmetry group operators. For instance in the abstract notation (2.6) above we have Ug1(Σd−p−1)Ug2(Σd−p−1) = Ug1g2(Σd−p−1) , (2.8) (2.8) which generalizes the additive nature of conserved charges. We can also view this as a rule for what happens when the defects Ug1 and Ug2 intersect: they produce Ug1g2. The fusion rule (2.8) encodes only the simplest possible collision of symmetry group operators. More generally we must permit multiple defects to intersect in generic conﬁgu- rations. When this happens we can encounter defects of various form degrees leading to a higher-group global symmetry. An example taken from [12] is shown in ﬁgure 1. When the resulting algebra of symmetry defects involves non-trivial intersections with form degrees which are not all equal we say there is higher-group global symmetry. Finally, we can also describe higher-group global symmetry in terms of background ﬁelds, which is the method that we employ below. We consider a collection of symmetries – 6 – and background ﬁelds A(1), A(2), · · · A(ℓ) that are sources for the higher-form symmetries of diﬀerent degrees as in (2.2). Each such background ﬁeld is subject to a gauge transformation as in (2.3), but now we allow for mixing of the gauge transformation between background ﬁelds of diﬀering form degree: A(p+1) 7−→A(p+1) + dΛ(p) + X k<p αk(A(i))Λ(k) + Schwinger terms . (2.9) (2.9) Here, the sum is over k < p mirroring the restrictions on p3 in equation (2.7). Meanwhile the Schwinger terms above refers to possible pieces of the gauge transformation rules that are non-linear in the gauge parameters. This sort of mixed gauge transformation rule involving forms of diﬀering degrees is familiar from the Green-Schwarz mechanism [7]. In that context one encounters such mixed transformations for dynamical ﬁelds whereas here it occurs for background ﬁelds and signiﬁes a higher-group global symmetry. Crucially, as we see in examples below, this means that the gauge invariant ﬂuxes for background ﬁelds are modiﬁed to include Chern-Simons-like corrections constructed out of the other gauge ﬁelds. JHEP02(2022)145 The normalization of J1 can be checked by noting that the charged operators are exp(ina/f) for integer n. 4In the presence of both A(1) and A(3) the action is not fully invariant under background gauge trans- formations but instead has a mixed ’t Hooft anomaly characterized by an inﬂow action 3The normalization of J1 can be checked by noting that the charged operators are exp(ina/f) for integer n. 4In the presence of both A(1) and A(3) the action is not fully invariant under background gauge trans 2.2 The symmetries of axion-Maxwell theory (2.12) (2.12) It is straightforward to couple these symmetries to background two-form gauge ﬁelds (B(2) e , B(2) m ): JHEP02(2022)145 S = 1 2g2 Z (F −B(2) e ) ∧∗(F −B(2) e ) + i Z B(2) m ∧F 2π . (2.13) ( ) (2.13) Under gauge transformations of B(2) e , the dynamical gauge ﬁeld F also shifts indicating that the photon is the Goldstone mode for the electric 1-form symmetry. We now couple the axion to the Maxwell ﬁeld leading to the action (without background ﬁelds): S = 1 2 Z da ∧∗da + 1 2g2 Z F ∧∗F −iK 8π2f Z aF ∧F , (2.14) (2.14) Here K ∈Z is a discrete coupling constant. It is quantized to ensure that the action respects periodicity of the axion. For some purposes below it is also useful to write the action in way that more manifestly respects both the periodicity of the action and gauge invariance. This can be done by choosing an auxiliary ﬁve-manifold X whose boundary is the physical four-dimensional spacetime Y and writing iK 8π2f Z Y aF ∧F = iK 8π2f Z X da ∧F ∧F , (2.15) (2.15) where now the quantized coupling K ensures that the action does not depend on the choice of X. For further mathematical details about the deﬁnition of the axion interaction see [14]. where now the quantized coupling K ensures that the action does not depend on the choice of X. For further mathematical details about the deﬁnition of the axion interaction see [14]. The interaction term in the action leads to a modiﬁcation of the equations of motion. It is convenient to describe this in terms of the currents introduced in equation (2.10) and (2.12): The interaction term in the action leads to a modiﬁcation of the equations of motion. It is convenient to describe this in terms of the currents introduced in equation (2.10) and (2.12): d ∗J3 = 0 , d ∗J2,m = 0 , d ∗J1 = K 8π2 F ∧F , d ∗J2,e = −K 4π2f d(aF) . (2.16) (2.16) The non-zero right-hand-side of d ∗J1 and d ∗J2,e above implies that in axion electro- dynamics, these symmetries are broken. In fact, the as we will see these symmetries are broken to discrete groups controlled by K: U(1)(0) −→Z(0) K , U(1)(1) e −→Z(1) K . 2.2 The symmetries of axion-Maxwell theory Let us now describe in more detail the symmetry structure of axion-Maxwell theory, i.e. an axion coupled to an abelian gauge ﬁeld. We ﬁrst ignore the coupling between a and the gauge ﬁelds and then incorporate modiﬁcations due to the interactions. The symmetries and higher-group structure present in these theories has also been discussed in [15, 16]. ( ) ( ) The axion ﬁeld a has two natural symmetries: a U(1)(0) shift symmetry and a U(1)(2) two-form symmetry which acts on strings. The associated currents are respectively:3 (2.10) ∗J1 = if ∗da , ∗J3 = 1 2πf da . (2.10) The choice of vacuum for a spontaneously breaks the shift symmetry. Meanwhile the two- form symmetry is always preserved. The action coupled to background ﬁelds (A(1), A(3)) is written as: The choice of vacuum for a spontaneously breaks the shift symmetry. Meanwhile the two- form symmetry is always preserved. The action coupled to background ﬁelds (A(1), A(3)) is written as: S = 1 2 Z (da −fA(1)) ∧∗(da −fA(1)) + i 2πf Z A(3) ∧da . (2.11) (2.11) In particular, the coupling of the axion to A(1) is standard for a Goldstone mode and under gauge transformations of A(1), a also shifts to ensure the action is invariant.4 3The normalization of J1 can be checked by noting that the charged operators are exp(ina/f) for integer n. 4In the presence of both A(1) and A(3) the action is not fully invariant under background gauge trans- formations but instead has a mixed ’t Hooft anomaly characterized by an inﬂow action Sinﬂow = i 2π Z X dA(1) ∧A(3) , where X is a ﬁve-dimensional manifold with boundary the physical spacetime Y . See e.g. [22, 23] and references therein for an overview of anomaly inﬂow. – 7 – Now let us consider an abelian gauge ﬁeld C with gauge invariant ﬁeld strength F = dC. We ﬁrst focus on the case without matter and later generalize to include charged matter ﬁelds. First, in the case of free Maxwell theory without coupling to a, there are 1-form symmetries U(1)(1) e ×U(1)(1) m , where the subscripts indicate the common association of these as “electric” and “magnetic” respectively. The associated conserved currents are: ∗J2,e = i g2 ∗F , ∗J2,m = F 2π . 2.2 The symmetries of axion-Maxwell theory (2.17) (2.17) In particular, in the case of minimal coupling, K = 1, these symmetries are completely broken. To demonstrate the above, we need to deﬁne topological operators generating discrete transformations. Those generating the discrete Z(0) K axion shift symmetry are given by: Uℓ(Σ3) = exp 2πiℓ K Z Σ3 ∗J1 −K 8π2 C ∧F , ℓ∈Z . (2.18) (2.18) – 8 – – 8 – By the equations of motion, the argument of the integral is formally closed. However, since it involves the dynamical gauge ﬁeld C explicitly it is not manifestly gauge invariant. The quantization condition ℓ∈Z ensures that the Chern-Simons correction term is well-deﬁned. Similarly the symmetry operators for the discrete electric one-form symmetry Z(1) K are: By the equations of motion, the argument of the integral is formally closed. However, since it involves the dynamical gauge ﬁeld C explicitly it is not manifestly gauge invariant. The quantization condition ℓ∈Z ensures that the Chern-Simons correction term is well-deﬁned. (1) Similarly, the symmetry operators for the discrete electric one-form symmetry Z(1) K ar Uℓ(Σ2) = exp 2πiℓ K Z Σ2 ∗J2,e + K 4π2f a F , ℓ∈Z . (2.19) (2.19) Again the equations of motion (2.16) imply that the integrand is closed so that the operator is topological and the quantization ℓ∈Z implies that the operator is well-deﬁned under the shift redundancy of the axion a →a + 2πf. Again the equations of motion (2.16) imply that the integrand is closed so that the operator is topological and the quantization ℓ∈Z implies that the operator is well-deﬁned under the shift redundancy of the axion a →a + 2πf. JHEP02(2022)145 2.2.1 Coupling to background ﬁelds We now explain the coupling of axion electrodynamics to background ﬁelds. To do so, we must introduce the appropriate backgrounds for the discrete symmetries Z(0) K and Z(1) K . These are discrete gauge ﬁelds. We represent them by standard gauge ﬁelds obeying constraints: KA(1) = dλ(0) , KB(2) e = dλ(1) , (2.20) (2.20) where λ(i) above are gauge parameters with the periods of dλ(i) valued in 2πZ. Thus the background ﬁelds are ﬂat gauge ﬁelds with holonomies that are K-th roots of unity. Mathematically one can also view the gauge invariant data in these background gauge ﬁelds in terms of cohomology classes. Under gauge transformation: A(1) 7−→A(1) + dΛ(0) , λ(0) 7−→λ(0) + KΛ(0) , (2.21) B(2) e 7−→B(2) e + dΛ(1) e , λ(1) 7−→λ(1) + KΛ(1) e . (2.21) Thus the invariant information is encoded in K 2πA(1) ∈H1(Y, ZK) , K 2πB(2) e ∈H2(Y, ZK) , (2.22) (2.22) where the square brackets indicate the gauge equivalence class, and Y is spacetime. At the linearized level, it is straightforward to couple the symmetries to background ﬁelds by introducing couplings of the form (2.2). At the non-linear level this is more subtle due to the periodicity of the axion and the appearance of Chern-Simons terms. A simple strategy is to recall that the axion a and the gauge ﬁeld C must still shift under the discrete symmetries. To account for the possible non-linear mixing between background ﬁelds we write the coupling to the U(1)(2) two-form symmetry of the axion and U(1)(1) magnetic symmetry of the gauge ﬁeld in terms of ﬁeld strengths: S ⊃ i 2πf Z a G(4) −i 2π Z H(3) ∧C , (2.23) (2.23) where G(4) and H(3) are gauge invariant ﬁeld strengths: where G(4) and H(3) are gauge invariant ﬁeld strengths: G(4) = dA(3) + · · · , H(3) = dB(2) m + · · · , (2.24) (2.24) – 9 – and the omitted terms above indicate possible corrections that are non-linear in the back- ground ﬁelds and will be determined below. Additionally, we ﬁnd it convenient to use the ﬁve-dimensional presentation of (2.15). 2.2.1 Coupling to background ﬁelds With these preliminaries the action coupled to all background ﬁelds is simply: S = 1 2 Z Y (da −fA(1)) ∧∗(da −fA(1)) + 1 2g2 Z Y (F −B(2) e ) ∧∗(F −B(2) e ) (2.25) + i 2πf Z Y a G(4) −i 2π Z Y H(3) ∧C −iK 8π2f Z X (da −fA(1)) ∧(F −B(2) e ) ∧(F −B(2) e ) . (2.25) Written as above, it is transparent that the theory is invariant under all background gauge transformation and enjoys the correct linear couplings to backgrounds.5 However, superﬁ- cially it would seem to depend on the explicit choice of ﬁve-manifold X. Demanding that this dependence drop our for all terms that involve dynamical ﬁelds ﬁxes the non-linear form of the ﬁeld strengths in (2.24) to be:6 Written as above, it is transparent that the theory is invariant under all background gauge transformation and enjoys the correct linear couplings to backgrounds.5 However, superﬁ- cially it would seem to depend on the explicit choice of ﬁve-manifold X. Demanding that this dependence drop our for all terms that involve dynamical ﬁelds ﬁxes the non-linear form of the ﬁeld strengths in (2.24) to be:6 JHEP02(2022)145 G(4) = dA(3) + K 4πB(2) e ∧B(2) e , H(3) = dB(2) m + K 2πA(1) ∧B(2) e . (2.26) (2.26) Notice that in general when B(2) e and A(1) are activated, the ﬂuxes G(4) and H(3) are fractional. Moreover, the non-linear terms above mean that the gauge transformations of background ﬁelds must also contain non-linear corrections as in the Green-Schwarz mechanism [7]. In particular this means that while the gauge transformations for A(1) and B(2) e are standard: A(1) 7−→A(1) + dΛ(0) , B(2) e 7−→B(2) e + dΛ(1) e , (2.27) (2.27) those of the other background ﬁelds contain non-linear terms:7 A(3) 7−→A(3) + dΛ(2) −K 2πB(2) e ∧dΛ(1) e −K 4πΛ(1) e ∧dΛ(1) e , (2.28) B(2) m 7−→B(2) m + dΛ(1) m + K 2πA(1) ∧Λ(1) e −K 2πB(2) e Λ(0) + K 2πdΛ(0) ∧Λ(1) e . (2.29) (2.28) (2.29) These mixed gauge transformations mean that the symmetries of axion-Maxwell theory form a non-trivial higher-group. Speciﬁcally, in the terminology of section 2.1.1 this is a three-group. 7Below we omit possible total derivatives. As usual in descent these can be absorbed into choices of local counterterms. 6Mathematically, 1 4π B(2) e ∧B(2) e is the Pontryagin square of Be embedded into a continuous variables in (2.20). 5Here we are ignoring shifts by terms involving only background ﬁelds i.e. ’t Hooft anomalies discuss below. 2.2.1 Coupling to background ﬁelds As noted in section 1 this means that at loci where the one-form symmetry defects intersect we ﬁnd localized G(4) ﬂux, while at the loci where the defects of Z(0) K and Z(1) K intersect we ﬁnd H(3) ﬂux. For more details on this algebra of symmetry defects see also [16]. To conclude this section, let us also note that we can immediately also determine the ’t Hooft anomaly from the action (2.25). As usual the anomaly is conveniently encoded t Hooft anomaly from the action (2.25). As usual the anomaly is conveniently encoded 5Here we are ignoring shifts by terms involving only background ﬁelds i.e. ’t Hooft anomalies discussed below. 5Here we are ignoring shifts by terms involving only background ﬁelds i.e. ’t Hooft anomalies discussed below. – 10 – in a ﬁve-dimensional local topological action for the background ﬁelds which encodes the anomaly via inﬂow: in a ﬁve-dimensional local topological action for the background ﬁelds which encodes the anomaly via inﬂow: Sinﬂow = i 2π Z X G(4) ∧A(1) + i 2π Z X H(3) ∧B(2) e . (2.30) (2.30) This is the appropriate generalization of the separate anomalies of Maxwell theory and the axion theory. This is the appropriate generalization of the separate anomalies of Maxwell theory and the axion theory. 8See [24] and references therein for a review of the quantum numbers of monopole operators. 2.2.2 Extension to axion-QED – 11 – We now proceed to couple the theory to background ﬁelds for these global symmetries. The background ﬁelds are: We now proceed to couple the theory to background ﬁelds for these global symmetries. The background ﬁelds are: Z(0) K ←→A(1) , SU(Nf) ZL (0) ←→X(1) , U(1)(1) m ←→B(2) m , U(1)(2) ←→A(3) . (2.33) (2.33) An important role will be played by the ZL quotient of the ﬂavor symmetry above. This quotient means that there are more possible backgrounds available than with the naive SU(Nf) global symmetry alone. The additional backgrounds have a topological invariant, the second Steifel-Whitney class w2 ∈H2(Y, ZL), which one can view as a discrete magnetic ﬂux. In many ways, this ﬂux plays an analogous role to the electric one-form symmetry background ﬁeld in the theory without matter. To emphasize this, we deﬁne a dependent background ﬁeld B(2) e JHEP02(2022)145 B(2) e ≡2πw2 L , (2.34) (2.34) where B(2) e behaves as a discrete ZL-valued two-form gauge ﬁeld as in (2.20). Importantly, when B(2) e is non-zero, the ﬂuxes of the dynamical gauge ﬁeld become fractional as: Z Σ2 (F −B(2) e ) ∈2πZ . (2.35) (2.35) This is because the ZL subgroup deﬁning the quotient in (2.32) is also a subgroup of dynamical U(1) gauge group. So in conﬁgurations with discrete magnetic ﬂavor ﬂux, the eﬀective gauge group is U(1)/ZL and the dynamical ﬂux must also be fractional. g g g p ( )/ L y With these preliminaries, we can write the action coupled to backgrounds as: S = 1 2 Z Y (da −fA(1)) ∧∗(da −fA(1)) + 1 2g2 Z Y F ∧∗F + Z Y iΨ( /D + /X(1))Ψ + i˜Ψ( /D −/X(1))˜Ψ + (m˜ΨΨ + c.c.) + i 2πf Z Y G(4) ∧a −i 2π Z Y H(3) ∧C −iK 8π2f Z X (da −fA(1)) ∧F ∧F . (2.36) (2.36) Demanding as before that the action depends only on the dynamical variables in the physical four-dimensional spacetime results in the gauge invariant ﬁeld strengths: G(4) = dA(3) + K 4πB(2) e ∧B(2) e , H(3) = dB(2) m + K 2πA(1) ∧B(2) e . (2.37) (2.37) Thus, as before we ﬁnd a higher-group structure, now involving the ﬂavor symmetry, when- ever L = gcd(K, Nf) is larger than one. We also ﬁnd an ’t Hooft anomaly in this theory with inﬂow action again given by (2.30). 2.2.2 Extension to axion-QED JHEP02(2022)145 Let us now discuss the extension of this analysis to theories with matter ﬁelds. We include NF massive fermion ﬂavors of electric charge one. (Similar analysis applies to theories with charged bosons.) The action is thus: S = 1 2 Z da∧∗da+ 1 2g2 Z F∧∗F−iK 8π2f Z aF∧F+ Z iΨ /DΨ+i˜Ψ /D ˜Ψ+(m˜ΨΨ+c.c.) , (2.31) (2.31) where Ψ and ˜Ψ are Weyl fermions of opposite electric charge and the ﬂavor indices on the fermions are suppressed. ( ) ( ) The Z(0) K shift symmetry of the axion is present as before, as is the U(1)(1) m magnetic one- form symmetry and the U(1)(2) two-form axion string symmetry. However, the presence of the charged matter ﬁelds means that Wilson lines can be screened and hence the Z(1) K one-form symmetry is broken. Instead, in this model we claim that there is a new zero-form ﬂavor global symmetry given by SU(Nf) ZL (0) , L = gcd(K, Nf) . (2.32) (2.32) The fact that there is a ﬂavor symmetry, which locally (i.e. at the Lie algebra level) is of the form SU(Nf) is obvious from the presence of multiple ﬂavors and the invariant mass. What remains to be explained is the ZL quotient above. The precise meaning of the quotient is that all operators are acted on trivially by the ZL subgroup of the center of SU(Nf). If we look at the sector of local operators that we can build out of Ψ’s alone this is clear: every gauge invariant local operator contains equal numbers of Ψ’s and Ψ’s and hence is neutral under the entire ZNf center of SU(Nf). To see that the quotient is in fact only by ZL and not ZNf we examine an axion domain wall across which a jumps by 2πf. The coupling between a and F then leads to a Chern-Simons theory at level K along the wall. In particular, this means that monopole operators embedded in the wall carry electric charge −K. We can dress such operators with K modes of Ψ to obtain a gauge invariant ﬁeld conﬁguration.8 This extended conﬁguration of operators transforms with charge K under the central ZNf subgroup of SU(Nf). This means that while ZNf in general acts non-trivially on such a dressed monopole, any ZL subgroup of ZNf where L divides K acts trivially. The largest such L is gcd(K, Nf). 2.3 The symmetries of axion-Yang-Mills We now discuss axion-Yang-Mills. For simplicity, we restrict ourselves to gauge group SU(N). The higher-group symmetries we uncover follow directly from the results of [13, 14]. To begin let us describe symmetries of the Yang-Mills sector independent of the axion. This theory has a one-form symmetry Z(1) N related to the center of the gauge group [4]. Physically this symmetry is present because the only dynamical ﬁelds are the adjoint gluons which – 12 – are neutral under the center of the gauge group. Thus, one cannot screen Wilson lines in representations which are charged under the center of the gauge group. This leads to a quantum number, conserved modulo N, that is carried by Wilson lines. In other words a one-form symmetry. In many ways this Z(1) N is a discrete analog of the U(1)(1) e symmetry of Maxwell theory. By contrast, the U(1)(1) m symmetry has no discrete avatar in SU(N) Yang-Mills.9 To couple the SU(N) theory to an appropriate background ﬁeld we follow the procedure described in [25]. The background is again represented by a constrained gauge ﬁeld B(2) obeying NB(2) = dλ(1) , N 2πB(2) ∈H2(Y, ZN) . (2.38) JHEP02(2022)145 (2.38) We ﬁrst promote the dynamical gauge ﬁeld to be a U(N) connection (for simplicity we denote this by the same symbol C now valued in the adjoint of U(N)). We then introduce a Lagrange multiplier ﬁeld ϕ which constrains the trace of C to be given by the background source: 1 S ⊃1 2π Z ϕ(Tr(F) −NB(2)) . (2.39) (2.39) Thus on the solutions to the equation of motion, the abelian part of the ﬁeld strength is ﬁxed by B(2) and so F = B(2)1+ ˜F, with 1 the N ×N identity matrix and ˜F traceless. As in the Maxwell example, C shifts under the one-form symmetry. This means that under gauge transformations: Thus on the solutions to the equation of motion, the abelian part of the ﬁeld strength is ﬁxed by B(2) and so F = B(2)1+ ˜F, with 1 the N ×N identity matrix and ˜F traceless. As in the Maxwell example, C shifts under the one-form symmetry. This means that under gauge transformations: B(2) 7−→B(2) + dΛ(1) , λ(1) 7−→λ(1) + NΛ(1) , C 7−→C + Λ(1)1 , (2.40) (2.40) so the interaction (2.39) is invariant under background gauge transformations. 9This situation is changed for other global forms of the gauge group. For instance in SU(N)/ZN Yang- Mills, there is no one-form charge carried by the Wilson lines, but instead the ’t Hooft lines are charged under a Z(1) N . 2.3 The symmetries of axion-Yang-Mills Notice in particular that if B(2) vanishes, F is traceless and the remaining background gauge redundancy can be used to shift C by an arbitrary ﬂat U(1) gauge ﬁeld. Thus, we are reduced to the original SU(N) theory. More generally, a U(N) Yang-Mills theory with the interaction (2.39) has the same number of dynamical degrees of freedom as SU(N) Yang-Mills. With these remarks it is straightforward to write the full action coupled to B(2) as: S = Z ϕ 2π(Tr(F) −NB(2)) + 1 g2 Tr h (F −B(2)1) ∧∗(F −B(2)1) i + iθ 8π2 Tr h (F −B(2)1) ∧(F −B(2)1) i . (2.41) As it will be important in the following, let us examine the θ term in more detail. We recall the formula for the integral second Chern number of a U(N) gauge ﬁeld: c2 = 1 8π2 Z Tr(F) ∧Tr(F) −Tr(F ∧F) , c2 ∈Z . (2.42) (2.42) 9This situation is changed for other global forms of the gauge group. For instance in SU(N)/ZN Yang- Mills, there is no one-form charge carried by the Wilson lines, but instead the ’t Hooft lines are charged under a Z(1) N . 9This situation is changed for other global forms of the gauge group. For instance in SU(N)/ZN Yang- Mills, there is no one-form charge carried by the Wilson lines, but instead the ’t Hooft lines are charged under a Z(1) N . – 13 – Then, using the constraint arising from (2.39) we have: Then, using the constraint arising from (2.39) we have: iθ 8π2 Tr h (F −B(2)1) ∧(F −B(2)1) i = iθ 8π2 Z h (Tr(F ∧F) −Tr(F) ∧Tr(F)) + N(N −1)B(2) ∧B(2)i =−iθc2 + iθN(N −1) 8π2 Z B(2) ∧B(2) . (2.43) (2.43) In particular, since B(2) has periods that are N-th roots of unity, we see that in the presence of background B(2), the periodicity of θ is enlarged to θ ∼θ + 2πN. This eﬀect will be responsible for the intricate symmetry structure present in axion Yang-Mills.10 It is clarifying to describe the manipulations above using characteristic classes. We are making use of the fact that U(N) ∼= U(1)×SU(N) ZN . The ZN quotient means that the allowed gauge bundles of U(N) are more general than products of U(1) bundles and SU(N) bundles. Instead, a U(N) bundle can be described as a U(1)/ZN bundle and a SU(N)/ZN bundle subject to a constraint. 2.3 The symmetries of axion-Yang-Mills This means that the U(1) ﬂux may be fractional (an N-th root of unity) while the discrete magnetic ﬂux of SU(N)/ZN, mathematically, the second Steifel- Whitney class w2 ∈H2(Y, ZN), may be non-zero as long as these quantities are related as: JHEP02(2022)145 1 2πTr(F) = w2 = N 2πB(2) ∈H2(Y, ZN) . (2.44) (2.44) In particular, as is well-known, bundles of SU(N)/ZN have fractional instanton number which is precisely encoded by formula (2.43).11 In particular, as is well-known, bundles of SU(N)/ZN have fractional instanton number which is precisely encoded by formula (2.43).11 11As in (2.26), we can express this in terms of the Pontryagin square, with the fractional part of the instanton number being given by N−1 2N P(w2). 10Here, we are restricting ourselves to spacetimes which are spin, appropriate to later applications in- volving fermions. More generally, on a non-spin manifold the periodicity of θ is 2πN for N odd and 4πN for N even. 2.3.1 Background ﬁelds for axion Yang-Mills We now describe the symmetries of axion-Yang-Mills. The analysis is similar to that of axion-Maxwell theory. In the absence of background ﬁelds, the action is 1 2 Z da ∧∗da + 1 g2 Z Tr(F ∧∗F) −iK 8π2f Z aTr(F ∧F) , (2.45) (2.45) where above, the coupling constant K is a positive integer. There are three symmetry groups of interest: Z(0) K , Z(1) N , and U(1)(2). We now activate background ﬁelds and account for possible non-linear corrections to the gauge transformations and ﬂuxes. As before, this is most clear by writing the axion term as a coupling on a ﬁve-manifold X with boundary the physical spacetime Y : S = 1 2 Z Y (da −fA(1)) ∧∗(da −fA(1)) + Z Y ϕ 2π(Tr(F) −NB(2)) + i 2πf Z Y a G(4) + 1 g2 Z Y Tr h (F −B(2)1) ∧∗(F −B(2)1) i −iK 8π2f Z X (da −fA(1))Tr h (F −B(2)1) ∧(F −B(2)1) i . (2.46) i (2.46) 10Here, we are restricting ourselves to spacetimes which are spin, appropriate to later applications in- volving fermions. More generally, on a non-spin manifold the periodicity of θ is 2πN for N odd and 4πN for N even for N even. 11As in (2.26), we can express this in terms of the Pontryagin square, with the fractional part of the instanton number being given by N−1 2N P(w2). – 14 – Demanding that, modulo possible anomalies, the action depends only on the physical spacetime Y implies that the gauge invariant ﬁeld strength must be: Demanding that, modulo possible anomalies, the action depends only on the physical spacetime Y implies that the gauge invariant ﬁeld strength must be: G(4) = dA(3) + KN(N −1) 4π B(2) ∧B(2) . (2.47) (2.47) In particular, this means that while the gauge transformations of A(1) and B(2) are stan- dard, the gauge transformation rules for A(3) are more interesting: In particular, this means that while the gauge transformations of A(1) and B(2) are stan- dard, the gauge transformation rules for A(3) are more interesting: A(3) 7−→A(3) + dΛ(2) −KN(N −1) 2π Λ(1) ∧B(2) −KN(N −1) 4π Λ(1) ∧dΛ(1) . (2.48) (2.48) JHEP02(2022)145 Again, this is an example of a higher-group global symmetry. We can also readily read oﬀthe anomaly inﬂow action Sinﬂow = i 2π Z X G(4) ∧A(1) . (2.49) (2.49) 3 Higher-groups and constraints on emergence The symmetries that we have discussed in the previous section are examples of the general phenomena of higher-group global symmetry described in section 2.1.1. Here we use this to deduce model independent inequalities on the scales of emergent symmetries in models of axions following the general logic of [11]. 2.3.2 Extension to axion-QCD (2.54) (2.54) JHEP02(2022)145 By varying the ﬂuxes of the ﬂavor gauge ﬁelds subject to (2.54), we can achieve any value of the w2(SU(N)) determined via (2.53), and therefore a general fractional instanton number as in (2.43). To match our previous discussion and notation we introduce a background ﬁeld B(2) dependent on the zero-form symmetry as: B(2) = 2π Nf w2(SU(Nf)) . (2.55) (2.55) Then the instanton number is fractional with Then the instanton number is fractional with 1 8π2 Z Tr(F ∧F) = N(N −1) 8π2 (Z −B) ∧(Z −B) . (2.56) (2.56) Hence proceding as before leads to a formula for the gauge invariant four-form ﬁeld strength coupling to the axion: G(4) = dA(3) + KN(N −1) 4π (Z −B) ∧(Z −B) . (2.57) (2.57) In particular, the ﬂuxes of G(4)/2π are fractional as long as N does not divide K. 2.3.2 Extension to axion-QCD As a ﬁnal model of axions, we consider axions coupled to Yang-Mills ﬁelds and Nf funda- mental quarks, all with equal mass. The results follow straightforwardly from the previous examples and our treatment is brief. The action is given by: 1 2 Z da∧∗da+ 1 g2 Z Tr(F ∧∗F)−iK 8π2f Z aTr(F ∧F)+ Z iΨ /DΨ+i˜Ψ /D ˜Ψ+(m˜ΨΨ+c.c.) , (2.50) (2.50) ( ) where the gauge and ﬂavor indices on the fermions Ψ are suppressed. The presence of fundamental matter means that the one-form symmetry of axion-Yang-Mills is broken. However, there is a new zero-form ﬂavor symmetry U(Nf) ZN (0) arising from rotating the quarks, where the ZN quotient is because it is identiﬁed with the center of the gauge group. In summary the symmetry is: Z(0) K × U(Nf) ZN (0) × U(1)(2) . (2.51) (2.51) We will now show that these symmetries are extended into a non-trivial higher group whenever N does not divide K. The key feature is that when global symmetry backgrounds that make use of the ZN quotient are activated, the ﬂux of the dynamical Yang-Mills ﬁeld is also fractional. Following [14], we can understand this by looking at the gauge and global group action on the fundamental fermions. The part that acts faithfully is given by SU(N) × U(Nf) ZN ∼= SU(N) ZN × SU(Nf) ZNf × U(1) ZM , M = lcm(N, Nf) , (2.52) (2.52) and above our convention is that the U(1) factor (under which baryons are charged) acts on the fundamental fermions with charge one. This means that in general, the discrete magnetic ﬂux of the dynamical gauge group w2(SU(N)) ∈H2(M, ZN) is non-zero, as is – 15 – that of the ﬂavor group w2(SU(Nf)) ∈H2(M, ZNf ) and the fractional part of the U(1) ﬂavor ﬂux (denoted by Z below). However they are related via: that of the ﬂavor group w2(SU(Nf)) ∈H2(M, ZNf ) and the fractional part of the U(1) ﬂavor ﬂux (denoted by Z below). However they are related via: Z Σ2 Z 2π = 1 N Z Σ2 w2(SU(N)) + 1 Nf Z Σ2 w2(SU(Nf)) + s , s ∈Z . (2.53) (2.53) Similarly, we can also write a formula relating the fractional ﬂux of the U(1) to the discrete magnetic ﬂux w2(SU(Nf)) for the ﬂavor bundles of U(Nf)/ZN as: N Z Σ2 Z 2π = N Nf Z Σ2 w2(SU(Nf)) + ℓ, ℓ∈Z . 3.1 Inequalities for axion-Yang-Mills Consider as an illustrative example the case of axion-SU(N)-Yang-Mills described in sec- tion 2.3. The action is given by (2.45) and the symmetries are Z(0) K × Z(1) N × U(1)(2). The key result is equation (2.47) for the gauge invariant four-form ﬁeld strength associated to the symmetry U(1)(2): G(4) = dA(3) + KN(N −1) 4π B(2) ∧B(2) , (3.1) (3.1) – 16 – where B(2) is the background ﬁeld for the Z(1) N one-form global symmetry. In particular notice that when B(2) is activated, there is in general also a non-trivial four-form ﬂux G(4) (typically fractional). Now imagine a renormalization group ﬂow where both the symmetries in question are emergent. Let Escreen be the scale below which the Z(1) N is a good symmetry of the eﬀective ﬁeld theory, and Estring the scale below which U(1)(2) is a good symmetry. Then below Escreen we can couple our theory to general backgrounds B(2) and hence, since G(4) is also in general activated, we must also be below Estring so that we can activate this background as well. Therefore we deduce that for any renormalization group trajectory which at low energies ﬂows to axion-Yang-Mills, we have model independent inequalities: JHEP02(2022)145 Escreen ≲Estring . (3.2) (3.2) Of course, as remarked in the introduction, the scales Escreen, Estring are only approximately deﬁned along the ﬂow. Therefore, we interpret the above as the statement that Escreen cannot be parametrically larger than Estring. Of course, as remarked in the introduction, the scales Escreen, Estring are only approximately deﬁned along the ﬂow. Therefore, we interpret the above as the statement that Escreen cannot be parametrically larger than Estring. 3.1.1 Pecci-Quinn example The fermions are then massive with mψ = λf and provided this is above the QCD scale so that the model remains weakly coupled we can integrate them out. However, since the chiral symmetry is anomalous, this chiral rotation generates an axion coupling between a and the remaining gauge ﬁelds. Thus the ﬁnal action is that of axion-Yang-Mills. Via a chiral rotation of the fermions we can eliminate the a-dependence of the fermion mass term. The fermions are then massive with mψ = λf and provided this is above the QCD scale so that the model remains weakly coupled we can integrate them out. However, since the chiral symmetry is anomalous, this chiral rotation generates an axion coupling between a and the remaining gauge ﬁelds. Thus the ﬁnal action is that of axion-Yang-Mills. JHEP02(2022)145 1 2 Z da ∧∗da + 1 g2 Z Tr(F ∧∗F) −iK 8π2f Z aTr(F ∧F) , (3.7) (3.7) The UV theory has neither a one-form nor a two-form global symmetry. Both are emergent at low energies and are associated to characteristic energy scales. Speciﬁcally: The UV theory has neither a one-form nor a two-form global symmetry. Both are emergent at low energies and are associated to characteristic energy scales. Speciﬁcally: • the one-form symmetry is associated to a scale • the one-form symmetry is associated to a scale Escreen ≈λf , (3.8) (3.8) which is the mass of the fundamental fermions. Indeed above this scale, the fermions ﬂuctuate and can screen the charged Wilson lines breaking the one-form symme- try Z(1) N . which is the mass of the fundamental fermions. Indeed above this scale, the fermions ﬂuctuate and can screen the charged Wilson lines breaking the one-form symme- try Z(1) N . which is the mass of the fundamental fermions. Indeed above this scale, the fermions ﬂuctuate and can screen the charged Wilson lines breaking the one-form symme- try Z(1) N . • Deﬁning the two-form symmetry scale Estring is more subtle. There are several scales naturally associated to the emergence of the axion. First, there is the string tension T 1/2 string = f.12 There is also the scale mρ ≈mf where the radial mode begins to ﬂuctuate. At weak coupling, m ≪1 these scales are well separated T 1/2 string ∼f ≫ mρ ∼mf. 3.1.1 Pecci-Quinn example It is useful to make the inequality more precise in the case of an explicit KSVZ axion model [18, 19], where the ﬁeld a emerges from a Pecci-Quinn mechanism. In the ultraviolet, we have SU(N) gauge theory with K Weyl fermions in the fundamental (ψ+) and anti- fundamental (ψ−) representation (ﬂavor indices suppressed) and an uncharged complex scalar ﬁeld ϕ. The action is: S = Z d4x ( 1 g2 Tr[F ∧∗F] + i ¯ψ± /Dψ± + 1 2dϕ ∧∗d ¯ϕ −V (ϕ) + λ ¯ϕψ+ψ−+ λϕ ¯ψ−¯ψ+ ) . (3 (3.3) We assume the coupling constants above are small, so that semiclassical analysis is reliabl We assume the coupling constants above are small, so that semiclassical analysis is reliable. At the classical level there is a U(1) symmetry where the diﬀerent ﬁelds have We assume the coupling constants above are small, so that semiclassical analysis is reliable. At the classical level, there is a U(1)PQ symmetry where the diﬀerent ﬁelds have charges We assume the coupling constants above are small, so that semiclassical analysis is reliable. At the classical level, there is a U(1)PQ symmetry where the diﬀerent ﬁelds have charges At the classical level, there is a U(1)PQ symmetry where the diﬀerent ﬁelds have charges Field U(1)PQ charge ψ± +1 ϕ +2 At the quantum level, this U(1)PQ is broken by an ABJ anomaly. Let the scalar poten- tial be: At the quantum level, this U(1)PQ is broken by an ABJ anomaly. Let the scalar poten- tial be: V (ϕ) = m2(\|ϕ\|2 −f2)2 , (3.4) (3.4) so that ϕ condenses and gives a mass to the fermions through the Yukawa couplings. When we ﬂow to energies below the mass of the radial mode of ϕ we will see that this theory is described by the gauge ﬁelds and a pseudo-goldstone mode a which is an axion. More explicitly, we write the complex scalar ﬁeld as More explicitly, we write the complex scalar ﬁeld as ϕ = ρeia/f . (3.5) (3.5) – 17 – – 17 – The radial mode ρ has a mass mρ ∼mf. Below this scale the action reduces to: S = Z d4x ( 1 g2 Tr[F ∧∗F] + i ¯ψ± /Dψ± + 1 2da ∧∗da + λfe−ia/fψ+ψ−+ λfeia/f ¯ψ+ ¯ψ− ) (3 (3.6) Via a chiral rotation of the fermions we can eliminate the a-dependence of the fermion mass term. 3.1.1 Pecci-Quinn example Note that just above the energy mρ, the scalar potential conﬁnes ϕ to valley surrounding the minimum and the scalar ﬁeld eﬀectively becomes S1 × I-valued where I is an interval. It is this eﬀective non-trivial topology in the space of ﬁelds, i.e. the presence of a circle factor, that ensures there is a two form-symmetry and a topologically conserved winding number. In particular just above the scale mρ the two-form symmetry persists. As we go to yet higher scales, we eventually reach the energy Estring where the ρ particle can traverse all the way to the origin in ﬁeld space 12The tension of the axion string can be estimated by approximating a charge P axion string solution ϕ = fθ(\|x\| −ℓcore)eiP φ , where ℓcore is the characteristic size of the string. The tension can be roughly determined by the potential energy: Tstring ∼ Z d2x m2(\|ϕ\|2 −f 2)2 = πm2f 4ℓ2 core . Then using ℓcore ∼ 1 mρ ∼ 1 mf , we ﬁnd that the tension goes like Tstring ∼f 2 Then using ℓcore ∼ 1 mρ ∼ 1 mf , we ﬁnd that the tension goes like Tstring ∼f 2. Then using ℓcore ∼ 1 mρ ∼ 1 mf , we ﬁnd that the tension goes like Tstring ∼f 2. – 18 – (a) (b) (c) (d) (e) Figure 2. This ﬁgure illustrates how the axion string can unwind. In (a), we show the winding solution for the axion string along a circle of ﬁxed radius. As we go around the blue circle (a circle linking the axion string (black) in spacetime) the scalar ﬁeld ϕ winds around the bottom of the mexican hat potential in red. (b) shows the excitations of the radial mode which is activated at the scale mρ ∼mf. (c-e) shows a process by which we can unwind the scalar ﬁeld. Here we deform solution in the radial direction over the top of the potential which costs energy √mf ≫mf. The resulting conﬁguration has no winding — indicating the decay of the string. (a) (b) (b) (a) (d) (e) (c) JHEP02(2022)145 (d) (c) (e) Figure 2. This ﬁgure illustrates how the axion string can unwind. In (a), we show the winding solution for the axion string along a circle of ﬁxed radius. 3.1.1 Pecci-Quinn example As we go around the blue circle (a circle linking the axion string (black) in spacetime) the scalar ﬁeld ϕ winds around the bottom of the mexican hat potential in red. (b) shows the excitations of the radial mode which is activated at the scale mρ ∼mf. (c-e) shows a process by which we can unwind the scalar ﬁeld. Here we deform solution in the radial direction over the top of the potential which costs energy √mf ≫mf. The resulting conﬁguration has no winding — indicating the decay of the string. and break the two-form symmetry. Physically this means that at this scale there are processes that allow strings to unwind (see ﬁgure 2). This scale is determined by the value of the potential at the origin V (0) ∼m2f4 so that Estring ≈√mf . (3.9) (3.9) At weak-coupling m ≪1 there is thus a hierarchy: At weak-coupling m ≪1 there is thus a hierarchy: T 1/2 string = f ≫ Estring = √mf ≫ mρ = mf . (3.10) (3.10) Having identiﬁed the emergent symmetry scales in the problem, we can now apply the general inequality (3.2) to deduce Having identiﬁed the emergent symmetry scales in the problem, we can now apply the general inequality (3.2) to deduce √m ≳λ ⇐⇒mρf ≳m2 ψ , (3.11) (3.11) where the left-hand-side expresses the constraint in terms of dimensionless coupling con- stants, and the right-hand side expresses the same constraint in terms of the physical masses and tensions in the problem. where the left-hand-side expresses the constraint in terms of dimensionless coupling con- stants, and the right-hand side expresses the same constraint in terms of the physical masses and tensions in the problem. It is interesting to explore what happens when the inequality (3.11) is violated. Since the scales in question are all approximate, we consider a parametric limit within a weakly- coupled setting where we retain technical control: 1 ≫λ2 ≫m. We will argue that the Pecci-Quinn ﬁeld theory (3.3) breaks down. To begin let us note that the action (3.3) does not deﬁne a UV complete theory: at very high energies the scalar quartic coupling runs to a Landau pole. We interpret the UV action as an eﬀective theory deﬁned with a cutoﬀscale ΛUV ≫f. We interpret the – 19 – inequality (3.11) as a statement about the couplings deﬁned at the scale ΛUV. 13Notice that in the limit of interest λ2 ≫m the logarithm is small and we expect the one-loop eﬀective potential to be reliable. 3.1.1 Pecci-Quinn example Thus, the inequality is parametrically violated when λ2(ΛUV) ≫m(ΛUV). In this limit, we will ﬁnd that the fermions signiﬁcantly modify the eﬀective scalar potential. It is straightforward to compute the eﬀective potential by integrating out the fermions in a background of constant ϕ. (For convenience, one can combine the pair of Weyl fermions ψ± into a single Dirac fermion, and express ϕ in terms of its real and imaginary parts.) Their contribution to the eﬀective scalar potential is then: ∆V = Z dλ∂λ Tr log(−i /D + λ(ϕR + iγ5ϕI)) (3.12) = λ4NK 8π2 \|ϕ\|4 log λ\|ϕ\| ΛUV −\|ϕ\|4 4 ! . (3.12) JHEP02(2022)145 The full eﬀective potential for the theory is therefore: Veﬀ= λ4NK 8π2 \|ϕ\|4 log λ\|ϕ\| ΛUV −\|ϕ\|4 4 ! + m2(\|ϕ\|2 −f2)2 , (3.13) (3.13) where all couplings are evaluated at the scale ΛUV. where all couplings are evaluated at the scale ΛUV. Wh λ2 i ll d i i l h h l l i l d i When λ2 is small compared to m, it is clear that the tree level potential dominates and one ﬁnds a minimum near ⟨\|ϕ\|⟩= f and the previous analysis leading to axion-Yang- Mills is valid. However, when the quartic coupling is non-negligible the minimum moves drastically. Working in the limit where the cutoﬀscale is large, ΛUV ≫λf, we ﬁnd that the implied vacuum is at: ⟨\|ϕ\|⟩≈ΛUV λ exp −8π2m2 NKλ4 ! . (3.14) (3.14) In particular when λ2 ≫m so that the inequality (3.11) is violated, the minimum is larger than the cutoﬀscale ΛUV.13 When this happens the eﬀective ﬁeld theory breaks down and we lose control. Indeed, in formulating the eﬀective ﬁeld theory, we have neglected irrelevant operators in the action suppressed by powers of the cutoﬀ. For instance, terms in the potential of the form \|ϕ\|L+4/ΛL UV for large L have been discarded. This analysis is correct provided that the minimum of \|ϕ\| is small, but when \|ϕ\| is of order the cutoﬀthe EFT analysis breaks down. Thus we see that the inequality (3.11), derived from general considerations of symmetry, is built in to the consistency of the eﬀective ﬁeld theory. In particular when λ2 ≫m so that the inequality (3.11) is violated, the minimum is larger than the cutoﬀscale ΛUV.13 When this happens the eﬀective ﬁeld theory breaks down and we lose control. 3.1.1 Pecci-Quinn example Indeed, in formulating the eﬀective ﬁeld theory, we have neglected irrelevant operators in the action suppressed by powers of the cutoﬀ. For instance, terms in the potential of the form \|ϕ\|L+4/ΛL UV for large L have been discarded. This analysis is correct provided that the minimum of \|ϕ\| is small, but when \|ϕ\| is of order the cutoﬀthe EFT analysis breaks down. Thus we see that the inequality (3.11), derived from general considerations of symmetry, is built in to the consistency of the eﬀective ﬁeld theory. 3.2 Inequalities for axion-QCD We can apply similar logic to SU(N) axion-QCD which has global symmetry Z(0) K × (U(Nf)/ZN)(0) × U(1)(2). As found in equation (2.57), the ﬂuxes of G(4) are necessarily activated (and fractional) provided that we activate general backgrounds of (U(Nf)/ZN)(0) that make use of the ZN quotient. Therefore, we again obtain constraints on the scales of symmetry emergence for any ﬂow that realizes axion-QCD in the IR: EZN ≲Estring . (3.15) (3.15) EZN ≲Estring . 13Notice that in the limit of interest λ2 ≫m the logarithm is small and we expect the one-loop eﬀective potential to be reliable. – 20 – Here, the scale Estring is again the string stability scale, while EZN is the energy at which operators charged under the ZN ⊂U(Nf) appear. Here, the scale Estring is again the string stability scale, while EZN is the energy at which operators charged under the ZN ⊂U(Nf) appear. As in our analysis of axion-Yang-Mills inequalities we expect that (3.15) is enforced by the consistency of any UV eﬀective ﬁeld theory which ﬂows to axion-QCD at long distances. As a simple example consider the action: S = Z d4x 1 g2 Tr[F ∧∗F]+i ¯ψ± /Dψ±+ 1 2dϕ∧∗d ¯ϕ−m2(\|ϕ\|2−f2)2+λ ¯ϕψ+ψ−+λϕ ¯ψ−¯ψ+ +i¯ΨI /DΨI +i˜Ψ I /D ˜ΨI +(M ˜ΨIΨI +c.c.) , (3.16) (3.16) JHEP02(2022)145 where the fermions ΨI (˜ΨI) transform in the fundamental (anti-fundamental) representa- tion of the gauge group SU(N). Note that in this EFT the relevant the ﬂavor symmetry is in fact U(Nf)(0) not (U(Nf)/ZN)(0). The reason is that we can construct the local, gauge invariant operators OI = ψ1ΨI , ˜OI = ψ2 ˜ΨI , (3.17) (3.17) which transform non-trivially under ZN. When we ﬂow to the IR, the Higgs ﬁeld gives the PQ-fermions a mass. This reduces the ﬂavor symmetry U(Nf) −→U(Nf)/ZN. Therefore, we identify the scale EZN ≈λf. Meanwhile, following the discussion of axion-Yang-Mills, the string stability scale is Estring ≈√mf. The inequality (3.15) reduces to λ ≲√m and is derived identically to the analysis of axion Yang-Mills above. Again, when the inequality is parametrically violated the PQ fermions make large contributions to the scalar potential and the EFT breaks down. 3.3 Inequalities for axion electrodynamics We then have: min{Eshift, EZL} ≲Emagnetic . (3.20) (3.20) JHEP02(2022)145 3.3 Inequalities for axion electrodynamics We can also apply our analysis to axion-electrodynamics. As discussed in section 2.2, there are several higher group structures to consider when the axion coupling constant K > 1. The implications of the higher group extension of U(1)(2) are identical to those of the previous section: the emergence scale Estring must be the largest scale of those involved in the higher group. Somewhat more novel is the higher group extension of the U(1)(1) m magnetic one-form symmetry which measures the charges of ’t Hooft lines. The gauge invariant ﬁeld strength derived in (2.26) and (2.37) is: H(3) = dB(2) m + K 2πA(1) ∧B(2) e . (3.18) (3.18) Here we recall that A(1) is the background ﬁeld for the Z(0) K shift symmetry of the axion. Meanwhile, in axion-Maxwell theory (with no charged matter) B(2) e is the background for the electric Z(1) K one-form symmetry, while it is a composite due to the quotient in the ﬂavor symmetry (SU(Nf)/ZL)(0) (L = gcd(K, Nf)) in the axion-QED. In particular, we see that if both A(1) and B(2) e are present then necessarily H(3) ﬂuxes are also present. Following similar logic to our analysis of axion-YM in section 3.1, this leads to a constraint on the scales of symmetry emergence. Here we recall that A(1) is the background ﬁeld for the Z(0) K shift symmetry of the axion. Meanwhile, in axion-Maxwell theory (with no charged matter) B(2) e is the background for the electric Z(1) K one-form symmetry, while it is a composite due to the quotient in the ﬂavor symmetry (SU(Nf)/ZL)(0) (L = gcd(K, Nf)) in the axion-QED. In particular, we see that if both A(1) and B(2) e are present then necessarily H(3) ﬂuxes are also present. Following similar logic to our analysis of axion-YM in section 3.1, this leads to a constraint on the scales of symmetry emergence. – 21 – In the case of axion Maxwell theory, we let Emagnetic be the scale of emergence of the U(1)(1) m magnetic one-form symmetry. We also denote by Eshift the scale where the Z(0) K shift symmetry emerges, and Escreen the scale where the Z(1) K electric one-form symmetry emerges. Then we obtain a universal inequality: min{Eshift, Escreen} ≲Emagnetic . (3.19) (3.19) Similarly in axion-QED the role of Eelectric is now played by EZL, above which there are operators charged under the ZL subgroup of the ﬂavor group SU(Nf). 3.3.1 Adjoint Higgsing example As with the inequalities derived for axion-QCD, we expect that in examples (3.19) and (3.20) are enforced automatically by self consistency. Let us illustrate this in a concrete ﬂow ending in axion-Maxwell theory. In the UV we consider a model of SU(2) axion-Yang-Mills coupled to a real adjoint Higgs ﬁeld Φ, and potential V (Φ). The action is: 1 2 Z da∧∗da+ 1 g2 Z Tr(F ∧∗F)− i 8π2f Z aTr(F ∧F)+ Z Tr (DΦ ∧∗DΦ)−V (Φ) . (3.2 Note in particular that we have set the integral coupling constant K = 1 above. This means that this UV action has symmetry Z(1) 2 (arising from the center of the gauge group), and U(1)(2), but no discrete shift symmetry of the axion. We now consider a potential such that ⟨Tr(Φ2)⟩= v2 thus Higgsing the gauge group from SU(2) to U(1). Importantly, this means that the minimum instanton number is multiplied by two: Tr(F ∧F) →2F ∧F, where the right-hand-side is the remaining abelian gauge ﬁeld. Therefore at low-energies we ﬁnd axion-Maxwell theory (plus an irrelevant decoupled scalar descending from Φ) with discrete coupling constant K = 2. Therefore, at low-energies we arrive in a theory with non-trivial higher group structure extending the emergent U(1)(1) m magnetic one-form symmetry. It is straightforward to determine the energy scales associated with the emergent sym- metries: • the Z(1) 2 electric symmetry is present in both the UV and IR. Thus we have Escreen →∞ ( ) • The U(1)(1) m magnetic one-form symmetry emerges at the Higgsing scale, so Emagnetic ≈ v. One way to understand this is that, in terms of UV non-abelian ﬁelds, the long- distance gauge ﬁeld is 1 vTr(ΦF). Therefore, the scale v is where its correlation func- tions begin to diﬀer from those of an Abelian gauge ﬁeld. • The Z(0) 2 emergent zero form symmetry is broken in the UV by instantons that gen- erate a potential for the axion. One can estimate this potential using an instanton gas approximation 2 ! • The Z(0) 2 emergent zero form symmetry is broken in the UV by instantons that gen- erate a potential for the axion. One can estimate this potential using an instanton gas approximation V (a) ≈g4αv4 exp −8π2 g2 ! cos(a/f) . 3.3.1 Adjoint Higgsing example (3.22) (3.22) – 22 – Here, g is the coupling evaluated at the scale v of Higgsing, and α is a constant ﬁxed by the β function of the theory. This formula arises by estimating the integral over instanton moduli space where the size modulus of the instanton is cut oﬀat the scale gv where the W-bosons become massive. The energy cost of a shift a →a+πf, then sets the scale Eshift: Eshift ≈gαv exp −2π2 g2 ! . (3.23) (3.23) We can now apply the inequality (3.19): Eshift ≲Emagnetic ⇐⇒gα exp −2π2 g2 ! ≲1 . (3.24) JHEP02(2022)145 (3.24) Thus, the inequality is simply the statement that at the Higgsing scale v, where g is evaluated, the theory is weakly coupled g(v) ≪1. Of course, if this is not true, the theory conﬁnes before Higgsing and our analysis leading to axion-maxwell theory in the IR breaks down. Therefore (3.19) is automatic as expected. Acknowledgments We thank T. Dumitrescu, Z. Komargodski, M. Reece, and L.T. Wang for discussions and comments. TDB is supported by the Mafalda and Reinhard Oehme Postdoctoral Fellow- ship in the Enrico Fermi Institute at the University of Chicago and in part by the U.S. Department of Energy DE-SC0009924. The work of CC is supported by the U.S. Depart- ment of Energy DE-SC0021432 and by the Simons collaboration on Global Categorical Symmetries. Open Access. This article is distributed under the terms of the Creative Commons Attribution License (CC-BY 4.0), which permits any use, distribution and reproduction in any medium, provided the original author(s) and source are credited. References [1] J.E. Kim and G. Carosi, Axions and the Strong CP Problem, Rev. Mod. Phys. 82 (2010) 557 [Erratum ibid. 91 (2019) 049902] [arXiv:0807.3125] [INSPIRE]. [2] D.J.E. Marsh, Axion Cosmology, Phys. Rept. 643 (2016) 1 [arXiv:1510.07633] [INSPIRE]. [3] P. Svrček and E. Witten, Axions In String Theory, JHEP 06 (2006) 051 [hep-th/0605206] [INSPIRE]. [4] D. Gaiotto, A. Kapustin, N. Seiberg and B. Willett, Generalized Global Symmetries, JHEP 02 (2015) 172 [arXiv:1412.5148] [INSPIRE]. [5] E. Lake, Higher-form symmetries and spontaneous symmetry breaking, arXiv:1802.07747 [INSPIRE]. [6] S.R. Coleman, There are no Goldstone bosons in two-dimensions, Commun. Math. Phys. 31 (1973) 259 [INSPIRE]. [1] J.E. Kim and G. Carosi, Axions and the Strong CP Problem, Rev. Mod. Phys. 82 (2010) 557 [Erratum ibid. 91 (2019) 049902] [arXiv:0807.3125] [INSPIRE]. [2] D.J.E. Marsh, Axion Cosmology, Phys. Rept. 643 (2016) 1 [arXiv:1510.07633] [INSPIRE]. [3] P. Svrček and E. Witten, Axions In String Theory, JHEP 06 (2006) 051 [hep-th/0605206 [INSPIRE]. [4] D. Gaiotto, A. Kapustin, N. Seiberg and B. Willett, Generalized Global Symmetries, JHEP 02 (2015) 172 [arXiv:1412.5148] [INSPIRE]. [4] D. Gaiotto, A. Kapustin, N. Seiberg and B. Willett, Generalized Global Symmetries, JHEP 02 (2015) 172 [arXiv:1412.5148] [INSPIRE]. [5] E. Lake, Higher-form symmetries and spontaneous symmetry breaking, arXiv:1802.07747 [INSPIRE]. [5] E. Lake, Higher-form symmetries and spontaneous symmetry breaking, arXiv:1802.07747 [INSPIRE]. [6] S.R. Coleman, There are no Goldstone bosons in two-dimensions, Commun. Math. Phys. 31 (1973) 259 [INSPIRE]. [6] S.R. Coleman, There are no Goldstone bosons in two-dimensions, Commun. Math. Phys. 31 (1973) 259 [INSPIRE]. – 23 – [7] M.B. Green and J.H. Schwarz, Anomaly Cancellation in Supersymmetric D = 10 Gauge Theory and Superstring Theory, Phys. Lett. B 149 (1984) 117 [INSPIRE]. [8] A. Kapustin and R. Thorngren, Higher symmetry and gapped phases of gauge theories, arXiv:1309.4721 [INSPIRE]. [9] A. Kapustin and R. Thorngren, Anomalies of discrete symmetries in various dimensions and group cohomology, arXiv:1404.3230 [INSPIRE]. [10] Y. Tachikawa, On gauging ﬁnite subgroups, SciPost Phys. 8 (2020) 015 [arXiv:1712.09542] [INSPIRE]. [11] C. Córdova, T.T. Dumitrescu and K. Intriligator, Exploring 2-Group Global Symmetries, JHEP 02 (2019) 184 [arXiv:1802.04790] [INSPIRE]. JHEP02(2022)145 [12] F. Benini, C. Córdova and P.-S. Hsin, On 2-Group Global Symmetries and their Anomalies, JHEP 03 (2019) 118 [arXiv:1803.09336] [INSPIRE]. [13] N. Seiberg, Y. Tachikawa and K. Yonekura, Anomalies of Duality Groups and Extended Conformal Manifolds, Prog. Theor. Exp. Phys. 2018 (2018) 073B04 [arXiv:1803.07366] [INSPIRE]. [14] C. Córdova, D.S. Freed, H.T. Lam and N. References Seiberg, Anomalies in the Space of Coupling Constants and Their Dynamical Applications II, SciPost Phys. 8 (2020) 002 [arXiv:1905.13361] [INSPIRE]. [15] Y. Hidaka, M. Nitta and R. Yokokura, Higher-form symmetries and 3-group in axion electrodynamics, Phys. Lett. B 808 (2020) 135672 [arXiv:2006.12532] [INSPIRE]. [16] Y. Hidaka, M. Nitta and R. Yokokura, Global 3-group symmetry and ’t Hooft anomalies in axion electrodynamics, JHEP 01 (2021) 173 [arXiv:2009.14368] [INSPIRE]. [17] C. Córdova, T.T. Dumitrescu and K. Intriligator, 2-Group Global Symmetries and Anomalies in Six-Dimensional Quantum Field Theories, JHEP 04 (2021) 252 [arXiv:2009.00138] [INSPIRE]. [18] J.E. Kim, Weak Interaction Singlet and Strong CP Invariance, Phys. Rev. Lett. 43 (1979) 103 [INSPIRE]. [19] M.A. Shifman, A.I. Vainshtein and V.I. Zakharov, Can Conﬁnement Ensure Natural CP Invariance of Strong Interactions?, Nucl. Phys. B 166 (1980) 493 [INSPIRE]. [20] R.D. Peccei and H.R. Quinn, CP Conservation in the Presence of Instantons, Phys. Rev. Lett. 38 (1977) 1440 [INSPIRE]. [21] R.D. Peccei and H.R. Quinn, Constraints Imposed by CP Conservation in the Presence of Instantons, Phys. Rev. D 16 (1977) 1791 [INSPIRE]. [22] C. Córdova, D.S. Freed, H.T. Lam and N. Seiberg, Anomalies in the Space of Coupling Constants and Their Dynamical Applications I, SciPost Phys. 8 (2020) 001 [arXiv:1905.09315] [INSPIRE]. [23] C. Córdova and K. Ohmori, Anomaly Obstructions to Symmetry Preserving Gapped Phases, arXiv:1910.04962 [INSPIRE]. [24] C. Córdova, P.-S. Hsin and N. Seiberg, Time-Reversal Symmetry, Anomalies, and Dualities in (2 + 1)d, SciPost Phys. 5 (2018) 006 [arXiv:1712.08639] [INSPIRE]. [25] D. Gaiotto, A. Kapustin, Z. Komargodski and N. Seiberg, Theta, Time Reversal, and Temperature, JHEP 05 (2017) 091 [arXiv:1703.00501] [INSPIRE]. – 24 –
https://openalex.org/W1825957253	https://link.springer.com/content/pdf/10.1007/s10543-015-0562-0.pdf	English	null	Trigonometric interpolation on lattice grids	BIT/BIT numerical mathematics	2,015	cc-by	7,501	BIT Numer Math (2016) 56:341–356 DOI 10.1007/s10543-015-0562-0 Trigonometric interpolation on lattice grids Tor Sørevik1 · Morten A. Nome1 Received: 26 June 2014 / Accepted: 19 May 2015 / Published online: 1 August 2015 © The Author(s) 2015. This article is published with open access at Springerlink.com Abstract In this paper we construct non-aliasing interpolation spaces and Lagrange functions for lattice grids. We argue that lattice grids are good for trigonometric inter- polation and support this claim by numerical experiments. A greedy algorithm allows us to embed hyperbolic crosses in our interpolation spaces, and numerical experiments indicate that lattice grids are at least as good as sparse grids for trigonometric interpo- lation. A straightforward FFT-algorithm for functions sampled on lattice grids allows for fast computation and good approximation. Keywords Trigonometric interpolation · Fourier coefﬁcients · Trigonometric polynomials Mathematics Subject Classiﬁcation 42A15 · 42A16 · 42A05 Communicated by Lothar Reichel. B Tor Sørevik tor.sorevik@math.uib.no Morten A. Nome morten.nome@math.uib.no 1 Department of Mathematics, University of Bergen, Bergen, Norway B Tor Sørevik tor.sorevik@math.uib.no B Tor Sørevik tor.sorevik@math.uib.no Morten A. Nome morten.nome@math.uib.no 1 Department of Mathematics, University of Bergen, Bergen, Norway 1 Introduction If N basis functions are needed for a sufﬁciently accurate approximation in one dimen- sion, then the naive extension to s dimensions is a tensor product of N s basis functions. This exponential growth in terms is a show-stopper for high dimensional computing, but in many applications the expansion terms of a large number of these basis functions Communicated by Lothar Reichel. 123 123 342 T. Sørevik, M. A. Nome are insigniﬁcant, and can be omitted without serious degradation of the accuracy. For example, the hyperbolic cross approximation space (whose name refers to the shape of its Fourier index set) has been suggested as an alternative [1,21,23] for functions of bounded mixed derivatives. are insigniﬁcant, and can be omitted without serious degradation of the accuracy. For example, the hyperbolic cross approximation space (whose name refers to the shape of its Fourier index set) has been suggested as an alternative [1,21,23] for functions of bounded mixed derivatives. Sampling grids and approximation spaces come in pairs, in the sense that one desires to be able to go quickly and accurately between function values and expansion coefﬁ- cients. For hyperbolic cross approximation spaces, a specially tailored version of the FFT-algorithm is available [5–7] for sparse grids [1], but Kämmerer and Kunis prove in [9] that the condition number for computing expansion coefﬁcients grows disturbingly with the number of grid points. Numerical experiments also show that the computa- tional overhead is much higher for sparse grid FFT than for regular multidimensional FFT [13]. For trigonometric interpolation, lattice grids have been suggested as an alternative to sparse grids [8,10,11,18]. By a simple variable transformation, sampling on lattice grids permits a standard multidimensional FFT-algorithm for computing the expan- sion coefﬁcients [18], and the condition number of the matrix of the discrete Fourier transform grows only modestly [10] with the number of grid points. For any combination of sampling grid and approximation space, the Lebesgue constant is a measure of the quality of this combination. One famous example is polynomial interpolation, which is terrible on the cartesian grid, but excellent on many different grids with grid points clustering near the end points of the interval. In this paper we study trigonometric interpolation on lattice grids in general, propose a conjecture regarding the Lebesgue constant of lattice grids, and conduct experiments indicating the behaviour of the Lebesgue constant for lattice grids and sparse grids. 1 Introduction S tti ll 1 f ll k d ﬁ i t t f ti – The tensor product space: T s d = {k\| ∥k ∥∞≤d} – The tensor product space: T s d = {k\| ∥k ∥∞≤d} – The tensor product space: T s d = {k\| ∥k ∥∞≤d} d – All trigonometric polynomials of degree d Ps d = {k\| ∥k ∥1≤d} g p y f g d { \| ∥ ∥ } – The Dyadic hyperbolic cross Hs d = {k\| ki ∈(−2 ji−1, 2 ji−1] ∩ZZ, ∥j ∥1≤d} – The Symmetric hyperbolic cross ˜Hs d = {k\| ki ∈[−2 ji−1, 2 ji−1]∩ZZ, ∥j ∥1≤d}. d – The Dyadic hyperbolic cross Hs d = {k\| ki ∈(−2 ji−1, 2 ji−1] ∩ZZ, ∥j – The Symmetric hyperbolic cross Hs d = {k\| ki ∈[−2 ji−1, 2 ji−1]∩ZZ, ∥j ∥1≤d}. Setting all ck = 1 for all k deﬁnes an important function. Setting all ck = 1 for all k deﬁnes an important function. Deﬁnition 1.1 The Dirichlet kernel of S on [0, 1)s is deﬁned as DS(x) = k∈S e2πik·x. DS(x) = k∈S e2πik·x. Another useful tool for interpolation are the Lagrange functions. Another useful tool for interpolation are the Lagrange functions. Deﬁnition 1.2 Let Ω be a set of grid points, unisolvent with respect to HS. The Lagrange functions of Ω with respect to HS are functions Li(x) ∈HS satisfying Li(x j) = δi j. If Ω is unisolvent with respect to HS, a set of N Lagrange functions can always be found, and the interpolation can be written I f = N i=1 f (xi)Li(x). If Ω is unisolvent with respect to HS, a set of N Lagrange functions can always be found, and the interpolation can be written I f = N i=1 f (xi)Li(x). A basic question is how well the interpolant approximates the interpolated function. Let f ∗be the best approximation of f onto HS. Since f ∗∈HS, we have ∥I f −f ∥≤∥I ( f −f ∗) ∥+ ∥f −f ∗∥≤(∥I ∥+1) ∥f −f ∗∥, and thus the norm of the interpolation operator tells us how far I f is from the best approximation of f in HS. From the deﬁnition of operator norm it follows that and thus the norm of the interpolation operator tells us how far I f is from the best approximation of f in HS. 1 Introduction We believe that the modest growth of the Lebesgue constants of lattice grids explains why these grids seem excellent for trigonometric approximation. In the current section we give the necessary background on multidimensional trigonometric interpolation, and introduce the Lebesgue constant. In Sect. 2 we com- pute the known Lebesgue constant for the cartesian grid in several dimensions. In Sects. 3 and 4 we discuss interpolation and Fast Fourier Transform on lattice grids. In Sect. 5 we illustrate our ﬁndings by numerical experiments. Let Ω be a set of N grid points x j ∈[0, 1)s. We are interested in approximating a periodic function f on [0, 1)s by an s−dimensional trigonometric polynomial f (x) ≈ k∈S cke2πik·x, such that such that f (x j) = k∈S cke2πik·x j ∀x j ∈Ω. (1.1) (1.1) Here k = (k1, k2, . . . , ks) is a multi-index, commonly called wave numbers or Fourier indices, and S ⊂ZZs is a ﬁnite set with \|S\| = N. Associated with S is the approxi- mation space HS = {e2πik·x}k∈S. We write I f = k∈S cke2πik·x, 123 123 Trigonometric interpolation on lattice grids 343 where I denotes the interpolation operator. The expansion coefﬁcients ck are usually found by Fast Fourier Transform. If the system (1.1) is non-singular, the grid is said to be unisolvent with respect to HS. Approximation spaces of interest are: where I denotes the interpolation operator. The expansion coefﬁcients ck are usually found by Fast Fourier Transform. If the system (1.1) is non-singular, the grid is said to be unisolvent with respect to HS. Approximation spaces of interest are: – The tensor product space: T s d = {k\| ∥k ∥∞≤d} – All trigonometric polynomials of degree d Ps d = {k\| ∥k ∥1≤d} – The Dyadic hyperbolic cross Hs d = {k\| ki ∈(−2 ji−1, 2 ji−1] ∩ZZ, ∥j ∥1≤d} – The Symmetric hyperbolic cross ˜Hs d = {k\| ki ∈[−2 ji−1, 2 ji−1]∩ZZ, ∥j ∥1≤d}. 1 Introduction From the deﬁnition of operator norm it follows that \|\|I\|\|∞= max \|\| f \|\|∞=1 \|\|I f \|\|∞= max \|\| f \|\|∞=1 max x∈[0,1]s \| N i=1 f (xi)Li(x)\| ≤ max \|\| f \|\|∞=1 max x∈[0,1]s N i=1 \| f (xi)\|\|Li(x)\| = max x∈[0,1]s N i=1 \|Li(x)\|. This motivates the deﬁnition of the Lebesgue constant. This motivates the deﬁnition of the Lebesgue constant. Deﬁnition 1.3 Let Ω be a set of grid points, unisolvent with respect to HS, and with Lagrange functions Li(x). The Lebesgue constant of Ω and HS is H = max x∈[0,1]s N i=1 \|Li(x)\|. 12 3 344 T. Sørevik, M. A. Nome 2 Dirichlet kernels and Lebesgue constants on equidistant grids In this section we compute Lebesgue constants for some equidistant grids. These are known results, but we include them for completeness, and for the intuition these proofs provide regarding our conjecture in Sect. 3. It was shown by Ehlich and Zeller [4] that the Lebesgue constant for trigonometric interpolation at N equidistant nodes equals the Lebesgue constant for standard polynomial interpolation at the Chebyshev points (the zeroes of the Chebyshev polynomials), which are known to grow as 2 π log N + C where C is a small constant [14]. To construct Lagrange functions, we use the 1-dimensional Dirichlet kernel Dn(x) = n k=−n eπikx = sin π(2n+1)x sin πx x ̸= 0 2n + 1 x = 0 This function is periodic, continuous, and has the 2n zeros x j = j (2n+1); j = 1, . . . , 2n on the interval [0, 1), its interval of periodicity. Its global maximum is Dn(0) = 2n+1. A simple scaling and translation thus creates trigonometric Lagrange functions on the grid points x j = j (2n+1); j = 0, . . . , 2n, This function is periodic, continuous, and has the 2n zeros x j = j (2n+1); j = 1, . . . , 2n on the interval [0, 1), its interval of periodicity. Its global maximum is Dn(0) = 2n+1. A simple scaling and translation thus creates trigonometric Lagrange functions on the grid points x j = j (2n+1); j = 0, . . . , 2n, L j(x) = 1 2n + 1 Dn(x −x j). On this grid the Lebesgue constant then becomes H = 1 2n + 1 max x∈[0,1] 2n j=0 \|Dn(x −x j)\|. Theorem 2.1 For trigonometric interpolation on the N = 2n + 1 grid points x j = j (2n+1) j = 0, . . . , 2n on [0, 1), the Lebesgue constant is bounded by H < π+4 π + 2 π log N. Theorem 2.1 For trigonometric interpolation on the N = 2n + 1 grid points x j = j (2n+1) j = 0, . . . , 2n on [0, 1), the Lebesgue constant is bounded by H < π+4 π + 2 π log N. 2 Dirichlet kernels and Lebesgue constants on equidistant grids Proof From 1 sin πx ≤ 1 πx ; 0 < \|x\| ≤1, Proof From 1 sin πx ≤ 1 πx ; 0 < \|x\| ≤1, and a bit of elementary calculus it follows that and a bit of elementary calculus it follows that and a bit of elementary calculus it follows that \|Dn(x)\| ≤ 2n + 1 for \|x\| < 1 4n+2 1 π\|x\| for \|x\| ≥ 1 4n+2 (2.1) (2.1) (2.1) Since Dn(x) is 1-periodic, and Dn(x −x j) is deﬁned by a regular shift of x j = j/(2n + 1), the sum n j=−n \|D(x −x j)\| is 1 2n+1-periodic, and we can write Since Dn(x) is 1-periodic, and Dn(x −x j) is deﬁned by a regular shift of x j = j/(2n + 1), the sum n j=−n \|D(x −x j)\| is 1 2n+1-periodic, and we can write 123 123 Trigonometric interpolation on lattice grids 345 H = 1 2n + 1 max x∈[0,1] 2n j=0 \|Dn(x −x j)\| = 1 2n + 1 max x∈[−1/2,1/2] n j=−n \|Dn(x −x j)\| = 1 2n + 1 max x∈ − 1 4n+2 , 1 4n+2 n j=−n \|Dn(x −x j)\| ≤ 1 2n + 1 n j=−n max x∈ − 1 4n+2 , 1 4n+2 \|Dn(x −x j)\|. H = 1 2n + 1 max x∈[0,1] 2n j=0 \|Dn(x −x j)\| = 1 2n + 1 max x∈[−1/2,1/2] n j=−n \|Dn(x −x j)\| = 1 2n + 1 max x∈ − 1 4n+2 , 1 4n+2 n j=−n \|Dn(x −x j)\| ≤ 1 2n + 1 n j=−n max x∈ − 1 4n+2 , 1 4n+2 \|Dn(x −x j)\|. For j = 0, Dn(0) = 2n + 1 is the maximum value. For j ̸= 0, bounds can be established by writing max x∈ − 1 4n+2 , 1 4n+2 \|Dn(x −x j)\| = max x∈ x j− 1 4n+2 ,x j+ 1 4n+2 \|Dn(x)\|, and using (2.1) evaluated at the interval endpoint closest to the origin, Dn 2\| j\| −1 4n + 2 ≤1 π 4n + 2 2\| j\| −1 = 2 π 2n + 1 2\| j\| −1. 2 Dirichlet kernels and Lebesgue constants on equidistant grids We can now write We can now write We can now write H ≤ 1 2n + 1 n j=−n max x∈[− 1 4n+2 , 1 4n+2 ] \|Dn(x −x j)\| ≤1 + 2 2n + 1 2 π n j=1 2n + 1 2 j −1 < 1 + 4 π + 4 π n 1 1 2x −1dx = 1 + 4 π + 2 π log(2n −1) < π + 4 π + 2 π log N, where we have used that \|Dn(x −x j)\| = \|Dn(x −x−j)\|. where we have used that \|Dn(x −x j)\| = \|Dn(x −x−j)\|. ⊓⊔ where we have used that \|Dn(x −x j)\| = \|Dn(x −x−j)\|. ⊓⊔ In general it is difﬁcult to ﬁnd closed form expressions for the Dirichlet kernel in several dimensions, but it is easy to show that DT s d factors into s univariate Dirichlet kernels. Lagrange functions can thus be constructed for the Cartesian grid by scaling and translation, and an application of the above theorem gives the following corollary. Corollary 2.1 For the Cartesian grid on [0, 1)s, with N grid points in each direction, the Lebesgue constant for interpolation in T s d satisﬁes H < π + 4 π + 2 π log N s . H < π + 4 π + 2 π log N s . 12 3 346 T. Sørevik, M. A. Nome 3 Interpolation on lattice grids An s-dimensional lattice Λ is simply the integer linear combinations of s linearly independent basis vectors. Arranging these vectors as rows in a matrix A, we have Λ = x \| xT = λT A, λ ∈ZZs . (3.1) (3.1) The matrix A is called a generator matrix for Λ. If Λ is periodic on [0, 1)s, the lattice will be denoted an integration lattice [19]. Associated with every lattice is the dual lattice, Λ⊥, deﬁned by k ∈Λ⊥⇔k · x ∈ZZ ∀x ∈Λ. We have to prove that this implies (k −h) ∈Λ⊥. Since N is prime, z/N may be used as generator for Λ (this is not true for any lattice point z/N unless N is prime), and the lattice points in [0, 1)s may be written By the deﬁnition of dual lattices, (k −h) ∈Λ⊥implies that (k −h) · z N is an integer, and hence k · z (mod N) = h · z (mod N). Conversely, if k · z (mod N) = h · z (mod N), then (k −h) · z N = m; m ∈ZZ. We have to prove that this implies (k −h) ∈Λ⊥. Since N is prime, z/N may be used as generator for Λ (this is not true for any lattice point z/N unless N is prime), and the lattice points in [0, 1)s may be written j z N = j z N − j z N ; j = 0, · · · , N −1. j z N − j z N · (k −h) = jm − j z N · (k −h) ⊥ But is an integer for all j, and hence k −h ∈Λ⊥. is an integer for all j, and hence k −h ∈Λ⊥. ⊓⊔ is an integer for all j, and hence k −h ∈Λ⊥. For a function f with convergent Fourier series it follows that for x j ∈Λ and S a non-aliasing index set, we have: f (x j) = k∈ZZs cke2πik·x j = k∈S m∈Λ⊥/0 ck+me2πi(k+m)·x j = k∈S ˜cke2πik·x j , f (x j) = k∈ZZs cke2πik·x j = k∈S m∈Λ⊥/0 ck+me2πi(k+m)·x j = k∈S ˜cke2πik·x j , where where where ˜ck = m∈Λ⊥\{0} ck+m. (3.2) ˜ck = m∈Λ⊥\{0} ck+m. (3.2) (3.2) This generalizes the standard result for trigonometric interpolation to functions sam- pled on lattice grids. This generalizes the standard result for trigonometric interpolation to functions sam- pled on lattice grids. Theorem 3.1 The trigonometric interpolating polynomial on a non-aliasing index set S is p(x) = k∈S ˜cke2πik·x, here ˜ck is deﬁned by Eq. 3.2. where ˜ck is deﬁned by Eq. 3.2. where ˜ck is deﬁned by Eq. 3.2. where ˜ck is deﬁned by Eq. 3.2. k ∈Λ⊥⇔k · x ∈ZZ ∀x ∈Λ. The dual lattice is itself a lattice, and has B = (A−1)T as a generator matrix. If Λ is an integration lattice, then Λ⊥is an integer lattice, B is an integer matrix [15], and the number of lattice points in [0, 1)s is N = \| det(B)\|. An alternative form for describing an integration lattice is by its tZ - form [16]. Let Z = (z1\| · · · \|zt) ∈ZZs×t and D = diag(d1, . . . , dt) ∈INt×t; di > 1, i = 1, . . . , t. Th x ∈Λ ⇔x = Z D−1j ; j ∈ZZt. By the notation {x} we understand the fractional part of x. With this form the lattice points in Λ ∩[0, 1)s can be written xj = t i=1 ji zi di . If di+1/di ∈IN and N = t i=i dit, t coincide with the rank of the lattice rule [16]. Consequently if N is prime there is a tZ-form with t = 1. Two Fourier modes e2πik·x and e2πih·x will be indistinguishable on the lattice grid points if k −h ∈Λ⊥. This phenomenon is called aliasing. Deﬁnition 3.1 For a given integration lattice Λ, a non-aliasing index set is a set S ∈ZZs of order N such that if k, h ∈S and k ̸= h then k −h /∈Λ⊥. Lemma 3.1 Let Λ be an integration lattice with N points in [0, 1)s, where N is prime. Let z, k, h ∈ZZs/{0}, with z N ∈Λ ∩[0, 1)s. Then k · z (mod N) = h · z (mod N) ⇔k −h ∈Λ⊥ k · z (mod N) = h · z (mod N) ⇔k −h ∈Λ⊥ Proof First, a simple computation yields k · z (mod N) = h · z (mod N) ⇕ (k −h) · z (mod N) = 0 123 Trigonometric interpolation on lattice grids 347 ⇕ (k −h) · z = Nm; m ∈ZZ ⇕ (k −h) · z N = m; m ∈ZZ. (k −h) · z N = m; m ∈ZZ. By the deﬁnition of dual lattices, (k −h) ∈Λ⊥implies that (k −h) · z N is an integer, and hence k · z (mod N) = h · z (mod N). Conversely, if k · z (mod N) = h · z (mod N), then (k −h) · z N = m; m ∈ZZ. k ∈Λ⊥⇔k · x ∈ZZ ∀x ∈Λ. Deﬁnition 3.2 (A lattice interpolation space) HS is the space spanned by {e2πik·x}k ∈ S, where S satisﬁes Deﬁnition 3.1. The following theorem proves that if N is prime, we can use shifted Dirichlet kernels as Lagrange functions for interpolation on the corresponding lattice grid. 123 123 348 T. Sørevik, M. A. Nome Theorem 3.2 If x j and xl are lattice points on an integration lattice Λ, with N points in [0, 1)s, N being prime, and HS is the associated lattice interpolation space as deﬁned in Deﬁnition 3.1, then 1 N DS(x j −xl) = δ jl. 1 N DS(x j −xl) = δ jl. Proof Since any linear combination of lattice points is also a lattice point, we have x j −xl = z/N, where z ∈ZZs. Thus k · (x j −xl) = k · z/N = m N for all k ∈S, and we can write DS(x j −xl) = k∈S e2πik·(x j−xl) = N j = l N−1 m=0 e 2πim N = 0 j ̸= l The lower entry on the right is a ﬁnite geometric sum, since the complex exponentials take at most N different values (by construction N = \|S\|), and none of the m’s are equal (mod N) (due to Lemma 3.1). ⊓⊔ The lower entry on the right is a ﬁnite geometric sum, since the complex exponentials take at most N different values (by construction N = \|S\|), and none of the m’s are equal (mod N) (due to Lemma 3.1). ⊓⊔ ⊓⊔ Remark 3.1 This theorem holds for general lattices, but cannot be proved with Lemma 3.1, since this does not hold unless N is prime. The general proof requires an approach based on group theory. We will describe this approach and present the general version of the theorem in a forthcoming paper. Corollary 3.1 For a given integration lattice Λ and a non-aliasing index set, a com- plete set of Lagrange functions for trigonometric interpolation on HS is L j(x) = 1 N DS(x −x j) ∀x j ∈Λ ∪[0, 1)s. The expression for the Lebesgue constant can now be written The expression for the Lebesgue constant can now be written H = 1 N max x∈[0,1]s x j∈Λ∩[0,1)s \|DS(x −x j)\|. k ∈Λ⊥⇔k · x ∈ZZ ∀x ∈Λ. Since every Lagrange function is a shifted Dirichlet kernel, computing H is equivalent to summing the local maxima of one Dirichlet kernel over the N unit cells of the lattice. Th h Since every Lagrange function is a shifted Dirichlet kernel, computing H is equivalent to summing the local maxima of one Dirichlet kernel over the N unit cells of the lattice. Thus we have H = 1 N U max x∈U \|DS(x)\|, where U is the parallellotope deﬁned by 2s adjacent lattice points. The last sum can be interpreted as an upper Riemann sum of the integral [0,1]s \|DS(x)\|, which gives the following lower bound on H H > [0,1]s \|DS(x)\|dx = \|\|DS(x)\|\|1. 123 123 Trigonometric interpolation on lattice grids 349 Similarly, we conjecture that the total volume of N ‘hyperpyramids’ with base U Similarly, we conjecture that the total volume of N ‘hyperpyramids’ with base U 1 (1 + s)N U max x∈U \|DS(x)\| is a lower bound for \|\|DS(x)\|\|1, leading to the following bounds on H \|\|DS(x)\|\|1 ≤H ≤(s + 1)\|\|DS(x)\|\|1. (3.3) (3.3) In Travaglini [22] proved the following bounds on the Dirichlet kernel for convex polyhedral interpolation spaces In Travaglini [22] proved the following bounds on the Dirichlet kernel for convex polyhedral interpolation spaces C1(ln N)s ≤\|\|DS(x)\|\|1 ≤C2(ln N)s. (3.4) (3.4) Here C1, C2 are constants depending on s, but not on N . Combining (3.3) and (3.4) establishes the following conjecture. Conjecture 3.1 The Lebesgue constant for trigonometric interpolation on a lattice grid Λ onto an associated non-aliasing interpolation space HS, is bounded by ˜C1(ln N)s ≤H ≤˜C2(ln N)s. We now proceed to the practical contruction of non-aliasing index sets for integra- tion lattices. Theorem 3.3 Let Λ be the integration lattice generated by A, ΛT generated by AT and B = (AT )−1. Then the set S = k : k = BT x ∀x ∈ΛT ∩[0, 1)s is a non-aliasing index set for Λ. is a non-aliasing index set for Λ. Proof By (3.1) k = BT x = BT Aλ, so k is an integer vector. Since B is non-singular and \| det B\| = \| det BT \|, there are N different k in S. Next let k, h ∈S. If k−h ∈Λ⊥, then k −h = BT λ for some λ ∈ZZs, and consequently A(k −h) ∈ZZs. Note that the transposed lattice ΛT of Theorem 3.3 changes with B. As illustrated in the ﬁgure, S-domains obtained using Theorem 3.3 are always parallellotopes, being afﬁne mappings of the unit cube. If one wishes instead to mimic the spaces listed in Sect. 1, a simple greedy algorithm can be utilized to produce an interpolation space HΛ containing any speciﬁc space as a sub-space. Given a lattice, choose N non-aliasing points k ∈ZZs of increasing order. If we want S to mimic a hyperbolic cross we use the “Zaremba”-index ρ(k) = s i=1 max(1, \|ki\|) as the ordering criterion. Choosing the 1-norm \|\|k\|\|1 = s i=1 \|ki\| as ordering criterion, gives an index-space resembling Ps d . These two ordering criterions are illustrated in Fig. 2, where we display two different versions of S for the lattice of Fig. 1. None of these sets are unique, as there are many integer points having equal Zaremba index or 1-norm. k ∈Λ⊥⇔k · x ∈ZZ ∀x ∈Λ. But by construction Ak, Ah ∈[0, 1)s, so A(k −h) ∈(−1, 1)s, and as A is non-singular, we have k = h. ⊓⊔ ⊓⊔ It easy to see that translation of all of S does not destroy its non-aliasing property. For practical applications, one would like the index set to be centered about the origin. Let m be the integer vector with components mi = ⌊(maxk∈S ki +mink∈S ki)/2⌋; i = 1, . . . s. We construct an origin centered index set SO by SO = {k′ : k′ = k −m, k ∈S}. Note that the interpolation space, as deﬁned by SO, is not unique, since the choice of generator matrix is not unique. If U is a unimodular matrix, then U B generates Note that the interpolation space, as deﬁned by SO, is not unique, since the choice of generator matrix is not unique. If U is a unimodular matrix, then U B generates 3 350 T. Sørevik, M. A. Nome −50 0 50 −5 0 5 S(B1) 0 0.5 1 0 0.2 0.4 0.6 0.8 1 Λ(A) −50 0 50 −5 0 5 S(B1) −10 −5 0 5 10 −10 −5 0 5 10 S(B2) Fig. 1 In the left frame we display the rank-1 lattice generated by z = (1, 55); N = 144. In the mid- dle frame we display the index set corresponding to B1 while in the right frame we show the index set corresponding to B2 −10 −5 0 5 10 −10 −5 0 5 10 S(B2) 0 0.5 1 0 0.2 0.4 0.6 0.8 1 Λ(A) 0.5 Fig. 1 In the left frame we display the rank-1 lattice generated by z = (1, 55); N = 144. In the mid- dle frame we display the index set corresponding to B1 while in the right frame we show the index set corresponding to B2 the same dual lattice as B, but a different non-aliasing set. An illustration of SO for different choices of B is shown in Fig. 1. A generator matrix for the integration lattice shown in the left frame is 1 A = 1 144 1 55 0 144 , while two different generator matrices for the dual lattice are B1 = 34 2 −55 1 and B2 = −8 8 13 5 . Note that the transposed lattice ΛT of Theorem 3.3 changes with B. 4 FFT on lattice grids On equidistant s-dimensional Cartesian grid, the standard multi-dimensional FFT- algorithm provides fast and stable computation of the Fourier coefﬁcients. On lattice grids, the FFT-algorithm is readily available, see [18]. We include it for completeness. Utilizing the t Z-form, the Fourier coefﬁcients can be written 123 Trigonometric interpolation on lattice grids 351 −15 −10 −5 0 5 10 15 −15 −10 −5 0 5 10 15 −10 −5 0 5 10 −10 −5 0 5 10 Fig. 2 In this ﬁgure we show two different index spaces for the lattice of Fig. 1. In the left frame the selection criterion has been the Zaremba-index, while in the right frame we have used the trigonometric degree −15 −10 −5 0 5 10 15 −15 −10 −5 0 5 10 15 −10 −5 0 5 10 −10 −5 0 5 10 Fig. 2 In this ﬁgure we show two different index spaces for the lattice of Fig. 1. In the left frame the selection criterion has been the Zaremba-index, while in the right frame we have used the trigonometric degree f (k) = 1 N x∈Λ∩[0,1)s f (x)e−2πik·x; k ∈S = 1 N j∈ZZD f (xj)e−2πikT Z D−1j; k ∈S = 1 N j∈ZZD f (xj)e−2πimT D−1j = f (m); m ∈ZZD, Here ZZD = ZZd1 × · · · × ZZdt where ZZdi = {0, 1, . . . , di −1}, xj = Z D−1j and Here ZZD = ZZd1 × · · · × ZZdt where ZZdi = {0, 1, . . . , di −1}, xj = Z D−1j and m = Z T k + Dl for some l ∈ZZt. (4.1) (4.1) The expression for f (m) is a standard t-dimensional DFT, straightforwardly computed by your favorite FFT-algorithm. The components are identiﬁed by the m-index which are related to the k-index by the relation (4.1). 5 Numerical experiments In this section we present some computational results in 2D and 3D. In Fig. 3 we show an estimate of the Lebesgue constants for lattice grids compared to sparse grids as a function of interpolation points. An estimate for the Lebesgue constant is obtained by evaluating the Lagrange functions on a ﬁne grid. In 2D we have used the rank-1 lattices generated by z = [1, 2d −1] and N = 2d2; d = 2 : 2 : 24. These are known to be optimal lattices for the space Ps d [2]. In 3D we have used lattices optimized for the space Ps d , based on skew-circulant generator matrices[17]. For the computation of sparse grids and hyperbolic spaces we have used the matlab package nhcfft-0.91 freely available from TU-Chemnitz [3]. 12 3 352 T. Sørevik, M. A. Nome 10 0 10 1 10 2 10 3 10 4 0 5 10 15 20 25 30 35 40 Estimated Lebesgue number Lattice grid Sparse grid (log n)s cs(log n)s 10 0 10 1 10 2 10 3 10 0 20 40 60 80 100 120 140 160 180 Estimated Lebesgue number Lattice grid Sparse grid (log n)s cs(log n)s Fig. 3 Estimate of Lebesgue constants for lattice grid and sparse grid. 2D in left frame and 3D to the right. For comparison we have plotted the curves (log n)s and cs(log n)s, where c2 = 3, c3 = 10 and n = N(1/s). N being the number of grid points 10 0 10 1 10 2 10 3 10 0 20 40 60 80 100 120 140 160 180 Estimated Lebesgue number Lattice grid Sparse grid (log n)s cs(log n)s 10 0 10 1 10 2 10 3 10 4 0 5 10 15 20 25 30 35 40 Estimated Lebesgue number Lattice grid Sparse grid (log n)s cs(log n)s Estimated Lebesgue number Estimated Lebesgue number Fig. 3 Estimate of Lebesgue constants for lattice grid and sparse grid. 2D in left frame and 3D to the right. For comparison we have plotted the curves (log n)s and cs(log n)s, where c2 = 3, c3 = 10 and n = N(1/s). N being the number of grid points We note that in 2D as well in 3D the Lebesgue constant is signiﬁcantly lower for lattice grids than for sparse grids. 5 Numerical experiments The Lebesgue constant seems for both grids to grow as Cs(log n)s, with lattice grids having signiﬁcantly lower constant Cs. To test the quality of the interpolation we have used the following test functions: f1(x) = s j=1 (x j −1)2x2 j f2(x) = s j=1 (esin 2πx j −1) gp(x) = s j=1 2 + sgn x j −1 2 sin 2πx j p ; p = 1, 2, 3. The ﬁrst two are found in [18], while the family gp is a scaled version of the fam- ily used for testing sparse grid interpolation in [6]. Note that in each direction the periodic extension of f1(xi) is in C3(R) while f2(xi) is in C∞(R) and gp(xi) is in C p(R). Since the asymptotic behavior of the Fourier coefﬁcients of a one dimen- sional C p(R) function is ck ∼\|k\|−(p+1), the s-dimensional coefﬁcient will behave as ck = s i=1 cki ∼ s i=1 \|ki\| −(p+1) for our functions, assuming all ki ̸= 0. This indicates that the hyperbolic cross is a good approximation space for product functions in C p(R), and we expect the sparse grid to work very well in this case. When f ∈C∞(R) the decay rate for the 1-dimensional Fourier coefﬁcients is expo- nential: \|ck\| ∼a\|k\|; 0 < a < 1. The corresponding s−dimensional coefﬁcient is ck = s i=1 cki ∼s i=1 a\|ki\| = a\|\|k\|\|1, and in this case Ps d should be a better approxi- mation space. 123 123 Trigonometric interpolation on lattice grids 353 The errors reported in Figs. 4, 5, 6, 7 are estimates of The errors reported in Figs. 4, 5, 6, 7 are estimates of The errors reported in Figs. 4, 5, 6, 7 are estimates of Err( f ) = \|\| f (x) −L f (x)\|\|∞ \|\| f (x)\|\|∞ computed by evaluating the exact value of fi as well as the approximation on a ﬁne, regular Cartesian grid. When comparing these results we have to keep in mind that the interpolation spaces are slightly different for the two grid types; for sparse grids we have used a dyadic hyperbolic cross, while for lattice grids we have used spaces which embed the largest possible symmetric hyperbolic cross. Nevertheless we ﬁnd these results encouraging; latticegridsseemtoperformsimilarlyto,orbetterthan,sparsegridsfortheseexamples, reﬂecting the fact that it has signiﬁcantly lower Lebesgue constant. 10 0 10 1 10 2 10 3 10 4 10 −6 10 −5 10 −4 10 −3 10 −2 10 −1 10 0 Relative error of Test function no.1, s=2 lattice grid sparse grid 101 102 103 104 10 −4 10 −3 10 −2 10 −1 10 0 Relative error of Test function no.1, s=3 lattice grid sparse grid Fig. 4 A comparison of the relative interpolation error for f1 when sampled on a lattice grid and on a sparse grid. Left 2D and right 3D 10 0 10 1 10 2 10 3 10 4 10 −6 10 −5 10 −4 10 −3 10 −2 10 −1 10 0 Relative error of Test function no.1, s=2 lattice grid sparse grid 4 101 102 103 104 10 −4 10 −3 10 −2 10 −1 10 0 Relative error of Test function no.1, s=3 lattice grid sparse grid Relative error of Test function no.1, s=3 Relative error of Test function no.1, s=2 Fig. 4 A comparison of the relative interpolation error for f1 when sampled on a lattice grid and on a sparse grid. Left 2D and right 3D 100 101 102 103 104 10 −14 10 −12 10 −10 10 −8 10 −6 10 −4 10 −2 10 0 10 2 Relative error of Test function no.2, s=2 lattice grid sparse grid 101 102 103 104 10−3 10−2 10−1 10 0 Relative error of Test function no.2, s=3 lattice grid sparse grid Fig. 123 5 A comparison of the relative interpolation error for f2 when sampled on a lattice grid and on a sparse grid. Left 2D and right 3D 100 101 102 103 104 10 −14 10 −12 10 −10 10 −8 10 −6 10 −4 10 −2 10 0 10 2 Relative error of Test function no.2, s=2 lattice grid sparse grid Relative error of Test function no.2, s=2 Relative error of Test function no.2, s=2 Relative error of Test function no.2, s=3 Relative error of Test function no.2, s=3 Fig. 5 A comparison of the relative interpolation error for f2 when sampled on a lattice grid and on a sparse grid. Left 2D and right 3D 12 3 3 354 T. Sørevik, M. A. Nome 10 0 10 2 10 4 10 −2 10 −1 10 0 p = 1 10 0 10 2 10 4 10 −4 10 −2 10 0 p = 2 10 0 10 2 10 4 10 −5 10 0 p = 3 lattice grid sparse grid lattice grid sparse grid lattice grid sparse grid Fig. 6 A comparison of the relative interpolation error for gp in 2D when sampled on a lattice grid and on a sparse grid. The value used for p is: p = 1, 2, 3 10 0 10 2 10 4 10 −4 10 −2 10 0 p = 2 lattice grid sparse grid 10 0 10 2 10 4 10 −5 10 0 p = 3 lattice grid sparse grid 10 0 10 2 10 4 10 −2 10 −1 10 0 p = 1 lattice grid sparse grid 10 2 Fig. 6 A comparison of the relative interpolation error for gp in 2D when sampled on a lattice grid and o a sparse grid. The value used for p is: p = 1, 2, 3 10 0 10 2 10 4 10 −2 10 −1 10 0 10 0 10 2 10 4 10 −3 10 −2 10 −1 10 0 10 0 10 2 10 4 10 −3 10 −2 10 −1 10 0 lattice grid sparse grid lattice grid sparse grid lattice grid sparse grid Fig. 7 A comparison of the relative interpolation error for gp in 3D when sampled on a lattice grid and on a sparse grid. 123 The most remarkable difference is the minor increase in Lebesgue constants for lattice grids, as opposed to the disturbing growth for sparse grids. We believe this to be the main reason why the relative error is smaller for lattice grids than sparse grids. 6 Conclusion In this paper we have constructed non-aliasing interpolation spaces and Lagrange functions for lattice grids. Our family of approximation spaces does not include the hyperbolic cross, which is the natural Fourier index set for approximating functions with bounded mixed derivatives, but a simple greedy algorithm allows us to embed the hyperbolic cross as a sub-space in our interpolation spaces. Both lattice grids and sparse grids seem to have quasi optimal Lebesgue con- stants.The quality of the lattice interpolation appears to be better than that of sparse grid interpolation, especially in several dimensions, as measured by estimation of the Lebesgue constant as well as in approximation of test functions. Acknowledgments We would like to thank the anonymous referee, whose constructive suggestions and careful proof reading have greatly improved the readability of this paper. Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 Interna- tional License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. 123 The value used for p is: p = 1, 2, 3 10 0 10 2 10 4 10 −3 10 −2 10 −1 10 0 lattice grid sparse grid 10 0 10 2 10 4 10 −3 10 −2 10 −1 10 0 lattice grid sparse grid 10 0 10 2 10 4 10 −2 10 −1 10 0 lattice grid sparse grid 10 4 Fig. 7 A comparison of the relative interpolation error for gp in 3D when sampled on a lattice grid and on a sparse grid. The value used for p is: p = 1, 2, 3 Where lattice grids really outperforms sparse grid is for f2. As explained above we contribute this to the fact that this function is in C∞(R). This is the same effect as seen for the test function gp. Increasing p means increasing the smoothness which should favor lattice grids. s Lattice grid Sparse grid N Err(f) H N Err(f) H 2 4001 4.910−7 26.4 6144 4.110−8 61.0 3 4001 2.610−4 31.2 4096 2.610−4 235.9 4 4001 1.110−3 33.8 4048 8.610−2 623.3 5 4001 7.810−3 34.7 2972 1.04 1027.9 6 4001 1.110−1 37.8 5336 1.05 2504.0 Our brute force approach for estimating the error and the Lebesgue constant becomes very expensive for large numbers of lattice points. This limits our possi- bilities for extensive experiments in higher dimensions. We have run some limited experiments where we have kept the number of lattice points constant and compared it with sparse grid interpolation of approximately the same size. The lattice rules we have used are rank-1 lattices constructed by the component-by-component technique 123 123 Trigonometric interpolation on lattice grids 355 introduced by Sloan and Reztsov [20]. Our lattices are those reported by Kuo et al. [12]. We have also restricted us to test function f1 only. The results are reported in the table above. The most remarkable difference is the minor increase in Lebesgue constants for lattice grids, as opposed to the disturbing growth for sparse grids. We believe this to be the main reason why the relative error is smaller for lattice grids than sparse grids. introduced by Sloan and Reztsov [20]. Our lattices are those reported by Kuo et al. [12]. We have also restricted us to test function f1 only. The results are reported in the table above. References 1. Bungartz, H., Griebel, M.: Sparse grids. Acta Numer. 13, 1–123 (2004) 2. Cools, R., Sloan, I.: Minimal cubature formulae of trigonometric degree. Math. Comput. 65, 1583–1600 (1996) 3. Döhler, M., Kämmerer L., Kunis, S., Potts, D.: NHCFFT, MATLAB toolbox for the nonequispaced hyperbolic cross FFT. http://www.tu-chemnitz.de/~lkae/nhcfft/nhcfft.php (2009) 4. Ehlich, H., Zeller, K.: Auswertung der Normen yon interpolationsoperatoren math. Annalen 16 105–112 (1966) 5. Gradinaru, V.: Fourier transform on sparse grids: code design and the time dependent Schrödinger equation. Computing 80, 1–22 (2007) 6. Griebel, M., Hamaekers, J.: Fast discrete Fourier transform on generalized sparse grids. Sparse grids and applications. Lect. Notes Comput. Sci. Eng. 97, 75–108 (2014) 7. Hallatschek, K.: Fourier transformation auf dünnen Gittern mit hierarchischen Basen. Numer. Mat 63, 83–97 (1992) 8. Kämmerer, L.: Reconstruction hyperbolic cross trigonometric polynomials by sampling along rank-1 lattices. SIAM J. Numer. Anal. 51(5), 2723–2796 (2013) 9. Kämmerer, L., Kunis, S.: On the stability of the hyperbolic cross discrete Fourier transform. Numer. Math. 117, 581–600 (2011) 10. Kämmerer, L., Kunis, S., Potts, D.: Interpolation lattices for hyperbolic cross trigonometric polyno- mials. J. Complex. 28(1), 76–92 (2012) 11. Kämmerer, L., Potts, D., Volkmer, T.: Approximation of multivariate functions by trigonometric poly- nomials based on sampling along rank-1 lattice with generating vector of Korobov form. J. Complex. 31, 424–456 (2015) 12 3 356 T. Sørevik, M. A. Nome 12. Kuo, F.Y., Sloan, I.H., Wozniakowski, H.: Lattice rule algorithms for multivariate approximation the average case setting. J. Complex. 24, 283–323 (2008) 13. Kupka, F.G.: Sparse grid spectral methods and some results from approximation theory. In: Lai, C.-H., Bjørstad, P.E., Cross, M., Widlund, O.B. (eds.) Domain Decomposition, 11th edn, pp. 57–64. Greenwich, London (1999) 14. Luttmann, F.W., Rivlin, T.J.: Some numerical experiments in the theory of polynomial interpolation. IBM J. Res. Dev. 9, 187–191 (1965) 15. Lyness, J.N.: An introduction to lattice rules and their generator matrices. IMA J. Numer. Anal. 405–419 (1989) ( ) 16. Lyness, J.N., Keast, P.: Application of the Smith normal form to the structure of lattice rules. SIAM J. Matrix Anal. Appl. 16(1), 218–231 (1995) 17. Lyness, J.N., Sørevik, T.: Four dimensional lattice rules generated by skew-circulant matrices. Math. Comput. 73(245), 279–295 (2003) 18. Munthe-Kaas, H., Sørevik, T.: Multidimensional pseudo-spectral methods on Lattice grids. Appl. Numer. Math. 62, 155–165 (2012) 19. Sloan, I.H.: Lattice methods for multiple integration. J. Comput. Appl. Math. 12, 13, 131–143 (198 19. References Sloan, I.H.: Lattice methods for multiple integration. J. Comput. Appl. Math. 12, 13, 131–143 (1985) 20. Sloan,I.H.,Reztsov,A.V.:Component-by-componentconstructionofgoodlatticerules.Math.Comput. 71(237), 263–273 (2002) . Sloan,I.H.,Reztsov,A.V.:Component-by-componentconstructionofgoodlatticerules.Math.Comput. 71(237), 263–273 (2002) 21. Temlyakov, V.N.: Approximation of functions with bounded mixed derivative. Proc. Steklov Inst. Math. vi+121 (Translation of Trudy Mat. Inst. Steklov, 178) (1986) y 22. Travaglini, G.: Polyhedra summability of multiple Fourier series (and explicit formulas for Dirichlet Kernels on T n and on compact lie groups). Colloq. Math. LXV, 103–116 (1993) 23. Zenger, C.: Sparse grids. Notes Numer. Fluid Dyn. 31, 241–251 (1991) 123
https://openalex.org/W3010707384	http://www.ajnr.org/content/ajnr/41/3/515.full.pdf	English	null	Early Detection of Cancer: Evaluation of MR Imaging Grading Systems in Patients with Suspected Nasopharyngeal Carcinoma	American journal of neuroradiology	2,020	cc-by	6,796	Received September 23, 2019; accepted after revision December 14. From the Departments of Imaging and Interventional Radiology (A.D.K., Q.Y.A., T.Y.S., K.W.N.Y.), Otorhinolaryngology, Head and Neck Surgery (J.K.S.W., A.C.V.), Clinical Oncology (F.K.F.M., E.P.H., B.B.Y.M., A.T.C.C.), and Chemical Pathology (W.K.J.L., I.O.L.T., R.W.K.C., Y.M.D.L., K.C.A.C.), and Li Ka Shing Institute of Health Sciences (F.K.F.M., W.K.J.L., I.O.L.T., E.P.H., B.B.Y.M., R.W.K.C., A.T.C.C., Y.M.D.L., K.C.A.C.), The Chinese University of Hong Kong, Prince of Wales Hospital, Hong Kong SAR, China; and State Key Laboratory of Translational Oncology (F.K.F.M., W.K.J.L., I.O.L.T., E.P.H., B.B.Y.M., R.W.K.C., A.T.C.C., Y.M.D.L., K.C.A.C.), The Chinese University of Hong Kong, Hong Kong SAR, China. This work was supported partially by Kadoorie Charitable Foundation and the Research Grants Council of the Hong Kong Special Administrative Region, China (Project No. SEG_CUHK02, 14107216 and T12-404/11). Early Detection of Cancer: Evaluation of MR Imaging Grading Systems in Patients with Suspected Nasopharyngeal Carcinoma Early Detection of Cancer: Evaluation of MR Imaging Grading Systems in Patients with Suspected Nasopharyngeal Carcinoma A.D. King, J.K.S. Woo, Q.-Y. Ai, F.K.F. Mo, T.Y. So, W.K.J. Lam, I.O.L. Tse, A.C. Vlantis, K.W.N. Yip, E.P. Hui, B.B.Y. Ma, R.W.K. Chiu, A.T.C. Chan, Y.M.D. Lo, and K.C.A. Chan A.D. King, J.K.S. Woo, Q.-Y. Ai, F.K.F. Mo, T.Y. So, W.K.J. Lam, I.O.L. Tse, A.C. Vlantis, K.W.N. Yip, E.P. Hui, B.B.Y. Ma, R.W.K. Chiu, A.T.C. Chan, Y.M.D. Lo, and K.C.A. Chan o, Q.-Y. Ai, F.K.F. Mo, T.Y. So, W.K.J. Lam, I.O.L. Tse, A.C. Vlantis, K.W.N. Yip, E.P. Hui, B.B.Y. Ma, R.W.K. Chiu, A.T.C. Chan, Y.M.D. Lo, and K.C.A. Chan Please address correspondence to Ann D. King, MD, Department of Imaging and Interventional Radiology, Chinese University of Hong Kong, Prince of Wales Hospital, 30-32 Ngan Shing St, Shatin, Hong Kong SAR, China; e-mail: king2015@cuhk.edu.hk Indicates open access to non-subscribers at www.ajnr.org Indicates article with supplemental on-line table. Indicates article with supplemental on-line photos. http://dx.doi.org/10.3174/ajnr.A6444 AJNR Am J Neuroradiol 41:515–21 Mar 2020 www.ajnr.org 515 of October 23, 2024. This information is current as of October 23, 2024. This information is current as http://www.ajnr.org/content/41/3/515 https://doi.org/10.3174/ajnr.A6444 doi: 2020, 41 (3) 515-521 AJNR Am J Neuroradiol ORIGINAL RESEARCH HEAD & NECK ORIGINAL RESEARCH HEAD & NECK ABSTRACT BACKGROUND AND PURPOSE: We evaluated modiﬁcations to our contrast-enhanced MR imaging grading system for symptomatic patients with suspected nasopharyngeal carcinoma, aimed at improving discrimination of early-stage cancer and benign hyperplasia. We evaluated a second non-contrast-enhanced MR imaging grading system for asymptomatic patients from nasopharyngeal carci- noma plasma screening programs. MATERIALS AND METHODS: Dedicated nasopharyngeal MR imaging before (plain scan system) and after intravenous contrast administration (current and modiﬁed systems) was reviewed in patients from a nasopharyngeal carcinoma–endemic region, compris- ing 383 patients with suspected disease without nasopharyngeal carcinoma and 383 patients with nasopharyngeal carcinoma. The modiﬁed and plain scan systems reﬁned primary tumor criteria, added a nodal assessment, and expanded the system from 4 to 5 grades. The overall combined sensitivity and speciﬁcity of the 3 systems were compared using the extended McNemar test (a x 2 value x 2 ð2Þ. 5.99 indicates signiﬁcance). RESULTS: The current, modiﬁed, and plain scan MR imaging systems yielded sensitivities of 99.74%, 97.91%, and 97.65%, respectively, and speciﬁcities of 63.45%, 89.56% and 86.42%, respectively. The modiﬁed system yielded signiﬁcantly better performance than the current (x 2 ð2Þ ¼ 122) and plain scan (x 2 ð2Þ ¼ 6.1) systems. The percentages of patients with nasopharyngeal carcinoma in grades 1–2, grade 3, and grades 4–5 for the modiﬁed and plain scan MR imaging systems were 0.42% and 0.44%; 6.31% and 6.96%; and 90.36% and 87.79%, respectively. No additional cancers were detected after contrast administration in cases of a plain scan graded 1–2. CONCLUSIONS: We propose a modiﬁed MR imaging grading system that improves diagnostic performance for nasopharyngeal carcinoma detection. Contrast was not valuable for low MR imaging grades, and the plain scan shows potential for use in screening programs. ABBREVIATION: NPC ¼ nasopharyngeal carcinoma This work was supported partially by Kadoorie Charitable Foundation and the Research Grants Council of the Hong Kong Special Administrative Region, China (Project No. SEG_CUHK02, 14107216 and T12-404/11). Received September 23, 2019; accepted after revision December 14. Contrast-enhanced/plain scan: spread outside the nasopharynx (superﬁcial or deep Contrast-enhanced/plain scan: metastatic retropharyngeal or upper cervical nodese Contrast-enhanced/plain scan: spread outside the nasopharynx (superﬁcial or deep) Contrast-enhanced/plain scan: metastatic retropharyngeal or upper cervical nodese Contrast-enhanced/plain scan: spread outside the nasopharynx (superﬁcial or deep) Contrast-enhanced/plain scan: metastatic retropharyngeal or upper cervical nodese Contrast-enhanced/plain scan: spread outside the nasopharynx (superﬁcial or deep) Contrast-enhanced/plain scan: metastatic retropharyngeal or upper cervical nodese Note: —CUHK indicates Chinese University of Hong Kong; NPC, nasopharyngeal carcinoma. Note: —CUHK indicates Chinese University of Hong Kong; NPC, nasopharyngeal carcinoma. a Modiﬁed MR imaging protocol: T2- and T1-weighted images without and with intravenous contrast enhancemen a Modiﬁed MR imaging protocol: T2- and T1-weighted images without and with intravenous contrast enhancement. b b Plain scan MR imaging protocol: T2- and T1-weighted images without intravenous contrast enhancement. ain scan MR imaging protocol: T2- and T1-weighted images without intravenous contrast enhancement. g g p g g c For the modiﬁed grading system, the grade for contrast-enhanced images outranks the grade for non-contrast-enhanced images. d Symmetry refers to a comparison of the right and left halves of the nasopharynx for size, signal intensity, and contour; cysts do not contribute to wall or adenoid asym- metry or to distortion of adenoidal septa. c For the modiﬁed grading system, the grade for contrast-enhanced images outranks the grade for non-contrast-enhanced images. d Symmetry refers to a comparison of the right and left halves of the nasopharynx for size, signal intensity, and contour; cysts do not metry or to distortion of adenoidal septa. c For the modiﬁed grading system, the grade for contrast-enhanced images outranks the grade for non-contrast-enhanced images. d Symmetry refers to a comparison of the right and left halves of the nasopharynx for size, signal intensity, and contour; cysts do not contribute to wall or adenoid asym- metry or to distortion of adenoidal septa. e Diagnosis of a metastatic node is based on size (minimum axial nodal diameter: retropharyngeal, $6 mm; jugulodigastric, $11 mm; all other nodes, $10 mm or groups of $3 nodes with a minimal axial diameter of $8 mm) or any node with necrosis or extracapsular spread. e Diagnosis of a metastatic node is based on size (minimum axial nodal diameter: retropharyngeal, $6 mm; jugulodigastric, $11 mm; all other nodes, $10 mm or groups of $3 nodes with a minimal axial diameter of $8 mm) or any node with necrosis or extracapsular spread. Contrast-enhanced/plain scan: spread outside the nasopharynx (superﬁcial or deep Contrast-enhanced/plain scan: metastatic retropharyngeal or upper cervical nodese regions,3,4 as well as nonendemic regions,5 have shown that MR imaging can detect NPCs hidden from endoscopic view. In an MR imaging study of predominantly symptomatic patients referred to the outpatient clinic with suspected NPC, we found that MR imaging detects the 10% of tumors that are hidden from endoscopic view.3 This figure rises to 17% in asymptomatic patients screened for NPC using per- sistently elevated plasma Epstein-Barr virus DNA.4 In both studies, MR imaging detected tumors up to 3 years before they became visible endoscopically.4,6 MR imaging, there- fore, has a complementary role in early NPC detection. could produce false-positive findings leading to unnecessary alarm and resource usage.4 Moreover, in a study of a large group of stage T1 NPCs confined to the nasopharynx, we observed that a small number of these early-stage tumors were symmetrically diffuse and overlapped in appearance with benign hyperplasia.8 These early-stage tumors could potentially reduce the high sensitivity of MR imaging. p y g y g g In the first part of the study, we modified the current con- trast-enhanced MR imaging grading system, which can be used worldwide as a complementary investigation in sympto- matic patients with suspected NPC when endoscopic exami- nation findings are negative or indeterminate. The aim was to improve the specificity of MR imaging and maintain a high sensitivity, by adding focal expansion in asymmetric diffuse thickening, absent or external distortion of contrast-enhanc- ing adenoidal septa, and the presence of metastatic nodes to indicate NPC. In the second part of the study, we evaluated a new plain scan MR imaging grading system involving a short protocol without intravenous contrast for use in screening programs in endemic regions such as east and southeast parts of Asia, where the incidence in middle-aged men is up to 35 per 100,000.10 Most patients with persistently elevated plasma or serology markers for NPC referred for investiga- tion do not have NPC. In these patients, it would be advanta- geous to limit the use of gadolinium MR imaging contrast We have shown that the standardized MR imaging grad- ing system used at our institution has a high sensitivity for NPC detection,3,4,7 which includes early-stage tumors hidden on the endoscopic examination that tend to form a focal mass in the roof or pharyngeal recess.3,4 However, some early-stage cancers confined within the nasopharynx are diffuse rather than focal. ABBREVIATION: NPC ¼ nasopharyngeal carcinoma Several studies investigating patients with suspected NPC from endemic 515 Table 1: CUHK MR imaging grading systems for detection of NPC using a modified system with a contrast-enhanced scana and a plain scan system with a noncontrast-enhanced scanb MR Imaging Gradec Walls Adenoid Grade 1: normal Contrast-enhanced/plain scan: thin wall, 1–3 mm Contrast-enhanced/plain scan: absent/vestigial tags/ nubbin Grade 2: probably benign hyperplasia Contrast-enhanced/plain scan: diffuse thickening (.3 mm), symmetricd size, signal intensity, and contour Contrast-enhanced scan: composed of Thornwaldt cyst/multiple cysts, OR symmetricd size, signal intensity, and contour with preserved symmetric contrast-enhancing septa perpendicular to the roof, separated by less enhancing columns (ie, stripped appearance) Plain scan: composed of Thornwaldt cyst/multiple cysts Grade 3: indeterminate Contrast-enhanced/plain scan: diffuse thickening (.3 mm); asymmetric size or signal intensity or contour, which is nonexpansile Contrast-enhanced scan: asymmetric size, signal intensity, OR contour with preserved or partial disruption/internal distortion of contrast-enhancing septa Plain scan: symmetricd size, signal intensity, and contour Grade 4: suspicious for NPC Contrast-enhanced/plain scan: diffuse thickening (.3 mm); asymmetric size or signal intensity or contour, which is expansile (superﬁcial or deep margins) Contrast-enhanced scan: absent contrast-enhancing septa in a focal adenoid, OR external distortion of contrast-enhancing septa by an adjacent roof mass Plain scan: asymmetric size, signal intensity, or contour Grade 5: probably NPC 5a Contrast-enhanced/plain scan: focal mass Contrast-enhanced scan: absent contrast-enhancing septa in an adenoid ﬁlling the whole roof on at least 1 section Plain scan: no grade 5b Contrast-enhanced/plain scan: spread outside the nasopharynx (superﬁcial or deep) 5c Contrast-enhanced/plain scan: metastatic retropharyngeal or upper cervical nodese Table 1: CUHK MR imaging grading systems for detection of NPC using a plain scan system with a noncontrast-enhanced scanb MR Imaging Gradec Walls Grade 1: normal Contrast-enhanced/plain scan: thin wall, 1–3 mm Grade 2: probably benign hyperplasia Contrast-enhanced/plain scan: diffuse thickening (.3 mm), symmetricd size, signal intensity, and contour Grade 3: indeterminate Contrast-enhanced/plain scan: diffuse thickening (.3 mm); asymmetric size or signal intensity or contour, which is nonexpansile Grade 4: suspicious for NPC Contrast-enhanced/plain scan: diffuse thickening (.3 mm); asymmetric size or signal intensity or contour, which is expansile (superﬁcial or deep margins) Grade 5: probably NPC 5a Contrast-enhanced/plain scan: focal mass 5b Contrast-enhanced/plain scan: spread outs 5c Contrast-enhanced/plain scan: metastatic r Table 1: CUHK MR imaging grading systems for detection of NPC using a modified system plain scan system with a noncontrast-enhanced scanb Adenoid Contrast-enhanced/plain scan: absent/vestigial tags/ nubbin Contrast-enhanced scan: composed of Thornwaldt cyst/multiple cysts, OR symmetricd size, signal intensity, and contour with preserved symmetric contrast-enhancing septa perpendicular to the roof, separated by less enhancing columns (ie, stripped appearance) Contrast-enhanced scan: composed of Thornwaldt cyst/multiple cysts, OR symmetricd size, signal intensity, and contour with preserved symmetric contrast-enhancing septa perpendicular to the roof, separated by less enhancing columns (ie, stripped appearance) Plain scan: composed of Thornwaldt cyst/multiple cysts Contrast-enhanced scan: asymmetric size, signal intensity, OR contour with preserved or partial disruption/internal distortion of contrast-enhancing septa p Plain scan: symmetricd size, signal intensity, and contour Contrast-enhanced/plain scan: diffuse thickening (.3 mm); asymmetric size or signal intensity or contour, which is expansile (superﬁcial or deep margins) Contrast-enhanced scan: absent contrast-enhancing septa in a focal adenoid, OR external distortion of contrast-enhancing septa by an adjacent roof mass Plain scan: asymmetric size, signal intensity, or contour Contrast-enhanced scan: absent contrast-enhancing septa in an adenoid ﬁlling the whole roof on at least 1 section Plain scan: no grade scan: spread outside the nasopharynx (superﬁcial or deep 16 King Mar 2020 www.ajnr.org ABBREVIATION: NPC ¼ nasopharyngeal carcinoma However, subsequent investigations must be able to detect these early cancers. T he early diagnosis of nasopharyngeal carcinoma (NPC) is essential to future improvements in patient survival and reduction of the long-term adverse effects of aggressive treatment regimens.1 Recently, we found that a high percentage of patients with early-stage NPC could be identified by population screening in Hong Kong using persistently elevated plasma Epstein-Barr vi- rus DNA (71% versus 20% compared with historical data).2 T Cancers of the nasopharynx are notorious for being submuco- sal, located deep in the pharyngeal recess, or masked by ade- noidal hyperplasia in the nasopharyngeal roof. Accordingly, endoscopic tumor detection is challenging at this site, not only for asymptomatic patients in the screening setting but also for symptomatic patients in the clinical setting. MATERIALS AND METHODS Patients This retrospective evaluation of MR imaging grading systems in patients who underwent MR imaging for the staging of known NPC or investigation of suspected NPC was approved by the institutional review board (the Chinese University of Hong Kong), which waived the requirement for written informed con- sent. Nasopharyngeal MR imaging data obtained from 766 patients before and after intravenous contrast were reviewed. Patients did not have a history of head and neck cancer before they underwent the MR imaging for suspected or biopsy-proved NPC. The first group comprised 383 patients without NPC (304 men and 79 women; mean age, 52 years; age range, 18–83years) who had undergone an endoscopic examination and MR imaging for suspected NPC (based on clinical symptoms, clinical signs, or elevated blood markers for NPC) between 2005 and 2016, but had not been diagnosed with head and neck cancer (minimum follow-up of 2 years). Details are shown in On-line Fig 1. These included 358 patients recruited for 2 previous prospective NPC- detection studies.3,4 The second group comprised a similar sample size of 383 patients referred for suspected NPC (based on clinical symptoms, clinical signs, or elevated blood markers for NPC) who had biopsy-proved undifferentiated NPC (282 men and 101 women; mean age, 53years; age range, 19–92years). These patients were ran- domly selected from consecutive patients with NPC scanned between 2005 and 2016 according to the expected T-stage11 Contrast-enhanced/plain scan: spread outside the nasopharynx (superﬁcial or deep Contrast-enhanced/plain scan: metastatic retropharyngeal or upper cervical nodese In these diffuse early-stage NPCs, the wall thick- ening is usually greater on one side of the nasopharynx, so discrimination of NPC from benign hyperplasia of the walls and adenoid relies heavily on finding asymmetry on bilateral nasopharyngeal MR imaging evaluations.3,4,8,9 Unfortunately, on scrutiny of the images for small early-stage tumors, benign hyperplasia is often slightly asymmetric. This characteristic is challenging in screened subjects with benign lesions and 516 516 King Mar 2020 www.ajnr.org Table 2: Patients with and without NPC in each MR imaging grade as determined using the current, modified, and plain scan MR imaging grading systems h and without NPC in each MR imaging grade as determined using the current, modified, and plain scan MR atients with and without NPC in each MR imaging grade as determined using the current, modified, and plai ading systems Table 2: Patients with and without NPC in each MR imaging grade as determined using the current, modified, and plain scan MR imaging grading systems imaging grading systems Grading Systems No. of Patients without NPC No. of Patients with NPC Nodes Not Assessed Nodes Assessed Nodes Not Assessed Nodes Assessed Grade 1 Current 69 – 0 – Modiﬁed 69 69 0 0 Plain scan 79 78 0 0 Grade 2 Current 174 – 1 – Modiﬁed 174 170 1 1 Plain scan 147 146 2 1 Grade 3 Current 132 – 47 – Modiﬁed 109 104 13 7 Plain scan 112 107 15 8 Grade 4 Current 8 – 335 – Modiﬁed 28 26 91 26 Plain scan 42 38 123 32 Grade 5 Current – – – – Modiﬁed 3 14 278 349 Plain scan 3 14 243 342 Note:—– indicates not applicable; NPC, nasopharyngeal carcinoma. Note:—– indicates not applicable; NPC, nasopharyngeal carcinoma. distribution in a screened population: stage T1 (nasopharynx), 246 patients (64.2%); T1 (nasal cavity/oropharynx), 13 (3.4%); T2, 33 (8.6%); T3, 80 (20.9%); and T4, 11 (2.9%). These include 134 patients with T1 cancer reported previously.8 agents. Furthermore, a short plain scan has greater potential for use in screening programs in NPC endemic regions to pri- oritize or even select patients for endoscopic examination. Finally, observers with different levels of experience in head and neck MR imaging tested the modified and plain scan MR imaging systems. AJNR Am J Neuroradiol 41:515–21 Mar 2020 www.ajnr.org MR Imaging Acquisition Targeted nasopharyngeal MR imaging was performed using a 1.5T or 3T whole-body MR imaging system (Philips Healthcare, Best, the Netherlands). During a scan duration of approximately 15–20 minutes, axial fat-suppressed T2-weighted, coronal T2- weighted, and axial T1-weighted images, and axial and coronal T1-weighted images were obtained after a bolus injection of 0.1mmol per kilogram of body weight of gadoterate meglumine (Dotarem; Guerbet, Aulnay-sous-Bois, France). Statistical Analysis The performance of MR imaging for NPC detection and classifi- cation of lesions as benign and malignant was assessed by calcu- lating the sensitivity, specificity, positive predictive value, negative predictive value, and accuracy based on groupings of grades. The concordance rates between MR imaging grading sys- tems were analyzed according to Hawass.14 The McNemar test was used to test the sensitivity or specificity individually. A calcu- lated x 2 value exceeding the critical value of x2 ð2Þ ¼ 3.84 indicated a significant difference. Moreover, the extended McNemar test was used to test the overall significance of the sensitivity and specificity with a fixed error rate; a calculated x 2 value exceeding the critical value of x 2 ð2Þ ¼ 5.99 indicated a significant difference. The Youden index was applied if one system exhibited significant differences in sensitivity while the other exhibited significant dif- ferences in specificity. The highest Youden index denoted the sys- tem with the best performance. MR Imaging Grading rmance of MR imaging after grouping of grades into benign or malignant according to grades in the cur- scan grading systems ble 3: Diagnostic performance of MR imaging after grouping of grades into benign or malignant according to t, modified, and plain scan grading systems predictive value. a Combination of grades in the respective current, modiﬁed, and plain scan systems in Table 2. 3, and 4–5 were 0.42% (1/240), 6.31% (7/111), and 90.36% (375/ 415), respectively, for the modified grading system, and 0.44% (1/ 225), 6.96% (8/115), 87.79% (374/426), respectively, for the plain scan MR imaging grading system. (observer 2) and a general radiologist with a 1-year postqualifica- tion in radiology (after training on 50 patients not included the study analysis comprising 25 without and 25 with NPC with a distribution of T-stages similar to that in the study) (observer 3). The diagnostic performance using the current, modified, and plain scan MR imaging grading systems for grouped grades are presented in Table 3, and the statistical comparisons of the sys- tems are shown in Table 4. The modified grading system yielded statistically significant better overall performance than the cur- rent and the plain scan grading systems. The modified grading system yielded statistically significant better overall performance using the indeterminate grade 3 to indicate benign versus malig- nant and without nodal assessment versus with nodal assessment. DISCUSSION We modified our MR imaging grading system for NPC detection to meet the challenge of using MR imaging to investigate patients with suspected NPC from both the clinical setting (symptomatic) and the screening setting (asymptomatic). We analyzed tumors with a T-stage distribution similar to that expected in a screening population—that is, a high percentage of early-stage primary tumors. Tumors with spread outside the nasopharynx were retained in the analysis to ensure that the grading system reflected the full range of T-stage tumors that may be encountered in both of these settings and to allow us to evaluate the new plain scan Observer Results The sensitivity, specificity, and accuracy using grades 4 and 5 to indicate malignancy (Table 1) for the plain scan grading system were 96.61%, 91.91%, and 94.26%, respectively, for observer 2, and 95.04%, 96.87%, and 95.95%, respectively, for observer 3; and for the modified grading system, they were 97.65%, 92.69%, and 95.17%, respectively, for observer 2 and 97.13%, 97.13%, and 97.13%, respectively, for observer 3. The respective interclass correlation coefficients (k and weighted k) for the modified systems, which were determined using 2 (grades 1–3 versus grades 4–5) and 3 scales (grades 1–2 versus grade 3 versus grades 4–5), were 0.87 and 0.81 between observers 1 and 2 and 0.88 and 0.76 between observers 1 and 3. The respective interclass correlation coefficients for the plain scan systems using 2 and 3 scales were 0.87 and 0.83 between observers 1 and 2 and 0.84 and 0.81 between observers 1 and 3. The k and weighted k analyses were calculated to determine the interobserver agreement among 3 observers using 2 scales (grades 1–3 versus grades 4–5) and 3 scales (grades 1–2 versus grade 3 versus grades 4–5), respectively. k values of #0.20, 0.21–0.40, 0.41–0.60, 0.61–0.80, and 0.81–1.00 indicated slight, fair, moderate, substantial, and almost perfect agreement, respectively.15 All statisti- cal analyses were 2-sided, and a P value , .05 was a statistically sig- nificant difference. The analyses were performed using SPSS software (Version 25.0; IBM, Armonk, New York) and SAS software (Version 9.4; SAS Institute, Cary North Carolina). MR Imaging Grading The current MR imaging grading system3 is shown in the On- line Table. The modified and plain scan MR imaging grading sys- tems are shown in Table 1 (which differ only in grading the adenoid because contrast-enhancing septa cannot be assessed on the plain scan) and are illustrated in On-line Figs 2 and 3, respec- tively. The diagnosis of a metastatic node was based on recog- nized imaging criteria shown in Table 1.12,13 The MR imaging data were graded by readers blinded to the diagnosis of NPC or no NPC. Initially, non-contrast-enhanced images (T2- and T1- weighted images) were graded according to the proposed plain scan MR imaging grading system. Subsequently, the full scan, including contrast-enhanced images (T2- and T1-weighted images pre- and post-contrast enhancement) was graded according to our current and proposed modified MR imaging grading systems. Results were from the MR imaging data assessed by a radiol- ogist with .20years of experience in MR imaging of NPC (ob- server 1). The grading systems were tested by 2 further observers, a researcher with 5 years of experience in MR imaging of NPC 517 Table 3: Diagnostic performance of MR imaging after grouping of grades into benign or malignant according to grades in the cur- rent, modified, and plain scan grading systems MR Imaging Grade Grouping to Indicate NPC FP FN TP TN Sen % Spec % PPV % NPV % Accuracy % Grades in the current system Primary tumor 3, 4a 140 1 382 243 99.74 63.45 73.18 99.59 81.59 Primary tumor only 4 8 48 335 375 87.47 97.91 97.67 88.65 92.69 Grades in the modiﬁed system Primary tumor 3, 4, 5 140 1 382 243 99.74 63.45 73.18 99.59 81.59 Primary tumor 4, 5 31 14 369 352 96.34 91.91 92.25 96.17 94.13 Primary tumor 4, 5 þ node 5a 40 8 375 343 97.91 89.56 90.36 97.72 93.73 Grades in the plain scan system Primary tumor 4, 5 45 17 366 338 95.56 88.25 89.05 95.21 91.91 Primary tumor 4, 5 þ node 5a 52 9 374 331 97.65 86.42 87.79 97.35 92.04 Note:—FP indicates false-positive; FN, false-negative; TP, true-positive; TN, true-negative; Sen, sensitivity; Spec, speciﬁcity; PPV, positive predictive value; NPV, negative predictive value. a Combination of grades in the respective current, modiﬁed, and plain scan systems in Table 2. 518 King Mar 2020 www.ajnr.org RESULTS Table 4: Differences in the diagnostic performances of the current, modified, and plain scan grading systems Grading Systems to Indicate NPC Sens Spec Combined Sens and Spec Youden Index v2 ð2Þ(Signiﬁcance, .3.84) v2 ð2Þ(Signiﬁcance, .5.99) Current system, primary tumor (3, 4)a vs Modiﬁed system using only primary tumor (4, 5)b 13a 109b 122b a ¼ 72.8%; b ¼ 88.4% Current system, primary tumor (3, 4)a vs modiﬁed system, primary tumor (4, 5) þ node (5)c 7a 21.3c 28.3c a ¼ 72.8%; c ¼ 88.1% Modiﬁed system using only primary tumor (4, 5)d vs modiﬁed system, primary tumor (4, 5) þ node (5)c 6c 9d 15d c ¼ 88.4%; d ¼ 88.1% Modiﬁed system, primary tumor (4, 5) þ node (5)c vs plain scan system, primary tumor (4, 5) þ node (5)e 1 5.14c 6.14c c ¼ 88.1%; e ¼ 85.2% Note:—Superscript refers to the grading system with the best performance; NPC indicates nasopharyngeal carcinoma; sens, sensitivity; spec, speciﬁcity. a Current system, primary tumor (3, 4). b Modiﬁed system using only primary tumor (4, 5). c Modiﬁed system, primary tumor (4, 5) þ node (5). d Modiﬁed system using only primary tumor (4, 5). e Plain-scan system primary tumor (4 5) þ node (5) Table 4: Differences in the diagnostic performances of the current, modified, and plain scan g rences in the diagnostic performances of the current, modified, and plain scan grading systems MR imaging grading system that would be advantageous in screening programs. the primary tumor criteria alone. However, nodal assessment also reduced the specificity because of an overlap in the size of enlarged reactive nodes and metastatic nodes. Most interesting, false-positive results for malignancy occurred in the retrophar- yngeal group despite using the 6-mm13 rather than 5-mm17 threshold for metastatic node diagnosis. The decrease in specific- ity outweighed the increase in sensitivity, but the difference in overall performance was small, so we believe that nodal assess- ment still has a role because it detects primary cancers that would otherwise be missed by MR imaging. Current versus Modified Grading Systems for Detection of the Primary Tumor The grading system was modified primarily to improve the speci- ficity for primary tumor detection. The current grading system had a low specificity for NPC detection (63.5%) because of false- positive results from cases of benign hyperplasia with asymmetry (grade 3). Using grade 4 only in the current grading system (focal mass or extension beyond the nasopharynx) to indicate NPC greatly improved the specificity to 97.9%, but this was at the expense of sensitivity, which decreased from 99.7% to 87.5%. Modified versus Plain Scan Grading Systems for NPC Detection As expected, the modified full-protocol grading system, which includes T2- and T1-weighted images before and after contrast enhancement, yielded better overall NPC diagnostic performance than the plain scan grading system. The plain scan grading system, nevertheless, performed well, with high sensitivity and specificity (97.7% and 86.4%, respectively) and high negative and positive predictive values (97.3% and 87.8%, respectively). Contrast-enhanced images detected only 1 additional primary cancer and, surprisingly, no additional spread beyond the nasopharynx; this finding is encouraging because it is an important MR imaging indicator of malig- nancy. Contrast-enhanced images did not detect any addi- tional metastatic nodes. In a screening setting, avoidance of intravenous contrast would enable more rapid, less expensive scans and eliminate the need for intravenous MR imaging contrast agents in healthy patients with false-positive blood tests. The modified grading system yielded a statistically significant better overall performance and better balance between specificity and sensitivity (91.9% and 96.3%, respectively), producing high positive and negative predictive values (90.4% and 97.7%, respec- tively). This result was achieved mainly by the subdivision of asymmetric diffuse wall thickening into asymmetry without focal expansion, which is less indicative of NPC (new indeterminate grade 3), and into asymmetry with focal expansion, which indi- cates a higher risk of NPC (new grade 4). Moreover, a focal mass or extension beyond the nasopharynx was elevated to grade 5, thus expanding the 4-grade system to 5 grades. The refined criteria also incorporated absent contrast-enhancing septa8 and external distortion of contrast-enhancing septa in the adenoid into the new grades 4–5 to indicate suspi- cious or probable NPC. AJNR Am J Neuroradiol 41:515–21 Mar 2020 www.ajnr.org RESULTS The results for each grade using the current, modified, and plain scan MR imaging systems are shown in Table 2. Metastatic nodes were present in 62.6% (154/246) of patients with stage T1 pri- mary tumors confined to the nasopharynx. All 6 cases with stage T1 NPC detected by nodal assessment alone had abnormal nodes in the upper internal jugular chain, whereas 7/9 patients with be- nign hyperplasia had abnormal nodes in the retropharyngeal group only. The percentage of patients with NPC in grades 1–2, 518 King Mar 2020 www.ajnr.org Table 4: Differences in the diagnostic performances of the current, modified, and plain scan grading systems Grading Systems to Indicate NPC Sens Spec Combined Sens and Spec Youden Index v2 ð2Þ(Signiﬁcance, .3.84) v2 ð2Þ(Signiﬁcance, .5.99) Current system, primary tumor (3, 4)a vs Modiﬁed system using only primary tumor (4, 5)b 13a 109b 122b a ¼ 72.8%; b ¼ 88.4% Current system, primary tumor (3, 4)a vs modiﬁed system, primary tumor (4, 5) þ node (5)c 7a 21.3c 28.3c a ¼ 72.8%; c ¼ 88.1% Modiﬁed system using only primary tumor (4, 5)d vs modiﬁed system, primary tumor (4, 5) þ node (5)c 6c 9d 15d c ¼ 88.4%; d ¼ 88.1% Modiﬁed system, primary tumor (4, 5) þ node (5)c vs plain scan system, primary tumor (4, 5) þ node (5)e 1 5.14c 6.14c c ¼ 88.1%; e ¼ 85.2% Note:—Superscript refers to the grading system with the best performance; NPC indicates nasopharyngeal carcinoma; sens, sensitivity; spec, speciﬁcity. a Current system, primary tumor (3, 4). b Modiﬁed system using only primary tumor (4, 5). c Modiﬁed system, primary tumor (4, 5) þ node (5). d Modiﬁed system using only primary tumor (4, 5). e Plain-scan system, primary tumor (4, 5) þ node (5). Observer Results This study aimed to produce a simple MR imaging grading sys- tem based on conventional sequences for a quick assessment by radiologists with differing levels of experience. Therefore, we did not include subtle abnormalities of malignancy such as loss of the white line sign8 or functional sequences such as diffusion- weighted imaging,18 which require quantification and use of thresholds that are not easily implemented across centers. Using our proposed modified and plain scan grading systems yielded substantial agreement between the most and the least experienced observers when assessing grades 1–2 versus grade 3 versus grades 4–5 on contrast MR imaging. All other agreement among the 3 observers was almost perfect. Therefore, we believe that these new MR imaging grading systems can be used by specialists and generalists. Most interesting, the number of false-positive cases increased with experience, which lowered the overall accuracy. Disclosures: Rossa W.K. Chiu—RELATED: Grant: Research Grants Council of Hong Kong Special Administrative Region Government Theme-Based Research Grant (T12-401/16-W); UNRELATED: Consultancy: GRAIL; Grants/Grants Pending: contract research agreement from GRAIL; Patents (Planned, Pending or Issued): both pending and issued patents on cancer diagnostics; Royalties: from patents on cancer diagnostics; Stock/Stock Options: GRAIL, Take2 Healthcare; Other: founder of Take2 Healthcare. K.C. Allen Chan—RELATED: Grant: GRAIL, Cirina, Comments: This work is partially supported by GRAIL/Cirina; Consulting Fee or Honorarium: GRAIL, Comments: I am a consultant to GRAIL; UNRELATED: Board Membership: Take2 Healthcare, DRA, Comments: I am a director of Take2 Healthcare and DRA; Consultancy: GRAIL, Comments: I am a consultant of GRAIL; Grants/Grants Pending: GRAIL/Cirina, Comments: My institution received funding support from GRAIL/Cirina; Patents (Planned, Pending or Issued): patents on molecular diagnostics; Royalties: GRAIL, Sequenom, Illumine, Take2, Xcelom; Stock/ Stock Options: GRAIL, DRA, Take2 Healthcare, Comments: I hold equities. Wai Kei Jacky Lam—UNRELATED: Patents (Planned, Pending or Issued): enhancement of can- cer screening using cell-free viral nucleic acids (pending); Stock/Stock Options: GRAIL. Edwin P. Hui—UNRELATED: Consultancy: Merck Sharp & Dohme; Payment for Lectures Including Service on Speakers Bureaus: Merck Serono, Merck Sharp & Dohme. Brigette Ma—UNRELATED: Grants/Grants Pending: Novartis, Boerhinger Ingelheim, Comments: research grant; Payment for Lectures Including Service on Speakers Bureaus: Merck Sharp & Dohme, Bristol-Myers Squibb.* Anthony Chan— UNRELATED: Consultancy: Merck Serono, Merck Sharp & Dohme, Cullinan Management Inc; Employment: The Chinese University of Hong Kong; Grants/grants pending: Boehringer Ingelheim, Bristol Myers Squibb, Amgen, Pﬁzer, Eli Lilly, Merck Serono, Merck Sharp & Dohme. Modified Grading System for NPC Detection with and without Metastatic Node Assessment We suggest that for plain scan MR imaging graded 1 or 2, the risk of NPC is low (0.44%) and the patient can be reassured. Intravenous contrast is of limited benefit and can be withheld, especially in the absence of abnormal findings on endoscopy. On the other hand, for plain scan MR imaging graded 4 or 5, the risk of NPC is high (87.80%) and intravenous contrast is suggested irrespective of the endoscopy findings. For plain scan MR imag- ing graded 3, the risk of NPC is intermediate (6.96%) and the role The incidence of regional spread to retropharyngeal and upper internal jugular chain nodes covered on the short MR imaging protocol was high for stage T1 tumors confined to the nasophar- ynx (62.6%) and is consistent with the reported incidence for early-stage NPC (71.1%).16 The inclusion of nodal assessment increased the sensitivity of the modified grading system by detect- ing 6 additional cases of NPC that would have been missed using 519 of intravenous contrast is unclear, though results suggest that contrast is more beneficial when the MR imaging is graded by a less experienced assessor or to increase the confidence that a sym- metric enlarged adenoid is due to benign hyperplasia (striped appearance with alternating septal and lymphoid columns). for NPC. Patients with a low-risk MR imaging grade can be spared an injection of intravenous contrast. We hope the grading system will help disseminate knowledge to other centers and act as the foundation for building a consensus on future guidelines for a Nasopharynx Imaging Reporting and Data System. Finally, the plain scan alone exhibited potential usefulness in future screening programs. Observer Results Yuk Ming Dennis Lo—RELATED: Grant: Kadoorie Charitable Foundation, GRAIL Inc, Li Ka Shing Foundation, Hong Kong Research Grants Council, Comments: Hong Kong Research Grants Council Theme-Based Research Grant (T12-401/16-W); Consulting fees or honorarium: GRAIL Inc, Decheng Capital; Other: Take2 Holdings Limited, GRAIL Inc, Comments: Take2 Holdings Limited - Co-founder, Shareholder, Chairman of the Board, GRAIL Inc. - Scientiﬁc Co-founder, shareholder, member of Scientiﬁc Advisory Board; UNRELATED: Board membership: DRA Limited, Comments: Chairman of the Board; Patents (Planned, Pending or Issued): Illumina, Sequenom, Xcelom Limited, DRA Limited, Comments: Licensed patents or patent applications in the area of non-invasive prenatal testing; Royalties: Illumina, Sequenom, Xcelom Limited, DRA Limited, Comments: Royalties for licensed patents or patent applications in the area of non-invasive prenatal test- ing; Stock/stock options: DRA Limited; Other: KingMed Future, Comments: Service agreement in the area of non-invasive prenatal testing. Money paid to the institution. Limitations The use of MR imaging in NPC detection is still in its infancy, so our results are from a single-center setting and include cases reported previously. Therefore, to ensure generalizability, the data should be validated at outside institutions, and we hope our proposed systems will be tested and developed by researchers in this field. Moreover, future studies should explore the potential use of only a non-contrast-enhanced MR imaging protocol and the plain scan grading system in screening settings. We envisage future refinements of the MR imaging grading systems, and our proposal regarding the use of intravenous contrast may evolve as advances in the circulating Epstein-Barr virus DNA evaluations reduce the number of healthy subjects referred for further investi- gation after a false-positive blood test.19 Finally, the grading sys- tems are proposed for use only in patients with suspected NPC who have a dedicated nasopharyngeal MR imaging and not for use in a general population. REFERENCES 1. Lee AW, Ng WT, Chan LL, et al. Evolution of treatment for naso- pharyngeal cancer: success and setback in the intensity-modu- lated radiotherapy era. Radiother Oncol 2014;110:377–84 CrossRef Medline CONCLUSIONS We refined our current MR imaging grading system for NPC detection, with the intent to meet the challenge of applying MR imaging not only in the clinical setting but also in a screening setting involving greater numbers of patients with early-stage pri- mary cancers that must be discriminated from benign hyperpla- sia. We added focal expansion in diffuse asymmetric wall thickening, absent or external distortion of contrast-enhancing adenoidal septa, and the presence of metastatic nodes as indicators of malignancy. The modified MR imaging grading system improved the overall diagnostic performance of MR imaging when compared with the current MR imaging grading system and produced sensitivity and specificity rates of 96.3% and 91.9%, respectively. The modified MR imaging grading system was used by specialists and generalists to classify patients into low (grades 1, 2), indeterminate (grade 3), and high-risk (grades 4, 5) categories 2. Chan KC, Woo JK, King A, et al. Analysis of plasma Epstein-Barr virus DNA to screen for nasopharyngeal cancer. N Engl J Med 2017;377:513–22 CrossRef Medline 3. King AD, Vlantis AC, Bhatia KS, et al. Primary nasopharyngeal car- cinoma: diagnostic accuracy of MR imaging versus that of endos- copy and endoscopic biopsy. Radiology 2011;258:531–37 CrossRef Medline 3. King AD, Vlantis AC, Bhatia KS, et al. Primary nasopharyngeal car- cinoma: diagnostic accuracy of MR imaging versus that of endos- copy and endoscopic biopsy. Radiology 2011;258:531–37 CrossRef Medline 4. King AD, Woo JK, Ai QY, et al. Complementary roles of MRI and endoscopic examination in the early detection of nasopharyngeal carcinoma. Ann Oncol 2019;30:977–82 CrossRef Medline 4. King AD, Woo JK, Ai QY, et al. Complementary roles of MRI and endoscopic examination in the early detection of nasopharyngeal carcinoma. Ann Oncol 2019;30:977–82 CrossRef Medline 5. Shayah A, Wickstone L, Kershaw E, et al. The role of cross-sectional imaging in suspected nasopharyngeal carcinoma. Ann R Coll Surg Engl 2019;101:325–27 CrossRef Medline 5. Shayah A, Wickstone L, Kershaw E, et al. The role of cross-sectional imaging in suspected nasopharyngeal carcinoma. Ann R Coll Surg Engl 2019;101:325–27 CrossRef Medline 6. King AD, Vlantis AC, Yuen TW, et al. Detection of nasopharyn- geal carcinoma by MR imaging: diagnostic accuracy of MRI compared with endoscopy and endoscopic biopsy based on 520 520 King Mar 2020 www.ajnr.org long-term follow-up. AJNR Am J Neuroradiol 2015;36:2380–85 CrossRef Medline 13. Zhang G, Liu L, Wei W, et al. AJNR Am J Neuroradiol 41:515–21 Mar 2020 www.ajnr.org CONCLUSIONS Radiologic criteria of retropharyngeal lymph node metastasis in nasopharyngeal carcinoma treated with radiation therapy. Radiology 2010;255:605–12 CrossRef Medline 7. King AD, Vlantis AC, Tsang RK, et al. Magnetic resonance imaging for the detection of nasopharyngeal carcinoma. AJNR Am J Neuroradiol 2006;27:1288–91 CrossRef Medline gy 14. Hawass N. Comparing the sensitivities and specificities of two diagnostic procedures performed on the same group of patients. Br J Radiol 1997;70:360–66 CrossRef Medline 8. King AD, Wong LY, Law BK, et al. MR imaging criteria for the detection of nasopharyngeal carcinoma: discrimination of early- stage primary tumors from benign hyperplasia. AJNR Am J Neuroradiol 2018;39:515–23 CrossRef Medline 15. Kundel HL, Polansky M. Measurement of observer agreement. Radiology 2003;228:303–08 CrossRef Medline 16. Guo R, Tang LL, Mao YP, et al. Proposed modifications and incor- poration of plasma Epstein-Barr virus DNA improve the TNM staging system for Epstein-Barr virus-related nasopharyngeal car- cinoma. Cancer 2019;125:79–89 CrossRef Medline 9. Wang ML, Wei XE, Yu MM, et al. Value of contrast-enhanced MRI in the differentiation between nasopharyngeal lymphoid hyperplasia and T1 stage nasopharyngeal carcinoma. Radiol Med 2017;122:743–51 CrossRef Medline 17. King AD, Ahuja AT, Leung SF, et al. Neck node metastases from na- sopharyngeal carcinoma: MR imaging of patterns of disease. Head Neck 2000;22:275–81 CrossRef Medline 10. Li K, Lin GZ, Shen JC, et al. Time trends of nasopharyngeal carci- noma in urban Guangzhou over a 12-year period (2000–2011): declines in both incidence and mortality. Asian Pac J Cancer Prev 2014;15:9899–903 CrossRef Medline 18. Ai QY, King AD, Chan JS, et al. Distinguishing early-stage nasopha- ryngeal carcinoma from benign hyperplasia using intravoxel incoher- ent motion diffusion-weighted MRI. Eur Radiol 2019;29:5627–34 CrossRef Medline 11. Amin MB, Edge SB, Greene FL, et al, eds. AJCC Cancer Staging Manual. 8th ed. New York: Springer-Verlag; 2017 19. Lam WK, Jiang P, Chan KCA, et al. Sequencing-based counting and size profiling of plasma Epstein–Barr virus DNA enhance popula- tion screening of nasopharyngeal carcinoma. Proc Natl Acad Sci U S A 2018;115:E5115–24 CrossRef Medline 12. van den Brekel MW, Stel HV, Castelijns JA, et al. Cervical lymph node metastasis: assessment of radiologic criteria. Radiology 1990; 177:379–84 CrossRef Medline 521
https://openalex.org/W2037030152	https://europepmc.org/articles/pmc4102246?pdf=render	English	null	Safety and efficacy of protease inhibitor based combination therapy in a single-center “real-life” cohort of 110 patients with chronic hepatitis C genotype 1 infection	BMC gastroenterology	2,014	cc-by	7,581	* Correspondence: j.schulze-zur-wiesch@uke.de †Equal contributors 1I. Medizinische Klinik und Poliklinik, Universitätsklinikum Hamburg-Eppendorf, Martinistr. 52, Hamburg 20246, Germany Full list of author information is available at the end of the article © 2014 Wehmeyer et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Safety and efficacy of protease inhibitor based combination therapy in a single-center “real-life” cohort of 110 patients with chronic hepatitis C genotype 1 infection Safety and efficacy of protease inhibitor based combination therapy in a single-center “real-life” cohort of 110 patients with chronic hepatitis C genotype 1 infection Malte H Wehmeyer1, Friederike Eißing1, Sabine Jordan1, Claudia Röder1, Annette Hennigs1, Olaf Degen2, Anja Hüfner2, Sandra Hertling2, Stefan Schmiedel1, Martina Sterneck1, Jan van Lunzen2, Ansgar W Lohse1, Julian Schulze zur Wiesch1† and Stefan Lüth1† Abstract Background: The combination of boceprevir or telaprevir with peginterferon-alfa and ribavirin for the treatment of patients infected with HCV genotype 1 has led to significantly increased rates of sustained virological response (SVR) in phase III trials. There is only limited data regarding the safety and efficacy in a “real-life” cohort. Methods: We analyzed a cohort of 110 unselected HCV patients who started triple therapy from September 2011 to February 2013 by chart review with focus on the individual course of treatment, complications and outcome. We excluded 8 patients from analysis because of HIV-coinfection (N = 6) or status post liver transplant (N = 2). Importantly, 41 patients displayed F3 or F4 fibrosis, 10 patients had a history of treatment with protease/polymerase inhibitors and 15 patients were prior partial- or null-responder. alyzed a cohort of 110 unselected HCV patients who started triple therapy from September 2011 by chart review with focus on the individual course of treatment, complications and outcome. We Results: SVR12 was achieved in 62 of the 102 patients (60.8%). A high rate of serious adverse events (N = 30) was observed in 22 patients including 2 fatalities in cirrhotic diabetes patients. Age >50 years, liver cirrhosis, bilirubin >1.1 mg/ dl (P < 0.01, each), platelets <100,000/μl (P = 0.01), ASAT >100 U/l (P = 0.03) and albumin ≤35 g/l (P = 0.04) at baseline were associated with occurence of a SAE. Conclusions: The frequency of SVR in a “real-life” treatment setting is slightly lower as compared to the results of the phase III trials for telaprevir or boceprevir. Importantly, we observed a high frequency of SAE in triple therapy, especially in patients with liver cirrhosis. Keywords: Boceprevir, Serious adverse events, SAE, Side effects, Sustained virological response, SVR, Telaprevir genotype 1 infection in the US, Canada and many European countries in phase III trials with sustained virological response (SVR) rates of 67 to 75% in treatment naive patients [3-5]. Even higher SVR rates have been achieved in patients with history of relapse following a previous therapy [6,7]. Interestingly, first “real-life” efficacy data revealed a significantly lower frequency of SVR [8]. On the other hand, treatment with a PI is associated with high rates of side-effects, such as fatigue, anemia and high- grade neutropenia [3-7]. Skin reactions and gastrointes- tinal disorders were frequently observed side-effects of TPR in the phase II and III trials [4,5,7]. Correspondence: j.schulze-zur-wiesch@uke.de † Wehmeyer et al. BMC Gastroenterology 2014, 14:87 http://www.biomedcentral.com/1471-230X/14/87 Wehmeyer et al. BMC Gastroenterology 2014, 14:87 http://www.biomedcentral.com/1471-230X/14/87 Background An estimated 170 million people are chronically infected with the hepatitis C virus (HCV) [1] and have an elevated risk for liver-related mortality [2]. Recently, introduction of the serine protease inhibitors (PI) boceprevir (BOC) and telaprevir (TPR) which are used in combination with peginterferon-alfa 2a or 2b (pegIFN) and ribavirin (RBV) has increased cure rates of patients with chronic HCV * Correspondence: j.schulze-zur-wiesch@uke.de †Equal contributors 1I. Medizinische Klinik und Poliklinik, Universitätsklinikum Hamburg-Eppendorf, Martinistr. 52, Hamburg 20246, Germany Full list of author information is available at the end of the article Wehmeyer et al. BMC Gastroenterology 2014, 14:87 http://www.biomedcentral.com/1471-230X/14/87 Page 2 of 10 Page 2 of 10 However, the overall safety-profile of the PIs was acceptable in phase III trials [3-7], which included highly selected patients. Most recently, preliminary “real-life” data covering the first 12 to 16 weeks of therapy revealed considerably increased risk for severe and in some instances even lethal complications of PI-based treatment in cirrhotic patients [9,10]. The most common cause of death was sepsis, with staphylococcus being the most frequent causative organism [10]. The same analysis was conducted for patients suffering from predefined SAEs. Thresholds for continuous vari- ables were defined according to the results from the CUPIC cohort [10] or by clinical judgement. Variables which reached P < 0.1 in univariate analysis were entered in a backward step logistic regression model. The same analysis was conducted for patients suffering from predefined SAEs. Thresholds for continuous vari- ables were defined according to the results from the CUPIC cohort [10] or by clinical judgement. Variables which reached P < 0.1 in univariate analysis were entered in a backward step logistic regression model. The respective grading of laboratory events, adverse event definitions and virological definitions are de- scribed in the supplementary materials (text document, Additional file 1). All analyses were performed using SPSS Version 20. The figures were created using GraphPad Prism 4. The study was approved at the local ethics board (Ethik-Kommission der Ärztekammer Hamburg). Despite the approval of alternative direct acting antivi- rals (DAA) in Northern America and Europe, TPR and BOC have just arrived in many parts of the world. Background Given the increased likelihood of serious adverse events (SAE) provided by the preliminary reports on “real-life” data [9,10], we examined the outcomes and complications of triple therapy throughout the treatment course within our local “real-life”, difficult-to-treat cohort, which includes a number of patients with comorbidities, cirrhosis or pre- vious DAA experience. Study population and chart review W l d li i l d l b We analyzed clinical and laboratory data of 110 unselected patients who were chronically infected with HCV genotype 1 and in whom treatment with pegIFN, RBV and TPR or BOC was initiated from September 2011 to February 2013 at the viral hepatitis clinics of the University Medical Center Hamburg-Eppendorf, which is representative of a tertiary care referral center for antiviral HCV therapy in Germany. Liver transplant recipients (N = 2) and patients coinfected with human immunodeficiency virus were excluded (N = 6). All patients received an abdominal ultrasound prior to the start of therapy. The grade of liver fibrosis was mea- sured in the majority of patients by transient elastogra- phy (Fibroscan, Echosens, France) [11] or liver biopsy before the initiation of treatment (Table 1). Patient charts were analyzed regarding demographics, clinical data, HCV genotype, interleukin 28B (IL28B) rs12979860 polymorphism, as well as laboratory values and HCV viral load at different time points. The lower detection limit of the HCV PCR was 15 IU/ml (COBAS TaqMan HCV Qualitative, v2.0, Roche). The Child-Pugh score and the MELD score were assessed in all cirrhotic patients at baseline using the established formula [12,13]. In 15 patients virologic failure occured during PI treat- ment (Figure 1). The PI was discontinued early in 25 of the remaining 87 patients (28.7%) for various reasons (e.g. patient’s wish, side-effects, provider’s individual decision; Figure 2). After PI withdrawal, 5 additional patients expe- rienced a viral breakthrough on dual therapy (Figure 1) and 24 of the remaining 82 patients (29.3%) discontinued pegIFN and RBV prematurely as compared to the Characterization of the study population We describe here the detailed clinical course and treatment outcome of 102 patients who started triple therapy from September 2011 to February 2013 at our university viral hepatitis clinics. Baseline characteristics of all patients are summarized in Table 1. Fibroscan or liver biopsy was performed in 92 patients (90.2%) and diagnosis of bridging fibrosis (F3) or cirrhosis (F4) was established in 41 patients (40.2%). The 10 remaining patients, who did not receive a transient elastography or a liver biopsy prior to the HCV therapy, did not have any laboratory or sonographical evidence for high grade fibrosis or cirrhosis, respectively. IL28B polymorphism was assessed in 70 patients (68.6%), of whom 16 individuals displayed the favorable C/C IL28 haplotype (22.9%). Fifty five patients (53.9%) were HCV treatment experienced (25 patients with a prior relapse, 15 patients with prior partial- or null-response), including 10 patients who previously received a DAA based therapy in clinical trials. The exclusion criteria for registration trials for TPR or BOC [3-7] were met by 65 patients (64%, e.g. history of hepatocellular carcinoma, history of stem cell transplantation, renal dialysis, Crohn’s disease, thalassaemia major, autoimmune hepatitis and primary biliary cirrhosis). Twenty patients (19.6%) suffered from concomittant psychiatric disorders such as major depression (N = 18), anxiety disorder (N = 2), borderline psychosis (N = 1), post traumatic stress disorder (N = 1) and anorexia (N = 1) and were treated with psychotropics. Additionally, 10 patients were DAA experienced as participants of several phase II and III trials performed at our center. Statistical analysis Table 1 Baseline characteristics All patients Telaprevir Boceprevir (N = 102) (N = 65) (N = 37) N (%); median (range) N (%); median (range) N (%); median (range Male sex 63 (62%) 43 (66%) 20 (54%) Exclusion criteria for appropriate phase III trials 65 (64%) 42 (65%) 23 (62%) Treatment naïve 47 (46%) 23 (35%) 24 (65%) Treatment experienced 55 (54%) 42 (65%) 13 (35%) Relapse 25 (25%) 18 (28%) 7 (19%) Null/partial response 15 (15%) 13 (20%) 2 (5%) Breakthrough 6 (6%) 5 (8%) 1 (3%) Discontinuation§ 5 (5%) 3 (5%) 2 (5%) Unknown outcome 4 (4%) 3 (5%) 1 (3%) DAA experienced 10 (10%) 9 (14%) 1 (3%) RVR 38 (37%) 21 (32%) 17 (46%) Genotype Genotype 1a 39 (38%) 20 (31%) 19 (51%) Genotype 1b 53 (52%) 37 (57%) 16 (43%) No subtype provided 9 (9%) 7 (11%) 2 (5%) Unknown 1 (1%) 1 (2%) 0 IL28B (N = 70) C/C 16 (24%) 10 (20%) 6 (29%) C/T 42 (60%) 32 (65%) 10 (48%) T/T 12 (17%) 7 (14%) 5 (24%) Stage of fibrosis (N = 92) No or mild fibrosis (F0-F2) 51 (55%) 32 (49%) 19 (59%) Bridging fibrosis (F3) 12 (13%) 6 (10%) 6 (19%) Liver cirrhosis 29 (32%) 22 (37%) 7 (22%) Hemoglobin [g/dl] 14.7 (10.3-18.8) 14.7 (11.1-18.8) 14.6 (10.3-17.6) Leukocytes [x10^9/l] 6.1 (2.7-13.1) 5.9 (2.7-13.1) 6.3 (3.9-12.4) Platelets [x10^9/l] 188 (48–377) 180 (48–377) 203 (67–338) ASAT [U/l] 50.5 (16–328) 52 (16–328) 44 (19–156) ALAT [U/l] 75.5 (16–271) 87 (16–271) 72 (19–227) γGT [U/l] 66.5 (25–1274) 67 (25–1274) 62 (25–217) Bilirubin [mg/dl] 0.5 (0.2-2.2) 0.6 (0.3-2.2) 0.5 (0.2-1.4) Albumin [g/l] 40 (25–50) 40 (25–47) 39 (30–50) Prothrombin time [INR] 1.00 (0.90-3.29) 1.02 (0.90-3.29) 1.00 (0.92-2.33) Creatinine [mg/dl] 0.8 (0.5-5.8) 0.8 (0.5-1.1) 0.8 (0.5-5.8) Viral load undetectable at EOT 82 (80%) 51 (78%) 31 (84%) SVR12 62 (61%) 40 (62%) 22 (59%) [N = number; DAA = direct acting antivirals; RVR = rapid virological response; IL28B = interleukin-28B polymorphism; EOT = end of treatment; SVR12 = sustained virological response 12 weeks after last ribavirin dose; § = cessation of prior therapy due to side-effects]. [N = number; DAA = direct acting antivirals; RVR = rapid virological response; IL28B = interleukin-28B polymorphism; EOT = end of treatment; SVR12 = sustained virological response 12 weeks after last ribavirin dose; § = cessation of prior therapy due to side-effects]. Statistical analysis Variables of patients with SVR12 were compared with those of patients experiencing a treatment failure by uni- variate analysis using Fisher’s exact text, t-test (for vari- ables with assumed Gaussian’ distribution, e.g. age) and Mann–Whitney-U-Test (for variables without assumed Gaussian’ distribution, e.g. laboratory values), respectively. Wehmeyer et al. BMC Gastroenterology 2014, 14:87 http://www.biomedcentral.com/1471-230X/14/87 Page 3 of 10 Page 3 of 10 Table 1 Baseline characteristics guidelines for PI based therapy [14,15] (Figure 2). Seventeen patients treated with TPR (26.2%) and all patients treated with BOC received a lead-in phase with pegIFN and RBV prior to triple therapy (mean duration 4.7 weeks (TPR, standard deviation (SD) = 1.2) and 5.3 weeks (BOC, SD = 5.2), Figure 2). The rationale for starting therapy with a lead-in phase in TPR patients was to avoid the administra tion of a PI after a possible RVR under pegIFN/RBV [14]. Statistical analysis standard deviation (SD) = 1.2) and 5.3 weeks (BOC, SD = 5.2), Figure 2). The rationale for starting therapy with a lead-in phase in TPR patients was to avoid the administra- tion of a PI after a possible RVR under pegIFN/RBV [14]. guidelines for PI based therapy [14,15] (Figure 2). Seventeen patients treated with TPR (26.2%) and all patients treated with BOC received a lead-in phase with pegIFN and RBV prior to triple therapy (mean duration 4.7 weeks (TPR, Wehmeyer et al. BMC Gastroenterology 2014, 14:87 http://www.biomedcentral.com/1471-230X/14/87 Page 4 of 10 Figure 1 Efficacy of triple therapy. SVR rates for different subgroups are displayed in A and B, a characterization of treatment failures for the total study population [C] and for BOC versus TPR [D] are given, too. Patients who died after discontinuation of therapy (N = 3) and patients who were lost to follow up (N = 8) were not regarded as being at risk for relapse in C and D. [SVR = sustained virological response; BOC = boceprevir; TPR = telaprevir; IL28B = interleukin 28B polymorphism; BT = breakthrough; PI = protease inhibitor]. Figure 1 Efficacy of triple therapy. SVR rates for different subgroups are displayed in A and B, a characterization of treatment failures for the total study population [C] and for BOC versus TPR [D] are given, too. Patients who died after discontinuation of therapy (N = 3) and patients who were lost to follow up (N = 8) were not regarded as being at risk for relapse in C and D. [SVR = sustained virological response; BOC = boceprevir; TPR = telaprevir; IL28B = interleukin 28B polymorphism; BT = breakthrough; PI = protease inhibitor]. Figure 1 Efficacy of triple therapy. SVR rates for different subgroups are displayed in A and B, a characterization of treatment failures for the total study population [C] and for BOC versus TPR [D] are given, too. Patients who died after discontinuation of therapy (N = 3) and patients who were lost to follow up (N = 8) were not regarded as being at risk for relapse in C and D. [SVR = sustained virological response; BOC = boceprevir; TPR = telaprevir; IL28B = interleukin 28B polymorphism; BT = breakthrough; PI = protease inhibitor]. d in A and B a characterization of treatment failures for the Figure 1 Efficacy of triple therapy. Statistical analysis SVR rates for different subgroups are displayed in A and B, a characterization of treatment failures for the total study population [C] and for BOC versus TPR [D] are given, too. Patients who died after discontinuation of therapy (N = 3) and patients who were lost to follow up (N = 8) were not regarded as being at risk for relapse in C and D. [SVR = sustained virological response; BOC = boceprevir; TPR = telaprevir; IL28B = interleukin 28B polymorphism; BT = breakthrough; PI = protease inhibitor]. As of now (April 2014), 7 patients (6.9%) are lost to follow-up and are therefore regarded as treatment fail- ures in this analysis. patients (see figure in Additional file 2, which includes an overview of the treatment regimen and clinical or virological outcome of the trial and the course of ther- apy with TPR or BOC for each DAA experienced patient, as well as the course of therapy of patients with special comorbidities. Further information on the clinical history of the DAA experienced patients are shown in the Additional file 1). Efficacy of triple therapy Overall, 62 patients (60.8%) were successfully treated and achieved a SVR12. As expected, prior relapsers displayed the highest SVR rate of 72% and prior partial- or null-responders were less likely to achieve SVR12 (46.7%). SVR12 was achieved by 57.4% of treatment naive patients (including cirrhotics) and by 51.2% of pa- tients with bridging fibrosis or liver cirrhosis (Figure 1). Finally, SVR 12 was achieved by 5/10 DAA experienced The HCV subtype was not significantly associated with SVR12 in our study. However, patients infected with HCV genotype 1a displayed an odds ratio (OR) of 0.4 to achieve SVR12 (95% confidence interval (95%CI) 0.2-1.0, P = 0.06), while patients with HCV genotype 1b Wehmeyer et al. BMC Gastroenterology 2014, 14:87 http://www.biomedcentral.com/1471-230X/14/87 Page 5 of 10 Page 5 of 10 Figure 2 Individualized courses of treatment in our “real-life” cohort. Patients are grouped according to the respective guidelines [14,15]. Each symbol represents one patient. Patients with partial response or breakthrough under BOC/TPR are not depicted in A and B, patients with partial response or breakthrough at any time are not depicted in C and D. All BOC patients and TPR patients who received a dual lead-in phase prior to BOC/TPR are included in E. [BOC = boceprevir; TPR = telaprevir; PR = pegylated interferon and ribavirin; SVR = sustained virological response; EOT = patient concluded therapy, but is short of 12 weeks after last ribavirin dose]. Figure 2 Individualized courses of treatment in our “real-life” cohort. Patients are grouped according to the respective guidelines [14,15]. Each symbol represents one patient. Patients with partial response or breakthrough under BOC/TPR are not depicted in A and B, patients with partial response or breakthrough at any time are not depicted in C and D. All BOC patients and TPR patients who received a dual lead-in phase prior to BOC/TPR are included in E. [BOC = boceprevir; TPR = telaprevir; PR = pegylated interferon and ribavirin; SVR = sustained virological response; EOT = patient concluded therapy, but is short of 12 weeks after last ribavirin dose]. Figure 2 Individualized courses of treatment in our “real-life” cohort. Patients are grouped according to the respective guidelines [14,15]. Each symbol represents one patient. Patients with partial response or breakthrough under BOC/TPR are not depicted in A and B, patients with partial response or breakthrough at any time are not depicted in C and D. Side-effects and complications of triple therapy Detailed information about all side-effects are shown in Table 2. Severe flu-like symptoms were reported by 44 patients (43.1%) and 43 patients (42.2%) showed gastro- intestinal symptoms. Patients receiving BOC suffered more often from dysgeusia (24.3%) and fatigue (40.5%) than patients receiving TPR (1.5% and 18.5%, P < 0.01 and P = 0.02), while TPR based treatment was associated with a high risk for anorectal dyscomfort (36.9% (TPR) versus 2.7% (BOC), P < 0.01). Thirty serious adverse events (SAE) occured in 22/102 patients (21.6%). Details on the nature and outcome of the SAE are shown in Table 2. Table 2 Side effects and serious adverse events in patients receiving triple therapy Number BOC TPR P-Value (N = 102) (N = 37) (N = 65) Grade 3/4 anemia 13 (12.7%)§ 4 (10.8%) 9 (13.8%) 0.77 Grade 3/4 neutropenia 25 (24.5%) 12 (32.4%) 13 (20%) 0.23 Grade 3/4 thrombopenia 14 (13.7%) 3 (8.1%) 11 (16.9%) 0.25 Flu-like symptoms 44 (43.1%) 20 (54.1%) 24 (36.9%) 0.10 GI disorders 43 (42.2%) 16 (43.2%) 27 (41.5%) 1 Grade 1/2 rash 35 (34.3%) 11 (29.7%) 24 (36.9%) 0.52 Psychiatric disorder 29 (28.4%)§§ 11 (29.7%) 18 (27.7%) 1 Fatigue 27 (26.5%) 15 (40.5%) 12 (18.5%) 0.02 Anorectal dyscomfort 25 (24.5%) 1 (2.7%) 24 (36.9%)§§§ <0.01 Insomnia 23 (22.5%) 7 (18.9%) 16 (24.6%) 0.63 Bronchopulmonal symptoms 20 (19.6%) 9 (24.3%) 11 (16.9%) 0.44 Dysgeusia 10 (9.8%) 9 (24.3%) 1 (1.5%) <0.01 Immunothyreoiditis 4 (3.9%) 0 4 (6.2%) 0.29 Serious adverse event Number Comments/Outcome Grade 4 anemia 2 Both patients received blood transfusions Grade 4 neutropenia 3 1 patient with neutropenic sepsis; all 3 recovered after dose reduction of peginterferon Grade 4 thrombopenia 6 1 patient received a platelet concentrate Grade 3 rash 4 TPR was discontinued early in 1 patient DRESS/SJS None - Neurological symptoms 2 Radial nerve paresis (recovery after physiotherapy) and urine incontinency (ongoing) Localized infections 3 All fully recovered (balanitis, epididymitis, perspiratory gland abscess) Sepsis 4 2 patients recovered; 2 patients died Hepatic decompensation 2 Both recovered, but 1 patient died 6 weeks after discontinuation of treatment Decompensation of autoimmune disorder 1 First manifestation of autoimmune diabetes mellitus. Ileus 2 Fully recovered after surgical intervention (1x) or conservative treatment (1x) Cardiac complications 1 Unstable angina pectoris, full recovery after PTCA with stent implantation 30 SAE occured in 22 individuals. Efficacy of triple therapy All BOC patients and TPR patients who received a dual lead-in phase prior to BOC/TPR are included in E. [BOC = boceprevir; TPR = telaprevir; PR = pegylated interferon and ribavirin; SVR = sustained virological response; EOT = patient concluded therapy, but is short of 12 weeks after last ribavirin dose]. displayed an OR of 2.2 to achieve SVR12 (95% CI 1.0- 5.0; P = 0.07; see table in Additional file 3). Further ana- lysis of our small cohort revealed that neither the grade of fibrosis, nor the IL28B haplotype, treatment experi- ence, diabetes mellitus type 2, psychiatric disorders, the occurence of a RVR or a reduction of the pegIFN or RBV dose were associated with SVR12 in the univariate analysis (see table in Additional file 3). No independent predictors for SVR12 were identified in the multivariate analysis. displayed an OR of 2.2 to achieve SVR12 (95% CI 1.0- 5.0; P = 0.07; see table in Additional file 3). Further ana- lysis of our small cohort revealed that neither the grade of fibrosis, nor the IL28B haplotype, treatment experi- ence, diabetes mellitus type 2, psychiatric disorders, the occurence of a RVR or a reduction of the pegIFN or RBV dose were associated with SVR12 in the univariate analysis (see table in Additional file 3). No independent predictors for SVR12 were identified in the multivariate analysis. We also examined the frequency and possible conse- quences of a shortening of the duration of PI medication from the recommended length [15,16]. This analysis re- vealed that a higher number of patients on BOC treatment (17/37, 45.9%, mean duration of shortening 3.8 weeks, SD = 8.1) compared to patients on TPR treatment (9/65, 13.8%, mean duration of shortening 0.7 weeks, SD = 2.0, P < 0.01) reduced the time of protease therapy. The mean shortening of TPR duration was 0.8 weeks (SD = 2.2) in patients with later SVR12 and 0.9 weeks (SD = 2.1) in patients who later Wehmeyer et al. BMC Gastroenterology 2014, 14:87 http://www.biomedcentral.com/1471-230X/14/87 Page 6 of 10 experienced a viral breakthrough or relapse (P = 0.78, Figure 2B). Patients who experienced virological failure after BOC discontinuation, displayed a mean reduction of BOC medication time of 5.4 weeks (SD = 10.0), com- pared to 4.0 weeks (SD = 8.1) in BOC patients who achieved SVR12 (P = 0.68, Figure 2A). Efficacy of triple therapy Patients who ex- perienced a relapse discontinued pegIFN and RBV 6.5 weeks prematurely (SD = 13.4) compared to patients who achieved SVR12 (0.4 weeks (SD = 10.7), P = 0.03, Figure 2C and D). An early termination of pegIFN/RBV frequently led to treatment failures in patients who qualified for shorter treatment duration (24 weeks for TPR or 28 weeks for BOC, respectively), as well as in pa- tients who were at need for 48 weeks of therapy (Figure 2C and D). Due to the small number of patients in each group, we did not perform a subgroup analysis. The dur- ation of the lead-in phase had no statistically significant impact on the treatment outcome, neither in patients treated with TPR (P = 0.30), nor in patients receiving BOC (P = 0.68, Figure 2E). Wehmeyer et al. BMC Gastroenterology 2014, 14:87 http://www.biomedcentral.com/1471-230X/14/87 Wehmeyer et al. BMC Gastroenterology 2014, 14:87 http://www.biomedcentral.com/1471-230X/14/87 Page 7 of 10 Page 7 of 10 Age above 50 years (P < 0.01), thrombocytes < 100,000/ μl (P = 0.01), serum albumin < 35 g/l (P = 0.04), ASAT > 100 U/l (P = 0.03) and bilirubin ≥1.2 mg/dl (P < 0.01) were significantly associated with the occurence of a SAE (Table 3). Bilirubin ≥1.2 mg/dl (OR 13.1; 95%CI 2.1-81.4; P < 0.01) and ASAT > 100 U/l (OR 4.6; 95%CI 1.4-15.1; P = 0.01) were independent predictors for a SAE in the multivariate analysis. Tragically, two patients died directly following therapy (mortality 1.9%). Both suffered from cirrhosis (baseline Child-Pugh score: 5 and 7; baseline MELD score: 7 and 12) and the patients received TPR after a four week lead-in phase with pegIFN and RBV. More notable, both patients had known diabetes mellitus type 2 with a base- line HbA1c of 7.1 and 7.8%. The patient with the Child- Pugh score of 7 had also a serum albumin < 35 g/l and a thrombocytopenia < 100,000/µl. The frequency of SAE in patients with liver cirrhosis was 41.4% (12/29). In the subgroup analysis of patients with liver cirrhosis, median MELD score of patients with a SAE was higher as compared to patients without compli- cations (9 versus 7; P < 0.01). Low thrombocytes and ele- vated INR (P < 0.01 and 0.01, respectively) were associated with a SAE, too (see Table in Additional file 4, which In the univariate analysis advanced age (P = 0.02) and existence of liver cirrhosis (P < 0.01) were both associated with incidence of a predefined SAE. Furthermore, low platelet count and high INR (P < 0.01, each), as well as high ASAT and bilirubin (P = 0.04 and 0.01, respectively) were associated with risk for an episode of a SAE. Side-effects and complications of triple therapy [BOC = boceprevir; TPR = telaprevir; N = number; GI = gastrointestinal; DRESS = drug induced rash and eosinophilia with systemic symptoms; SJS = Stevens Johnson syndrome; PTCA = percutaneous coronary angiography; § = 5 patients received transfusion; §§ = 9 patients with history of psychiatric illness; §§§ = hemorrhagic proctitis in 5 patients]. Table 2 Side effects and serious adverse events in patients receiving triple therapy [SAE = serious adverse event; N = number; SD = standard deviation; ASAT = aspartate aminotransferase; ALAT = alanine aminotransferase; γGT = gamma- glutamyltransferase; INR = international normalized ratio]. umber; SD = standard deviation; ASAT = aspartate aminotransferase; ALAT = alanine aminotransferase; γGT = gamma- onal normalized ratio]. verse event; N = number; SD = standard deviation; ASAT = aspartate aminotransferase; ALAT = alanine aminotransferase; γGT = gamma- ase; INR = international normalized ratio] Discussion Additionally, every cirrhotic patient is seen by the transplant team before treatment initiation and listed for liver transplantation if deemed necessary. At the same time – and in concordance with previous reports [9,10] – we saw a high incidence of SAEs especially in patients with liver cirrhosis and even two fatal outcomes in our cohort. Counterintuitively, patients with psychiatric disorders displayed neither a higher rate of treatment fail- ures, nor a higher risk for the incidence of SAE. g The frequency of treatment failure at week 12 of PI administration was reported to be as high as 29% in pre- vious “real-life” reports [9]. Interestingly, in our cohort only 14.7% of patients experienced a treatment failure until week 12 of PI (13 patients with partial response and 2 patients stopping therapy due to moderate side- effects, Figures 1 and 2). Furthermore, the frequency of SVR was higher as compared to the SVR rate in a most recently published “real-life” cohort [8]. However, our observations might (at least partly) be explained by a higher rate of patients with bridging fibrosis or cirrhosis in these studies, as compared to our cohort [8,9]. On the one hand, the reduction of the pegIFN or RBV dose due to side-effects, as well as an early PI withdrawal (when appropriate) were not associated with lower chances for SVR in our small cohort. On the other hand, an early termination of pegIFN and RBV was determined by us as a risk factor for later relapse. Future prospective studies have to determine whether treatment individualization and de-escalation are indeed a valid option in difficult-to- Our retrospective study has certain limitations. First, in a minority of less than 10% of patients liver cirrhosis was not formally excluded, although none of the patients had any clinical or laboratory signs of liver cirrhosis. Second, we only recorded the fact of RBV dose reduction rather than the individual RBV dose. Since we are associated with a viral hepatitis study center, 81 additional patients (many of them treatment naive) recruited to clinical phase II or III trials during the study period. Finally, our cohort in- cluded patients who were previously treated with DAA. Discussion Our study aimed to extend the data obtained by the registration trials of TPR and BOC and to describe our “real-life” experiences of triple therapy in a large cohort of more than 100 patients including “difficult-to-treat” patients (including patients suffering from autoimmune disorders, as well as patients with a major depression) as well as a great number of patients with advanced liver disease. The frequency of F3 or F4 fibrosis (40.2%) was substantially higher as compared to the frequency of bridging fibrosis or cirrhosis in the participants of the registration trials (F3/F4 in 10 to 28% of patients) [3-7]. Of note, almost two-thirds of our patients would have been ineligible for the various registration trials of TPR or BOC [3-7], 15 patients (14.7%) had a history of partial- or null-response in previous treatment and 10 patients were DAA experienced. Astonishingly, these patients displayed a reasonable chance for SVR and the frequency of SVR12 (60.8%) in our “real-life” cohort was only slightly lower as compared to the results from the registration trials [3-7], but still higher than previous “real-life” data on dual treatment with pegIFN/RBV by us and others [17]. Whilst IL28B polymorphism is the strongest pretreat- ment predictor for SVR in pegIFN/RBV based treatment [18], our results confirm previous reports of limited practical value of IL28B polymorphism for prediction of SVR in patients treated with BOC or TPR [19,20]. Elevated ASAT and bilirubin at baseline were the only independent predictors of SAE in our cohort. However, thrombocytopenia and low serum albumin, which have been identified as the key risk factors for hepatic decom- pensation and death under triple therapy before [10], were also associated with occurence of a SAE in the uni- variate analysis in our cohort. Since all deceased patients had cirrhosis and diabetes mellitus type 2, cirrhotic pa- tients with diabetes should be treated with special care since they are most likely to experience severe complica- tions. Furthermore, we recommend that patients who display risk factors for complications, should be referred to an experienced viral hepatitis center. As a consequence of the treatment complexity of triple therapy with an increased risk for relevant and potentially lethal side-effects, we significantly remodelled procedures at our clinics. Every patient is discussed in a multidiscip- linary hepatitis board before HCV triple therapy is initi- ated. Wehmeyer et al. BMC Gastroenterology 2014, 14:87 http://www.biomedcentral.com/1471-230X/14/87 treat patients to manage side-effects and to achieve a rea- sonable chance for SVR as seen in this retrospective study. Our data also indicate, that triple therapy may be a reasonable option for certain DAA-experienced patients, too (Additional file 2). This is an important finding, since the number of DAA-experienced patients will rap- idly increase in the future. However, in the future testing of protease inhibitor escape mutations before initiation of re-treatment might be useful in these cases. Wehmeyer et al. BMC Gastroenterology 2014, 14:87 http://www.biomedcentral.com/1471-230X/14/87 Table 3 Risk factors for occurence of serious adverse events in patients receiving triple therapy No SAE (N = 80) SAE (N = 22) P-Value Number (%) Number (%) Mean (±SD) Mean (±SD) Median (range) Median (range) Male sex 50 (62.5%) 13 (59.1%) 0.81 Age [years] 46.7 (±12.4) 53.5 (±7.3) 0.02 Age > 50 years 32 (40%) 16 (72.7%) <0.01 Liver cirrhosis 17 (23.9%) 12 (57.1%) <0.01 Diabetes mellitus type 2 8 (10%) 5 (22.7%) 0.15 Psychiatric disorders 16 (20%) 4 (18.2%) 1 Exclusion criteria for registration trials 48 (60%) 17 (77.3%) 0.210 Drug 0.45 Telaprevir 49 (61.3%) 16 (72.7%) Boceprevir 31 (38.8%) 6 (27.3%) Baseline laboratory Hemoglobin [g/dl] 14.7 (10.3-18.8) 14.6 (12.3-18.0) 0.69 Leukocytes [x10^9/l] 6.2 (3.1-13.1) 5.9 (2.7-9.5) 0.79 Platelets [x10^9/l] 199.5 (85–377) 127 (48–329) <0.01 ASAT [U/l] 49.5 (16–156) 67.5 (26–328) 0.04 ALAT [U/l] 70 (16–255) 91 (22–271) 0.28 γGT [U/l] 62.5 (57–1274) 72 (25–459) 0.18 Bilirubin [mg/dl] 0.5 (0.2-1.5) 0.75 (0.2-2.2) 0.01 Albumin [g/l] 40 (29–50) 39 (25–44) 0.11 Prothrombin time [INR] 1.00 (0.9-3.3) 1.13 (1.0-2.3) <0.01 Creatinine [mg/dl] 0.8 (0.5-5.8) 0.8 (0.5-1.1) 0.97 Platelets < 100,000/μl 5 (6.3%) 6 (27.3%) 0.01 ASAT > 100 U/l 11 (13.8%) 8 (36.4%) 0.03 Bilirubin ≥1.2 mg/dl 2 (2.5%) 6 (27.3%) <0.01 Albumin ≤35 g/l 3 (3.8%) 4 (18.2%) 0.04 Prothrombin time [INR] > 1.2 4 (5.3%) 3 (13.6%) 0.17 [SAE = serious adverse event; N = number; SD = standard deviation; ASAT = aspartate aminotransferase; ALAT = alanine aminotransferase; γGT = gamma- glutamyltransferase; INR = international normalized ratio]. r occurence of serious adverse events in patients receiving triple therapy Wehmeyer et al. BMC Gastroenterology 2014, 14:87 http://www.biomedcentral.com/1471-230X/14/87 Wehmeyer et al. BMC Gastroenterology 2014, 14:87 http://www.biomedcentral.com/1471-230X/14/87 Page 8 of 10 treat patients to manage side-effects and to achieve a rea- sonable chance for SVR as seen in this retrospective study. Our data also indicate, that triple therapy may be a reasonable option for certain DAA-experienced patients, too (Additional file 2). This is an important finding, since the number of DAA-experienced patients will rap- idly increase in the future. However, in the future testing of protease inhibitor escape mutations before initiation of re-treatment might be useful in these cases. displays risk factors for a SAE in cirrhotic patients). The multivariate analysis did not identify any inde- pendent predictors for a SAE in the subgroup of cir- rhotic patients. Conclusions Shouval D, Sola R, Terg RA, Yoshida EM, Adda N, Bengtsson L, Sankoh AJ, Kieffer TL, George S, Kauffman RS, Zeuzem S: Telaprevir for previously untreated chronic hepatitis C virus infection. N Engl J Med 2011, 364:2405–16. In conclusion, triple therapy with first generation PI pro- vides a reasonable chance for SVR even in “difficult-to- treat” patients, as presented here. However, considering high rates of complications as reported from us and others [9,10], careful patient selection, extensive patient educa- tion and precise monitoring are essential, especially in pa- tients with liver cirrhosis. 5. Sherman KE, Flamm SL, Afdhal NH, Nelson DR, Sulkowski MS, Everson GT, Fried MW, Adler M, Reesink HW, Martin M, Sankoh AJ, Adda N, Kauffman RS, George S, Wright CI, Poordad F: Response-guided telaprevir combination treatment for hepatitis C virus infection. N Engl J Med 2011, 365:1014–24. 6. Bacon BR, Gordon SC, Lawitz E, Marcellin P, Vierling JM, Zeuzem S, Poordad F, Goodman ZD, Sings HL, Boparai N, Burroughs M, Brass CA, Albrecht JK, Esteban R: Boceprevir for previously treated chronic HCV genotype 1 infection. N Engl J Med 2011, 364:1207–17. Discussion Although 50% of the DAA experienced patients achieved SVR12, our study was too small to identify patients who should receive TPR or BOC-based treatment after a viro- logical failure in a DAA based therapy and no assessment of protease excape mutations was performed. However, we believe that this study reflects the “real-life” situation in many large tertiary referral centers and our study provides important learning points in these “challenging-to-treat” patients for other HCV therapy providers worldwide. Wehmeyer et al. BMC Gastroenterology 2014, 14:87 http://www.biomedcentral.com/1471-230X/14/87 Page 9 of 10 Page 9 of 10 Competing interests l h h Malte H. Wehmeyer has served as a speaker for BMS. Sabine Jordan, Olaf Degen, Martina Sterneck, Jan van Lunzen, Ansgar W. Lohse, Julian Schulze zur Wiesch and Stefan Lüth have served as speakers for Roche, Janssen-Cilag and MSD. Annette Hennigs has served as a speaker for Janssen-Cilag. Friederike Eißing, Claudia Röder, Anja Hüfner, Sandra Hertling and Stefan Schmiedel declare no conflict of interest. 10. Hézode C, Fontaine H, Dorival C, Larrey D, Zoulim F, Canva V, de Ledinghen V, Poynard T, Samuel D, Bourlière M, Zarski JP, Raabe JJ, Alric L, Marcellin P, Riachi G, Bernard PH, Loustaud-Ratti V, Métivier S, Tran A, Serfaty L, Abergel A, Causse X, Di Martino V, Guyader D, Lucidarme D, Grando-Lemaire V, Hillon P, Feray C, Dao T, Cacoub P, et al: Triple therapy in treatment- experienced patients with hcv-cirrhosis in a multicentre cohort from the french early access programme (anrs co20-CUPIC) – NCT01514890. J Hepatol 2013, 59:434–441. 10. Hézode C, Fontaine H, Dorival C, Larrey D, Zoulim F, Canva V, de Ledinghen V, Poynard T, Samuel D, Bourlière M, Zarski JP, Raabe JJ, Alric L, Marcellin P, Riachi G, Bernard PH, Loustaud-Ratti V, Métivier S, Tran A, Serfaty L, Abergel A, Causse X, Di Martino V, Guyader D, Lucidarme D, Grando-Lemaire V, Hillon P, Feray C, Dao T, Cacoub P, et al: Triple therapy in treatment- experienced patients with hcv-cirrhosis in a multicentre cohort from the french early access programme (anrs co20-CUPIC) – NCT01514890. J Hepatol 2013, 59:434–441. Authors’ contributions MHW drafted the originial manuscript, contributed to study design, performed the statistical analysis, interpreted the results and collected the data; FE performed additional statistical analysis and collected the data. SJ, AHe, OD, AHü, SH, SS and MS collected the data; CR contributed to the study design; JvL contributed to study design and data collection; AWL critically revised the manuscript; JSzW and SL contributed to study design, collected data and critically revised the manuscript. All authors read and approved the final manuscript. 11. Castera L, Forns X, Alberti A: Non-invasive evaluation of liver fibrosis using transient elastography. J Hepatol 2008, 48:835–47. 12. Pugh RNH, Murray-Lyon IM, Dawson JL, Pietroni MC, Williams R: Transection of the oesophagus for bleeding oesophageal varices. Br J Surg 1973, 60:646–9. 13. Kamath PS, Wiesner RH, Malinchoc M, Kremers W, Therneau TM, Kosberg CL, D’Amico G, Dickson ER, Kim WR: A model to predict survival in patients with end-stage liver disease. Hepatology 2001, 33:464–70. 14. Fried MW, Hadziyannis SJ, Shiffman ML, Messinger D, Zeuzem S: Rapid virological response is the most important predictor of sustained virological response across genotypes in patients with chronic hepatitis C virus infection. J Hepatol 2011, 55:69–75. Additional files 7. Zeuzem S, Andreone P, Pol S, Lawitz E, Diago M, Roberts S, Focaccia R, Younossi Z, Foster GR, Horban A, Ferenci P, Nevens F, Müllhaupt B, Pockros P, Terg R, Shouval D, van Hoek B, Weiland O, Van Heeswijk R, De Meyer S, Luo D, Boogaerts G, Polo R, Picchio G, Beumont M: Telaprevir for retreatment of HCV infection. N Engl J Med 2011, 364:2417–28. Additional file 1: Virological and adverse events definitions, grading of laboratory events and further information on the DAA experienced patients. Additional file 2: Course of therapy in DAA-experienced patients, as well as in patients with special comorbidities. Additional file 3: Evaluation of predictors for SVR12. Additional file 4: Evaluation of risk factors for SAE in cirrhotic patients. Additional file 1: Virological and adverse events definitions, grading of laboratory events and further information on the DAA experienced patients. Additional file 2: Course of therapy in DAA-experienced patients, as well as in patients with special comorbidities. Additional file 3: Evaluation of predictors for SVR12. Additional file 4: Evaluation of risk factors for SAE in cirrhotic patients. Additional file 1: Virological and adverse events definitions, grading of laboratory events and further information on the DAA experienced patients. Additional file 2: Course of therapy in DAA-experienced patients, as well as in patients with special comorbidities. Additional file 2: Course of therapy in DAA-experienced patients, as well as in patients with special comorbidities. 8. Backus LI, Belperio PS, Shahoumian TA, Cheung R, Mole LA: Comparative effectiveness of the hepatitis C virus protease inhibitors boceprevir and telaprevir in a large U.S. cohort. Aliment Pharmacol Ther 2014, 39:93–103. 9. Maasoumy B, Port K, Markova AA, Serrano BC, Rogalska-Taranta M, Sollik L, Mix C, Kirschner J, Manns MP, Wedemeyer H, Cornberg M: Eligibility and safety of triple therapy for hepatitis C: lessons learned from the first experience in a real world setting. PLoS ONE 8(2):e55285. doi:10 1371/journal pone 0055285 Acknowledgements Besides the named authors of this manuscript, no further persons contributed to the conception, design, acquisition of data, data analysis, interpretation of data, or was involved in drafting of the manuscript or revising it, respectively. 15. Ghany MG, Nelson DR, Strader DB, Thomas DL, Seeff LB: An update on treatment of genotype 1 chronic hepatitis C virus infection: 2011 practice guideline by the American association for the study of liver diseases. Hepatology 2011, 54:1433–44. g y We declare no sources of funding regarding this study. We declare no sources of funding regarding this study. Author details 1 1I. Medizinische Klinik und Poliklinik, Universitätsklinikum Hamburg-Eppendorf, Martinistr. 52, Hamburg 20246, Germany. 2Ambulanzzentrum des UKE, Fachbereich Infektiologie, Hamburg, Germany. 16. Sarrazin C, Berg T, Cornberg M, Dollinger M, Ferenci P, Hinrichsen H, Klinker H, Kraus M, Manns M, Mauss S, Peck-Radosavljevic M, Schmidt H, Spengler U, Wedemeyer H, Wirth S, Zeuzem S: Expert opinion on boceprevir- and telaprevir-based triple therapies of chronic hepatitis C. Z Gastroenterol 2012, 50:57–72. 1I. Medizinische Klinik und Poliklinik, Universitätsklinikum Hamburg-Eppendorf, Martinistr. 52, Hamburg 20246, Germany. 2 1I. Medizinische Klinik und Poliklinik, Universitätsklinikum Hamburg-Eppendorf, Martinistr. 52, Hamburg 20246, Germany g pp g y 2Ambulanzzentrum des UKE, Fachbereich Infektiologie, Hamburg, Germany. Received: 13 December 2013 Accepted: 28 April 2014 Published: 5 May 2014 Received: 13 December 2013 Accepted: 28 April 2014 Received: 13 December 2013 Accepted: 28 April 2014 Received: 13 December 2013 Accepted: 28 April 2014 Published: 5 May 2014 17. Schulze zur Wiesch J, Pudelski N, Hoepner L, Supplieth M, Buggisch P, Lohse AW, Lüth S: “Real-life” comparison of pegylated-interferon 2a versus 2b combination therapy of chronic hepatitis c virus. Hepatology 2011, 53:1405–1406. Published: 5 May 2014 19. Pol S, Aerssens J, Zeuzem F, Andreone P, Lawitz EJ, Roberts S, Younossi Z, Foster GR, Focaccia R, Horban A, Pockros PJ, Van Heeswijk RP, De Meyer S, Luo D, Botfield M, Beumont M, Picchio G: Limited impact of IL28B References 1. Global surveillance and control of hepatitis C: report of a WHO consultation organized in collaboration with the viral hepatitis prevention board, Antwerp, Belgium. J Viral Hepat 1999, 6:35–47. 18. Thompson AJ, Muir AJ, Sulkowski MS, Ge D, Fellay J, Shianna KV, Urban T, Afdhal NH, Jacobson IM, Esteban R, Poordad F, Lawitz EJ, McCone J, Shiffman ML, Galler GW, Lee WM, Reindollar R, King JW, Kwo PY, Ghalib RH, Freilich B, Nyberg LM, Zeuzem S, Poynard T, Vock DM, Pieper KS, Patel K, Tillmann HL, Noviello S, Koury K, et al: Interleukin-28B polymorphism improves viral kinetics and is the strongest pretreatment predictor of sustained virological response in genotype 1 hepatitis C virus. Gastroenterology 2010, 139:120–129. 2. Amin J, Law MG, Bartlett M, Kaldor JM, Dore GJ: Causes of death after diagnosis of hepatitis B or hepatitis C infection: a large community- based linkage study. Lancet 2006, 368:938–945. g y 3. Poordad F, McCone J, Bacon BR, Bruno S, Manns MP, Sulkowski MS, Jacobson IM, Reddy KR, Goodman ZD, Boparai N, DuNubile MJ, Sniukiene V, Brass CA, Albrecht JK, Bronowicki JP: Boceprevir for untreated chronic HCV genotype 1-infection. N Engl J Med 2011, 364:1195–206. 19. Pol S, Aerssens J, Zeuzem F, Andreone P, Lawitz EJ, Roberts S, Younossi Z, Foster GR, Focaccia R, Horban A, Pockros PJ, Van Heeswijk RP, De Meyer S, Luo D, Botfield M, Beumont M, Picchio G: Limited impact of IL28B 19. Pol S, Aerssens J, Zeuzem F, Andreone P, Lawitz EJ, Roberts S, Younossi Z, Foster GR, Focaccia R, Horban A, Pockros PJ, Van Heeswijk RP, De Meyer S, Luo D, Botfield M, Beumont M, Picchio G: Limited impact of IL28B 4. Jacobson IM, McHutchison JG, Dusheiko G, Di Bisceglie AM, Reddy KR, Bzowej NH, Marcellin P, Muir AJ, Ferenci P, Flisiak R, George J, Rizzetto M, Page 10 of 10 Page 10 of 10 Wehmeyer et al. BMC Gastroenterology 2014, 14:87 http://www.biomedcentral.com/1471-230X/14/87 genotype on response rates in telaprevir-treated patients with prior treatment failure. J Hepatol 2013, 58:883–889. 20. Poordad F, Bronowicki JP, Gordon SC, Zeuzem S, Jacobson IM, Sulkowski MS, Poynard T, Morgan TR, Molony C, Pedicone LD, Sings HL, Burroughs MH, Sniukiene V, Boparai N, Goteti VS, Brass CA, Albrecht JK, Bacon BR: Factors that predict response of patients with hepatitis C virus infection to boceprevir. Gastroenterology 2012, 143:608–618. References doi:10.1186/1471-230X-14-87 Cite this article as: Wehmeyer et al.: Safety and efficacy of protease inhibitor based combination therapy in a single-center “real-life” cohort of 110 patients with chronic hepatitis C genotype 1 infection. BMC Gastroenterology 2014 14:87. doi:10.1186/1471-230X-14-87 Cite this article as: Wehmeyer et al.: Safety and efficacy of protease inhibitor based combination therapy in a single-center “real-life” cohort of 110 patients with chronic hepatitis C genotype 1 infection. BMC Gastroenterology 2014 14:87. Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color ﬁgure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color ﬁgure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit
https://openalex.org/W2130449444	https://cdr.lib.unc.edu/downloads/n583z3850	English	null	Disentangling the effects of climate and people on Sahel vegetation dynamics	Biogeosciences	2,009	cc-by	7,507	Disentangling the effects of climate and people on Sahel vegetation dynamics J. W. Seaquist1, T. Hickler1, L. Eklundh1, J. Ard¨o1, and B. W. Heumann2 1Department of Physical Geography and Ecosystems Analysis, Geobiosphere Science Centre, Lund University, S¨olvegatan 12, 223 62, Lund, Sweden 2Department of Geography, University of North Carolina at Chapel Hill, Saunders Hall, Campus Box 3220, Chapel Hill, NC 27599-3220, USA J. W. Seaquist1, T. Hickler1, L. Eklundh1, J. Ard¨o1, and B. W. Heumann2 1Department of Physical Geography and Ecosystems Analysis, Geobiosphere Science Centre, Lund University, S¨olvegatan 12, 223 62, Lund, Sweden 2Department of Geography, University of North Carolina at Chapel Hill, Saunders Hall, Campus Box 3220, Chapel Hill, NC 27599-3220, USA Received: 1 July 2008 – Published in Biogeosciences Discuss.: 7 August 2008 Revised: 20 February 2009 – Accepted: 25 February 2009 – Published: 26 March 2009 interactions, and/or limitations in the data used for this study. We do not exclude the possibility of a greater human inﬂu- ence on vegetation dynamics over the coming decades with changing land use. interactions, and/or limitations in the data used for this study. We do not exclude the possibility of a greater human inﬂu- ence on vegetation dynamics over the coming decades with changing land use. Abstract. The Sahel belt of Africa has been the focus of in- tensive scientiﬁc research since the 1960s, spurred on by the chronic vulnerability of its population to recurring drought and the threat of long-term land degradation. But satellite sensors have recently shown that much of the region has experienced signiﬁcant increases in photosynthetic activity since the early 1980s, thus re-energizing long-standing de- bates about the role that people play in shaping land surface status, and thus climate at regional scales. In this paper, we test the hypothesis that people have had a measurable impact on vegetation dynamics in the Sahel for the period 1982– 2002. We compare potential natural vegetation dynamics predicted by a process-based ecosystem model with satellite- derived greenness observations, and map the agreement be- tween the two across a geographic grid at a spatial resolution of 0.5◦. As aggregated data-model agreement is very good, any local differences between the two could be due to human impact. We then relate this agreement metric to state-of-the- art data sets on demographics, pasture, and cropping. Our ﬁndings suggest that demographic and agricultural pressures in the Sahel are unable to account for differences between simulated and observed vegetation dynamics, even for the most densely populated areas. Disentangling the effects of climate and people on Sahel vegetation dynamics But we do identify a weak, positive correlation between data-model agreement and pas- ture intensity at the Sahel-wide level. This indicates that herding or grazing does not appreciably affect vegetation dy- namics in the region. Either people have not had a signiﬁcant impact on vegetation dynamics in the Sahel or the identiﬁca- tion of a human “footprint” is precluded by inconsistent or subtle vegetation response to complex socio-environmental Biogeosciences, 6, 469–477, 2009 www.biogeosciences.net/6/469/2009/ © Author(s) 2009. This work is distributed under the Creative Commons Attribution 3.0 License. Biogeosciences 1 Introduction The Sahel belt of Africa has been the focus of intensive sci- entiﬁc research in recent decades, spurred on by the chronic vulnerability of its population to recurring drought and the threat of long-term land degradation. Century-long records of annual rainfall totals show that a lengthy period of desic- cation began in the mid-1960s, accompanied by a marked decrease in interannual rainfall variability (Hulme, 2001). Evidence is mounting that the rainfall climate of the Sahel is ultimately driven by changes in ocean circulation and sea surface temperatures and further ampliﬁed by internal non- linear biogeophysical feedback mechanisms that operate be- tween the vegetation and the atmosphere (e.g. Zeng et al., 1999; Wang and Eltahir, 2000; Giannini et al., 2003). This feedback is thought to have led to an abrupt change to a per- sistently drier climate compared to the ﬁrst half of the 20th century (Foley et al, 2003). Recent research has re-invigorated long-standing debates about the direct impact people have on observed changes in the Sahel’s land cover status (e.g. Hein and de Ridder, 2006; Prince et al., 2007). The paradigm of geographically exten- sive and irreversible land degradation has been challenged by satellite observations of vegetation greenness (NDVI – Normalized Difference Vegetation Index) that showed the return of vegetation in years that received favourable rain- fall (Helld´en, 1991; Tucker et al., 1991). Even more recent ﬁndings using data from the same satellite sensors show a Correspondence to: J. W. Seaquist (jonathan.seaquist@nateko.lu.se) Correspondence to: J. W. Seaquist (jonathan.seaquist@nateko.lu.se) Published by Copernicus Publications on behalf of the European Geosciences Union. Published by Copernicus Publications on behalf of the European Geosciences Union. 470 J. W. Seaquist et al.: Disentangling the effects on Sahel vegetation dynamics J. W. Seaquist et al.: Disentangling the effects on Sahel vegetation dynamics widespread increase in photosynthetic activity since the early 1980s (Eklundh and Olsson, 2003; Olsson et al., 2005; Her- mann et al., 2005, Seaquist et al. 2006; Heumann et al., 2007). Proposed explanations have included increasing rain- fall, CO2 enhancement effects, agricultural intensiﬁcation, extensiﬁcation, and land abandonment stemming from socio- economic change and conﬂict. approach by distributing model output across a geographic grid. We then relate a data-model agreement metric to state- of-the-art data sets on demographics, pasture, and cropping intensity on a cell-by-cell basis. 1available at: http://www.pik-potsdam.de/research/cooperations/lpjweb 1 Introduction If human impact has been signiﬁcant, we should then expect to ﬁnd an association be- tween data-model differences and land use/population pres- sures. g Using a process-based vegetation model, LPJ-DGVM (Lund Potsdam Jena-Dynamic Global Vegetation Model), Hickler et al. (2005) identiﬁed rainfall change as the primary regulator of vegetation dynamics in the Sahel between 1982 and 1998 (with CO2 having a minor positive inﬂuence), al- beit at the aggregated level. They closely reproduced the satellite-observed greening trend and its interannual variabil- ity by holding all driving variables in the model constant ex- cept rainfall. In another study, Hermann et al. (2005) used a data-driven approach to reach a similar conclusion by statis- tically relating gridded, satellite-derived rainfall data (at 2.5◦ and 1.0◦spatial resolution) with satellite-based NDVI obser- vations (at 8 km spatial resolution) for the period 1982–2003. After removing the inﬂuence of rainfall from the NDVI data, spatially coherent, statistically signiﬁcant residual trends re- mained, showing where the land surface has either been greening faster (possibly indicating land “rehabilitation”) or slower (possibly indicating land degradation) than can be ac- counted for by rainfall alone. These ﬁndings are echoed by Evans and Geerken (2004) and Wessels et al. (2007) who used similar NDVI-rainfall residual trend analyses (at NDVI resolution of 8 km and interpolated rainfall data) in order to isolate possible human-induced degradation signals in the drylands of Syria and the rangelands of South Africa respec- tively. Additionally, Li et al. (2004) applied NDVI-rainfall regression diagnostics to assess land cover performance in Senegal. 2.1 Satellite-derived greenness data We used the Global Inventory Modeling and Mapping Stud- ies (GIMMS) 10-day maximum value composite NDVI dataset derived from the National Oceanic and Atmospheric Administration’s (NOAA) series of Advanced Very High Resolution Radiometer (AVHRR) instruments with a spa- tial resolution of 8 km (Tucker et al., 2005). The NDVI is computed from the red and near infrared bands of the AVHRR sensor and is correlated with photosynthetic activity and leaf area index of green vegetation (Sellers, 1989). The NDVI also reduces unwanted variations in the data due to to- pography, sensor calibration, and viewing geometry, while maximum value compositing reduces the effects of cloud cover and the atmosphere (Holben, 1986). The GIMMS data set has been corrected for sensor differences, sensor drift, and the effects of volcanic eruptions. We extracted peak NDVI observations (NDVImax) for each growing sea- son from 1982–2002. We then projected them to a geo- graphic grid while re-sampling to 0.5◦(using bi-linear in- terpolation) in order to match the resolution of the climate input data used for driving LPJ-DGVM. Data were extracted for 12◦N to 20◦N and 19◦W to 40◦E (Fig. 1). A southern limit of 12◦N was chosen to avoid saturation of the NDVI observations which would make vegetation changes difﬁcult to discern. g Though Hickler et al. (2005) presented convincing evi- dence that rainfall governs vegetation dynamics in the Sa- hel at the aggregated level, the spatial details of data-model agreement were not resolved. It is possible that minor dif- ferences in aggregated simulated versus observed vegetation dynamics stem from substantial differences in sub-regions where human impact may be measurable. Since the LPJ- DGVM simulates potential natural vegetation while satellite- derived greenness represents actual vegetation, differences between the two could be due to anthropogenic pressures. The possibility therefore exists that the human “footprint” in dryland regions, including the Sahel, may be signiﬁcant enough to alter broad-scale biogeophysical processes, car- bon cycling, and climate through direct impact on vegetation dynamics. But the human impact hypothesis remains largely untested for the Sahel at broad spatial extents. Biogeosciences, 6, 469–477, 2009 3available at: http://na.unep.net/globalpop/africa/Africa index.html 4available at: http://www.geog.mcgill.ca/∼nramankutty/Datasets/Datasets.html 2available at: http://cdiac.esd.ornl.gov/trends/co2/contents.htm 2.2 Vegetation model LPJ-DGVM is a coupled biogeography-biogeochemistry model, which incorporates process-based representations of terrestrial vegetation dynamics and biogeochemical cycling. A full description of the model can be found in Sitch et al. (2003). Processes governing water uptake by vegetation were updated by Gerten et al. (2004). The model is used by a broad community of global-change researchers1 and has, for example, been shown to successfully reproduce the distribution of global biomes (Sitch et al., 2003; Hickler et al., 2006), variations in net primary production (NPP) across biomes (Hickler et al., 2006; Zaehle et al., 2005), satellite- derived high-latitude greening trends in the 1980s and 1990s (Lucht et al., 2002), and site-scale and global patterns in the In this paper, we test the hypothesis that people have had a measurable impact on vegetation dynamics in the Sahel for the period 1982–2002. Like Hickler et al. (2005), we com- pare vegetation dynamics predicted by the LPJ-DGVM with satellite-based greenness observations, but we extend their www.biogeosciences.net/6/469/2009/ Biogeosciences, 6, 469–477, 2009 www.biogeosciences.net/6/469/2009/ J. W. Seaquist et al.: Disentangling the effects on Sahel vegetation dynamics 471 Fig. 1. Study area, restricted to 12◦N to 20◦N and 19◦W to 40◦E. water balance of terrestrial ecosystems (Gerten et al., 2004). We used the C++ implementation of LPJ-DGVM within the generalized modelling framework LPJ-GUESS (Smith et al., 2001). 2.4 Data-model agreement and greenness trends slope coefﬁcient of the regression was used for determining the sign and magnitude of the trends. It is worth noting that vegetation change, particularly in drylands, may not always change linearly, but most studies that analyze Sahel green- ing at the broad extent apply OLS (e.g. Eklundh and Olsson, 2003; Olsson et al., 2005; Hermann et al., 2005; Heumann et al., 2007). We therefore used an OLS analysis for contextu- alizing our results. Spearman’s rank-order correlation coefﬁcient was used to gauge data-model agreement (rs[NDVI−LAI]). Spearman’s co- efﬁcient varies between −1 and +1, where large positive values indicate strong agreement, large negative values in- dicate strong disagreement, and values near 0 indicate ran- dom agreement. The purpose of this coefﬁcient is to mea- sure the association between two variables (e.g. LAImax and NDVImax) whose scores are in rank order (McGrew Jr. and Monroe, 2000). The reason for choosing this coefﬁcient was to remove any non-linearity that could undermine the detection of a monotonic relationship between LAImax and NDVImax. 2.3 Modelling protocol and climate data LPJ-DGVM was driven with monthly data for air tempera- ture, precipitation, number of rain days, and sunshine hours from the global CRU05 dataset for the region bounded by 12◦N to 20◦N and 19◦W to 40◦E at a 0.5◦resolution (Mitchell and Jones, 2005). Monthly temperature and sun- shine data were interpolated to provide quasi-daily time se- ries. Daily precipitation ﬁelds were generated with a stochas- tic weather generator (Gerten et al., 2004). Annual global atmospheric CO2 concentrations were obtained courtesy of the Carbon Cycle Model Linkage Project (McGuire et al., 2001) and TRENDS2. Climate data for a 1000-year model spin-up were derived from the same data by repeating de- trended values of the climate input for 1901–1930, using pre-industrial CO2 content of the atmosphere. The model was run as to simulate the potential natural vegetation; in- formation on land cover was not included. Growing season maximum LAI (LAImax) were then extracted for the period 1982–2002. Fig. 1. Study area, restricted to 12◦N to 20◦N and 19◦W to 40◦E. Fig. 1. Study area, restricted to 12◦N to 20◦N and 19◦W to 40◦E. 2.5 Population density and land use Gridded data on population density (persons km−2) for the years 1980, 1990, and 2000 were downloaded from the Africa Population Distribution Database3, resampled to a 0.5◦resolution and subset to the same spatial extent as all other layers (Fig. 2a). These grids were derived from census data, transportation networks (roads, railroad, and navigable rivers) and information on the location of urban centres. The underlying assumption for the generation of these data is that population density is strongly dependent on accessibility via transportation routes. We chose to compare LAImax with NDVImax because the use of other vegetation parameters would require a number of additional, potentially complicating assumptions. For ex- ample, while the comparison of integrated NDVI and annual NPP may seem better for capturing the impact of land use on vegetation (as the NDVI integral is often assumed to be a proxy for NPP), integrated NDVI more realistically repre- sents the amount of photosynthetically active radiation ab- sorbed by the plant canopy rather than plant growth, rather than NPP (Prince, 1991). We used the Agricultural Lands in the Year 2000 (AL2000) data set described in Ramankutty et al. (20084). This is a state-of-the-art global data set depicting cropland use and pasture use intensity (values ranging between 0 and For comparison, we also applied a temporal trend anal- ysis the NDVImax observations. Curves were ﬁtted to ev- ery cell using OLS (ordinary least-squares regression). The Biogeosciences, 6, 469–477, 2009 www.biogeosciences.net/6/469/2009/ J. W. Seaquist et al.: Disentangling the effects on Sahel vegetation dynamics J. W. Seaquist et al.: Disentangling the effects on Sahel vegetation dynamics 472 Fig. 2. Population density for year 2000 (Nelson, 2004) (a), cropping intensity ca. year 2000 (Ramankutty et al., 2008) (b), and grazing intensity ca. year 2000 (Ramankutty et al., 2008) (c). Fig. 2. Population density for year 2000 (Nelson, 2004) (a), cropping intensity ca. year 2000 (Ramankutty et al., 2008) (b), and grazing intensity ca. year 2000 (Ramankutty et al., 2008) (c). 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 −3 −2 −1 0 1 2 3 Year Standardized NDVI and LAI anomalies Model LAI Max. NDVI YLAI = 0.117 X − 234 YNDVI = 0.098 X − 195 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 −3 −2 −1 0 1 2 3 Year Standardized NDVI and LAI anomalies Model LAI Max. 2.5 Population density and land use NDVI YLAI = 0.117 X − 234 YNDVI = 0.098 X − 195 Fig. 3. Aggregated standardized anomalies of modelled LAImax and observed NDVImax 1982–2002, and linear regression lines for the data vs. time. Fig. 3. Aggregated standardized anomalies of modelled LAImax and observed NDVImax 1982–2002, and linear regression lines for the data vs. time. 1) at a spatial resolution of 5 min (approximately 10 km). The AL2000 was produced by merging agricultural inven- tory data for different countries with satellite-derived land cover data derived from Boston University’s Moderate Res- olution Imaging Spectrometer (MODIS) land cover product and Satellite Pour l’Observation de la Terre (SPOT) Vegeta- tion’s Global Land Cover 2000 product. This is the ﬁrst data set of its kind to attach statistical conﬁdence intervals to land use estimates. We extracted data for the study area, while coarsening the resolution to 0.5◦using bi-linear interpolation (see Fig. 2b and c). www.biogeosciences.net/6/469/2009/ Biogeosciences, 6, 469–477, 2009 J. W. Seaquist et al.: Disentangling the effects on Sahel vegetation dynamics 473 Fig. 4. Spearman rank correlation coefﬁcient between (a) simulated LAImax and observed NDVImax between 1982–2002 (strongly posi- tive/negative and positive/negative denote statistical signiﬁcance thresholds of α=0.01, rs[NDVI−LAI]=0.55, and α=0.05, rs[NDVI−LAI]=0.44, respectively), and (b) NDVImax trend change 1982–2002 expressed as a percentage. Fig. 4. Spearman rank correlation coefﬁcient between (a) simulated LAImax and observed NDVImax between 1982–2002 (strongly posi- tive/negative and positive/negative denote statistical signiﬁcance thresholds of α=0.01, rs[NDVI−LAI]=0.55, and α=0.05, rs[NDVI−LAI]=0.44, respectively), and (b) NDVImax trend change 1982–2002 expressed as a percentage. 3.1 Data-model comparison On an aggregated level, LPJ-DGVM generally reproduced the overall trend and pattern of interannual variation in observed vegetation greenness between 1982 and 2002 (Fig. 3). The simulated mean LAImax was 3.15 while the mean NDVImax was 0.36, averaged over all cells for all years. Regressing the yearly spatial mean NDVImax anomaly against the yearly spatial mean LAImax anomaly yields a Pearson correlation coefﬁcient 0.77 with a small bias (LAImax=0.77∗NDVImax−0.004). rs[NDVI−LAI] tends to be higher for areas with strong pos- itive NDVImax trend correlation coefﬁcients (Fig. 5a). Poor data-model agreement is associated with negative trends in observed vegetation indicating poor model performance for these areas. However, negative changes in NDVImax are gen- erally much smaller than positive trends when expressed as changes in percent (Fig. 5b). 2.6 Data-model agreement vs. population density and land use niﬁcant negative correlations, corresponding to 3.4% of the total land area (mean rs[NDVI−LAI]=−0.56, α=0.05, thresh- old rs[NDVI−LAI] for statistical signiﬁcance=−0.44). Finally, we related level of data model agreement (rs[NDVI−LAI]) to estimates of population density and its change from the Africa Population Distribution Database, as well as cropping and pasture from the AL2000 for all terres- trial surfaces (n=1780). For comparison, Fig. 4b shows the strength of the trends in NDVImax for 1982–2002 in percent (α=0.05). Statistically signiﬁcant increases in NDVImax occur in a relatively unin- terrupted band stretching from the Atlantic coast to the Red Sea. Strongest trends occur in Central Chad and Western Su- dan with increases in excess of 100% for the 1982 to 2002 period. 683 cells or 38.3% of the area show trends that are statistically signiﬁcant. Negative trends are conﬁned to three cells in the southwest corner of Niger with decreases around 20%. J. W. Seaquist et al.: Disentangling the effects on Sahel vegetation dynamics Solid line in (d) indicates regression for Spearman rank correlation coefﬁcient vs. pasture intensity; y=0.078+0.513x; r=0.46. −1 −0.5 0 0.5 1 −1 −0.5 0 0.5 1 NDVI trend corr. coeff. rs[NDVI−LAI] a) 0 50 100 −1 −0.5 0 0.5 1 NDVI change (%) rs[NDVI−LAI] b) Fig. 5. Plots of Spearman rank correlation coefﬁcient (rs[NDVI−LAI]) against (a) NDVImax trend correlation coefﬁcients, and (b) change in NDVImax over time. Dashed grey lines indicate thresholds of statistical signiﬁcance (α=0.05) for data-model com- parison and temporal least-squares regression for NDVImax. 0 100 200 300 400 −1 −0.5 0 0.5 1 Population density year 2000 (pers. / km2) rs[NDVI−LAI] a) −50 0 50 −1 −0.5 0 0.5 1 Rel. population change 1980−2000 (%) rs[NDVI−LAI] b) 0 5 0 0.5 1 rs[NDVI−LAI] c) 0 5 0 0.5 1 rs[NDVI−LAI] d) −1 −0.5 0 0.5 1 −1 −0.5 0 0.5 1 NDVI trend corr. coeff. rs[NDVI−LAI] a) 0 50 100 −1 −0.5 0 0.5 1 NDVI change (%) rs[NDVI−LAI] b) Fig. 5. Plots of Spearman rank correlation coefﬁcient (rs[NDVI−LAI]) against (a) NDVImax trend correlation coefﬁcients, and (b) change in NDVImax over time. Dashed grey lines indicate thresholds of statistical signiﬁcance (α=0.05) for data-model com- parison and temporal least-squares regression for NDVImax. 0 50 100 −1 −0.5 0 0.5 1 NDVI change (%) rs[NDVI−LAI] b) −1 −0.5 0 0.5 1 −1 −0.5 0 0.5 1 NDVI trend corr. coeff. rs[NDVI−LAI] a) 0 100 200 300 400 −1 −0.5 0 0.5 1 Population density year 2000 (pers. / km2) rs[NDVI−LAI] a) −50 0 50 −1 −0.5 0 0.5 1 Rel. population change 1980−2000 (%) rs[NDVI−LAI] b) rs[NDVI−LAI] 0 0.2 0.4 0.6 0.8 1 −1 −0.5 0 0.5 1 Pasture intensity rs[NDVI−LAI] d) 0 0.2 0.4 0.6 0.8 1 −1 −0.5 0 0.5 1 Cropping intensity rs[NDVI−LAI] c) Fig. 5. Plots of Spearman rank correlation coefﬁcient (rs[NDVI−LAI]) against (a) NDVImax trend correlation coefﬁcients, and (b) change in NDVImax over time. Dashed grey lines indicate thresholds of statistical signiﬁcance (α=0.05) for data-model com- parison and temporal least-squares regression for NDVImax. rs[NDVI−LAI] agreement versus the absolute change in population density between 1980 and 2000 was very similar to Fig. 6a. Rates of population growth throughout the period ranged from −0.54 to 26.5 persons km−2 yr−1 with a mean of 0.36 km−2 yr−1. Relative change in population ranges from −39.3% to 492.4%, with a mean of 43.3%. J. W. Seaquist et al.: Disentangling the effects on Sahel vegetation dynamics However, rs[NDVI−LAI] in- creases with pasture intensity (Fig. 6d), indicating that the greater proportion of pasture in an area, the better the model simulates actual vegetation. For this relationship, r=0.46. Log-transforming population and agricultural parameters do not result in increased levels of association with data-model agreement, and there were no strong associations between population, cropping intensity, and grazing intensity. Fig. 6. Plots of Spearman rank correlation coefﬁcient (rs[NDVI−LAI]) against (a) population density in the year 2000, (b) relative population change from 1980 to 2000, (c) cropping inten- sity, and (d) pasture intensity. Dashed grey lines indicate thresh- olds of statistical signiﬁcance for data-model comparison (α=0.05). Solid line in (d) indicates regression for Spearman rank correlation coefﬁcient vs. pasture intensity; y=0.078+0.513x; r=0.46. ﬂuences), this suggests that livestock grazing is generally not associated with variations in greenness in the Sahel, at least at the scale of our study. Once possible interpreta- tion is that the pastoralists, along with their herds, slowly migrated to the areas that were receiving plentiful rainfall over the period. As mentioned in Sect. 3.2, data-model agreement is always positive for areas experiencing depop- ulation (Fig. 6b), but most of these cells already have very low population densities and occur within the broad area of greening that roughly corresponds with the pastoral zone. And since there was a moderate relationship between data- model agreement and pasture with high population growth rates (2.4 persons km−2 yr−1 n=58) from between 1980 and 2000 (r=0.58), the implication is that areas near urban cen- tres did not undergo degradation, but on the contrary, became greener. It also re-enforces the hypothesis that animal density does not appreciably impact vegetation dynamics, at least at this scale. This ﬁnding is consistent with the non-equilibrium hypothesis of rangeland ecology (e.g. Behnke and Scoones, 1993). Since most of the densely populated areas occur within or nearby urban centres, we also investigated the hypothesis that data-model agreement would be most affected by population and land use only in those areas undergoing the most rapid population growth. We therefore isolated areas with different thresholds in population growth between 1980 and 2000, and retained the threshold that gave an acceptable sample size (2.4 persons km−2 yr−1 n=58). We then repeated our analy- sis, relating data-model agreement to population/population change, cropping, and pasture. r for rs[NDVI−LAI] versus pas- ture was 0.58. r for all other relationships was below 0.07. J. W. Seaquist et al.: Disentangling the effects on Sahel vegetation dynamics J. W. Seaquist et al.: Disentangling the effects on Sahel vegetation dynamic 474 −1 −0.5 0 0.5 1 −1 −0.5 0 0.5 1 NDVI trend corr. coeff. rs[NDVI−LAI] a) 0 50 100 −1 −0.5 0 0.5 1 NDVI change (%) rs[NDVI−LAI] b) Fig. 5. Plots of Spearman rank correlation coefﬁcient (rs[NDVI−LAI]) against (a) NDVImax trend correlation coefﬁcients, and (b) change in NDVImax over time. Dashed grey lines indicate thresholds of statistical signiﬁcance (α=0.05) for data-model com- parison and temporal least-squares regression for NDVImax. agreement versus the absolute change in population density between 1980 and 2000 was very similar to Fig. 6a. Rates of population growth throughout the period ranged from −0.54 to 26.5 persons km−2 yr−1 with a mean of 0.36 km−2 yr−1. Relative change in population ranges from −39.3% to 492.4%, with a mean of 43.3%. However, rs[NDVI−LAI] in- creases with pasture intensity (Fig. 6d), indicating that the greater proportion of pasture in an area, the better the model 0 100 200 300 400 −1 −0.5 0 0.5 1 Population density year 2000 (pers. / km2) rs[NDVI−LAI] a) −50 0 50 −1 −0.5 0 0.5 1 Rel. population change 1980−2000 (%) rs[NDVI−LAI] b) 0 0.2 0.4 0.6 0.8 1 −1 −0.5 0 0.5 1 Cropping intensity rs[NDVI−LAI] c) 0 0.2 0.4 0.6 0.8 1 −1 −0.5 0 0.5 1 Pasture intensity rs[NDVI−LAI] d) Fig. 6. Plots of Spearman rank correlation coefﬁcient (rs[NDVI−LAI]) against (a) population density in the year 2000, (b) relative population change from 1980 to 2000, (c) cropping inten- sity, and (d) pasture intensity. Dashed grey lines indicate thresh- olds of statistical signiﬁcance for data-model comparison (α=0.05). Solid line in (d) indicates regression for Spearman rank correlation 0 100 200 300 400 −1 −0.5 0 0.5 1 Population density year 2000 (pers. / km2) rs[NDVI−LAI] a) −50 0 50 −1 −0.5 0 0.5 1 Rel. population change 1980−2000 (%) rs[NDVI−LAI] b) 0 0.2 0.4 0.6 0.8 1 −1 −0.5 0 0.5 1 Cropping intensity rs[NDVI−LAI] c) 0 0.2 0.4 0.6 0.8 1 −1 −0.5 0 0.5 1 Pasture intensity rs[NDVI−LAI] d) Fig. 6. Plots of Spearman rank correlation coefﬁcient (rs[NDVI−LAI]) against (a) population density in the year 2000, (b) relative population change from 1980 to 2000, (c) cropping inten- sity, and (d) pasture intensity. Dashed grey lines indicate thresh- olds of statistical signiﬁcance for data-model comparison (α=0.05). J. W. Seaquist et al.: Disentangling the effects on Sahel vegetation dynamics The results did not vary appreciably for neighbouring popu- lation growth thresholds, either higher or lower. 3.2 Relationship between rs[NDVI−LAI] and population density and land use Geographically, rs[NDVI−LAI] is fair-to-strong for a rela- tively contiguous band stretching from the coastal area of Northern Senegal, Southern Mauritania, Central Mali, and Western Niger (Fig. 4a). Agreement drops off in Chad before picking up again throughout Central Sudan and into Eritrea. 542 cells (0.5◦) show statistically signiﬁcant positive corre- lations, corresponding to 30.4% of the total land area (mean rs[NDVI−LAI]=0.58, α=0.05, threshold rs[NDVI−LAI] for statis- tical signiﬁcance=0.44). A large patch of strong disagree- ment is found in Northern Niger, with smaller patches in Southwest Mali and Ethiopia. 61 cells show statistically sig- There is no strong association between rs[NDVI−LAI] and ei- ther total population (Fig. 6a), relative changes in population density between 1980 and 2000 (Fig. 6b), or cropping in- tensity (Fig. 6c); Pearson correlation coefﬁcients r, <0.12 (i.e. less than 2% of the variance explained). But for ar- eas experiencing depopulation, data-model agreement is al- ways positive (Fig. 6b). The mean population density was 17.5 persons km−2 with a maximum of 994.1 persons km−2. Though not reproduced here, a scatter-plot of data-model www.biogeosciences.net/6/469/2009/ Biogeosciences, 6, 469–477, 2009 J. W. Seaquist et al.: Disentangling the effects on Sahel vegetation dynamics 4 Discussion Furthermore, the NDVI is fairly sensitive to vegetation changes since NDVI values are below the point of saturation for large parts of the growing season. Systematic shifts due to orbital drift and sensor differences during the observation period are concerns but a recent analysis showed that trends in a related AVHRR data product (NOAA/NASA Pathﬁnder AVHRR Land Data Set) were not affected by solar zenith angle variations (Lind- str¨om et al., 2006). Detecting trends over short time peri- ods using generally noisy satellite sensor data with ordinary least-squares regression is problematic due to the low num- ber of observations, but a spatio-temporal statistical model that more efﬁciently analyzes the trends has been tested over the Sahel and generates homogeneous areas of strong and statistically signiﬁcant areas of change (Bolin et al., 2008). Finally, locust swarms are a possible reason for reduced data-model agreement in some areas but the effect is likely marginal. Tratalos and Cheke (2006) emphasize that locusts are usually found for areas receiving above normal rainfall and are associated with good harvests. Moreover, such oc- currences are likely to have only short-term impacts in indi- vidual years (Hermann et al., 2005). Our analysis covers the years 1982–2002, a period for which rainfall appears to have been the main driver of change in the Sahel. It is conceivable that the impact that people have on vegetation dynamics will become greater in the coming decades as the landscape is transformed by human activities. For example Taylor et al. (2002), used a detailed land use model to project accelerating change in the composition of land cover between the mid-1990s and 2015 based on shift- ing demographics, extensiﬁcation, livestock population, etc. They also showed that by 2015, such changes in land cover could lead to a reduction in rainfall by almost 9% (this esti- mate does not take into consideration potential magniﬁcation of this reduction to feedbacks between the land surface and atmosphere). Our experimental design carries with it the assumption that random or negative data-model agreement would be at- tributed to the direct impact that people or animals have had on the landscape, but other factors may also be important in certain areas. For example, low data-model agreement could be caused by poor performance of the vegetation model and error in model input in terms of soil texture and climate data. 4 Discussion Demographic and agricultural pressures in the Sahel were largely unable to account for differences between simulated and observed vegetation dynamics for the period 1982–2002. However, we did identify a positive relationship between data-model agreement and pasture intensity at the Sahel- wide level (Fig. 6d). As the model predicts potential natural vegetation dynamics (without animal grazing and human in- It is important to keep in mind that human modiﬁcation of vegetation over large areas occurs within a social context and can be the result of a myriad of decisions and behaviours that may either act to suppress or augment vegetation amount and its variability (Turner, 2003). If consistent vegetation Biogeosciences, 6, 469–477, 2009 www.biogeosciences.net/6/469/2009/ 475 J. W. Seaquist et al.: Disentangling the effects on Sahel vegetation dynamics 2002). The situation is compounded by scattered mountain ranges where a larger number of climate stations would be necessary to derive reliable gridded data. However, the good agreement between modelled and observed trends at the ag- gregated level (Fig. 2) and across large areas of the study area (Fig. 3b), and various tests of the LPJ model in general (see model description) suggest that LPJ is capable of reproduc- ing changes in vegetation greenness in the Sahel when driven with the CRU05 climate data. Nevertheless it is impossible to quantify this potential source of error. response to human modiﬁcation is dependent on the local context, then relationships between data-model agreement and human activity measures might be embedded within the scatter-plots shown in Fig. 6a–d. We provisionally tested this hypothesis by focusing on relationships for an area of large, positive data-model agreement extending from West- ern Mali to Western Niger (n=190), and an area of random- to-negative data-model agreement stretching from Eastern Niger to Eastern Chad (n=196). No appreciable differences in the strength of the relationships were observed comparing them with Fig. 6a–d, except for the Mali area, which showed no relationship between data-model agreement and pasture. This is not surprising as the bulk of the rs[NDVI−LAI] values was >0.50. In contrast, the bulk of the rs[NDVI−LAI] values in the low agreement zone in Chad occupied a much wider range. q y p Regarding satellite data, NDVI observations from NOAA are affected by various sources of noise, but observations over the Sahel are less affected by cloud contamination than more humid areas (Chappell et al., 2001). 4 Discussion Soils in the LPJ-DGVM are derived from Food and Agri- culture Administration’s Soil Map of the World. The soils are divided into two layers of constant depth and textures are assigned according to Zobler (1986). For a complete descrip- tion of the hydrology, see Sitch et al. (2003). The Soil Map of the World is a highly generalized product where vector poly- gons are used for mapping soil attributes. Therefore the true heterogeneity of soil attributes is not captured. The absolute values of some of the hydrological output (and therefore as- sociated vegetation predictions) may be inaccurate, but we expect ﬂuctuations in such variables are to be more accu- rate. As we have applied a parametric measure of data-model agreement in the temporal domain (that would the effect of non-linearity, for example) we would expect to minimize this error. The climate data might be a particular source of error in central and Northern Niger and Northern and Central Chad where low or negative data-model agreement coincides with a relatively low number of precipitation stations (New et al., y We have conducted this study using highly generalized data sets with respect to information content and spatio- temporal resolution. Though the AL2000 is the most robust, comprehensive, widely available characterization of agricul- tural activity at a global extent, it summarizes agricultural status for roughly the year 2000 (near the end of the study period), while our study considers vegetation dynamics for the period 1982 to 2002 (Ramankutty et al., 2008). Factors that may cause land degradation such as shortening of fal- low cycles, decreased fertilizer, and limited herd mobility in times of drought, etc. are not captured. We have also aggre- gated cell size to 0.5◦in order to be able to compare observed and simulated vegetation dynamics which resulted in a loss of spatial ﬁdelity. Additionally, the deﬁnition of pasture and cropping in the AL2000 product follows the deﬁnition from the Food and Agricultural Organization, but there has been confusion regarding the characterization of both crop- lands and pasture, especially in census data (Ramankutty et al., 2008). For example, areas designated as cropland may actually be fallow for long periods of time. Much of the Biogeosciences, 6, 469–477, 2009 www.biogeosciences.net/6/469/2009/ 476 J. W. Seaquist et al.: Disentangling the effects on Sahel vegetation dynamics sibility of a greater human inﬂuence on vegetation dynamics over the coming decades with changing land use. References Behnke, R. H. and Scoones, I.: Rethinking range ecology implica- tions for range management in Africa, edited by: Behnke, R. H., Scoones, I., and Kerven, C., Range Ecology at Disequilibrium, New Models of Natural Variability and Pastoral Adaptation in African Savannas, Overseas Development Institute, London, 1– 30, 1993. In summary, there exist a number of potential sources of error in our analyses, but the consistency and robustness of our results suggest that the general conclusions of our study hold true. Bolin, D., Lindstr¨om, J., Eklundh, L., and Lindgren, F.: Fast es- timation of spatially dependent temporal vegetation trends us- ing Gaussian Markov Random Fields, Comput. Stat. Data An., doi:10.1016/j.csda.2008.09.017, in press, 2008. Chappell, A., Seaquist, J. W., and Eklundh, L.: Improving the es- timation of noise from NOAA AVHRR NDVI for Africa using geostatistics, Int. J. Remote Sens., 22, 1067–1080, 2001. 4 Discussion agricultural land is also multifunctional, though it has not been possible to incorporate this type of information into these data products. Regarding population data, they have not been evaluated thoroughly for Africa, though evaluation of the Asia Population Distribution Database (derived using the same methods) shows good agreement for relatively ho- mogeneous regions, but agreement drops where population distribution is scattered (Nelson, 2004). Acknowledgements. This work was partly funded by a Discovery Grant to J. W. Seaquist from the Natural Sciences and Engineering Research Council of Canada. T. Hickler acknowledges funding from the European Commission’s 6th Framework Programme ENSEMBLES (contract number GOCE-CT-2003-505539) while L. Eklundh and J. Ard¨o were supported by the Swedish National Space Board (contracts 87/07 and 120/06 respectively). We thank C. J. Tucker and co-investigators at the NASA/GSFC Global Inven- tory Modeling and Mapping Studies (GIMMS) group for providing access to the AVHRR NDVI data. The CRU05 climate data were supplied courtesy of the Climatic Research Unit, University of East Anglia. We also thank B. Smith for providing the LPJ code and P. Miller for formatting the climate input data. Finally, we are indebted to E. Lambin and N. Zeng for their valuable insights and comments. LAI and NDVI are widely used variables for characteriz- ing vegetation structure, which are also highly relevant for vegetation functioning (e.g. photosynthesis, NPP, transpira- tion and interception). But some aspects of vegetation dy- namics might not be covered by these variables. Structurally very different vegetation types, such as grassland, shrubland or savanna, can have similar LAI and NDVI, and there can be substantial turnover of species without any effects on veg- etation structure. For example, new plant species gaining a foothold through agricultural extensiﬁcation or land aban- donment could respond in a similar way to rainfall than the vegetation they are replacing, and subtle differences in veg- etation response may not be able to be picked up using our methodology at this level of spatial aggregation. Along simi- lar lines, an invasion of locally exotic plant species may also result in little or no change in peak growing-season biomass levels (Evans and Geerken, 2004). However, if humans were a major driver of vegetation dynamics in the Sahel, one would also expect some inﬂuence on vegetation greenness. Edited by: V. Brovkin Edited by: V. Brovkin 5 Conclusions W., Olsson, L., Ard¨o, J., and Eklundh, L.: Broad-scale increase in NPP quantiﬁed for the African Sahel 1982–1999, Int. J. Remote Sens., 27, 5115–5122, 2006. Hickler, T., Eklundh, L., Seaquist, J. W., Smith, B., Ard¨o, J., Ols- son, L., Sykes, M. T., and Sj¨ostr¨om, M.: Precipitation con- trols Sahel greening trend, Geophys. Res. Lett., 32, L21415, doi:10.1029/2005GL024370, 2005. Sellers, P. J.: Vegetation canopy spectral reﬂectance and biophysi- cal processes, edited by: Asrar, G., Theory and Applications of Optical Remote Sensing, Wiley, New York, 297–335, 1989. Hickler, T., Prentice, I. C., Smith, B., Sykes, M. T., and Zaehle, S.: Implementing plant hydraulic architecture within the LPJ Dynamic Global Vegetation Model, Global Ecol. Biogeogr., 15, 567–577, 2006. Sitch, S., Smith, B., Prentice, I. C., Arneth, A., Bondeau, A., Cramer, W., Kaplan, J., Levis, S., Lucht, W., Sykes, M., Thon- icke, K., and Venevsky, S.: Evaluation of ecosystem dynamics, plant geography and terrestrial carbon cycling in the LPJ Dy- namic Global Vegetation Model, Glob. Change Biol., 9, 161– 185, 2003. Holben, B. N.: Characteristics of maximum-value composite im- ages from temporal AVHRR data, Int. J. Remote Sens., 7, 1417– 1437, 1986. Hulme, M.: Climatic perspectives on Sahelian desiccation, Global Environ. Chang., 11, 19–29, 2001. Smith, B., Prentice, I. C., and Sykes, M. T.: Representation of vege- tation dynamics in modelling of terrestrial ecosystems: compar- ing two contrasting approaches within European climate space, Global Ecol. Biogeogr., 10, 621–637, 2001. Lindstr¨om, J., Eklundh, L., Holst, J., and Holst, U.: Inﬂuence of solar zenith angles on observed trends in the NOAA/NASA 8 km Pathﬁnder normalized difference vegetation index over the African Sahel, Int. J. Remote Sens., 27, 1973–1991, 2006. Taylor, C. M., Lambin, E. F., Stephenne, N., Harding, R. J., and Essery, R. L. H.: The inﬂuence of land use change on climate in the Sahel, J. Climate, 15, 3615–3629. 2002. Li, J., Lewis, J., Rowland, J., Tappan, G., and Tieszen, L. L.: Evalu- ation of land peformance in Senegal using multi-temporal NDVI and rainfall series, J. Arid Environ., 59, 463–480, 2004. Tratalos, J. A. and Cheke, R. A.: Can NDVI GAC imagery be used to monitor desert locust breeding areas?, J. Arid Environ., 64, 342–356, 2006. Lucht, W., Prentice, I. C., Myneni, R. B., Sitch, S., Friedlingstein, P., Cramer, W., Bousquet, P., Buermann, W., and Smith, B.: Cli- matic control of the high-latitude vegetation greening trend and Pinatubo effect, Science, 296, 1687–1689, 2002. Tucker, C. 5 Conclusions J., Dregne, H. E., and Newcomb, W. W.: Expansion and contraction of the Sahara Desert from 1980 to 1990, Science, 253, 299–301, 1991. McGuire, A. D., Sitch, S., Clein, J. S., Dargaville, R., Esser, G., Fo- ley, J., Heimann, M., Joos, F., Kaplan, J., Kicklighter, D. W., Meier, R. A., Melillo, J. M., Moore, B., Prentice, I. C., Ra- mankutty, N., Reichenaum, T., Schloss, A., Tian, H., Williams, L. J., and Wittenberg, U.: Carbon balance of the terrestrial biosphere in the twentieth century: analyses of CO2, climate and land use effects with four process-based ecosystem models, Global Biogeochem. Cy., 15, 183–206, 2001. Tucker, C. J., Pinzon, J. E., Brown, M. E., Slayback, D. A., Pak, E. W., Mahoney, R., Vermote, E. F., and El Saleous, N.: An extended AVHRR 8 km NDVI dataset compatible with MODIS and SPOT vegetation NDVI data, Int. J. Remote Sens., 26, 4485– 4498, 2005. Turner, M. D.: Methodological reﬂections on the use of remote sensing and geographic information science in human ecologi- cal research, Hum. Ecol., 31, 255–279, 2003. McGrew Jr., J. C. and Monroe, C. B.: An Introduction to Statis- tical Problem Solving in Geography, McGraw Hill, New York, 254 pp., 2000. Wang, G. and Elthahir, E. A. B.: Role of vegetation dynamics in en- hancing the low frequency variability of the Sahel rainfall, Water Resour. Res., 36, 1013–1021, 2000. Mitchell, T. D. and Jones, P. D.: An improved method of construct- ing a database of monthly climate observations and associated high-resolution grids, Int. J. Climatol., 25, 693–712, 2005. Wessels, K. J., Prince, S. D., Malherbe, J., Small, J., Frost, P. E., and VanZyl, D.: Can human-induced land degradation be distin- guished from the effects of rainfall variability? A case study in South Africa, J. Arid Environ., 68, 271–297, 2007. Nelson, A.: African Population Database Documentation, http://na. unep.net/globalpop/africa/Africa index.html, 2004. Zaehle, S., Smith, B., and Hatterman, F.: Effects of pa- rameter uncertainties on the modelling of terrestrial bio- sphere dynamics, Global Biogeochem. Cy., 19, GB3020, doi:10.1029/2004GB002395, 2005. New, M., Lister, D., Hulme, M., and Makin, I.: A high-resolution data set of surface climate over global land areas, Climate Res., 21, 1–25, 2002. Olsson, L., Eklundh, L., and Ard¨o, J.: A recent greening of the Sahel – trends, patterns and potential causes, J. Arid Environ., 63, 556–566, 2005. Zeng, N., Neelin, J. D., Lau, K. M., and Tucker, C. 5 Conclusions Eklundh, L. and Olsson, L.: Vegetation index trends for the African Sahel, Geophys. Res. Lett., 30, 1430–1433, 2003. We tested the hypothesis that people have had a measurable impact on vegetation dynamics across the Sahel for the pe- riod 1982–2002. To the best of our knowledge, this is the ﬁrst study to tackle this question at the Sahel-wide extent. Our methodology rests on the assumptions that low or neg- ative agreement between simulated and observed vegetation dynamics would be related to the impact that people have on vegetation. But our results indicate that there is gener- ally no relationship between data-model agreement and the best available estimates of population density and agricul- tural activity. However, a positive relationship was estab- lished between level of data-model agreement and pasture intensity, indicating that animal presence did not apprecia- bly affect vegetation dynamics in the region over the period. Either people have not had a signiﬁcant impact on vegeta- tion dynamics across the Sahel, or the identiﬁcation of a hu- man “footprint” is precluded by inconsistent or subtle vege- tation response to complex socio-environmental interactions, and/or data limitations. Finally, we do not exclude the pos- Evans, J. and Geerken, R.: Discrimination between climate and human-induced dryland degradation, J. Arid Environ., 57, 535– 554, 2004. Foley, J. A., Coe, M. T., Scheffer, M., and Wang, G.: Regime shifts in the Sahara and Sahel: interactions between ecological and cli- mate systems in northern Africa, Ecosystems, 6, 524–539, 2003. Gerten, D., Schabhoff, S., Haberlandt, U., Lucht, W., and Sitch, S.: Terrestrial vegetation and water balance – hydrological eval- uation of a dynamic global vegetation model, J. Hydrol., 286, 249–270, 2004. Hein, L. and de Ridder, N.: Desertiﬁcation in the Sahel: a reinter- pretation, Glob. Change Biol., 12, 751–758, 2006. Helld´en, U.: Desertiﬁcation: time for an assessment?, Ambio, 20, 372–383, 1991. Hermann, S. M., Anyamba, A., and Tucker, C. J.: Recent trends in vegetation dynamics in the African Sahel and their relationship to climate, Global Environ. Chang., 15, 394–404, 2005. Heumann, B. W., Seaquist, J. W., Eklundh, L., and J¨onsson, P.: AVHRR derived phenological change in the Sahel and Soudan, Africa, 1982–2005, Remote Sens. Environ., 108, 385–392, 2007. www.biogeosciences.net/6/469/2009/ Biogeosciences, 6, 469–477, 2009 www.biogeosciences.net/6/469/2009/ J. W. Seaquist et al.: Disentangling the effects on Sahel vegetation dynamics J. W. Seaquist et al.: Disentangling the effects on Sahel vegetation dynamics 477 Seaquist, J. 5 Conclusions J.: Enhancement of interdecadal climate variability in the Sahel by vegetation in- teraction, Science, 286, 1537–1540, 1999. Prince, S. D.: A model of regional primary production for use with coarse resolution satellite data, Int. J. Remote Sens., 12, 1313– 1330. Zobler, L.: A World Soil File for Global Climate Modelling, NASA Technical Memorandum 87802, NASA Goddard Institute for Space Studies, New York, New York, USA, 1986. Prince, S. D., Wessels, K. J., Tucker, C. J., and Nicholson, S. E.: Desertiﬁcation in the Sahel: a reinterpretation of an interpreta- tion, Glob. Change Biol., 13, 1308–1313, 2007. Ramankutty, N., Evan, A. T., Monfreda, C., and Foley, J. A.: 1. Geographic distribution of global agricultural lands in the year 2000, Global Biogeochem. Cy., 22, GB1003, doi:10.1029/2007GB002952, 2008. Biogeosciences, 6, 469–477, 2009 www.biogeosciences.net/6/469/2009/
https://openalex.org/W2569614730	https://aacr.figshare.com/articles/journal_contribution/Supplementary_Table_S3_from_APC_C_Dysfunction_Limits_Excessive_Cancer_Chromosomal_Instability/22531350/1/files/39994527.pdf	English	null	APC/C Dysfunction Limits Excessive Cancer Chromosomal Instability	Cancer discovery	2,017	cc-by	214	Supplementary Table S3 Supplementary Table S3 Supplementary Table S3 Supplementary Table S3 TCGA STUDY NAME No. of cases altered Total no. of cases Percentage altered Adrenocortical Carcinoma 9 90 10% Bladder Urothelial Carcinoma 17 130 13.10% Brain Lower Grade Glioma 5 286 1.70% Breast Invasive Carcinoma 29 982 3% Cervical Squamous Cell Carcinoma and Endocervical Adenocarcinoma 16 194 8.20% Cholangiocarcinoma 8 35 22.90% Colorectal Adenocarcinoma 27 223 12.10% Esophageal Carcinoma 28 185 15.10% Glioblastoma Multiforme 13 290 4.50% Head and Neck Squamous Cell Carcinoma 35 512 6.80% Kidney Chromophobe 12 66 18.20% Kidney Renal Clear Cell Carcinoma 26 451 5.80% Kidney Renal Papillary Cell Carcinoma 12 282 4.30% Liver Hepatocellular Carcinoma 20 373 5.40% Lung Adenocarcinoma 26 230 11.30% Lung Squamous Cell Carcinoma 18 177 10.20% Lymphoid Neoplasm Diffuse Large B-cell Lymphoma 6 48 12.50% Ovarian Serous Cystadenocarcinoma 10 316 3.20% Pancreatic Adenocarcinoma 6 150 4% Pheochromocytoma and Paraganglioma 6 184 3.30% Prostate Adenocarcinoma 26 499 5.20% Sarcoma 6 247 2.40% Skin Cutaneous Melanoma 69 368 18.80% Stomach Adenocarcinoma 58 395 14.70% Testicular Germ Cell Cancer 15 155 9.70% Thymoma 2 123 1.60% Thyroid Carcinoma 3 405 0.70% Uterine Carcinosarcoma 3 57 5.30% Uterine Corpus Endometrial Carcinoma 35 248 14.10% Uveal Melanoma 1 80 1.30% average 8% median 6% range 0.7-22.9% standard deviation 0.058968568
https://openalex.org/W2916520161	http://www.scielo.br/pdf/coluna/v18n1/2177-014X-coluna-18-01-0047.pdf	English	null	IMPACT OF THE MIS-TLIF TECHNIQUE ON THE SPINOPELVIC PARAMETERS OF PATIENTS WITH LUMBAR ARTHRODESIS	Coluna/Columna	2,019	cc-by	4,946	IMPACTO DE LA TÉCNICA MIS-TLIF SOBRE LOS PARÁMETROS ESPINOPÉLVICOS DE PACIENTES CON ARTRODESIS LUMBAR Cristiano Magalhães Menezes,1,2 Leonardo Fernandes Aguiar,1,2 André de Oliveira Arruda,1 Rodrigo Vieira Cardoso,1 Germano Senna Oliveira do Valle,2 Rodrigo Souza Lima,1 Mário Leite Bringel,1 Felipe Miranda Mendonça Fernandes,1 Joint Halley Guimbard Pérez1 1. Columna Instituto Dr. Cristiano Menezes, Belo Horizonte, MG, Brazil. 2. Spine Surgery Group of the Hospital Ortopédico Lifecenter, Belo Horizonte, MG, Brazil. 1. Columna Instituto Dr. Cristiano Menezes, Belo Horizonte, MG, Brazil. 2. Spine Surgery Group of the Hospital Ortopédico Lifecenter, Belo Horizonte, MG, Brazil. ABSTRACT Objective: To evaluate the influence of the MIS-TLIF technique on the spinopelvic parameters of patients submitted to lumbar arthrodesis up to three levels for the treatment of vertebral degenerative conditions without deformity. Methods: Retrospective radiographic evaluation of 52 patients submitted to the surgical treatment of lumbar arthrodesis using the MIS-TLIF technique in up to three levels. The spinopelvic parameters – pelvic incidence (PI), pelvic tilt(PT), lumbar lordosis (LL), segmental lordosis (Lseg), and the difference between lumbar lordosis and pelvic incidence (LL-PI mismatch) were analyzed in orthostatic lateral radiographs in the pre- and postoperative periods, with a minimum follow-up of 1 year. The patients were divided into three groups: PI <45°, PI between 45° and 55° and PI >55°. Results: Sixty-nine operated levels were evaluated in 15 patients with PI <45°, 19 with PI between 45° and 55° and 18 with PI >55°. The mean value of the pelvic incidence was 52.3° (± 11.5), lumbar lordosis 46.1° (pre)/45.6° (post); segmental lordosis 20.3° (pre)/20.6° (post); pelvic tilt 18.5° (pre)/18.2° (post); “mismatch” (PI-LL) 7° (pre)/ 6.6° (post), with no statistical difference among all parameters (p>0.05). Conclusions: The MIS-TLIF technique had no influence on postoperative spinopelvic parameters of patients undergoing lumbar arthrodesis surgery. Level of evidence: III. Retrospective comparative study. Keywords: Spinal curvatures; Spinal diseases; Spinal fusion; Arthrodesis. RESUMO Objetivos: Avaliar a influência da técnica MIS-TLIF nos parâmetros espinopélvicos de pacientes submetidos à artrodese lombar até três níveis para tratamento de condições degenerativas vertebrais sem deformidade. Métodos: Avaliação radiográfica retrospectiva de 52 pacientes submetidos a tratamento cirúrgico de artrodese lombar pela técnica MIS-TLIF em até três níveis. Os parâmetros espinopélvicos - incidência pélvica (PI), versão pélvica (PT), lordose lombar (LL), lordose segmentar (Lseg), diferença entre lordose lombar e incidência pélvica (“mismatch” PI-LL) foram analisados em radiografias em perfil ortostático no pré e pós-operatório, com segmento mínimo de 1 ano. Os pacientes foram divididos em três grupos: PI<45°; PI entre 45° e 55°e PI> 55°. Resultados: Foram avaliados 69 níveis operados - PI < 45°, 15 pacientes; PI entre 45° e 55°, 19 e PI > 55°, 18. O valor médio da incidência pélvica foi 52,3° (+\- 11,5), lordose lombar 46,1° (pré) / 45,6° (pós); lordose segmentar 20,3° (pré) / 20,6° (pós); rotação pélvica 18,5° (pré) / 18,2° (pós); “mismatch” (PI-LL) 7° (pré) / 6,6°(pós), sem diferença estatística entre todos os parâmetros (p>0,05). Conclusão: A técnica MIS-TLIF não apresentou influência nos parâmetros espinopélvicos no pós-operatório de pacientes submetidos à cirurgia de artrodese lombar. Nível de evidência III. Estudo retrospectivo comparativo. Descritores: Curvaturas da coluna vertebral; Doenças da coluna vertebral; Fusão vertebral; Artrodese. Original Article/Artigo Original/Artículo Original Original Article/Artigo Original/Artículo Original METHODS The same parameters were analyzed with the division of patients according to the pelvic incidence values. For the group of patients with PI < 45º (n = 15), a mean pelvic incidence of 40° (standard deviation: 3.40) was obtained. The other radiographic parameters are presented in Table 3. Following approval by the Institutional Review Board (no. 1.341.609), we conducted a retrospective, observational study, reviewing the medical records of patients who were submitted to spinal arthrodesis performed by a single surgeon from January 2008 to December 2012. p The comparison between the pre- and postoperative parameters can be confirmed according to the graphic representation in Figure 2. In the group of patients with PI between 45º and 55º (n = 19), the mean value of PI was 49.8° (standard deviation: 3.40). The other radiographic parameters are presented in Table 4. The comparison between the pre- and postoperative parameters can be confirmed according to the graphic representation in Figure 2. We included patients 18 years of age and above who underwent arthrodesis of from one to three levels and satisfied at least one of the following conditions: degenerative disc disease with or without disc herniation, canal stenosis, low-grade spondylolisthesis, and post-lami nectomy/discectomy syndrome. The minimum follow-up for inclusion was 12 months. Patients with traumatic conditions, tumors, and severe associated deformities, with possible surgical indications in which the de formity would be the primary justification for intervention, were excluded. The surgical technique used was described by Foley10,11 and also published previously in this journal.11-13 g g p p g In the group of patients with PI between 45º and 55º (n = 19), the mean value of PI was 49.8° (standard deviation: 3.40). The other radiographic parameters are presented in Table 4. Table 1. List of number of patients by level analyzed. Table 1. List of number of patients by level analyzed. Levels Number of patients evaluated L3-L4 (1 level) 1 L4-L5 (1 level) 21 L5-S1 (1 level) 15 L3-L5 (2 levels) 2 L4-S1 (2 levels) 11 L3-S1 (3 levels) 2 Table 1. List of number of patients by level analyzed. y p y j The surgical technique used was described by Foley10,11 and also published previously in this journal.11-13 INTRODUCTION In order to verify possible differences in the MIS-TLIF under di fferent initial conditions, the patients were divided into three groups by range of pelvic incidence, namely, less than 45º (low), between 45º and 55º (medium), and greater than 55° (high). The loss of lordosis and consequent change in the spinopel vic parameters are common in degenerative conditions that affect the lumbar spine.1-4 Their analysis becomes highly important when considering surgical intervention, especially in terms of arthrodesis planning. The restoration of these parameters is related to better clinical and radiological results, with higher rates of fusion and a decrease in the incidence of adjacent level disease.5 ( ) g ( g ) All the radiographic measurements were performed by an in dependent spine surgeon and dubious cases were evaluated by a third-party researcher. The conventional terminology used in the literature was maintained, considering lordotic curves as negative values and kyphotic curves as positive. j There are several lumbar arthrodesis techniques, each with di fferent possible approaches, such as anterior, lateral, and/or pos terior. Among the interbody arthrodeses, the transforaminal (TLIF) is the most widely used,6,7 and over the last decade, the minimally invasive technique (MIS-TLIF) has gained in popularity. Among its advantages, as compared to the traditional technique, less intrao perative bleeding, a shortened hospital stay, and early postopera tive recovery stand out, in addition to interesting and compelling cost-effectiveness indices.8,9 Despite this, there are controversies in the literature around the ability of this technique to maintain and/ or increase segmental lordosis, exercising potential influence on the final spinopelvic parameters and thus affecting the clinical outcome.5 RESULTS p p p g In this context, the objective of this study is to evaluate the im pact of the standard MIS-TLIF technique on the radiographic spi nopelvic parameters of patients with degenerative conditions as the therapeutic indication and who underwent surgical treatment for degenerative conditions of the lumbar spine at up to three levels. The medical records of 52 patients were evaluated, of whom 22 were men and 30 were women, with a mean age of 53.5 years (ranging from 27 to 82 years), totaling 69 operated levels. (Table 1) ( g g y ) g p ( ) The mean value of pelvic incidence was 52.3° (median 51.0 SD ± 11.5). The values obtained for the other radiographic parameters follow. (Table 2) Statistical analysis As regards the statistical tests, absolute and relative frequencies were used to describe the qualitative variables, while measures of central tendency, position, and dispersion were used to describe the quantitative variables. For the comparison of the spinopelvic indices between the MIS-TLIF subgroups divided by ranges of PI, the Student’s t test was used for the normally distributed indicators and the Mann-Whitney test for the non-normal variables. Normality was verifying using the Shapiro-Wilk test. RESUMEN Objetivo: Evaluar la influencia de la técnica MIS-TLIF sobre los parámetros espinopélvicos de pacientes sometidos a artrodesis lum bar de hasta tres niveles para el tratamiento de condiciones vertebrales degenerativas sin deformidad. Métodos: Evaluación radiográfica retrospectiva de 52 pacientes sometidos a tratamiento quirúrgico de artrodesis lumbar por la técnica MIS-TLIF en hasta tres niveles. Los parámetros espinopélvicos como incidencia pélvica (IP), inclinación pélvica (PT), lordosis lumbar (LL), lordosis segmentaria (Lseg) diferencia entre lordosis lumbar e incidencia pélvica (“mismatch” LL-) se analizaron en radiografías laterales ortostáticas en el pre y postoperatorio, con seguimiento mínimo de 1 año. Los pacientes se dividieron en tres grupos: IP < 45°, IP entre 45° y 55° e IP > 55°. Resultados: Se evaluaron 69 niveles operados en 15 pacientes con IP < 45°, 19 con IP entre 45° y 55° y 18 con IP > 55°. El valor promedio de la incidencia pélvica fue 52,3° (± 11,5), lordosis lumbar 46,1° (pre)/45,6° (post), lordosis segmentaria 20,3° (pre)/20,6° (post), rotación pélvica 18,5° (pre)/18,2° (post), “mismatch” IP-LL 7° (pre)/6,6° (post), sin diferencia estadística entre todos los parámetros (p > 0,05). Conclusiones: La técnica MIS-TLIF no influenció los parámetros espinopélvicos en el postoperatorio de pacientes sometidos a cirugía de artrodesis lumbar. Nivel de evidencia: III. Estudio retrospectivo comparativo. Curvaturas de la columna vertebral; Enfermedades de la columna vertebral; Fusión vertebral; Artrodesis. Descriptores: Curvaturas de la columna vertebral; Enfermedades de la columna vertebral; Fusión vertebral; Artrodesis. Descriptores: Curvaturas de la columna vertebral; Enfermedades de la columna vertebral; Fusión vertebral; Artrodesis. Received on 11/28/2017 accepted on 07/25/2018 http://dx.doi.org/10.1590/S1808-185120191801188450 Study conducted at the Hospital Lifecenter, Belo Horizonte, MG, The Hospital Ortopédico BH, Belo Horizonte - MG, The Hospital Vera Cruz, Belo Horizonte, MG, and the Instituto Columna, Belo Horizonte, MG, Brasil. Correspondence: Instituto Columna, Rua Conde de Linhares, 278 –Cidade Jardim, Belo Horizonte, MG, 30380-030. cristiano@columnainstituto.com Coluna/Columna. 2019;18(1):47-50 Study conducted at the Hospital Lifecenter, Belo Horizonte, MG, The Hospital Ortopédico BH, Belo Horizonte - MG, The Hospital Vera Cruz, Belo Horizonte MG Brasil Study conducted at the Hospital Lifecenter, Belo Horizonte, MG, The Hospital Ortopédico BH, Belo Horizonte - MG, The Hospital Vera Cruz, Belo Horizonte, MG, and the Instituto Columna, Belo Horizonte, MG, Brasil. , Correspondence: Instituto Columna, Rua Conde de Linhares, 278 –Cidade Jardim, Belo Horizonte, MG, 30380-030. cristiano@columnainstituto.com Received on 11/28/2017 accepted on 07/25/2018 48 DISCUSSION In contrast, several authors have shown that a significant amount of endpoint segmental lordosis can be achieved placing the cage anteriorly and performing a bilateral facetectomy,14-16 while still maintaining the TLIF technique. Jong-Tae et al.19 analyzed the influence of straight and lordotic cages (interbody spacer) in 68 patients, concluding that the lordotic form is superior in relation to the increase in segmental lordosis and in maintaining disc height, providing better spinopelvic parameter results. In this study, the MIS-TLIF technique was not capable of significantly altering the parameters evaluated, effecting neither gains nor losses in lordosis following surgery. 1 Figure 2. Comparison of pre- and postoperative spinopelvic parameters, PI < 45°. Table 4. Radiographic parameters for the population with PI between 45 and 55º. Radiographic parameter Pre value (º) (n considered / standard deviation) Post value (º) (n considered / standard deviation) P value LL 43.5 (n=15 / 5.8) 42.1 (n=19 / 8.5) > 0.05 Lseg 18 (n=34 / 6.5) 19.2 (n=47 / 6.8) > 0.05 PT 17.5 (n=14 / 4.7) 17.8 (n=18 / 6.2) > 0.05 Mismatch (PI-LL) 6.2 (n=15 / 7.1) 7.7 (n=19 / 9.2) > 0.05 n* considered the total of the levels evaluated. Table 4. Radiographic parameters for the population with PI between 45 and 55º. Table 4. Radiographic parameters for the population with PI between 45 and 55º. Radiographic parameter Pre value (º) (n considered / standard deviation) Post value (º) (n considered / standard deviation) P value LL 43.5 (n=15 / 5.8) 42.1 (n=19 / 8.5) > 0.05 Lseg 18 (n=34 / 6.5) 19.2 (n=47 / 6.8) > 0.05 PT 17.5 (n=14 / 4.7) 17.8 (n=18 / 6.2) > 0.05 Mismatch (PI-LL) 6.2 (n=15 / 7.1) 7.7 (n=19 / 9.2) > 0.05 n* considered the total of the levels evaluated. Figure 3. Comparison of the pre- and postoperative spinopelvic parameters, PI between 45° and 55º. LL Lseg PT Mismatch 50 40 30 20 10 0 43.5 42.1 18 19,2 17,6 17,9 6,2 7,7 Spinopelvic parameters PI - 45 to 55° Preoperative Postoperative Figure 3. Comparison of the pre- and postoperative spinopelvic parameters, PI between 45° and 55º. LL Lseg PT Mismatch 50 40 30 20 10 0 43.5 42.1 18 19,2 17,6 17,9 6,2 7,7 Spinopelvic parameters PI - 45 to 55° Preoperative Postoperative Figure 3. Comparison of the pre- and postoperative spinopelvic parameters, PI between 45° and 55º. Radiographical analysis Panoramic radiographs of the spine and/or the lumbosacral spine with the inclusion of the femoral heads in standing anterior -posterior (AP) and lateral (L) views were analyzed. The following spinopelvic parameters were evaluated: pelvic incidence (PI), pelvic tilt (PT), lumbar lordosis (LL) (Figure 1), seg mental lordosis (Lseg), and the difference between pelvic incidence and lumbar lordosis (mismatch = PI-LL). Table 2. Radiographic parameters for the total population. Radiographic parameter Pre value (º) (n considered / standard deviation) Post value (º) (n considered / standard deviation) P value LL 46.1 (n = 35 / 10.2) 45.6 (n = 52 / 10.1) > 0.05 Lseg 20.3 (n* = 47 / 18.8) 20.6 (n* = 69 / 7.5) > 0.05 PT 18.5 (n = 34 / 7.3) 18.2 (n = 51 / 8.6) > 0.05 Mismatch (PI-LL) 7 (n = 35 / 10.7) 6.6 (n = 52 / 10.7 > 0.05 * considered the total number of levels evaluated. Table 2. Radiographic parameters for the total population. Figure 1. Schematic representation of the spinopelvic parameters. SS - sacral slope, PI - pelvic incidence, PT- pelvic tilt. SS PI PT Table 3. Radiographic parameters for the population with PI<45º. Radiographic parameter Pre value (º) (n considered / standard deviation) Post value (º) (n considered / standard deviation) P value LL 42.5 (n=9 / 9.5) 40.8 (n=15 / 9.7) > 0.05 Lseg 20.3 (n=12 / 6.3) 19.7 (n=21 / 8.5) > 0.05 PT 12.8 (n=9 / 5.2) 12.1 (n=15 / 6.1) > 0.05 Mismatch (PI-LL) -1.5 (n=9 / 10.6) -0.8 (n=15 / 10.3) > 0.05 n* considered the total number of levels evaluated. Table 3. Radiographic parameters for the population with PI<45º. Figure 1. Schematic representation of the spinopelvic parameters. Figure 1. Schematic representation of the spinopelvic parameters. IMPACT OF THE MIS-TLIF TECHNIQUE ON THE SPINOPELVIC PARAMETERS OF PATIENTS WITH LUMBAR ARTHRODESIS 49 The comparison between the pre- and postoperative values can be seen in Figure 3. Figure 4. Comparison of the pre- and postoperative spinopelvic parameters, PI > 55°. LL Lseg PT Mismatch 60 50 40 30 20 10 0 52.5 53.6 23.3 23.7 24.5 23.8 15.2 11.6 Spinopelvic parameters PI > 55° Preoperative Postoperative In the group of patients with PI > 55° (n = 18), we obtained a mean pelvic incidence value of 65.1° (standard deviation: 7.69). The other radiographic parameters are presented in Table 5 below. Radiographical analysis g p p p The comparison between the pre- and postoperative values can be seen in the following graph. (Figure 4) DISCUSSION The TLIF fusion technique, first described in 1998 by Harms and widely used since then, provides anterior support for the spine, posterior-lateral stabilization with pedicular instrumentation, direct visualization of the space to be approached, and decompression of Figure 4. Comparison of the pre- and postoperative spinopelvic parameters, PI > 55°. Figure 2. Comparison of pre- and postoperative spinopelvic parameters, PI < 45°. LL Lseg PT Mismatch 50 40 30 20 10 0 -10 42.3 40.8 20.3 19.8 12.9 12.1 -1.6 -0.8 Spinopelvic parameters - PI < 45° Preoperative Postoperative Spinopelvic parameters - PI < 45° the nerve roots.14-17 Given these considerations, TLIF is recognized as a viable and feasible procedure for various degenerative spinal diseases. Degenerative spinal conditions are for the most part as sociated with segmental hypolordosis, mainly at levels L4-L5 and L5-S1, generating the use of compensatory mechanisms such as an increase in pelvic tilt (PT).15-19 Knowing that some patients can not tolerate the smallest loss of lordosis when submitted to lumbar arthrodesis, the preoperative analysis of the spinopelvic parameters becomes essential, helping to determine the correction required and to choose the most appropriate surgical strategy. Minimally invasive TLIF was first described by Foley et al.10 and, since then, the use of the technique has grown exponentially. With the support of scientific literature, the less invasive technique offers results comparable to those of the traditional technique in terms of fusion rates and clinical results, with significant advantages in postoperative pain control, making possible early rehabilitation, shortened hospital stays, and lower complication rates.20-23 However, there are contradictions in the literature around the capacity of posterior arthrodesis techniques to correct the spinopelvic parameters, and even more so in relation to minimally invasive surgeries. In general, literature data are conflicting in relation to the capacity for the TLIF technique to both restore and maintain segmental lordosis. Hsieh et al.12 reported the mean loss of segmental lordosis following TLIF and attributed this lack of im provement to the difficulty in placing the interbody cage as anteriorly as possible, in addition to the presence of the intact contralateral facet joint, which prevents posterior compression. Kwon et al.,15 in turn, reported the occurrence of TLIF-induced segmental kyphosis, justified by the more posterior placement of the cage in the disc interspace. DISCUSSION In contrast, several authors have shown that a significant amount of endpoint segmental lordosis can be achieved placing the cage anteriorly and performing a bilateral facetectomy,14-16 while still maintaining the TLIF technique. Jong-Tae et al.19 analyzed the influence of straight and lordotic cages (interbody spacer) in 68 patients, concluding that the lordotic form is superior in relation to the increase in segmental lordosis and in maintaining disc height, providing better spinopelvic parameter results. In this study, the MIS-TLIF technique was not capable of significantly altering the parameters evaluated, effecting neither gains nor losses in lordosis following surgery. the nerve roots.14-17 Given these considerations, TLIF is recognized as a viable and feasible procedure for various degenerative spinal diseases. Degenerative spinal conditions are for the most part as sociated with segmental hypolordosis, mainly at levels L4-L5 and L5-S1, generating the use of compensatory mechanisms such as an increase in pelvic tilt (PT).15-19 Knowing that some patients can not tolerate the smallest loss of lordosis when submitted to lumbar arthrodesis, the preoperative analysis of the spinopelvic parameters becomes essential, helping to determine the correction required and to choose the most appropriate surgical strategy. Minimally invasive TLIF was first described by Foley et al.10 and, since then, the use of the technique has grown exponentially. With the support of scientific literature, the less invasive technique offers results comparable to those of the traditional technique in terms of fusion rates and clinical results, with significant advantages in postoperative pain control, making possible early rehabilitation, shortened hospital stays, and lower complication rates.20-23 However, there are contradictions in the literature around the capacity of posterior arthrodesis techniques to correct the spinopelvic parameters, and even more so in relation to minimally invasive surgeries. In general, literature data are conflicting in relation to the capacity for the TLIF technique to both restore and maintain segmental lordosis. Hsieh et al.12 reported the mean loss of segmental lordosis following TLIF and attributed this lack of im provement to the difficulty in placing the interbody cage as anteriorly as possible, in addition to the presence of the intact contralateral facet joint, which prevents posterior compression. Kwon et al.,15 in turn, reported the occurrence of TLIF-induced segmental kyphosis, justified by the more posterior placement of the cage in the disc interspace. DISCUSSION Our study, similarly to other publications5,17 shows that MIS-TLIF does not have the ability to affect the angles of the spine and pelvis. Lafage et al.23 reported that the increase in PT correlates directly to worse functional results in patients with spinal deformity and revea led that an increase in PT is directly linked to pain following lumbar arthrodesis. Khoi D. Than et al.20 showed that better clinical results are associated with the restoration of the spinopelvic parameters, with greatest significance when the SVA (sagittal vertical axis) < 5 cm and the PI-LL mismatch is less than 10. Table 5. Radiographic parameters for the population with PI>55º. Radiographic parameter Pre value (º) (n considered / standard deviation) Post value (º) (n considered / standard deviation) P value LL 52.5 (n=11 / 12.8) 53.5 (n=18 / 9.0) > 0.05 Lseg 23.1 (n=13 / 6.9) 23.6 (n=22 / 8.1) > 0.05 PT 24.4 (n=11 / 7.4) 23.7 (n=18 / 9.1) > 0.05 Mismatch (PI-LL) 15.1 (n=11 / 9.3) 11.6 (n=18 / 9.4) > 0.05 n* considered the total of levels evaluated. Table 5. Radiographic parameters for the population with PI>55º. To date, few studies have evaluated the radiological results following TLIF and/or MIS-TLIF procedures, with emphasis on 50 Statistically, MIS-TLIF was not able to restore lordosis or even cause curve loss in any of the different PI range groups. postoperative sagittal alignment and on spinopelvic parameters. In their case series of 45 patients, Mourad Ould-Slimane et al.5 showed that TLIF was capable of acting locally (effect on the segment), increasing disc height and segmental lordosis, with a correction of pelvic rotation, but that there was no significant modification, i.e., improvement in sagittal alignment, at the global level. Thus, in cases of severe deformity associated with important sagittal changes, the use of other techniques to achieve correction should be considered. postoperative sagittal alignment and on spinopelvic parameters. In their case series of 45 patients, Mourad Ould-Slimane et al.5 showed that TLIF was capable of acting locally (effect on the segment), increasing disc height and segmental lordosis, with a correction of pelvic rotation, but that there was no significant modification, i.e., improvement in sagittal alignment, at the global level. Thus, in cases of severe deformity associated with important sagittal changes, the use of other techniques to achieve correction should be considered. REFERENCES Harms J, Rolinger H. A one-stage procedure in operative treatment of spondylolistheses: dorsal traction-reposition and anterior fusion [in German]. Z Orthop Ihre Grenzgeb. 1982;120(3):343–7. eral transforaminal lumbar interbody fusion. J Neurosurg Spine. 2012;17 7. Rosenberg WS, Mummaneni PV. Transforaminal lumbar interbody fusion: technique, com plications, and early results. Neurosurgery. 2001;48(3):569–74. 19. Jong-Tae K, Myung-Hoon S, Ho-Jin L, Du-Yong C. Restoration of lumbopelvic sagi tal alignmentand its maintenance following transforaminal lumbar interbody fusion (TLIF): comparison between straight type versus curviliner type cage. Eur Spine J. 2015;24(11):2588–96. plications, and early results. Neurosurgery. 2001;48(3):569–74. 8. Foley KT, Lefkowitz MA. Advances in minimally invasive spine surgery. Clin Neurosurg. 2002;49:499-517. 2002;49:499-517. 9. Goldstein CL, Phillips FM, Rampersaud YR. Comparative Effectiveness and Economic Evaluations of Open Versus Minimally Invasive Posterior or Transforaminal Lumbar Interbody Fusion. A Systematic Review. Spine (Phila Pa 1976). 2016;41 Suppl 8):74–89. 10. Foley KT, Holly LT, Schwender JD. Minimally Invasive Lumbar Fusion. Spine. 2003;28:S26–35. 11. Menezes CM, Junior MAF, Falcon RS, Oliveira DA, Freire SG. Experiência inicial com a técnica de artrodese lombar minimamente invasiva por via transforaminal (MIS TLIF). Coluna/Columna. 2007;6(3):141-8. 20. Than KD, Park P, Fu KM, Nguyen S, Wang MY, Chou D, et al. Clinical and radiograph ic parameters associated with best versus worst clinical outcomes in minimally invasive spinal deformity surgery. J Neurosurg Spine. 2016; 25(1):21-5. 9. Goldstein CL, Phillips FM, Rampersaud YR. Comparative Effectiveness and Economic Evaluations of Open Versus Minimally Invasive Posterior or Transforaminal Lumbar Interbody Fusion. A Systematic Review. Spine (Phila Pa 1976). 2016;41 Suppl 8):74–89. Evaluations of Open Versus Minimally Invasive Posterior or Transforaminal Lumbar Interbody Fusion. A Systematic Review. Spine (Phila Pa 1976). 2016;41 Suppl 8):74–89. 10. Foley KT, Holly LT, Schwender JD. Minimally Invasive Lumbar Fusion. Spine. 2003;28:S26–35. 11 Menezes CM Junior MAF Falcon RS Oliveira DA Freire SG Experiência inicial com a 10. Foley KT, Holly LT, Schwender JD. Minimally Invasive Lumbar Fusion. Spine. 2003;28:S2 21. Groth AT, Kuklo TR, Klemme WR, Polly DW, Schroeder TM. Comparison of sagittal contour and posterior disc height following interbody fusion: threaded cylindrical cages versus structural allograft versus vertical cages. J Spinal Disord Tech. 2005;18(4):332–6. 11. Menezes CM, Junior MAF, Falcon RS, Oliveira DA, Freire SG. Experiência inicial com a técnica de artrodese lombar minimamente invasiva por via transforaminal (MIS TLIF). Coluna/Columna. 2007;6(3):141-8. 22. Peng CW, Yue WM, Poh SY, Yeo W, Tan SB. CONCLUSION All the patients evaluated in this study were operated by the same surgeon with extensive experience in minimally invasive tech niques, since these techniques require a considerable learning curve and the clinical and radiological results can be directly influenced according to the surgeon’s familiarity with the materials and proces ses involved in surgical planning and execution. The positioning of the cage in the interbody space was not analyzed, given that during the study period the oblique-type cage was routinely positioned in the center of the disc space. No posterior release procedures or associated osteotomies were performed. The standard MIS-TLIF technique had no restorative effect on lumbar lordosis, nor was it responsible for any loss of lor dosis in the operated segments. There was no postoperative influence on the spinopelvic parameters and the preoperative values were maintained. All authors declare no potential conflict of interest related to this article. CONTRIBUTION OF THE AUTHORS: Each author made significant individual contributions to this manuscript. CMM (0000-0001-6670-5159)* performed the surgeries and contributed to the design of the study; LFA (0000-0002-1347-2758), AOA (0000-0001-6579-8234), RVC (0000-0002-8377-2887), GSOV (0000-0003-2253-7450), RSL (0000-0002-7778-2616), MLB (0000-0001-5531-5569), FMMF (0000-0002-5692-788X), and JHGP (0000-0002- 8887-8947) followed-up on the patients and compiled the clinical data; CMM (0000-0001-6670-5159)* and AOA (0000-0001-6579-8234)* reviewed the statistical analysis data; RVC (0000-0002-8377-2887), CMM (0000-0001-6670-5159), and AOA (0000-0001-6579-8234)* performed the bibliographical review and reviewed the manuscript. ORCID (Open Researcher and Contributor ID). DISCUSSION The sample used in this study was comparable to those of other works published in the literature. The main limitations are the absence of a control group and the lack of functional results measurements. REFERENCES 14. Hsieh P, Koski T, O’Shaughnessy BA, Sugrue P, Salehi S, Ondra S, et al. Anterior lumbar interbody fusion in comparison with transforaminal lumbar interbody fusion: implications for the restoration of foraminal height, local discangle, lumbar lordosis and sagittal balan ce. J Neurosurg Spine. 2007;7(4):379–86. 1. Jang JS, Lee SH, Min JH, Han KM. Lumbar degenerative kyphosis: radiographic analysis and classifications. Spine (Phila Pa 1976). 2007;32(24):2694-9. 1. Jang JS, Lee SH, Min JH, Han KM. Lumbar degenerative kyphosis: radiographic analysis and classifications. Spine (Phila Pa 1976). 2007;32(24):2694-9. 2. Takemitsu Y, Harada Y, Iwahara T, Miyamoto M, Miyatake Y. Lumbar degenerative ky phosis. Clinical, radiological and epidemiological studies. Spine (Phila Pa 1976). 1988;13(11):1317–26. 15. Kwon B, Berta S, Daffner S, Vaccaro A, Hilibrand A, Grauer J, et al. Radiographic analysis of transforaminal lumbar interbody fusion for the treatment of adult isthmic spondylolis thesis. J Spinal Disord Tech. 2003;16(5):469–76. 3. Glassman S, Berven S, Bridwell K, Horton W, Dimar J. Correlation of radiographic param eters and clinical symptoms in adult scoliosis. Spine (Phila Pa 1976). 2005;30(6):682–8. 3. Glassman S, Berven S, Bridwell K, Horton W, Dimar J. Correlation of radiographic param eters and clinical symptoms in adult scoliosis. Spine (Phila Pa 1976). 2005;30(6):682–8. 4. Mehta V, Amin A, Omeis I, Gokaslan Z, Gottfried O. Implications of spinopelvic alignment for the spine surgeon. Neurosurgery. 2012;70(3):707–21. y thesis. J Spinal Disord Tech. 2003;16(5):469–76. y p p 4. Mehta V, Amin A, Omeis I, Gokaslan Z, Gottfried O. Implications of spinopelvic alignment for the spine surgeon. Neurosurgery. 2012;70(3):707–21. 16. Jagannathan J, Sansur C, Oskouian R, Fu KM, Shaffrey C. Radiographic restora tion of lumbar alignment after transforaminal lumbar interbody fusion. Neurosurgery. 2009;64(5):955–63. 5. Ould-Silmane M, Lenoir T, Dauzac C, Rillardon L, Hoffmann E, Guigui P, et al. Inﬂuence of transforaminal lumbar interbody fusion procedures on spinal and pelvic parameters of sagittal balance. Eur Spine J. 2012;21(6):1200-6. 17. Lee DY, Jung TG, Lee SH. Single-level instrumented mini-open transforaminal lumbar interbody fusion in elderly patients. J Neurosurg Spine. 2008;9(2):137–44. g p 6. Harms J, Rolinger H. A one-stage procedure in operative treatment of spondylolistheses: dorsal traction-reposition and anterior fusion [in German]. Z Orthop Ihre Grenzgeb. 1982;120(3):343–7. 7. Rosenberg WS, Mummaneni PV. Transforaminal lumbar interbody fusion: technique, com plications, and early results. Neurosurgery. 2001;48(3):569–74. 18. Yson S, Santos ERG, Sembrano J, Polly D. Segmental lumbar sagital correction after bilateral transforaminal lumbar interbody fusion. J Neurosurg Spine. 2012;17(1):37–4. 6. CONTRIBUTION OF THE AUTHORS: Each author made significant individual contributions to this manuscript. CMM (0000-0001-6670-5159) performed the surgeries and contributed to the design of the study; LFA (0000-0002-1347-2758), AOA (0000-0001-6579-8234), RVC (0000-0002-8377-2887), GSOV (0000-0003-2253-7450), RSL (0000-0002-7778-2616), MLB (0000-0001-5531-5569), FMMF (0000-0002-5692-788X), and JHGP (0000-0002- 8887-8947) followed-up on the patients and compiled the clinical data; CMM (0000-0001-6670-5159)* and AOA (0000-0001-6579-8234)* reviewed the statistical analysis data; RVC (0000-0002-8377-2887), CMM (0000-0001-6670-5159), and AOA (0000-0001-6579-8234)* performed the bibliographical review and reviewed the manuscript. *ORCID (Open Researcher and Contributor ID). Coluna/Columna. 2019;18(1):47-50 REFERENCES Clinical and Radiological Outcomes of Mini mally Invasive Versus Open Transforaminal Lumbar Interbody Fusion. Spine (Phila Pa 1976). 2009;34(13):1385–9. 12. Menezes CM, Junior MAF, Falcon RS, Oliveira DA. Artrodese minimamente invasiva para espondilolisteses de baixo grau. Coluna/Columna. 2008;7(3)241-5. 23. Lafage V, Blondel B, Smith JS, Broachie-Adiei O, Hostin RA, Burton D, et al.. Preop erative planning for pedicle subtraction osteotomy: Does pelvic tilt matter? Spine Deform. 2014;2(5):358-66. 13. Menezes CM, Junior MAF, Falcon RS, Alencar J. Avaliação clínica radiológica da artro dese lombar transforaminal aberta versus minimamente invasiva. Coluna/Columna. 2009;8(3):297-302.
https://openalex.org/W2087532731	https://europepmc.org/articles/pmc4133046?pdf=render	English	null	MODMatcher: Multi-Omics Data Matcher for Integrative Genomic Analysis	PLOS computational biology/PLoS computational biology	2,014	cc-by	11,174	Introduction the sample mis-labeling error. In other integrative analyses, such as the genetic gene expression studies [2,3], in which the aim is to discover how DNA variations or single nucleotide polymorphisms (SNPs) regulate gene expression changes, sample errors could have a larger effect. In one study, mis-matching of 20% of samples between genotype and gene expression data decreased the number of cis-eSNPs by 70% [4]. Cells employ multiple levels of regulation that enable them to respond to genetic, epigenetic, genomic, and environmental perturbations. With advances in high-throughput technologies, comprehensive data sets have been generated to measure multiple aspects of biological regulation, such as genetics, transcriptomics, metabolomics, glycomics, and proteomics. To elucidate the complexity of cell regulation, diverse types of data from these different technologies must be integrated. To fully understand biological systems, it is necessary to elucidate how genetic and epigenetic perturbations lead to transcriptomic and proteomic changes, which in turn contribute to the disease phenotype. Simultaneously considering different types of biological data can result a better understanding of biological systems [2,5–8]. Sample errors, including sample swapping, mis-labeling, and improper data entry are inevitable during large-scale data generation. Some of these errors can be detected during quality control (QC) on each type of data; however, others are more elusive and may affect integrative data analysis, depending on the integration methods used. In some integrative analyses, signature sets are first defined by each data type individually, for example signatures for gene expression, methylation, or copy number variation (CNV). Then, the signatures are overlapped to identify high-confidence changes [1]. In such analyses, potential sample inconsistencies may have a limited effect on results. For example, assume that samples A and B are swapped in gene expression data. If both samples are involved in the same subgroup (e.g., normal control or disease), the derived signatures will not be affected by With recent advances in high-throughput technologies, multiple layers of molecular phenotypes have been measured in the same sample for comprehensive survey of biological systems. To maximally utilize these data, it is necessary to properly match different types of data pertaining to the same sample or individual before integrative analyses. Here we present a sample mapping procedure called Multi-Omics Data matcher (MODMatcher), which not only identifies mis-matched omics profile pairs, but also properly assigns them to the correct samples based on other omics data. Abstract * Email: jun.zhu@mssm.edu Seungyeul Yoo1,2, Tao Huang1,2, Joshua D. Campbell3, Eunjee Lee1,2, Zhidong Tu1,2, Mark W. Geraci4, Charles A. Powell5, Eric E. Schadt1,2, Avrum Spira3, Jun Zhu1,2* Seungyeul Yoo1,2, Tao Huang1,2, Joshua D. Campbell3, Eunjee Lee1,2, Zhidong Tu1,2, Mark W. Geraci4, Charles A. Powell5, Eric E. Schadt1,2, Avrum Spira3, Jun Zhu1,2* 1 Icahn Institute of Genomics and Multiscale Biology, Icahn School of Medicine at Mount Sinai, New York, New York, United States of America, 2 Department of Genetics and Genomic Sciences, Icahn School of Medicine at Mount Sinai, New York, New York, United States of America, 3 Division of Computational Biomedicine, Department of Medicine, Boston University School of Medicine, Boston, Massachusetts, United States of America, 4 Division of Pulmonary Sciences and Critical Care Medicine, University of Colorado Denver, Aurora, Colorado, United States of America, 5 Division of Pulmonary, Critical Care and Sleep Medicine, Icahn School of Medicine at Mount Sinai, New York, New York, United States of America Abstract Errors in sample annotation or labeling often occur in large-scale genetic or genomic studies and are difficult to avoid completely during data generation and management. For integrative genomic studies, it is critical to identify and correct these errors. Different types of genetic and genomic data are inter-connected by cis-regulations. On that basis, we developed a computational approach, Multi-Omics Data Matcher (MODMatcher), to identify and correct sample labeling errors in multiple types of molecular data, which can be used in further integrative analysis. Our results indicate that inspection of sample annotation and labeling error is an indispensable data quality assurance step. Applied to a large lung genomic study, MODMatcher increased statistically significant genetic associations and genomic correlations by more than two-fold. In a simulation study, MODMatcher provided more robust results by using three types of omics data than two types of omics data. We further demonstrate that MODMatcher can be broadly applied to large genomic data sets containing multiple types of omics data, such as The Cancer Genome Atlas (TCGA) data sets. ation: Yoo S, Huang T, Campbell JD, Lee E, Tu Z, et al. (2014) MODMatcher: Multi-Omics Data Matcher for Integrative Genomic A l 10(8): e1003790. doi:10.1371/journal.pcbi.1003790 T, Campbell JD, Lee E, Tu Z, et al. (2014) MODMatcher: Multi-Omics Data Matcher for Integrative Genomic Analysis. PLoS Comput 10.1371/journal.pcbi.1003790 Editor: Xianghong Jasmine Zhou, University of Southern California, United States of America Editor: Xianghong Jasmine Zhou, University of Southern California, United States of America Editor: Xianghong Jasmine Zhou, University of Southern California, United States of America Received February 21, 2014; Accepted June 26, 2014; Published August 14, 2014 Received February 21, 2014; Accepted June 26, 2014; Published August 14, 2014 Copyright: 2014 Yoo et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: This work is partially supported by the NCI grant CA170722,NIH AG046170, and a fund from Canary Foundation (Palo Alto, CA) (www.canaryfoundation. org). AS is supported by the NIH grant 1RC2HL101715 (Lung Genomics Research Consortium). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. * Email: jun zhu@mssm edu Competing Interests: The authors have declared that no competing interests exist. August 2014 \| Volume 10 \| Issue 8 \| e1003790 PLOS Computational Biology \| www.ploscompbiol.org Author Summary Many human diseases are complex with multiple genetic and environmental causal factors interacting together to give rise to disease phenotypes. Such factors affect biological systems through many layers of regulations, including transcriptional and epigenetic regulation, and protein changes. To fully understand their molecular mechanisms, complex diseases are often studied in diverse dimensions including genetics (genotype variations by single nucleotide polymorphism (SNP) arrays or whole exome sequencing), transcriptomics, epigenetics, and proteomics. However, errors in sample annotation or labeling often occur in large-scale genetic and genomic studies and are difficult to avoid completely during data generation and management. Identifying and correcting these errors are critical for integrative genomic studies. In this study, we developed a computational approach, Multi- Omics Data Matcher (MODMatcher), to identify and correct sample labeling errors based on multiple types of molecular data before further integrative analysis. Our results indicate that signals increased more than 100% after correction of sample labeling errors in a large lung genomic study. Our method can be broadly applied to large genomic data sets with multiple types of omics data, such as TCGA (The Cancer Genome Atlas) data sets. Next, we iteratively matched SNP, gene expression, and methylation profiles using multi-omics identity similarity scores (Figure 4). We started with three sets of profile pairs with consistent inferred gender information: 179 pairs (50 CTRL and 129 COPD) for genotype and gene expression data, 182 pairs (51 CTRL 131 COPD) for genotype and methylation data, and 209 pairs (61 CTRL and 148 COPD) for methylation and gene expression profiling data. Cis regulation pairs (i.e. cis-eSNPs, cis- mSNPs, and cis methy-mRNA probes) were identified separately for CTRL and COPD samples. Sample identity similarity scores Sge, Sgm, and Sme based on identified cis regulation pairs were calculated for all possible profile pairs. Sge and Sgm were calculated from the distance between predicted and measured SNP genotypes. Sme was measured by correlation of rank- transformed methylation and gene expression levels in samples (Figure 5, see Methods). The similarity scores for matched profiles were 3.8, 3.2, and 1.8 standard deviations better than the mean similarity scores for Sge, Sgm, and Sme, respectively (Figure 6A–C). Thus, SNP-mRNA sample matches were more reliable than SNP- methylation or methylation-mRNA sample matches, perhaps because methylation data tends to be noisy due to intrinsic technical design [9,10]. Application to LGRC data Application to LGRC data The LGRC is a consortium for studying chronic lung diseases including chronic obstructive pulmonary disease (COPD). Clinical information and gene expression and methylation profiling data were obtained from the LGRC data portal (http://www.lung- genomics.org). Genotype data was provided by the LGRC consortium. The data set consists of gene expression profiles of lung tissues from 219 patients with COPD and 108 non-disease controls (CTRL), and methylation profiles of lung tissues from 173 COPD patients and 76 controls. First, the gender of each sample was inferred based on three types of data and compared to the gender annotated in clinical data. There was no ambiguity in gender prediction based on each individual type of data; the molecular profiles of different genders were clearly separated (Figures 1–3). However, we identified several mismatches between the predicted genders based on omics data and the clinically annotated genders. Among genders predicted by X-chromosome heterozygosity, we detected 4 mismatches in CTRL and 5 in COPD samples, corresponding to a mismatch error rate of 3.5% (9/256) for SNP genotype profiles (Figure 1). While there was no gender mismatch in CTRL samples, as judged by the expression level of Y-chromosome specific gene RPS4Y1, we detected 5 gender mismatches in COPD, corresponding to a mismatch error rate of 1.5% (5/327) for gene expression profiles (Figure 2). Among genders predicted from the intensity of the Y-chromosome specific methyl probe close to FAM197Y2P (see Methods), we found 1 gender inconsistency in CTRL samples and 15 in COPD samples, corresponding to a mismatch error rate of 6.4% (16/249) for methylation profiles (Figure 3). Overall, for 21 unique When all three types of data are available, the source of any sample labeling errors can be identified. It is also possible to remove or identify additional matched profiles that may be ambiguous as judged from Sge, Sgm, or Sme alone. Since cis-eSNPs pairs provided the best alignment signal, we started with sample matching by cis-eQTL. Then, samples were further matched by cis-mQTL and mRNA-methylation. For the SNP-mRNA profile match, we tested whether there was a methylation profile that matches well with both SNP and mRNA profiles in the matched pair. After each round of sample matching, the quality of sample alignment was assessed by counting the number cis pairs identified. For all pairs among these three data types, sample mapping correction significantly increased numbers of cis pairs identified (Figure 7). Author Summary Based on the gender-matching results, methylation profiles have a higher mis-label rate than other profile data, also contributing to the uncertainties of sample matching of methylation profiles. (LGRC) and one from The Cancer Genome Atlas (TCGA). In both cases, adjustment for mis-matched samples improved data consistency and increased statistic power to identify biological regulations. All software programs and scripts are available at http://research.mssm.edu/integrative-network-biology/Software. html. (LGRC) and one from The Cancer Genome Atlas (TCGA). In both cases, adjustment for mis-matched samples improved data consistency and increased statistic power to identify biological regulations. All software programs and scripts are available at http://research.mssm.edu/integrative-network-biology/Software. html. Next, we determined whether mis-aligned samples could be matched with other unmatched samples by reciprocal best matching, based on one type of identity similarity score. In other words, we tested whether mis-aligned genotype profile Gi had the highest similarity with an unmatched mRNA profile Ej among all mRNA profiles, and the unmatched mRNA profile Ej had the highest similarity with Gi among all genotype profiles as well. For the sample pair with a reciprocal best match, sample labels can be updated by comparison with mapping results based on other identity similarities. Introduction We applied MODMatcher to two large-scale public multi- omics datasets: one from the Lung Genomic Research Consortium 1 August 2014 \| Volume 10 \| Issue 8 \| e1003790 PLOS Computational Biology \| www.ploscompbiol.org MODMatcher: Multi-Omics Data Matcher individuals (Table S1), the gender information inferred from different sources did not match either with one another or with clinical annotation, indicating sample alignment problems. According to the error rate of gender mismatches, gene expression profiling data was least likely to be mis-labeled, and methylation profiling data was most likely to be mis-labeled in the LRGC data set. Application to LGRC data The number of cis-eSNPs stabilized within the first 5 rounds (Figure 7A). However, the number of cis-mSNP pairs stabilized in much later rounds (about 15–17), as expected because of the higher mis-label error rate and greater noise in the methylation data. Nonetheless, the numbers of cis-pairs involving methylation profiles increased substantially with the improved sample matching (Figure 7B and 7C). In COPD samples, the number of cis-eSNPs increased by ,100% and the number of cis August 2014 \| Volume 10 \| Issue 8 \| e1003790 PLOS Computational Biology \| www.ploscompbiol.org 2 Figure 1. Gender prediction based on genotype data. The inbreeding coefficient F, the X chromosome heterozygosity rate, is used to infer the gender of samples. F is around 0 in most female samples and around 1 in most male samples. For 9 samples, the inferred genders were inconsistent with clinically annotated genders (error rate 3.5%). doi:10.1371/journal.pcbi.1003790.g001 MODMatcher: Multi-Omics Data Matcher MODMatcher: Multi-Omics Data Matcher PLOS Computational Biology \| www.ploscompbiol.org Figure 1. Gender prediction based on genotype data. The inbreeding coefficient F, the X chromosome heterozygosity rate, is used to infer the gender of samples. F is around 0 in most female samples and around 1 in most male samples. For 9 samples, the inferred genders were inconsistent with clinically annotated genders (error rate 3.5%). doi:10.1371/journal.pcbi.1003790.g001 we simulated sample labeling errors by randomly assigning sample labels using only these 65 COPD samples. As in the empirical data, we kept low error rates in SNP and gene expression profiling data. We increased the number of mis-labeled methylation profiles from 0 to 24 (corresponding error rate 0% to 37%). At each error rate, we simulated 5 independent data sets and used the average for comparison. In both of duo and trio alignment, all three data types were used but in different ways. In duo alignment, we identified the sample pairs from each pair of data types independently and summarized them to have final pairs. For an example, a methylation profile can be matched with an mRNA profile directly based on the identity similarity score Sme or through a chain of matches, in which the methylation profile is matched to an SNP profile which matches the mRNA profile. In trio alignment, there is an additional three-way identity similarity score that considers all three data types simultaneously (as mRNA-methylation pairs increased by ,200%. Consistently, fewer cis pairs were identified in the CTRL data set than in the COPD data set. This difference likely reflects disease biology. Although there were fewer CTRL than COPD samples and thus less statistical power, the trend of difference was the same when we sampled equal numbers of COPD samples to CTRL samples (Figure S1). Using a series of simulated data sets, we demonstrated that trio alignment (considering three types of data simultaneously) resulted in better alignment than duo alignments (considering two types of data at a time) combined. From the sample alignment of the LGRC data as describe above, we identified 76 COPD samples with aligned genotype, gene expression and methylation profiles. Among these 76 samples, only 65 could be correctly matched when individual similarity scores such as cis mRNA-methylation pairs were used. For a fair comparison of trio and duo alignment, August 2014 \| Volume 10 \| Issue 8 \| e1003790 PLOS Computational Biology \| www.ploscompbiol.org 3 MODMatcher: Multi-Omics Data Matcher Figure 2. Trio alignment recovered more samples pairs and predicted sample pairs more accurately than alignments considering similarity scores independently. In trio alignment, the additional data type provided more bridging information for matching mis-aligned samples pairs. Thus, at the same mis-labeling error rate, trio alignment yielded a higher true positive rate and better coverage (Figure 8). As error rates increased, the benefit of using trio alignment became clearer. Thus, in correcting sample mis-matches, sample alignment considering three types of data simultaneously in sample alignment may have advantages over combining three independent duo-alignments. These simulation results confirm that sample alignment using multi-omics data is a critical QC step. Alignment that considers three types of omics data simultaneously is strongly recommended if applicable. Nevertheless, duo alignment is still useful for identifying and correcting mis-aligned pairs. described in Methods). Both trio and duo alignment identified mis- matches and improved data quality. However, trio alignment was more robust and superior, especially when mis-labeling error rates were high (Figure 8). Trio alignment recovered more samples pairs and predicted sample pairs more accurately than alignments considering similarity scores independently. In trio alignment, the additional data type provided more bridging information for matching mis-aligned samples pairs. Thus, at the same mis-labeling error rate, trio alignment yielded a higher true positive rate and better coverage (Figure 8). As error rates increased, the benefit of using trio alignment became clearer. Thus, in correcting sample mis-matches, sample alignment considering three types of data simultaneously in sample alignment may have advantages over combining three independent duo-alignments. These simulation results confirm that sample alignment using multi-omics data is a critical QC step. Alignment that considers three types of omics data simultaneously is strongly recommended if applicable. Nevertheless, duo alignment is still useful for identifying and correcting mis-aligned pairs. Gender prediction based on expression of the Y-chromosome specific gene RPS4Y1. The log2 transformed values of RPS4Y1 expression level are clearly separated between male and female samples both in CTRL and patients with COPD (.10 in male samples and ,10 in female samples). There were no gender mismatched samples in the CTRL and 5 mismatched samples (2 in females and 3 in males) in the COPD set (error rate of 1.5%). doi:10.1371/journal.pcbi.1003790.g002 Figure 2. Gender prediction based on expression of the Y-chromosome specific gene RPS4Y1. The log2 transformed values of RPS4Y1 expression level are clearly separated between male and female samples both in CTRL and patients with COPD (.10 in male samples and ,10 in female samples). There were no gender mismatched samples in the CTRL and 5 mismatched samples (2 in females and 3 in males) in the COPD set (error rate of 1.5%). doi:10.1371/journal.pcbi.1003790.g002 profiles. We detected one tumor sample whose predicted gender was inconsistent based on gene expression and methylation profiles. After removal of the gender mismatched sample, cis methylation-mRNA probe pairs were redefined for both normal and tumor samples. At p-value,0.01, 9195 pairs were identified for the tumor (FDR,0.02 based on permutation tests) and 537 pairs for normal samples (FDR,0.35). The identity similarity score Sme based on these cis probe pairs were normally distributed; one outlier had a higher similarity score (red star in Figure 9A), indicating a match of profiles of the same patient. There were 8 mis-aligned profile pairs among the tumor samples. Three profile pairs were matched by reciprocal mapping. Two of them, ‘‘TCGA-BH-A18K-01’’, and ‘‘TCGA-BH-A18T-01’’, were cross-aligned to each other in methylation and gene expression profiles (Figure 9B). Interestingly, the barcodes of two samples had only one difference (K vs. T), suggesting a sample swap in either the mRNA or methylation profiles. Further comparison with miRNA profiles of these tumor samples suggested that the swap was in the mRNA profiles (Figure S2). The updated sample alignment resulted in more cis pairs (9252 at p-value,0.01) and also stronger statistical p-values for the cis-correlations. For example, the p-value for the cis correlation of methylation and gene expression levels of TMEM139 was 1.6610267 before alignment and 3.8610274 after alignment. described in Methods). Both trio and duo alignment identified mis- matches and improved data quality. However, trio alignment was more robust and superior, especially when mis-labeling error rates were high (Figure 8). Application to TCGA data 1) TCGA BRCA samples. The same sample alignment approach was applied to another publically available dataset, TCGA breast cancer samples. There were 317 tumor samples and 20 adjacent normal samples with both gene expression and methylation profiles (Table 1). Genders of samples were inferred from molecular markers in gene expression and methylation 2) TCGA GBM samples. TCGA glioblastoma multiforme (GBM) is the first cancer data set in TCGA consisting of CNV, 2) TCGA GBM samples. TCGA glioblastoma multiforme (GBM) is the first cancer data set in TCGA consisting of CNV, August 2014 \| Volume 10 \| Issue 8 \| e1003790 PLOS Computational Biology \| www.ploscompbiol.org 4 MODMatcher: Multi-Omics Data Matcher Figure 3. Gender prediction based on methylation intensity. The raw intensity of a Y-chromosome methyl probe corresponding to FAM197Y2P is clearly different between genders. One error was identified in the CTRL and 15 errors were identified in the COPD set (6 in females, 9 in males) (error rate of 6.4%). doi:10.1371/journal.pcbi.1003790.g003 Figure 3. Gender prediction based on methylation intensity. The raw intensity of a Y-chromosome methyl probe corresponding to FAM197Y2P is clearly different between genders. One error was identified in the CTRL and 15 errors were identified in the COPD set (6 in females, 9 in males) (error rate of 6.4%). doi:10 1371/journal pcbi 1003790 g003 Figure 3. Gender prediction based on methylation intensity. The raw intensity of a Y-chromosome methyl probe corresponding to FAM197Y2P is clearly different between genders. One error was identified in the CTRL and 15 errors were identified in the COPD set (6 in females, 9 in males) (error rate of 6.4%). doi:10.1371/journal.pcbi.1003790.g003 gene expression, and methylation profiles. There were 470 GBM tumor samples with both CNV and mRNA profiles. We identified 24 mis-aligned profile pairs. Two of them were cross-aligned between CNV and gene expression profiles (TCGA-32-2632-01A, and TCGA-12-3652-01A) (Figure 10A). When we aligned meth- ylation and gene expression profiles based on the identity similarity score Sme calculated by using cis methylation-mRNA pairs, they were cross-aligned to each other as well, indicating that the labels of mRNA profiles are problematic (Figure 10B). Additionally, the two samples were self-aligned between CNV and methylation profiles (Figure 10C). These findings indicate that the sample labels of the two mRNA profiles were swapped. Application to TCGA data This example shows how sample alignment using three different molecular data sets can be useful for both correcting alignment errors in sample pairs and identifying the source of the errors. calculate sample identity similarity scores. When applied to two large public data sets, LGRC and TCGA, MODMatcher not only identified mis-aligned profile pairs but also corrected and rescued mis-labeled samples. In a simulation study of COPD samples in the LGRC set, sample alignment with three types of data (trio matching) performed better than alignment with two types of data (duo matching). When applied to the GBM data set in TCGA, trio matching unambiguously identified the source of sample labeling errors. Thus, MODMatcher can rescue mis-aligned or mis-labeled samples to maximize statistical power in integrative analysis in large-scale genetic and genomic studies. Indeed, correction of mis- aligned samples increased both the number of cis pairs identified and the statistical significance. Sample labeling errors are not unique to a few data sets, but are inevitable for any large data sets, despite intensive efforts in QCing each type of data individually. Our methods based on methylation profiles for gender inference and alignment with other omics profiles are novel and have not been included in standard data QC procedures. We applied our methylation-based gender inference method to more TCGA data sets and demonstrated that gender can be unambiguously inferred from methylation profiles (Figure S3). We identified 1, 4, 1, 2 gender mis-match errors in methylation profiles in data sets for colon adenocarcinoma (COAD), kidney renal papillary carcinoma (KIRC), acute myeloid leukemia (LAML), and lung adenocarcinoma (LUAD), respective- ly (Table S3). We also applied our methylation-gene expression PLOS Computational Biology \| www.ploscompbiol.org Discussion In large-scale genetic and genomic studies, errors in sample annotation or labeling are common and difficult to avoid completely. Identifying and correcting these errors is critical for statistical analysis, especially for integrative analysis. In this study, we introduce an iterative computational procedure, MOD- Matcher, that uses multiple types of molecular data (e.g., genotype, CNV, gene expression, and methylation profiles) for sample alignment by using cis regulation pairs of each pair of data types to August 2014 \| Volume 10 \| Issue 8 \| e1003790 PLOS Computational Biology \| www.ploscompbiol.org 5 MODMatcher: Multi-Omics Data Matcher Figure 4. Sample alignment with MODMatcher. Initial labels of samples are used to determine cis pairs, which are then used to calculate similarity scores. Based on the similarity scores determined with three data types, the molecular data are matched with each other (1) by gender, (2) by cis-eSNPs, (3) by cis-mSNPs, (4) by cis mRNA-methylation pairs, and (5) by all trio mapping. Then, updated sample pairs are used to calculate new cis pairs for another round of alignment. Rounds of alignment are repeated until there are no further changes. doi:10.1371/journal.pcbi.1003790.g004 Figure 4. Sample alignment with MODMatcher. Initial labels of samples are used to determine cis pairs, which are then used to calculate similarity scores. Based on the similarity scores determined with three data types, the molecular data are matched with each other (1) by gender, (2) by cis-eSNPs, (3) by cis-mSNPs, (4) by cis mRNA-methylation pairs, and (5) by all trio mapping. Then, updated sample pairs are used to calculate new cis pairs for another round of alignment. Rounds of alignment are repeated until there are no further changes. doi:10.1371/journal.pcbi.1003790.g004 profile matching method to additional TCGA data sets, COAD and lung squamous cell carcinoma (LUSC), and identified multiple mis-label errors (examples shown in Figure S4). Thus, checking sample alignment is a critical and necessary QC step before integrative analysis. only two types of omics profiles. MixupMapper and MOD- Matcher can only be compared for their ability to match genotype and mRNA profiles. We applied MODMatcher to 8 data sets examined by MixupMapper (downloaded from http://genenetwork.nl/wordpress/mixupmapper/#additional) and compared alignment results based on the two methods (Table S4). MODMatcher results completely agreed with MixupMapper results in 6 of 8 data sets. Discussion For the two datasets in which the MODMatcher and MixupMapper results are different, we further assessed sample alignment quality by counting cis-eQLs identified based on the final matching results. We input final matching pairs identified by each method and their corresponding profiles to the same program, MatrixEQTL [11], to identify cis-eQTLs. In both cases, more cis-eQTLs were identified with MODMatcher results than with MixupMapper results (Table S5). It is worth to note that the sample identity similarity scores, Sge, Sgm, and Sme, are calculated by using cis regulation pairs. Therefore, like the method of Westra et al. [4], MODMatcher depends on initial sample alignments to generate cis regulation pairs. However, MODMatcher is more robust and can tolerate extra noise, as shown in the simulation study. If the error rate of initial alignment is too high (e.g., .30% mis-alignment), we may not be able to identify enough cis-regulation pairs to accurately align samples on the basis of a single identity score. But based on three-way similarity, more accurate matching pairs can still be identified. After labeling errors in omics profiles are identified and corrected by leveraging information from multiple omics profiles, the corrected profiles can be compared with clinical information to answer many biological questions, such as what genes’ expression levels correlate with blood lipid level, and what genes’ methylation levels correlate with survival of cancer patients. To accomplish these tasks, we assume that all clinical data are correct, which may not always be true. There could be errors in clinical data files, such as missing data, and row or column shifts. It is more challenging to identify and correct errors in clinical data files than it is to identify labeling errors in omics profiles. More research efforts are warranted for checking potential errors in the links between clinical data and omics profiles. MODMatcher has several features not found in existing sample alignment methods such as MixupMapper [4]. First, we proposed novel methods for methylation profile based gender inference and sample alignment, and MODMatcher can be applied to diverse types of data, including genotype, gene expression, methylation, and CNV. MixupMapper can only be applied to genotype and gene expression data. PLOS Computational Biology \| www.ploscompbiol.org Discussion Second, by using more than two types of omics profiles, MODMatcher can not only identify potential mis- labeled omics profile pairs, but also pinpoint which profiles in the pairs are mis-labeled (Figures 9 and 10), and do so more robustly than when only two types of omics profiles are used (Figure 8) Even though MODMatcher is not designed for matching two types of omics profiles, it can be applied to data sets consisting of August 2014 \| Volume 10 \| Issue 8 \| e1003790 August 2014 \| Volume 10 \| Issue 8 \| e1003790 PLOS Computational Biology \| www.ploscompbiol.org 6 MODMatcher: Multi-Omics Data Matcher Figure 5. Sample similarity measurement based on cis methylation-mRNA pairs. After cis methylation-mRNA pairs are identified, t methylation and gene expression levels were rank-transformed. In this figure, there are M samples and i cis pairs. Then Pearson correlation calculated and used as sample similarity, Sme, between one methylation profile and all gene expression profiles. If both methylation and ge expression profiles are from the same individual, self-self correlation coefficient is expected to be significantly higher than correlation coefficien with other samples. doi:10.1371/journal.pcbi.1003790.g005 PLOS Computational Biology \| www.ploscompbiol.org 7 August 2014 \| Volume 10 \| Issue 8 \| e10037 Figure 5. Sample similarity measurement based on cis methylation-mRNA pairs. After cis methylation-mRNA pairs are identified, the methylation and gene expression levels were rank-transformed. In this figure, there are M samples and i cis pairs. Then Pearson correlation is calculated and used as sample similarity, Sme, between one methylation profile and all gene expression profiles. If both methylation and gene expression profiles are from the same individual, self-self correlation coefficient is expected to be significantly higher than correlation coefficients with other samples. doi:10.1371/journal.pcbi.1003790.g005 August 2014 \| Volume 10 \| Issue 8 \| e1003790 PLOS Computational Biology \| www.ploscompbiol.org PLOS Computational Biology \| www.ploscompbiol.org 7 MODMatcher: Multi-Omics Data Matcher Figure 6. Distribution of similarity scores. (A) The distribution of similarity scores of one profile versus all profiles of other omics data type based on cis-eQTLs. (B) The distribution of similarity scores of one profile versus all profiles of other data type based on cis-mQTLs. (C) The distribution of similarity scores of one profile versus all profiles of other data type based on cis mRNA-methylation pairs. If two profiles pertain to the same sample (self-consistent), their similarity scores (red stars) are expected to be higher than those of cross-matched pairs. Materials and Methods samples by quantile normalization and adjusted for covariance (e.g., batch number, plate number, center ID, and source site ID of sample). Samples were initially matched according to their labels as shown in Table 1. Comprehensive data sets generally consist of clinical or phenotypic data and multiple types of high-throughput data. For example, the LGRC data set consists of clinical, genotype, gene expression, and methylation data. The TCGA tumor data sets consist of clinical, genotype, gene expression, CNV, methylation, miRNA, and protein array data. Our procedure is applicable to data sets with clinical data and at least two different types of omics data. Here we illustrate our procedure on data sets with clinical data and genetic, gene expression, and methylation data. Gender inference Gender information is generally included in clinical data. We also inferred gender information from genotype, gene expression, and methylation profiling data. The gender of samples can be predicted from X-chromosome heterozygosity rates determined with PLINK [15]. An individual is predicted to be male if the estimated inbreeding coefficient F is . 0.8 and female if F,0.2 [16]. There were inconsistencies between gender inferred from genotype data and gender provided in clinical data for the LGRC samples (Figure 1). Datasets 1) LGRC data set. Clinical, gene expression and methylation data were obtained from the LGRC data portal (http://www. lung-genomics.org). The genotypic data was provided by the LGRC consortium. The gene expression data was generated with Agilent V2 human whole genome arrays. The processed mRNA array data was downloaded from the LGRC website. DNA methylation data was generated with Nimblegen 2.1M Whole- Genome Tiling Arrays. Raw DNA methylation data was downloaded from the LGRC website. The quality of each probe was compared with background probe signals, and probes with low quality were excluded from further analysis. Then DNA methylation level (b value) of each tiling probe was estimated with the CHARM method [9,12]. We confirmed that the estimated methylation level for each sample is almost identical with the processed methylation level data from the LGRC website. There were gene expression arrays for lung tissues from 219 COPD patients and 108 non-disease controls (CTRL) and methylation arrays for lung tissues from 173 COPD patients and 76 controls. Gene expression levels of Y-chromosome specific genes can also be used to reliably predict gender information. RPS4Y1 (ribosomal protein S4, Y-linked 1) is highly expressed in male [17]. Its expression level can robustly classify samples into male and female [6]. Figure 2 shows gender mismatches between clinical and gene expression data in the LGRC data set. Raw intensity data in methylation profiling was used to determine whether probes mapped to Y-chromosome DNA fragments can be used to classify samples into male and female. Raw intensities of probes representing the Y-chromosome specific genes FAM197Y2P, TTTY15, and TBL1Y were significantly associated with genders in the LGRC data set (t-test p- values = 3.25610228, 1.79610227, and 8.71610226, respectively). A methyl probe, ‘‘chrY:9994006’’, representing FAM197Y2P is the best methyl probe for gender prediction and was used to classify samples in the LGRC data set into male and female. Figure 3 shows that a higher mismatch rate between clinical and methylation profiling data than other pairs of data matching in the LGRC data set (Table S1). 2) TCGA data set. Different types of clinical and molecular data of various cancers are publicly available at the TCGA data portal (https://tcga-data.nci.nih.gov/tcga/). To illustrate our sample alignment procedure, we selected BRCA (one of the newest cancer data sets) [13] and GBM (the oldest cancer data set) [14]. Gene expression for the GBM and BRCA data sets was measured with microarrays. Discussion doi:10.1371/journal.pcbi.1003790.g006 Figure 6. Distribution of similarity scores. (A) The distribution of similarity scores of one profile versus all profiles of other omics data type based on cis-eQTLs. (B) The distribution of similarity scores of one profile versus all profiles of other data type based on cis-mQTLs. (C) The distribution of similarity scores of one profile versus all profiles of other data type based on cis mRNA-methylation pairs. If two profiles pertain to the same sample (self-consistent), their similarity scores (red stars) are expected to be higher than those of cross-matched pairs. doi:10.1371/journal.pcbi.1003790.g006 Datasets Methylation profiles were measured with Illumina HumanMethylation27 BeadChips. CNV data were generated with Affymetrix Genome-Wide Human SNP Array 6.0. Bulk data on BRCA and GBM samples was downloaded and processed individually. Each type of data was normalized between cis pair mapping The true positive rate is defined as the number o aligned pairs divided by all aligned pairs. doi:10.1371/journal.pcbi.1003790.g008 Figure 8. Comparison of sample alignment procedures based on three or two data types in simulated datasets. A total 65 COPD samples with all three types of data (gene expression, genotype, and methylation) were used. The mis-labeling error rates were fixed at 3% between gene expression and genotypes. The number of mis- aligned pairs was varied from 0 to 24 (corresponding error rate, 0% to 37%). Two sample alignment procedures were applied to the simulated data sets and final aligned pairs were compared with the true alignment. Triangles, duo-alignment results; circles, trio-alignment results. Numbers inside triangles or circles indicate the number of mis-aligned samples in each simulation. Coverage is defined as the number of correctly aligned pairs divided by 65 (the number of original pairs). The true positive rate is defined as the number of correctly aligned pairs divided by all aligned pairs. doi:10.1371/journal.pcbi.1003790.g008 Figure 7. Assessment of sample alignment quality. The number of cis pairs is counted after each round of alignment. The number of cis pairs increased markedly after alignment in both the CTRL and COPD sets. The exact numbers of cis-pairs are listed in Table S2. A) cis-eQTLs. B) cis-mQTLs. C) cis mRNA-methylation pairs. doi:10.1371/journal.pcbi.1003790.g007 PLOS Computational Biology \| www.ploscompbiol.org Figure 8. Comparison of sample alignment procedures based on three or two data types in simulated datasets. A total 65 COPD samples with all three types of data (gene expression, genotype, and methylation) were used. The mis-labeling error rates were fixed at 3% between gene expression and genotypes. The number of mis- aligned pairs was varied from 0 to 24 (corresponding error rate, 0% to 37%). Two sample alignment procedures were applied to the simulated data sets and final aligned pairs were compared with the true alignment. Triangles, duo-alignment results; circles, trio-alignment results. Numbers inside triangles or circles indicate the number of mis-aligned samples in each simulation. Coverage is defined as the number of correctly aligned pairs divided by 65 (the number of original pairs). The true positive rate is defined as the number of correctly aligned pairs divided by all aligned pairs. doi:10.1371/journal.pcbi.1003790.g008 Figure 8. Comparison of sample alignment procedures based on three or two data types in simulated datasets. A total 65 COPD samples with all three types of data (gene expression, genotype, and methylation) were used. cis pair mapping The mis-labeling error rates were fixed at 3% between gene expression and genotypes. The number of mis- aligned pairs was varied from 0 to 24 (corresponding error rate, 0% to 37%). Two sample alignment procedures were applied to the simulated data sets and final aligned pairs were compared with the true alignment. Triangles, duo-alignment results; circles, trio-alignment results. Numbers inside triangles or circles indicate the number of mis-aligned samples in each simulation. Coverage is defined as the number of correctly aligned pairs divided by 65 (the number of original pairs). The true positive rate is defined as the number of correctly aligned pairs divided by all aligned pairs. doi:10.1371/journal.pcbi.1003790.g008 infer sample genotypes from gene expression profiling data [4,19]. To identify cis-eSNPs, we used an efficient eSNP mapping program, MatrixEQTL [11]. Assuming that genotype had an additive and linear effect on gene expression, we calculated the t statics for each SNP and gene expression pair to evaluate the significance of association. Cis-eSNPs are defined as SNPs within 1 Mb of the genome region of the associated genes. The FDR (False Discovery Rate) was estimated from p-values with the procedure of Benjamini and Hochberg [20]. After cis-eSNPs were identified, the genotype gi of the cis-eSNP for a particular sample i is inferred from the associated gene expression level ei as follows. First, the mean gene expression level eJ for each genotype J (J~ 0, 1 for haploid cells and J~ 0, 1, 2 for diploid cells) is estimated by using all samples except sample i. Second, genotype at the cis-eSNP for sample i is inferred by comparing its gene expression level ei with the mean expression level of each genotype; the genotype whose mean is the closest to ei is assigned as the inferred genotype of sample i at the cis-eSNP location, noted as gei. Given sample genotypes measured by SNP array and inferred from cis-eSNPs, the sample identity similarity between the two genotypes is defined as Sge~ 1 { 1 N X N n~ 1 gn, i { gen, i j j , genotypes is defined as Sge~ 1 { 1 N X N n~ 1 gn, i { gen, i j j , Figure 7. Assessment of sample alignment quality. The number of cis pairs is counted after each round of alignment. The number of cis pairs increased markedly after alignment in both the CTRL and COPD sets. cis pair mapping 1) cis-eSNP mapping. An eSNP is a single nucleotide polymorphism (SNP) whose genotype associates with variation in the expression of a particular gene. If that gene and its corresponding eSNP are in proximity, the eSNP is called a cis- eSNP. Cis-eSNPs have been extensively studied for their association with disease risks [2,6,18], and have been used to August 2014 \| Volume 10 \| Issue 8 \| e1003790 PLOS Computational Biology \| www.ploscompbiol.org 8 MODMatcher: Multi-Omics Data Matcher infer sample genotypes from gene expression profiling d To identify cis-eSNPs, we used an efficient eSNP program, MatrixEQTL [11]. Assuming that genotyp additive and linear effect on gene expression, we calcu statics for each SNP and gene expression pair to ev significance of association. Cis-eSNPs are defined as SN 1 Mb of the genome region of the associated genes. (False Discovery Rate) was estimated from p-values procedure of Benjamini and Hochberg [20]. After cis-eSNPs were identified, the genotype gi of the for a particular sample i is inferred from the associ expression level ei as follows. First, the mean gene expre eJ for each genotype J (J~ 0, 1 for haploid cells and J for diploid cells) is estimated by using all samples excep Second, genotype at the cis-eSNP for sample i is in comparing its gene expression level ei with the mean level of each genotype; the genotype whose mean is the c is assigned as the inferred genotype of sample i at the location, noted as gei. Given sample genotypes measured by SNP array an from cis-eSNPs, the sample identity similarity betwee genotypes is defined as Sge~ 1 { 1 N X N n~ 1 gn, i { j Figure 7. Assessment of sample alignment quality. The number Figure 8. Comparison of sample alignment procedu on three or two data types in simulated datasets. COPD samples with all three types of data (gene expression and methylation) were used. The mis-labeling error rates w 3% between gene expression and genotypes. The numb aligned pairs was varied from 0 to 24 (corresponding error 37%). Two sample alignment procedures were applied to the data sets and final aligned pairs were compared with alignment. Triangles, duo-alignment results; circles, trio results. Numbers inside triangles or circles indicate the mis-aligned samples in each simulation. Coverage is defi number of correctly aligned pairs divided by 65 (the numbe pairs). cis pair mapping The exact numbers of cis-pairs are listed in Table S2. A) cis-eQTLs. B) cis-mQTLs. C) cis mRNA-methylation pairs. doi:10.1371/journal.pcbi.1003790.g007 where gn, i and gen, i are the observed genotype based on the given sample labels (which may be incorrect due to sample mis- labeling) and the inferred genotype at the nth cis-eSNP for sample i, respectively, and N is the total number of cis-eSNPs. where gn, i and gen, i are the observed genotype based on the given sample labels (which may be incorrect due to sample mis- labeling) and the inferred genotype at the nth cis-eSNP for sample i, respectively, and N is the total number of cis-eSNPs. August 2014 \| Volume 10 \| Issue 8 \| e1003790 PLOS Computational Biology \| www.ploscompbiol.org August 2014 \| Volume 10 \| Issue 8 \| e1003790 9 2) cis-mSNP mapping. Similar to eSNPs, genotypes of SNPs program MatrixEQTL [11], with input changed from gene Figure 9. Examples of sample alignment in the TCGA BRCA data set. (A) A similarity score distribution of a correctly labeled profile. The red star indicates the similarity score between self-matched profile pairs (gene expression and methylation data profiles are labeled as pertaining to the same sample). (B) Similarity scores of self-matched pairs (red stars) between gene expression and methylation profiles for two samples are lower than the similarity scores of cross-matched pairs (blue stars). doi:10.1371/journal.pcbi.1003790.g009 MODMatcher: Multi-Omics Data Matcher MODMatcher: Multi-Omics Data Matcher Figure 9. Examples of sample alignment in the TCGA BRCA data set. (A) A similarity score distribution of a correctly labeled profile. The red star indicates the similarity score between self-matched profile pairs (gene expression and methylation data profiles are labeled as pertaining to the same sample). (B) Similarity scores of self-matched pairs (red stars) between gene expression and methylation profiles for two samples are lower than the similarity scores of cross-matched pairs (blue stars). doi:10.1371/journal.pcbi.1003790.g009 Figure 9. Examples of sample alignment in the TCGA BRCA data set. (A) A similarity score distribution of a correctly labeled profile. The red star indicates the similarity score between self-matched profile pairs (gene expression and methylation data profiles are labeled as pertaining to the same sample). (B) Similarity scores of self-matched pairs (red stars) between gene expression and methylation profiles for two samples are lower than the similarity scores of cross-matched pairs (blue stars). doi:10.1371/journal.pcbi.1003790.g009 2) cis-mSNP mapping. PLOS Computational Biology \| www.ploscompbiol.org Multi-omics data matching procedure To determine the value of ‘‘n’’, we calculated the z-score of a genotype profile Gi mapped to a gene expression profile Ej and vice versa as Sge( i, j) { mean( Sge( i, ) ) p p y 3) cis methylation-mRNA mapping. DNA methylation is a common epigenetic signal that regulates gene expression levels. Increased methylation at CpGs sites near gene promoter region is associated with gene repression [23,24]. Transcript annotation of hg18 was fetched from UCSC database and further processed with the Bioconductor GenomicFeature package. Each methyl probe was mapped to a transcript whose starting site is within 10 Kb from the genomic position of the methyl probe. A methyl probe that is potentially mapped to multiple transcripts on the basis of the above criterion is assigned to the transcript whose start site is closest to the genomic position of the methyl probe. Methyl probes that can’t be mapped to any transcript based on the above criterion were excluded from further analysis. To identify cis- regulation pairs, we calculated the Spearman correlation between the methylation level of a methyl probe and the expression level of the corresponding gene at p-value,0.01. If multiple methyl probes were mapped to the same genes, the probe with the best p-value was selected. Therefore, in subsequent analyses, there was at most a single cis methylation-mRNA pair for each gene. Thus, any potential bias driven by a single gene was avoided. z( i ? j) ~ Sge( i, j) mean( Sge( i, ) ) std( Sge( i, ) ) and z( i / j) g ~ Sge( i, j) { mean( Sge( , j) ) std( Sge( , j) ) , and then compared the g z-score distribution of all top 1 similarity scores with the distribution of z-scores of all top ‘‘n’’ similarity scores. If the z- score distribution of all top 1 similarity scores is statistically different (t-test p-value,0.01) from the distribution of z-scores of all top 2 similarity scores, then ‘‘n’’ is set to 1. Otherwise, ‘‘n’’ is set to 2. In this fashion, we also compared the distributions of top 1 and top 3 similarity scores. For SNP- mRNA matching in the LGRC data set, top ‘‘n’’ was set as 1. cis pair mapping Similar to eSNPs, genotypes of SNPs are also associated with DNA methylation patterns and are called mSNPs [21,22]. To identify association between SNP genotype and methylation level, we used the SNP association mapping program MatrixEQTL [11], with input changed from gene expression profiles to DNA methylation profiles. Similarly, cis- mSNPs are defined as mSNPs within 1 Mb from the genomic regions of the associated methylation probes. PLOS Computational Biology \| www.ploscompbiol.org August 2014 \| Volume 10 \| Issue 8 \| e1003790 10 MODMatcher: Multi-Omics Data Matcher Table 1. Profile pairs used in TCGA dataset. Data type pairs BRCA GBM CNV-mRNA 165 tumor, 13 normal 470 tumor CNV-methylation 149 tumor, 0 normal 294 tumor mRNA-methylation 317 tumor, 20 normal 221 tumor doi:10.1371/journal.pcbi.1003790.t001 Table 1. Profile pairs used in TCGA dataset. 4) cis CNV-mRNA (or methylation) mapping. Copy number variations (CNVs) of genome regions are commonly associated with diseases and may be inherited or occur by de novo mutations. Increasing the DNA copy number of a gene can increase its expression level. Instead of genotype, CNV data was aligned to gene expression and DNA methylation profiles in TCGA data sets. Cis regulation pairs and identity similarity scores were defined similarly as described above for methylation-mRNA pairs. Multi-omics data matching procedure The identity similarity Sge of SNP and gene expression profiles of the same individual is significantly higher than that of random pairs of profiles (Figure 6A). If the self similarity score Sge( i, i) is within the top ‘‘n’’ similarity scores of all possible pairs reciprocally (the genotype profile Gi mapping to gene expression profiles and the gene expression profile Ei mapping to all genotype profiles), the pair of profiles is designated as correctly aligned. The ‘‘n’’ is #3 depending on the data set. To determine the value of ‘‘n’’, we calculated the z-score of a genotype profile Gi mapped to a gene expression profile Ej and vice versa as S ( i j) { mean( Sge( i ) ) (2) Match by SNP-mRNA based identity similarity Sge. After cis-eSNPs are identified, identity similarity Sge is calculated for all possible pairs of SNP-mRNA profiles based on the identified cis-eSNPs. The identity similarity Sge of SNP and gene expression profiles of the same individual is significantly higher than that of random pairs of profiles (Figure 6A). If the self similarity score Sge( i, i) is within the top ‘‘n’’ similarity scores of all possible pairs reciprocally (the genotype profile Gi mapping to gene expression profiles and the gene expression profile Ei mapping to all genotype profiles), the pair of profiles is designated as correctly aligned. The ‘‘n’’ is #3 depending on the data set. To determine the value of ‘‘n’’, we calculated the z-score of a genotype profile Gi mapped to a gene expression profile Ej and vice versa as ( ) Sge( i, j) { mean( Sge( i, ) ) ( ) (2) Match by SNP-mRNA based identity similarity Sge. After cis-eSNPs are identified, identity similarity Sge is calculated for all possible pairs of SNP-mRNA profiles based on the identified cis-eSNPs. The identity similarity Sge of SNP and gene expression profiles of the same individual is significantly higher than that of random pairs of profiles (Figure 6A). If the self similarity score Sge( i, i) is within the top ‘‘n’’ similarity scores of all possible pairs reciprocally (the genotype profile Gi mapping to gene expression profiles and the gene expression profile Ei mapping to all genotype profiles), the pair of profiles is designated as correctly aligned. The ‘‘n’’ is #3 depending on the data set. Multi-omics data matching procedure After cis-mSNPs are identified, the genotype gi of the cis-mSNP for a particular sample i can be inferred from the associated probe methylation level b i as follows. First, the mean probe methylation level bJ for each genotype J (J~ 0, 1 for haploid cells and J~ 0, 1, 2 for diploid cells) is estimated by using all samples except sample i. Second, genotype at the cis-mSNP for sample i is inferred by comparing its methylation level b i with bJ the mean methylation level of each genotype J; the genotype whose mean is the closest to b i is assigned as the genotype of sample i at the cis- mSNP location, noted as gmi. Multiple omics data surveying different molecular traits pertaining to the same set of samples were mapped according to the flow diagram in Figure 4. SNP genotype, gene expression, and methylation data are used for illustration purposes. Other types of data can be used as well. For example, CNV data was used instead of SNP data in the TCGA data sets. First, significant cis regulation (cis-eSNPs, cis-mSNPs, and cis methyl-mRNA) pairs were identified, and sample identity similarities were calculated based on these cis pairs as outlined above. Then, matches and mismatches between omics data were identified in the following steps (ordered by confidence of each test): Given sample genotypes measured by SNP and inferred from cis- mSNPs, the sample identify similarity between the two genotypes is (1) Match by gender. There is no ambiguity for the gender inferred from an omics profile. Any matched pair of omics profiles should have consistent gender information. (1) Match by gender. There is no ambiguity for the gender inferred from an omics profile. Any matched pair of omics profiles should have consistent gender information. defined as Sgm~ 1{ 1 N X N n~ 1 gn, i { gmn, i j j , where gn, i and gmn, i are observed and inferred genotype at the nth cis-mSNP for sample i, respectively, and N is the total number of cis-mSNPs. gmn, i are observed and inferred genotype at the nth cis-mSNP for sample i, respectively, and N is the total number of cis-mSNPs. (2) Match by SNP-mRNA based identity similarity Sge. After cis-eSNPs are identified, identity similarity Sge is calculated for all possible pairs of SNP-mRNA profiles based on the identified cis-eSNPs. Multi-omics data matching procedure For mis-aligned profiles, we further explored whether they could be matched with other unmatched samples by reciprocal matching, in which we determine whether a mis- aligned genotype profile Gi has the highest similarity with an unmatched mRNA profile Ej among all mRNA profiles, and the unmatched mRNA profile Ej has the highest similarity with Gi among all genotype profiles. If there is a reciprocal best match, then the SNP and mRNA profiles are linked and sample labels are updated by comparison with mapping results based on other identity similarities. Before aligning methylation and mRNA profiling data, we rank transformed both gene expression and methylation profiling data for each methyl probe or gene expression probe as RT( b n, i) and RT( en, i) , where RT( x) [ ½1, 2, :::, M is the rank transforma- tion function and M is the number of samples (Figure 5). Given a set of cis methylation-mRNA pairs n~ ( 1, 2, . . . , N) , the sample identity similarity between the two types of data is defined as ~ X N n~1 RT(bn,i) X N n~1 RT(en,i){N X N n~1 RT(bn,i) RT(en,i) ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ N X N n~1 RT(bn,i)2{( X N n~1 RT(bn,i))2 v u u t ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ N X N n~1 RT(en,i)2{( X N n~1 RT(en,i))2 v u u t : PLOS Computational Biology \| www.ploscompbiol.org PLOS Computational Biology \| www.ploscompbiol.org August 2014 \| Volume 10 \| Issue 8 \| e1003790 11 MODMatcher: Multi-Omics Data Matcher putational Biology \| www.ploscompbiol.org 12 August 2014 \| Volume 10 \| Issue 8 \| e PLOS Computational Biology \| www.ploscompbiol.org August 2014 \| Volume 10 \| Issue 8 \| e1003790 August 2014 \| Volume 10 \| Issue 8 \| e1003790 12 MODMatcher: Multi-Omics Data Matcher Figure 10. Identification of potential source of mis-labeling error by trio alignment in TCGA GBM samples. (A) The two GBM tumor samples were cross-matched between CNV and gene expression profiles. Red stars, similarity scores of self-matched pairs; blue stars, similarity scores of cross-matched profiles. (B) The two samples were also cross-matched between gene expression and methylation profiles. (C) Sample labels were consistent between CNV and methylation profiles. These results together indicate that the sample labeling error lay in the gene expression profiles. doi:10.1371/journal.pcbi.1003790.g010 (3) Match by SNP-methylation based identity similarity Sgm. Supporting Information Figure S1 Numbers of cis methyl-mRNA pairs in CTRL and COPD samples when equal numbers of samples were used. (TIF) Figure S2 The cross-aligned sample pair (TCGA-BH-A18T- 01A and TCGA-BH-A18K-01A) identified by methylation- mRNA comparison was cross-aligned based on miRNA and mRNA comparison. Similarity scores based on cis miRNA-mRNA were around zero for the same labels but similarity scores for swapped pairs were the highest in both samples. Combined with the results shown in Figure 9B in main text, mRNA labeling for these two samples was likely to be problematic. (TIF) (5) (5) Match by trio (simultaneously considering Sge, Sgm, and Sme). For the samples with all three types of data available, the source of any sample label mis-matches can be identified (Figure 4). For example, if we identify a sample mapping between the gene expression profile of individual A and the methylation profile of individual B, it is difficult to know which profile data is mis-labeled or both. If the gene expression profile of individual A matches the SNP profile of individual A based on Sge and the methylation profile of individual B is mapped with the SNP profiles of individual A, then it is certain that the methylation profile of individual B is mis-labeled. It is also possible to resolve matching conflicts and to identify additional matched profiles that may be ambiguous based on a single identity similarity score Sge, Sgm, or Sme alone. For example, if data quality is low or the initial profile labeling error rate is high for methylation data, then Sgm and Sme cannot be accurately calculated. If an SNP-mRNA sample match exists (SNP profile Gi matches gene expression profile Ej, then we can search whether there is a methylation profile Mk that matches Gi and Ej by a three way identity similarity score as Sgem( i, j, k) ~ Sgm( i, k) { mean( Sgm( i, ) ) std( Sgm( i, ) ) Figure S3 Gender prediction based on methylation probe intensity in 12 cancer types in the TCGA dataset. The raw intensity of a y-chromosome probe was estimated by summation of the methylated and unmethylated channel. The methyl probe ‘‘cg20401529’’ corresponding to PRKY was used as a gender marker for Illumina HumanMethylation27 Beadarray. For PRAD, for which only the HM450 platform is available, the methyl probe ‘‘cg04042030’’ corresponding to TBL1Y was used. Multi-omics data matching procedure After cis-mQTL are identified, the identity similarity score Sgm is calculated for all possible SNP and methylation profile pairs based on the set of identified cis-mSNPs. The identity similarity score Sgm between SNPs and methylation profiles of the individual is higher than other random pairs (shown in Figure 6B). As above, if the self similarity score Sgm( i, i) is within the top ‘‘n’’ similarity scores of all possible pairs, the pair of profiles is designated as correctly aligned. Top ‘‘n’’ was set to 3 in the LGRC data set. For mis-aligned profiles, we again further explored whether they could be matched with other unmatched samples by the reciprocal best matching procedure described above. each profile under each similarity measurement. To declare methylation profile Mk a match with Gi and Ej, both Sgm( i, k) and Sme( k, j) are required to be within top 3 among all possible similarity scores Sgm( i, ) and Sme( , j) , respectively, and Sgem( i, j, k) is $2.5. After label mis-matches between different types of omics data are identified and sample labeling errors are corrected by comparing multiple identity similarity measurements, the quality of sample alignment is re-assessed by counting the numbers of cis regulation pairs according to the updated data annotation. We iterate this process until data annotations are stable. (4) Match by mRNA-methylation based identity simi- larity Sme. After cis methylation-mRNA probes are identified, the identity similarity score Sme is calculated for all possible pairs of methylation-mRNA profiles based on the set of identified cis methylation–mRNA probes. The identity simi- larity score Sme of methylation-mRNA probes pairs of the same sample is higher than random pairs (Figure 6C). If the self similarity score Sme( i, i) is within the top ‘‘n’’ similarity scores of all possible pairs, the pair of profiles is designated as correctly aligned. Top ‘‘n’’ was set to 3 in the LGRC data set similar as above. For mis-aligned profiles, we used the reciprocal best matching procedure described above to determine whether they could be matched with other unmatched samples. Supporting Information Red, sample predicted to be female; blue, sample predicted to be male. The consistency between clinical and predicted gender is reported in Table S3. p (TIF) Figure S4 Examples of mis-aligned pairs of mRNA and methylation profiles in the TCGA COAD and LUSC datasets. The similarity score for the same sample pairs based on cis methylation-mRNA pairs was not significantly higher than that of other pairs, indicating mis-alignment. (TIF) (TIF) Table S1 Samples of mismatched gender information between clinical annotation and inference from multi-omics data (genotype, mRNA, and methylation profiles). Red ones are mismatched with respect to clinical annotation. (XLSX) z v Sme( k, j) { mean( Sme( , j) ) std( Sme( , j) ) , where v is the weight of similarity Sme relative to Sgm. v was set as 1.2 for the LGRC data set, reflecting the fact that the matching signal between genotype and methylation data is stronger than the matching signal between methylation and gene expression data. v can be estimated ( max( Sgm( i, ) ) { mean( Sgm( i, ) ) ) of similarity Sme relative to Sgm. v was set as 1.2 for the LGRC data set, reflecting the fact that the matching signal between genotype and methylation data is stronger than the matching signal between methylation and gene expression data. v can be estimated ( max( Sgm( i, ) ) { mean( Sgm( i, ) ) ) (XLSX) (XLSX) Table S2 Numbers of cis pairs in each round of alignment corresponding Figure 7. (XLSX) a s sqrt( mean( std( Sgm( i, ) ) ) mean( max( Sme( i, ) ) { mean( Sme( i, ) ) std( Sme( i, ) ) ) ) , Table S3 Gender inference based on methylation probe intensity in multiple cancer data sets in TCGA. The prediction (Figure S3) is compared with the clinically annotated gender. which is the square root of the ratio of mean maximum z-scores of August 2014 \| Volume 10 \| Issue 8 \| e1003790 PLOS Computational Biology \| www.ploscompbiol.org PLOS Computational Biology \| www.ploscompbiol.org 13 MODMatcher: Multi-Omics Data Matcher Sample pairs identified in the two data sets by MODMatcher are listed in Tables S6 and S7. (XLSX) There are gender mismatched samples in four datasets, COAD, KIRC, LAML, and LUAD. (XLSX) Table S4 Comparison of MODMatcher and MixupMapper sample alignments between SNP and mRNA profiles on the same dataset. MODMatcher was applied into 8 dataset including genotype and mRNA profiles examined by MixupMapper. MODMatcher and MixupMapper generated the same result for 6 dataset and there are small differences in for other two dataset. (XLSX) Table S6 Samples pairs in the Choy CHB+JPT data set identified by MODMatcher. (XLSX) Table S7 Samples pairs in the Choy YRI data set identified by MODMatcher. (XLSX) Table S5 Qualities of sample matching results based on MixupMapper and MODMatcher. For the two datasets (Choy CHB+JPT and Choy YRI) where ModMatcher and MixupMap- per results were different as shown in Table S4, the numbers of cis- eQTL pairs identified by each alignment method were compared and MODMatcher identified more cis-eQTLs in both dataset. References 1. Selamat SA, Chung BS, Girard L, Zhang W, Zhang Y, et al. (2012) Genome- scale analysis of DNA methylation in lung adenocarcinoma and integration with mRNA expression. Genome Res 22: 1197–1211. 13. TCGA-Network (2012) Comprehensive molecular portraits of human breast tumours. Nature 490: 61–70. 14. TCGA-Network (2008) Comprehensive genomic characterization defines human glioblastoma genes and core pathways. Nature 455: 1061–1068. p 2. Schadt EE, Molony C, Chudin E, Hao K, Yang X, et al. (2008) Mapping the genetic architecture of gene expression in human liver. PLoS Biol 6: e107. p 2. Schadt EE, Molony C, Chudin E, Hao K, Yang X, et al. (2008) Mapping the genetic architecture of gene expression in human liver. PLoS Biol 6: e107. 15. Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, et al. (2007) PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet 81: 559–575. 3. Morley M, Molony CM, Weber TM, Devlin JL, Ewens KG, et al. (2004) Genetic analysis of genome-wide variation in human gene expression. Nature 430: 743–747. 16. Qu C, Schuetz JM, Min JE, Leach S, Daley D, et al. (2011) Cost-effective prediction of gender-labeling errors and estimation of gender-labeling error rates in candidate-gene association studies. Front Genet 2: 31. 4. Westra HJ, Jansen RC, Fehrmann RS, te Meerman GJ, van Heel D, et al. (2011) MixupMapper: correcting sample mix-ups in genome-wide datasets increases power to detect small genetic effects. Bioinformatics 27: 2104–2111. 17. Karyagyna AS, Vassiliev MO, Ershova AS, Nurtdinov RN, Lossev IS (2010) Probe-Level Universal Search (PLUS) algorithm for gender differentiation in affymetrix datasets. J Bioinform Comput Biol 8: 553–577. p g 5. Chen Y, Zhu J, Lum PY, Yang X, Pinto S, et al. (2008) Variations in DNA elucidate molecular networks that cause disease. Nature 452: 429–435. 6. Emilsson V, Thorleifsson G, Zhang B, Leonardson AS, Zink F, et al. (2008) Genetics of gene expression and its effect on disease. Nature 452: 423–428. 18. Montgomery SB, Dermitzakis ET (2009) The resolution of the genetics of gene expression. Hum Mol Genet 18: R211–215. Genetics of gene expression and its effect on disease. Nature 452 7. Hsu YH, Zillikens MC, Wilson SG, Farber CR, Demissie S, et al. (2010) An integration of genome-wide association study and gene expression profiling to prioritize the discovery of novel susceptibility Loci for osteoporosis-related traits. PLoS Genet 6: e1000977. 19. Author Contributions Conceived and designed the experiments: SY JZ. Performed the experiments: SY TH JZ. Analyzed the data: SY TH JZ. Contributed reagents/materials/analysis tools: TH JDC EL ZT MWG CAP EES AS. Wrote the paper: SY JZ. References Schadt EE, Woo S, Hao K (2012) Bayesian method to predict individual SNP genotypes from gene expression data. Nat Genet 44: 603–608. genotypes from gene expression data. Nat Genet 44: 603–608. 20. Benjamini Y, Hochberg Y (1995) Controlling the False Discovery Rate - a Practical and Powerful Approach to Multiple Testing. Journal of the Royal Statistical Society Series B-Methodological 57: 289–300. 8. Zhong H, Beaulaurier J, Lum PY, Molony C, Yang X, et al. (2010) Liver and adipose expression associated SNPs are enriched for association to type 2 diabetes. PLoS Genet 6: e1000932. 21. Bell JT, Pai AA, Pickrell JK, Gaffney DJ, Pique-Regi R, et al. (2011) DNA methylation patterns associate with genetic and gene expression variation in HapMap cell lines. Genome Biol 12: R10. 9. Aryee MJ, Wu Z, Ladd-Acosta C, Herb B, Feinberg AP, et al. (2011) Accurate genome-scale percentage DNA methylation estimates from microarray data. Biostatistics 12: 197–210. 22. Gibbs JR, van der Brug MP, Hernandez DG, Traynor BJ, Nalls MA, et al. (2010) Abundant quantitative trait loci exist for DNA methylation and gene expression in human brain. PLoS Genet 6: e1000952. 10. Siegmund KD (2011) Statistical approaches for the analysis of DNA methylation microarray data. Hum Genet 129: 585–595. expression in human brain. PLoS Genet 6: e1000952 23. DiNardo DN, Butcher DT, Robinson DP, Archer TK, Rodenhiser DI (2001) Functional analysis of CpG methylation in the BRCA1 promoter region. Oncogene 20: 5331–5340. y 11. Shabalin AA (2012) Matrix eQTL: ultra fast eQTL analysis via large matrix operations. Bioinformatics 28: 1353–1358. 12. Irizarry RA, Ladd-Acosta C, Carvalho B, Wu H, Brandenburg SA, et al. (2008) Comprehensive high-throughput arrays for relative methylation (CHARM). Genome Res 18: 780–790. 24. Suzuki MM, Bird A (2008) DNA methylation landscapes: provocative insights from epigenomics. Nat Rev Genet 9: 465–476. August 2014 \| Volume 10 \| Issue 8 \| e1003790 PLOS Computational Biology \| www.ploscompbiol.org 14
https://openalex.org/W2075549177	https://aip.scitation.org/doi/pdf/10.1063/1.4770325	English	null	Influence of defect mobility on electrostrain in acceptor-doped Ba0.80Sr0.20TiO3	AIP advances	2,012	cc-by	2,567	RESEARCH ARTICLE \| DECEMBER 03 2012 RESEARCH ARTICLE \| DECEMBER 03 2012 Influence of defect mobility on electrostrain in acceptor- doped Ba0.80Sr0.20TiO3 K. Vani; Viswanathan Kumar AIP Advances 2, 042177 (2012) https://doi.org/10.1063/1.4770325 AIP Advances 2, 042177 (2012) https://doi.org/10.1063/1.4770325  View Online  Export Citation AIP ADVANCES 2, 042177 (2012) Inﬂuence of defect mobility on electrostrain in acceptor-doped Ba0.80Sr0.20TiO3 K. Vani and Viswanathan Kumara Center for Materials for Electronics Technology, Scientiﬁc Society, Department of Information Technology, Ministry of Communications and Information Technology, Government of India, Thrissur 680 581, Kerala, India (Received 18 July 2012; accepted 26 November 2012; published online 3 December 2012) This study reports signiﬁcant differences in the defect-mediated electrostrain in B- site trivalent ions (Fe3+ and Mn3+) doped Barium titanate based system. Electron Paramagnetic Resonance (EPR) Spectroscopy has been employed as a structural probe for understanding the symmetry of defects. Differences in the reorientation of the defect dipoles have been correlated with the electrostrain. Mechanism for the higher strain in Mn-doped system has also been explained. Copyright 2012 Author(s). This article is distributed under a Creative Commons Attribution 3.0 Unported License. [http://dx.doi.org/10.1063/1.4770325] In ferroelectric oxides having perovskite ABO3 structure, on substitution of trivalent acceptors such as Mn3+ at the B4+-site, oxygen vacancies are created to maintain charge neutrality1 as represented below; 2Ti O2 M2O3 −−→2M′ Ti + 3Ox o + ¨VO (1) (1) where M = Fe3+, Mn3+; M′ Ti, a negatively charged acceptor at the B-site and ¨VO, a doubly positively charged oxygen vacancy with respect to the neutral lattice according to Kr¨oger and Vink. The neg- atively charged trivalent dopant at the B-site forms a defect-dipole with the positively charged ¨VO, [M′ Ti −¨VO].2–4 B-site acceptor doped BaTiO3 ceramics have been reported to exhibit large recover- able electrostrains due to the symmetry conforming property of defect dipoles.5 The most important factor in this defect dipole-mediated electrostrain is the mobility of the point defects. In PZT based systems, recent studies in our laboratory have shown drastic differences in the acceptor-nature of Mn3+ and Fe3+ dopants at the B-site.6,7 The inherent differences in the polarization switching characteristics is therefore expected to signiﬁcantly inﬂuence the defect-mediated electrostrain. This has not yet been reported. Since PZT based systems are intrinsically p-type and also contain lead vacancies, [V ′′ pb], the objective of the present study is to study the differences in the electrostrain in such acceptor doped Barium titanate-based systems. where M = Fe3+, Mn3+; M′ Ti, a negatively charged acceptor at the B-site and ¨VO, a doubly positively charged oxygen vacancy with respect to the neutral lattice according to Kr¨oger and Vink. ae-mail: vkumar10@yahoo.com Inﬂuence of defect mobility on electrostrain in acceptor-doped Ba0.80Sr0.20TiO3 The neg- atively charged trivalent dopant at the B-site forms a defect-dipole with the positively charged ¨VO, [M′ Ti −¨VO].2–4 B-site acceptor doped BaTiO3 ceramics have been reported to exhibit large recover- able electrostrains due to the symmetry conforming property of defect dipoles.5 The most important factor in this defect dipole-mediated electrostrain is the mobility of the point defects. In PZT based systems, recent studies in our laboratory have shown drastic differences in the acceptor-nature of Mn3+ and Fe3+ dopants at the B-site.6,7 The inherent differences in the polarization switching characteristics is therefore expected to signiﬁcantly inﬂuence the defect-mediated electrostrain. This has not yet been reported. Since PZT based systems are intrinsically p-type and also contain lead vacancies, [V ′′ pb], the objective of the present study is to study the differences in the electrostrain in such acceptor doped Barium titanate-based systems. Barium titanate-based composition, (Ba0.80Sr0.20)TiO3 [BST] having tetragonal structure was chosen as the basis for doping with Mn3+ and Fe3+ to yield the composition with the general formula [(Ba0.8Sr0.2)Ti1-yMy3+ O3; y= 0.005]. BST was synthesized by the sol–gel method as per the proce- dure adapted from our earlier work.8 Barium acetate, (ACS reagent grade, Merck, India), strontium acetate, (99.9%, Sigma-Aldrich, Germany), manganese (III) acetylacetonate (tech.-grade, Sigma- Aldrich, Germany), iron (III) acetylacetonate (tech.-grade, Sigma-Aldrich, Germany) and titanium tetra isopropoxide (97%, Sigma-Aldrich, USA) were used as precursors. The gel-derived powders, obtained after calcination at 800◦C for 2 hrs, were isostatically pressed at 200 MPa into pellets of diameter 10 mm and thickness 1.0 mm and were sintered at a temperature of 1300◦C for 3 hrs. All X-ray Diffraction (XRD) measurements were performed on sintered and powdered samples on a X-ray diffractometer (Model D5005, Bruker, Germany) with CuKα radiation. The sintered BST C⃝Author(s) 2012 2, 042177-1 2158-3226/2012/2(4)/042177/4 042177-2 K. Vani and V. Kumar AIP Advances 2, 042177 (2012) FIG. 1. P-E and S-E curves of (a) Fe3+ and (b) Mn3+ doped (Ba,Sr)TiO3. FIG. 2. (i) Observed and (ii) Simulated EPR spectra of (a) Fe3+ and (b) Mn3+ doped (Ba,Sr)TiO3. Inset shows the resonance corresponding to [Fe′ Ti −¨VO]. dipole. FIG. 1. P-E and S-E curves of (a) Fe3+ and (b) Mn3+ doped (Ba,Sr)TiO3. FIG. 1. P-E and S-E curves of (a) Fe3+ and (b) Mn3+ doped (Ba,Sr)TiO3. FIG. 1. P-E and S-E curves of (a) Fe3+ and (b) Mn3+ doped (Ba,Sr)TiO3. FIG. 2. Inﬂuence of defect mobility on electrostrain in acceptor-doped Ba0.80Sr0.20TiO3 (i) Observed and (ii) Simulated EPR spectra of (a) Fe3+ and (b) Mn3+ doped (Ba,Sr)TiO3. Inset shows the resonance corresponding to [Fe′ Ti −¨VO]. dipole. FIG. 2. (i) Observed and (ii) Simulated EPR spectra of (a) Fe3+ and (b) Mn3+ doped (Ba,Sr)TiO3. Inset shows the resonance corresponding to [Fe′ Ti −¨VO]. dipole. specimens were polished and electroded with silver paste. The piezoelectric characteristics were determined using a Piezo evaluation System (Model 2000, aixACCT, Germany) after ageing the sample for ninety days. Electron Paramagnetic Resonance spectrum (EPR) of the aged specimens were recorded at liquid nitrogen temperature and X band frequency on an EPR spectrometer (Model E-112, Varian, USA). Raman spectra of the samples were recorded using a Raman spectrometer (Model LabRam 1B, Varian, USA) with a laser line of 514 nm and a laser power of 1.5 mW. p Figure 1 shows the ﬁeld dependence of polarisation and strain in Fe3+ and Mn3+ doped BST, aged for the same duration. Compared to Mn3+-doped BST, pronounced differences are observed in the P-E and S-E loops of Fe3+-doped system. P-E and S-E curves are not well developed in the case of Fe3+-doped system (Fig. 1(a)) whereas in the case of Mn3+-doped BST, they are well developed (Fig. 1(b)). This is attributed to inhibition of the switching of ferroelectric dipoles by the [Fe′ Ti −¨VO] . defect dipoles. For defect-mediated reversible switching of ferroelectric dipoles, it is required that the defect dipoles should attain the symmetry of the crystalline lattice. As defect dipole symmetry attains the crystal symmetry in aged specimens 5,9,10 Electron Paramagnetic Resonance (EPR) spectroscopy was employed as structural probe for understanding the defect symmetry.11–14 In the case of Fe3+-doped system, as a result of very high internal electric ﬁeld (Ei),7 [Fe′ Ti −¨VO] . defect dipoles do not attain the lattice symmetry as revealed by the symmetric center, having g = 2.01 in the EPR spectrum [Fig. 2(a)(i)].15,16 In case of Mn3+-doped system, due to relatively lower Ei, the [Mn′ Ti −¨VO] . defect dipoles are more labile and therefore, on ageing, attain the tetragonal symmetry of the lattice [Fig. 2(b)(i)]. Such fully reoriented [Mn′ Ti −¨VO] . defect dipoles therefore provide a strong restoring force for the complete reversibility of the switched ferroelectric dipoles [Fig. 1(b)]. In order to understand the symmetry of defects, EPR Spectroscopy was employed. The 042177-3 K. Vani and V. Inﬂuence of defect mobility on electrostrain in acceptor-doped Ba0.80Sr0.20TiO3 Kumar AIP Advances 2, 042177 (2012) FIG. 3. (a) XRD patterns of (Ba,Sr)Ti1-yMyO3 (a) y = 0, (ii) y = [Fe3+] = 0.005 and (iii) y = [Mn3+] = 0.005. Inset shows Raman spectrum of (ii). FIG. 3. (a) XRD patterns of (Ba,Sr)Ti1-yMyO3 (a) y = 0, (ii) y = [Fe3+] = 0.005 and (iii) y = [Mn3+] = 0.005. Inset show Raman spectrum of (ii). FIG. 4. Alteration of orbital energies of Mn3+ (d4) in octahedral symmetry due to z-out type J-T distortion, O ≫δ1> δ2. FIG. 4. Alteration of orbital energies of Mn3+ (d4) in octahedral symmetry due to z-out type J-T distortion, O ≫δ1> δ2 observed and simulated EPR patterns of acceptor-doped BST are shown in Figure 2. The EPR spectra was simulated using the EasySpin4.0.017 program. For Mn3+- doped BST, the spectrum [Fig. 2(b)(ii)] can be simulated using the spin Hamiltonian parameters corresponding to Mn2+ (S=5/2) center, g =2.004, hyperﬁne splitting constant, A=243 MHz and zero ﬁeld splitting param- eter, D= 6800 MHz; Ti vacancy, [V ′′′ Ti] at g=2.00418 and singly ionized Barium vacancy, [V ′ Ba] at g=2.0028.19 In Mn3+- doped BST, Manganese is known to occur in Mn2+ and Mn4+ states due to its facile disproportionation. However their concentration is considerably less than that of Mn3+. The observed spectrum of Fe-doped BST shows two centers. The signal at g ≈4.7 [Fig. 2(a)(i) inset] is characteristic of [Fe′ Ti −¨VO] . defect dipole and the one having g value ≈2.01 corresponds to a cubic Fe′ Ti centre. The simulated pattern obtained is shown in [Fig. 2(a)(ii)]. Spin Hamiltonian parameters, g = 2.0107; zero ﬁeld splitting parameter, D = 5000 MHz correspond to defect dipole centre and g = 2.01 for cubic center. Since the defect dipole, [M′ Ti −¨VO] . is positively charged, for overall charge compensation, the following incorporation reaction is proposed. observed and simulated EPR patterns of acceptor-doped BST are shown in Figure 2. The EPR spectra was simulated using the EasySpin4.0.017 program. For Mn3+- doped BST, the spectrum [Fig. Inﬂuence of defect mobility on electrostrain in acceptor-doped Ba0.80Sr0.20TiO3 2(b)(ii)] can be simulated using the spin Hamiltonian parameters corresponding to Mn2+ (S=5/2) center, g =2.004, hyperﬁne splitting constant, A=243 MHz and zero ﬁeld splitting param- eter, D= 6800 MHz; Ti vacancy, [V ′′′ Ti] at g=2.00418 and singly ionized Barium vacancy, [V ′ Ba] at g=2.0028.19 In Mn3+- doped BST, Manganese is known to occur in Mn2+ and Mn4+ states due to its facile disproportionation. However their concentration is considerably less than that of Mn3+. The observed spectrum of Fe-doped BST shows two centers. The signal at g ≈4.7 [Fig. 2(a)(i) inset] is characteristic of [Fe′ Ti −¨VO] . defect dipole and the one having g value ≈2.01 corresponds to a cubic Fe′ Ti centre. The simulated pattern obtained is shown in [Fig. 2(a)(ii)]. Spin Hamiltonian parameters, g = 2.0107; zero ﬁeld splitting parameter, D = 5000 MHz correspond to defect dipole centre and g = 2.01 for cubic center. Since the defect dipole, [M′ Ti −¨VO] . is positively charged, for overall charge compensation, the following incorporation reaction is proposed. BaO + 2M2O3 →Bax Ba + 4(M′ Ti −¨VO) . + 7Ox o + V ′ Ba + V ′′′ Ti (2) (2) where 2TiO2 is substituted by M2O3. Higher tetragonality as a result of z-out Jahn-Teller (J-T) distortion, in Mn3+-doped BST (Fig. 3), is also responsible for the increased strain. The alteration in the d-energy levels due to J-T distortion is shown in Figure 4. J-T distortion was conﬁrmed from 042177-4 K. Vani and V. Kumar AIP Advances 2, 042177 (201 AIP Advances 2, 042177 (2012) 042177-4 K. Vani and V. Kumar FIG. 5. Absorption spectra (a) Fe3+ and (b) Mn3+ doped (Ba,Sr)Ti0.995M0.005O3. FIG. 5. Absorption spectra (a) Fe3+ and (b) Mn3+ doped (Ba,Sr)Ti0.995M0.005O3. the absorption spectra from the two peaks at 16662 cm-1 and 10928 cm-1 [Fig. 5(b)(ii), (iii)] in the case of Mn3+-doped system corresponding to 5B1 ←5B2 and 5B1 ←5A1 transitions20 respectively whereas in case of Fe3+-doped system, in the absence of any distortion, only one peak corresponding to t2g →eg transition [Fig. 5(a)(i)] is observed. g g In summary, we have shown that electrostrain in Fe3+ and Mn3+-doped BST exhibit pronounced differences. Higher internal electric ﬁeld in Fe3+-doped system is found not only to inhibit the switching of ferroelectric dipoles but also to restrict the [Fe′ Ti −¨VO] . defect dipole from attaining the crystal symmetry of the lattice, on ageing. 1 B. Jaffe, W. R. Cook, Jr, and H. Jaffe, Piezoelectric Ceramics (Chapter 7) (Academic Press, New York, 1971). 1 B. Jaffe, W. R. Cook, Jr, and H. Jaffe, Piezoelectric Ceramics (Chapter 7) (Academic Press, New York, 1971). 2 E. S. Kirkpatrick, K. A. Muller, and R. S. Rubins, Phy. Rev. 135, A86 (1964). 3 K. A. Muller, W. Berlinger, and K. W. Blazey, Solid State Commun. 61, 21 (1987). 4 E. Siegel and K. A. Muller, Phys. Rev. B 19, 109 (1979). 5 X. Ren, Nature Mater. 3, 91 (2004). 6 K. P. Rema and V. Kumar, J. Am. Ceram. Soc. 91, 164 (2008). 7 K. P. Rema, K. E. Vinod, and V. Kumar, J. Mat. Sci:Mat. Elec. 21, 1149 (2009). 8 P. V. Divya and V. Kumar, J. Am. Ceram. Soc. 90, 472 (2007). 9 X. Ren and K. Otsuka, Phy. Rev. Lett. 85, 1016 (2000). 10 Z. Feng and X. Ren, Phy. Rev. B 77, 134115–1 (2008). 11 R. Merkle and J. Maier, Phys. Chem. Chem. Phys. 5, 2297 (2003). 12 W. L. Warren, D. Dimos, G. E. Pike, K. Vanheusden, and R. Ramesh, Appl. Phys. Lett. 67, 1689 (1995). 13 W. L. Warren, G. E. Pike, K. Vanheusden, D. Dimos, B. A. Tuttle, and J. Robertson, J. Appl. Phys 79, 9250 (1996). 14 L. Zheng, E. Erdem, X. Ren, and R. A. Eichel, Appl. Phys. Lett. 93, 202901–1 (2008). 15 R. S. Drago, Physical Methods in Inorganic Chemistry (Chapter 10) (East-West Press Pvt. Ltd. New Delhi, 1965). 16 R. A. Eichel, J. Am. Ceram. Soc. 91, 691 (2008). 17 S. Stoll and A. Schweiger, J. Magn. Reson. 178, 42 (2006). 18 T. Kolodiazhnyi and A. Petric, J.Phys. Chem. Solids 64, 953 (2003). 19 B. Milsch, Phys. Stat. Sol. (a) 133, 455 (1992). 20 T. S. Davis, J. P. Fackler, and M. J. Weeks, Inorg. Chem. 7, 1994 (1968). , , , , ( p ) ( 2 E. S. Kirkpatrick, K. A. Muller, and R. S. Rubins, Phy. Rev. 135, A86 (1964). 20 T. S. Davis, J. P. Fackler, and M. J. Weeks, Inorg. Chem. 7, 1994 (1968). p , , , y , ( ) 3 K. A. Muller, W. Berlinger, and K. W. Blazey, Solid State Commun. 61, 21 (1987). E. S. Kirkpatrick, K. A. Muller, and R. S. Rubins, Phy. Rev. 135, A86 (1964). 3 K. A. Muller, W. Berlinger, and K. W. Blazey, Solid State Commun. 61, 21 (1987). Inﬂuence of defect mobility on electrostrain in acceptor-doped Ba0.80Sr0.20TiO3 In addition to increased mobility of the [Mn′ Ti −¨VO] . defect dipole, structural distortion in Mn3+-doped BST is also responsible for higher electrostrain. One of the authors KV is grateful to Council of Scientiﬁc and Industrial Research [CSIR], Indi for the Junior Research Fellowship. , , ( ) 7 K. P. Rema, K. E. Vinod, and V. Kumar, J. Mat. Sci:Mat. Elec. 21, 1149 (2009). 12 W. L. Warren, D. Dimos, G. E. Pike, K. Vanheusden, and R. Ramesh, Appl. Phys. Lett. 67, 1689 (1995). 13 g pp y 15 R. S. Drago, Physical Methods in Inorganic Chemistry (Chapter 10) (East-West Press Pvt. Ltd. New Delhi, 1965). 16 7 S. Stoll and A. Schweiger, J. Magn. Reson. 178, 42 (2006). g g 18 T. Kolodiazhnyi and A. Petric, J.Phys. Chem. Solids 64, 953 (2003). , y ( ) , ( ) 20 T. S. Davis, J. P. Fackler, and M. J. Weeks, Inorg. Chem. 7, 1994 (1968).

Subsets and Splits

No community queries yet

The top public SQL queries from the community will appear here once available.